diff options
| author | root <root@artemis.panaceas.org> | 2015-12-25 04:40:36 +0000 |
|---|---|---|
| committer | root <root@artemis.panaceas.org> | 2015-12-25 04:40:36 +0000 |
| commit | 849369d6c66d3054688672f97d31fceb8e8230fb (patch) | |
| tree | 6135abc790ca67dedbe07c39806591e70eda81ce /tools | |
| download | linux-3.0.35-kobo-849369d6c66d3054688672f97d31fceb8e8230fb.tar.gz linux-3.0.35-kobo-849369d6c66d3054688672f97d31fceb8e8230fb.tar.bz2 linux-3.0.35-kobo-849369d6c66d3054688672f97d31fceb8e8230fb.zip | |
initial_commit
Diffstat (limited to 'tools')
309 files changed, 66929 insertions, 0 deletions
diff --git a/tools/firewire/Makefile b/tools/firewire/Makefile new file mode 100644 index 00000000..81767ada --- /dev/null +++ b/tools/firewire/Makefile @@ -0,0 +1,19 @@ +prefix = /usr +nosy-dump-version = 0.4 + +CC = gcc + +all : nosy-dump + +nosy-dump : CFLAGS = -Wall -O2 -g +nosy-dump : CPPFLAGS = -DVERSION=\"$(nosy-dump-version)\" -I../../drivers/firewire +nosy-dump : LDFLAGS = -g +nosy-dump : LDLIBS = -lpopt + +nosy-dump : nosy-dump.o decode-fcp.o + +clean : + rm -rf *.o nosy-dump + +install : + install nosy-dump $(prefix)/bin/nosy-dump diff --git a/tools/firewire/decode-fcp.c b/tools/firewire/decode-fcp.c new file mode 100644 index 00000000..e41223b6 --- /dev/null +++ b/tools/firewire/decode-fcp.c @@ -0,0 +1,213 @@ +#include <linux/firewire-constants.h> +#include <stdio.h> +#include <stdlib.h> + +#include "list.h" +#include "nosy-dump.h" + +#define CSR_FCP_COMMAND 0xfffff0000b00ull +#define CSR_FCP_RESPONSE 0xfffff0000d00ull + +static const char * const ctype_names[] = { + [0x0] = "control", [0x8] = "not implemented", + [0x1] = "status", [0x9] = "accepted", + [0x2] = "specific inquiry", [0xa] = "rejected", + [0x3] = "notify", [0xb] = "in transition", + [0x4] = "general inquiry", [0xc] = "stable", + [0x5] = "(reserved 0x05)", [0xd] = "changed", + [0x6] = "(reserved 0x06)", [0xe] = "(reserved 0x0e)", + [0x7] = "(reserved 0x07)", [0xf] = "interim", +}; + +static const char * const subunit_type_names[] = { + [0x00] = "monitor", [0x10] = "(reserved 0x10)", + [0x01] = "audio", [0x11] = "(reserved 0x11)", + [0x02] = "printer", [0x12] = "(reserved 0x12)", + [0x03] = "disc", [0x13] = "(reserved 0x13)", + [0x04] = "tape recorder/player",[0x14] = "(reserved 0x14)", + [0x05] = "tuner", [0x15] = "(reserved 0x15)", + [0x06] = "ca", [0x16] = "(reserved 0x16)", + [0x07] = "camera", [0x17] = "(reserved 0x17)", + [0x08] = "(reserved 0x08)", [0x18] = "(reserved 0x18)", + [0x09] = "panel", [0x19] = "(reserved 0x19)", + [0x0a] = "bulletin board", [0x1a] = "(reserved 0x1a)", + [0x0b] = "camera storage", [0x1b] = "(reserved 0x1b)", + [0x0c] = "(reserved 0x0c)", [0x1c] = "vendor unique", + [0x0d] = "(reserved 0x0d)", [0x1d] = "all subunit types", + [0x0e] = "(reserved 0x0e)", [0x1e] = "subunit_type extended to next byte", + [0x0f] = "(reserved 0x0f)", [0x1f] = "unit", +}; + +struct avc_enum { + int value; + const char *name; +}; + +struct avc_field { + const char *name; /* Short name for field. */ + int offset; /* Location of field, specified in bits; */ + /* negative means from end of packet. */ + int width; /* Width of field, 0 means use data_length. */ + struct avc_enum *names; +}; + +struct avc_opcode_info { + const char *name; + struct avc_field fields[8]; +}; + +struct avc_enum power_field_names[] = { + { 0x70, "on" }, + { 0x60, "off" }, + { } +}; + +static const struct avc_opcode_info opcode_info[256] = { + + /* TA Document 1999026 */ + /* AV/C Digital Interface Command Set General Specification 4.0 */ + [0xb2] = { "power", { + { "state", 0, 8, power_field_names } + } + }, + [0x30] = { "unit info", { + { "foo", 0, 8 }, + { "unit_type", 8, 5 }, + { "unit", 13, 3 }, + { "company id", 16, 24 }, + } + }, + [0x31] = { "subunit info" }, + [0x01] = { "reserve" }, + [0xb0] = { "version" }, + [0x00] = { "vendor dependent" }, + [0x02] = { "plug info" }, + [0x12] = { "channel usage" }, + [0x24] = { "connect" }, + [0x20] = { "connect av" }, + [0x22] = { "connections" }, + [0x11] = { "digital input" }, + [0x10] = { "digital output" }, + [0x25] = { "disconnect" }, + [0x21] = { "disconnect av" }, + [0x19] = { "input plug signal format" }, + [0x18] = { "output plug signal format" }, + [0x1f] = { "general bus setup" }, + + /* TA Document 1999025 */ + /* AV/C Descriptor Mechanism Specification Version 1.0 */ + [0x0c] = { "create descriptor" }, + [0x08] = { "open descriptor" }, + [0x09] = { "read descriptor" }, + [0x0a] = { "write descriptor" }, + [0x05] = { "open info block" }, + [0x06] = { "read info block" }, + [0x07] = { "write info block" }, + [0x0b] = { "search descriptor" }, + [0x0d] = { "object number select" }, + + /* TA Document 1999015 */ + /* AV/C Command Set for Rate Control of Isochronous Data Flow 1.0 */ + [0xb3] = { "rate", { + { "subfunction", 0, 8 }, + { "result", 8, 8 }, + { "plug_type", 16, 8 }, + { "plug_id", 16, 8 }, + } + }, + + /* TA Document 1999008 */ + /* AV/C Audio Subunit Specification 1.0 */ + [0xb8] = { "function block" }, + + /* TA Document 2001001 */ + /* AV/C Panel Subunit Specification 1.1 */ + [0x7d] = { "gui update" }, + [0x7e] = { "push gui data" }, + [0x7f] = { "user action" }, + [0x7c] = { "pass through" }, + + /* */ + [0x26] = { "asynchronous connection" }, +}; + +struct avc_frame { + uint32_t operand0:8; + uint32_t opcode:8; + uint32_t subunit_id:3; + uint32_t subunit_type:5; + uint32_t ctype:4; + uint32_t cts:4; +}; + +static void +decode_avc(struct link_transaction *t) +{ + struct avc_frame *frame = + (struct avc_frame *) t->request->packet.write_block.data; + const struct avc_opcode_info *info; + const char *name; + char buffer[32]; + int i; + + info = &opcode_info[frame->opcode]; + if (info->name == NULL) { + snprintf(buffer, sizeof(buffer), + "(unknown opcode 0x%02x)", frame->opcode); + name = buffer; + } else { + name = info->name; + } + + printf("av/c %s, subunit_type=%s, subunit_id=%d, opcode=%s", + ctype_names[frame->ctype], subunit_type_names[frame->subunit_type], + frame->subunit_id, name); + + for (i = 0; info->fields[i].name != NULL; i++) + printf(", %s", info->fields[i].name); + + printf("\n"); +} + +int +decode_fcp(struct link_transaction *t) +{ + struct avc_frame *frame = + (struct avc_frame *) t->request->packet.write_block.data; + unsigned long long offset = + ((unsigned long long) t->request->packet.common.offset_high << 32) | + t->request->packet.common.offset_low; + + if (t->request->packet.common.tcode != TCODE_WRITE_BLOCK_REQUEST) + return 0; + + if (offset == CSR_FCP_COMMAND || offset == CSR_FCP_RESPONSE) { + switch (frame->cts) { + case 0x00: + decode_avc(t); + break; + case 0x01: + printf("cal fcp frame (cts=0x01)\n"); + break; + case 0x02: + printf("ehs fcp frame (cts=0x02)\n"); + break; + case 0x03: + printf("havi fcp frame (cts=0x03)\n"); + break; + case 0x0e: + printf("vendor specific fcp frame (cts=0x0e)\n"); + break; + case 0x0f: + printf("extended cts\n"); + break; + default: + printf("reserved fcp frame (ctx=0x%02x)\n", frame->cts); + break; + } + return 1; + } + + return 0; +} + diff --git a/tools/firewire/list.h b/tools/firewire/list.h new file mode 100644 index 00000000..41f4bdad --- /dev/null +++ b/tools/firewire/list.h @@ -0,0 +1,62 @@ +struct list { + struct list *next, *prev; +}; + +static inline void +list_init(struct list *list) +{ + list->next = list; + list->prev = list; +} + +static inline int +list_empty(struct list *list) +{ + return list->next == list; +} + +static inline void +list_insert(struct list *link, struct list *new_link) +{ + new_link->prev = link->prev; + new_link->next = link; + new_link->prev->next = new_link; + new_link->next->prev = new_link; +} + +static inline void +list_append(struct list *list, struct list *new_link) +{ + list_insert((struct list *)list, new_link); +} + +static inline void +list_prepend(struct list *list, struct list *new_link) +{ + list_insert(list->next, new_link); +} + +static inline void +list_remove(struct list *link) +{ + link->prev->next = link->next; + link->next->prev = link->prev; +} + +#define list_entry(link, type, member) \ + ((type *)((char *)(link)-(unsigned long)(&((type *)0)->member))) + +#define list_head(list, type, member) \ + list_entry((list)->next, type, member) + +#define list_tail(list, type, member) \ + list_entry((list)->prev, type, member) + +#define list_next(elm, member) \ + list_entry((elm)->member.next, typeof(*elm), member) + +#define list_for_each_entry(pos, list, member) \ + for (pos = list_head(list, typeof(*pos), member); \ + &pos->member != (list); \ + pos = list_next(pos, member)) + diff --git a/tools/firewire/nosy-dump.c b/tools/firewire/nosy-dump.c new file mode 100644 index 00000000..f93b7763 --- /dev/null +++ b/tools/firewire/nosy-dump.c @@ -0,0 +1,1031 @@ +/* + * nosy-dump - Interface to snoop mode driver for TI PCILynx 1394 controllers + * Copyright (C) 2002-2006 Kristian Høgsberg + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + */ + +#include <byteswap.h> +#include <endian.h> +#include <fcntl.h> +#include <linux/firewire-constants.h> +#include <poll.h> +#include <popt.h> +#include <signal.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <sys/ioctl.h> +#include <sys/time.h> +#include <termios.h> +#include <unistd.h> + +#include "list.h" +#include "nosy-dump.h" +#include "nosy-user.h" + +enum { + PACKET_FIELD_DETAIL = 0x01, + PACKET_FIELD_DATA_LENGTH = 0x02, + /* Marks the fields we print in transaction view. */ + PACKET_FIELD_TRANSACTION = 0x04, +}; + +static void print_packet(uint32_t *data, size_t length); +static void decode_link_packet(struct link_packet *packet, size_t length, + int include_flags, int exclude_flags); +static int run = 1; +sig_t sys_sigint_handler; + +static char *option_nosy_device = "/dev/nosy"; +static char *option_view = "packet"; +static char *option_output; +static char *option_input; +static int option_hex; +static int option_iso; +static int option_cycle_start; +static int option_version; +static int option_verbose; + +enum { + VIEW_TRANSACTION, + VIEW_PACKET, + VIEW_STATS, +}; + +static const struct poptOption options[] = { + { + .longName = "device", + .shortName = 'd', + .argInfo = POPT_ARG_STRING, + .arg = &option_nosy_device, + .descrip = "Path to nosy device.", + .argDescrip = "DEVICE" + }, + { + .longName = "view", + .argInfo = POPT_ARG_STRING, + .arg = &option_view, + .descrip = "Specify view of bus traffic: packet, transaction or stats.", + .argDescrip = "VIEW" + }, + { + .longName = "hex", + .shortName = 'x', + .argInfo = POPT_ARG_NONE, + .arg = &option_hex, + .descrip = "Print each packet in hex.", + }, + { + .longName = "iso", + .argInfo = POPT_ARG_NONE, + .arg = &option_iso, + .descrip = "Print iso packets.", + }, + { + .longName = "cycle-start", + .argInfo = POPT_ARG_NONE, + .arg = &option_cycle_start, + .descrip = "Print cycle start packets.", + }, + { + .longName = "verbose", + .shortName = 'v', + .argInfo = POPT_ARG_NONE, + .arg = &option_verbose, + .descrip = "Verbose packet view.", + }, + { + .longName = "output", + .shortName = 'o', + .argInfo = POPT_ARG_STRING, + .arg = &option_output, + .descrip = "Log to output file.", + .argDescrip = "FILENAME" + }, + { + .longName = "input", + .shortName = 'i', + .argInfo = POPT_ARG_STRING, + .arg = &option_input, + .descrip = "Decode log from file.", + .argDescrip = "FILENAME" + }, + { + .longName = "version", + .argInfo = POPT_ARG_NONE, + .arg = &option_version, + .descrip = "Specify print version info.", + }, + POPT_AUTOHELP + POPT_TABLEEND +}; + +/* Allow all ^C except the first to interrupt the program in the usual way. */ +static void +sigint_handler(int signal_num) +{ + if (run == 1) { + run = 0; + signal(SIGINT, SIG_DFL); + } +} + +static struct subaction * +subaction_create(uint32_t *data, size_t length) +{ + struct subaction *sa; + + /* we put the ack in the subaction struct for easy access. */ + sa = malloc(sizeof *sa - sizeof sa->packet + length); + sa->ack = data[length / 4 - 1]; + sa->length = length; + memcpy(&sa->packet, data, length); + + return sa; +} + +static void +subaction_destroy(struct subaction *sa) +{ + free(sa); +} + +static struct list pending_transaction_list = { + &pending_transaction_list, &pending_transaction_list +}; + +static struct link_transaction * +link_transaction_lookup(int request_node, int response_node, int tlabel) +{ + struct link_transaction *t; + + list_for_each_entry(t, &pending_transaction_list, link) { + if (t->request_node == request_node && + t->response_node == response_node && + t->tlabel == tlabel) + return t; + } + + t = malloc(sizeof *t); + t->request_node = request_node; + t->response_node = response_node; + t->tlabel = tlabel; + list_init(&t->request_list); + list_init(&t->response_list); + + list_append(&pending_transaction_list, &t->link); + + return t; +} + +static void +link_transaction_destroy(struct link_transaction *t) +{ + struct subaction *sa; + + while (!list_empty(&t->request_list)) { + sa = list_head(&t->request_list, struct subaction, link); + list_remove(&sa->link); + subaction_destroy(sa); + } + while (!list_empty(&t->response_list)) { + sa = list_head(&t->response_list, struct subaction, link); + list_remove(&sa->link); + subaction_destroy(sa); + } + free(t); +} + +struct protocol_decoder { + const char *name; + int (*decode)(struct link_transaction *t); +}; + +static const struct protocol_decoder protocol_decoders[] = { + { "FCP", decode_fcp } +}; + +static void +handle_transaction(struct link_transaction *t) +{ + struct subaction *sa; + int i; + + if (!t->request) { + printf("BUG in handle_transaction\n"); + return; + } + + for (i = 0; i < array_length(protocol_decoders); i++) + if (protocol_decoders[i].decode(t)) + break; + + /* HACK: decode only fcp right now. */ + return; + + decode_link_packet(&t->request->packet, t->request->length, + PACKET_FIELD_TRANSACTION, 0); + if (t->response) + decode_link_packet(&t->response->packet, t->request->length, + PACKET_FIELD_TRANSACTION, 0); + else + printf("[no response]"); + + if (option_verbose) { + list_for_each_entry(sa, &t->request_list, link) + print_packet((uint32_t *) &sa->packet, sa->length); + list_for_each_entry(sa, &t->response_list, link) + print_packet((uint32_t *) &sa->packet, sa->length); + } + printf("\r\n"); + + link_transaction_destroy(t); +} + +static void +clear_pending_transaction_list(void) +{ + struct link_transaction *t; + + while (!list_empty(&pending_transaction_list)) { + t = list_head(&pending_transaction_list, + struct link_transaction, link); + list_remove(&t->link); + link_transaction_destroy(t); + /* print unfinished transactions */ + } +} + +static const char * const tcode_names[] = { + [0x0] = "write_quadlet_request", [0x6] = "read_quadlet_response", + [0x1] = "write_block_request", [0x7] = "read_block_response", + [0x2] = "write_response", [0x8] = "cycle_start", + [0x3] = "reserved", [0x9] = "lock_request", + [0x4] = "read_quadlet_request", [0xa] = "iso_data", + [0x5] = "read_block_request", [0xb] = "lock_response", +}; + +static const char * const ack_names[] = { + [0x0] = "no ack", [0x8] = "reserved (0x08)", + [0x1] = "ack_complete", [0x9] = "reserved (0x09)", + [0x2] = "ack_pending", [0xa] = "reserved (0x0a)", + [0x3] = "reserved (0x03)", [0xb] = "reserved (0x0b)", + [0x4] = "ack_busy_x", [0xc] = "reserved (0x0c)", + [0x5] = "ack_busy_a", [0xd] = "ack_data_error", + [0x6] = "ack_busy_b", [0xe] = "ack_type_error", + [0x7] = "reserved (0x07)", [0xf] = "reserved (0x0f)", +}; + +static const char * const rcode_names[] = { + [0x0] = "complete", [0x4] = "conflict_error", + [0x1] = "reserved (0x01)", [0x5] = "data_error", + [0x2] = "reserved (0x02)", [0x6] = "type_error", + [0x3] = "reserved (0x03)", [0x7] = "address_error", +}; + +static const char * const retry_names[] = { + [0x0] = "retry_1", + [0x1] = "retry_x", + [0x2] = "retry_a", + [0x3] = "retry_b", +}; + +enum { + PACKET_RESERVED, + PACKET_REQUEST, + PACKET_RESPONSE, + PACKET_OTHER, +}; + +struct packet_info { + const char *name; + int type; + int response_tcode; + const struct packet_field *fields; + int field_count; +}; + +struct packet_field { + const char *name; /* Short name for field. */ + int offset; /* Location of field, specified in bits; */ + /* negative means from end of packet. */ + int width; /* Width of field, 0 means use data_length. */ + int flags; /* Show options. */ + const char * const *value_names; +}; + +#define COMMON_REQUEST_FIELDS \ + { "dest", 0, 16, PACKET_FIELD_TRANSACTION }, \ + { "tl", 16, 6 }, \ + { "rt", 22, 2, PACKET_FIELD_DETAIL, retry_names }, \ + { "tcode", 24, 4, PACKET_FIELD_TRANSACTION, tcode_names }, \ + { "pri", 28, 4, PACKET_FIELD_DETAIL }, \ + { "src", 32, 16, PACKET_FIELD_TRANSACTION }, \ + { "offs", 48, 48, PACKET_FIELD_TRANSACTION } + +#define COMMON_RESPONSE_FIELDS \ + { "dest", 0, 16 }, \ + { "tl", 16, 6 }, \ + { "rt", 22, 2, PACKET_FIELD_DETAIL, retry_names }, \ + { "tcode", 24, 4, 0, tcode_names }, \ + { "pri", 28, 4, PACKET_FIELD_DETAIL }, \ + { "src", 32, 16 }, \ + { "rcode", 48, 4, PACKET_FIELD_TRANSACTION, rcode_names } + +static const struct packet_field read_quadlet_request_fields[] = { + COMMON_REQUEST_FIELDS, + { "crc", 96, 32, PACKET_FIELD_DETAIL }, + { "ack", 156, 4, 0, ack_names }, +}; + +static const struct packet_field read_quadlet_response_fields[] = { + COMMON_RESPONSE_FIELDS, + { "data", 96, 32, PACKET_FIELD_TRANSACTION }, + { "crc", 128, 32, PACKET_FIELD_DETAIL }, + { "ack", 188, 4, 0, ack_names }, +}; + +static const struct packet_field read_block_request_fields[] = { + COMMON_REQUEST_FIELDS, + { "data_length", 96, 16, PACKET_FIELD_TRANSACTION }, + { "extended_tcode", 112, 16 }, + { "crc", 128, 32, PACKET_FIELD_DETAIL }, + { "ack", 188, 4, 0, ack_names }, +}; + +static const struct packet_field block_response_fields[] = { + COMMON_RESPONSE_FIELDS, + { "data_length", 96, 16, PACKET_FIELD_DATA_LENGTH }, + { "extended_tcode", 112, 16 }, + { "crc", 128, 32, PACKET_FIELD_DETAIL }, + { "data", 160, 0, PACKET_FIELD_TRANSACTION }, + { "crc", -64, 32, PACKET_FIELD_DETAIL }, + { "ack", -4, 4, 0, ack_names }, +}; + +static const struct packet_field write_quadlet_request_fields[] = { + COMMON_REQUEST_FIELDS, + { "data", 96, 32, PACKET_FIELD_TRANSACTION }, + { "ack", -4, 4, 0, ack_names }, +}; + +static const struct packet_field block_request_fields[] = { + COMMON_REQUEST_FIELDS, + { "data_length", 96, 16, PACKET_FIELD_DATA_LENGTH | PACKET_FIELD_TRANSACTION }, + { "extended_tcode", 112, 16, PACKET_FIELD_TRANSACTION }, + { "crc", 128, 32, PACKET_FIELD_DETAIL }, + { "data", 160, 0, PACKET_FIELD_TRANSACTION }, + { "crc", -64, 32, PACKET_FIELD_DETAIL }, + { "ack", -4, 4, 0, ack_names }, +}; + +static const struct packet_field write_response_fields[] = { + COMMON_RESPONSE_FIELDS, + { "reserved", 64, 32, PACKET_FIELD_DETAIL }, + { "ack", -4, 4, 0, ack_names }, +}; + +static const struct packet_field iso_data_fields[] = { + { "data_length", 0, 16, PACKET_FIELD_DATA_LENGTH }, + { "tag", 16, 2 }, + { "channel", 18, 6 }, + { "tcode", 24, 4, 0, tcode_names }, + { "sy", 28, 4 }, + { "crc", 32, 32, PACKET_FIELD_DETAIL }, + { "data", 64, 0 }, + { "crc", -64, 32, PACKET_FIELD_DETAIL }, + { "ack", -4, 4, 0, ack_names }, +}; + +static const struct packet_info packet_info[] = { + { + .name = "write_quadlet_request", + .type = PACKET_REQUEST, + .response_tcode = TCODE_WRITE_RESPONSE, + .fields = write_quadlet_request_fields, + .field_count = array_length(write_quadlet_request_fields) + }, + { + .name = "write_block_request", + .type = PACKET_REQUEST, + .response_tcode = TCODE_WRITE_RESPONSE, + .fields = block_request_fields, + .field_count = array_length(block_request_fields) + }, + { + .name = "write_response", + .type = PACKET_RESPONSE, + .fields = write_response_fields, + .field_count = array_length(write_response_fields) + }, + { + .name = "reserved", + .type = PACKET_RESERVED, + }, + { + .name = "read_quadlet_request", + .type = PACKET_REQUEST, + .response_tcode = TCODE_READ_QUADLET_RESPONSE, + .fields = read_quadlet_request_fields, + .field_count = array_length(read_quadlet_request_fields) + }, + { + .name = "read_block_request", + .type = PACKET_REQUEST, + .response_tcode = TCODE_READ_BLOCK_RESPONSE, + .fields = read_block_request_fields, + .field_count = array_length(read_block_request_fields) + }, + { + .name = "read_quadlet_response", + .type = PACKET_RESPONSE, + .fields = read_quadlet_response_fields, + .field_count = array_length(read_quadlet_response_fields) + }, + { + .name = "read_block_response", + .type = PACKET_RESPONSE, + .fields = block_response_fields, + .field_count = array_length(block_response_fields) + }, + { + .name = "cycle_start", + .type = PACKET_OTHER, + .fields = write_quadlet_request_fields, + .field_count = array_length(write_quadlet_request_fields) + }, + { + .name = "lock_request", + .type = PACKET_REQUEST, + .fields = block_request_fields, + .field_count = array_length(block_request_fields) + }, + { + .name = "iso_data", + .type = PACKET_OTHER, + .fields = iso_data_fields, + .field_count = array_length(iso_data_fields) + }, + { + .name = "lock_response", + .type = PACKET_RESPONSE, + .fields = block_response_fields, + .field_count = array_length(block_response_fields) + }, +}; + +static int +handle_request_packet(uint32_t *data, size_t length) +{ + struct link_packet *p = (struct link_packet *) data; + struct subaction *sa, *prev; + struct link_transaction *t; + + t = link_transaction_lookup(p->common.source, p->common.destination, + p->common.tlabel); + sa = subaction_create(data, length); + t->request = sa; + + if (!list_empty(&t->request_list)) { + prev = list_tail(&t->request_list, + struct subaction, link); + + if (!ACK_BUSY(prev->ack)) { + /* + * error, we should only see ack_busy_* before the + * ack_pending/ack_complete -- this is an ack_pending + * instead (ack_complete would have finished the + * transaction). + */ + } + + if (prev->packet.common.tcode != sa->packet.common.tcode || + prev->packet.common.tlabel != sa->packet.common.tlabel) { + /* memcmp() ? */ + /* error, these should match for retries. */ + } + } + + list_append(&t->request_list, &sa->link); + + switch (sa->ack) { + case ACK_COMPLETE: + if (p->common.tcode != TCODE_WRITE_QUADLET_REQUEST && + p->common.tcode != TCODE_WRITE_BLOCK_REQUEST) + /* error, unified transactions only allowed for write */; + list_remove(&t->link); + handle_transaction(t); + break; + + case ACK_NO_ACK: + case ACK_DATA_ERROR: + case ACK_TYPE_ERROR: + list_remove(&t->link); + handle_transaction(t); + break; + + case ACK_PENDING: + /* request subaction phase over, wait for response. */ + break; + + case ACK_BUSY_X: + case ACK_BUSY_A: + case ACK_BUSY_B: + /* ok, wait for retry. */ + /* check that retry protocol is respected. */ + break; + } + + return 1; +} + +static int +handle_response_packet(uint32_t *data, size_t length) +{ + struct link_packet *p = (struct link_packet *) data; + struct subaction *sa, *prev; + struct link_transaction *t; + + t = link_transaction_lookup(p->common.destination, p->common.source, + p->common.tlabel); + if (list_empty(&t->request_list)) { + /* unsolicited response */ + } + + sa = subaction_create(data, length); + t->response = sa; + + if (!list_empty(&t->response_list)) { + prev = list_tail(&t->response_list, struct subaction, link); + + if (!ACK_BUSY(prev->ack)) { + /* + * error, we should only see ack_busy_* before the + * ack_pending/ack_complete + */ + } + + if (prev->packet.common.tcode != sa->packet.common.tcode || + prev->packet.common.tlabel != sa->packet.common.tlabel) { + /* use memcmp() instead? */ + /* error, these should match for retries. */ + } + } else { + prev = list_tail(&t->request_list, struct subaction, link); + if (prev->ack != ACK_PENDING) { + /* + * error, should not get response unless last request got + * ack_pending. + */ + } + + if (packet_info[prev->packet.common.tcode].response_tcode != + sa->packet.common.tcode) { + /* error, tcode mismatch */ + } + } + + list_append(&t->response_list, &sa->link); + + switch (sa->ack) { + case ACK_COMPLETE: + case ACK_NO_ACK: + case ACK_DATA_ERROR: + case ACK_TYPE_ERROR: + list_remove(&t->link); + handle_transaction(t); + /* transaction complete, remove t from pending list. */ + break; + + case ACK_PENDING: + /* error for responses. */ + break; + + case ACK_BUSY_X: + case ACK_BUSY_A: + case ACK_BUSY_B: + /* no problem, wait for next retry */ + break; + } + + return 1; +} + +static int +handle_packet(uint32_t *data, size_t length) +{ + if (length == 0) { + printf("bus reset\r\n"); + clear_pending_transaction_list(); + } else if (length > sizeof(struct phy_packet)) { + struct link_packet *p = (struct link_packet *) data; + + switch (packet_info[p->common.tcode].type) { + case PACKET_REQUEST: + return handle_request_packet(data, length); + + case PACKET_RESPONSE: + return handle_response_packet(data, length); + + case PACKET_OTHER: + case PACKET_RESERVED: + return 0; + } + } + + return 1; +} + +static unsigned int +get_bits(struct link_packet *packet, int offset, int width) +{ + uint32_t *data = (uint32_t *) packet; + uint32_t index, shift, mask; + + index = offset / 32 + 1; + shift = 32 - (offset & 31) - width; + mask = width == 32 ? ~0 : (1 << width) - 1; + + return (data[index] >> shift) & mask; +} + +#if __BYTE_ORDER == __LITTLE_ENDIAN +#define byte_index(i) ((i) ^ 3) +#elif __BYTE_ORDER == __BIG_ENDIAN +#define byte_index(i) (i) +#else +#error unsupported byte order. +#endif + +static void +dump_data(unsigned char *data, int length) +{ + int i, print_length; + + if (length > 128) + print_length = 128; + else + print_length = length; + + for (i = 0; i < print_length; i++) + printf("%s%02hhx", + (i % 4 == 0 && i != 0) ? " " : "", + data[byte_index(i)]); + + if (print_length < length) + printf(" (%d more bytes)", length - print_length); +} + +static void +decode_link_packet(struct link_packet *packet, size_t length, + int include_flags, int exclude_flags) +{ + const struct packet_info *pi; + int data_length = 0; + int i; + + pi = &packet_info[packet->common.tcode]; + + for (i = 0; i < pi->field_count; i++) { + const struct packet_field *f = &pi->fields[i]; + int offset; + + if (f->flags & exclude_flags) + continue; + if (include_flags && !(f->flags & include_flags)) + continue; + + if (f->offset < 0) + offset = length * 8 + f->offset - 32; + else + offset = f->offset; + + if (f->value_names != NULL) { + uint32_t bits; + + bits = get_bits(packet, offset, f->width); + printf("%s", f->value_names[bits]); + } else if (f->width == 0) { + printf("%s=[", f->name); + dump_data((unsigned char *) packet + (offset / 8 + 4), data_length); + printf("]"); + } else { + unsigned long long bits; + int high_width, low_width; + + if ((offset & ~31) != ((offset + f->width - 1) & ~31)) { + /* Bit field spans quadlet boundary. */ + high_width = ((offset + 31) & ~31) - offset; + low_width = f->width - high_width; + + bits = get_bits(packet, offset, high_width); + bits = (bits << low_width) | + get_bits(packet, offset + high_width, low_width); + } else { + bits = get_bits(packet, offset, f->width); + } + + printf("%s=0x%0*llx", f->name, (f->width + 3) / 4, bits); + + if (f->flags & PACKET_FIELD_DATA_LENGTH) + data_length = bits; + } + + if (i < pi->field_count - 1) + printf(", "); + } +} + +static void +print_packet(uint32_t *data, size_t length) +{ + int i; + + printf("%6u ", data[0]); + + if (length == 4) { + printf("bus reset"); + } else if (length < sizeof(struct phy_packet)) { + printf("short packet: "); + for (i = 1; i < length / 4; i++) + printf("%s%08x", i == 0 ? "[" : " ", data[i]); + printf("]"); + + } else if (length == sizeof(struct phy_packet) && data[1] == ~data[2]) { + struct phy_packet *pp = (struct phy_packet *) data; + + /* phy packet are 3 quadlets: the 1 quadlet payload, + * the bitwise inverse of the payload and the snoop + * mode ack */ + + switch (pp->common.identifier) { + case PHY_PACKET_CONFIGURATION: + if (!pp->phy_config.set_root && !pp->phy_config.set_gap_count) { + printf("ext phy config: phy_id=%02x", pp->phy_config.root_id); + } else { + printf("phy config:"); + if (pp->phy_config.set_root) + printf(" set_root_id=%02x", pp->phy_config.root_id); + if (pp->phy_config.set_gap_count) + printf(" set_gap_count=%d", pp->phy_config.gap_count); + } + break; + + case PHY_PACKET_LINK_ON: + printf("link-on packet, phy_id=%02x", pp->link_on.phy_id); + break; + + case PHY_PACKET_SELF_ID: + if (pp->self_id.extended) { + printf("extended self id: phy_id=%02x, seq=%d", + pp->ext_self_id.phy_id, pp->ext_self_id.sequence); + } else { + static const char * const speed_names[] = { + "S100", "S200", "S400", "BETA" + }; + printf("self id: phy_id=%02x, link %s, gap_count=%d, speed=%s%s%s", + pp->self_id.phy_id, + (pp->self_id.link_active ? "active" : "not active"), + pp->self_id.gap_count, + speed_names[pp->self_id.phy_speed], + (pp->self_id.contender ? ", irm contender" : ""), + (pp->self_id.initiated_reset ? ", initiator" : "")); + } + break; + default: + printf("unknown phy packet: "); + for (i = 1; i < length / 4; i++) + printf("%s%08x", i == 0 ? "[" : " ", data[i]); + printf("]"); + break; + } + } else { + struct link_packet *packet = (struct link_packet *) data; + + decode_link_packet(packet, length, 0, + option_verbose ? 0 : PACKET_FIELD_DETAIL); + } + + if (option_hex) { + printf(" ["); + dump_data((unsigned char *) data + 4, length - 4); + printf("]"); + } + + printf("\r\n"); +} + +#define HIDE_CURSOR "\033[?25l" +#define SHOW_CURSOR "\033[?25h" +#define CLEAR "\033[H\033[2J" + +static void +print_stats(uint32_t *data, size_t length) +{ + static int bus_reset_count, short_packet_count, phy_packet_count; + static int tcode_count[16]; + static struct timeval last_update; + struct timeval now; + int i; + + if (length == 0) + bus_reset_count++; + else if (length < sizeof(struct phy_packet)) + short_packet_count++; + else if (length == sizeof(struct phy_packet) && data[1] == ~data[2]) + phy_packet_count++; + else { + struct link_packet *packet = (struct link_packet *) data; + tcode_count[packet->common.tcode]++; + } + + gettimeofday(&now, NULL); + if (now.tv_sec <= last_update.tv_sec && + now.tv_usec < last_update.tv_usec + 500000) + return; + + last_update = now; + printf(CLEAR HIDE_CURSOR + " bus resets : %8d\n" + " short packets : %8d\n" + " phy packets : %8d\n", + bus_reset_count, short_packet_count, phy_packet_count); + + for (i = 0; i < array_length(packet_info); i++) + if (packet_info[i].type != PACKET_RESERVED) + printf(" %-24s: %8d\n", packet_info[i].name, tcode_count[i]); + printf(SHOW_CURSOR "\n"); +} + +static struct termios saved_attributes; + +static void +reset_input_mode(void) +{ + tcsetattr(STDIN_FILENO, TCSANOW, &saved_attributes); +} + +static void +set_input_mode(void) +{ + struct termios tattr; + + /* Make sure stdin is a terminal. */ + if (!isatty(STDIN_FILENO)) { + fprintf(stderr, "Not a terminal.\n"); + exit(EXIT_FAILURE); + } + + /* Save the terminal attributes so we can restore them later. */ + tcgetattr(STDIN_FILENO, &saved_attributes); + atexit(reset_input_mode); + + /* Set the funny terminal modes. */ + tcgetattr(STDIN_FILENO, &tattr); + tattr.c_lflag &= ~(ICANON|ECHO); /* Clear ICANON and ECHO. */ + tattr.c_cc[VMIN] = 1; + tattr.c_cc[VTIME] = 0; + tcsetattr(STDIN_FILENO, TCSAFLUSH, &tattr); +} + +int main(int argc, const char *argv[]) +{ + uint32_t buf[128 * 1024]; + uint32_t filter; + int length, retval, view; + int fd = -1; + FILE *output = NULL, *input = NULL; + poptContext con; + char c; + struct pollfd pollfds[2]; + + sys_sigint_handler = signal(SIGINT, sigint_handler); + + con = poptGetContext(NULL, argc, argv, options, 0); + retval = poptGetNextOpt(con); + if (retval < -1) { + poptPrintUsage(con, stdout, 0); + return -1; + } + + if (option_version) { + printf("dump tool for nosy sniffer, version %s\n", VERSION); + return 0; + } + + if (__BYTE_ORDER != __LITTLE_ENDIAN) + fprintf(stderr, "warning: nosy has only been tested on little " + "endian machines\n"); + + if (option_input != NULL) { + input = fopen(option_input, "r"); + if (input == NULL) { + fprintf(stderr, "Could not open %s, %m\n", option_input); + return -1; + } + } else { + fd = open(option_nosy_device, O_RDWR); + if (fd < 0) { + fprintf(stderr, "Could not open %s, %m\n", option_nosy_device); + return -1; + } + set_input_mode(); + } + + if (strcmp(option_view, "transaction") == 0) + view = VIEW_TRANSACTION; + else if (strcmp(option_view, "stats") == 0) + view = VIEW_STATS; + else + view = VIEW_PACKET; + + if (option_output) { + output = fopen(option_output, "w"); + if (output == NULL) { + fprintf(stderr, "Could not open %s, %m\n", option_output); + return -1; + } + } + + setvbuf(stdout, NULL, _IOLBF, BUFSIZ); + + filter = ~0; + if (!option_iso) + filter &= ~(1 << TCODE_STREAM_DATA); + if (!option_cycle_start) + filter &= ~(1 << TCODE_CYCLE_START); + if (view == VIEW_STATS) + filter = ~(1 << TCODE_CYCLE_START); + + ioctl(fd, NOSY_IOC_FILTER, filter); + + ioctl(fd, NOSY_IOC_START); + + pollfds[0].fd = fd; + pollfds[0].events = POLLIN; + pollfds[1].fd = STDIN_FILENO; + pollfds[1].events = POLLIN; + + while (run) { + if (input != NULL) { + if (fread(&length, sizeof length, 1, input) != 1) + return 0; + fread(buf, 1, length, input); + } else { + poll(pollfds, 2, -1); + if (pollfds[1].revents) { + read(STDIN_FILENO, &c, sizeof c); + switch (c) { + case 'q': + if (output != NULL) + fclose(output); + return 0; + } + } + + if (pollfds[0].revents) + length = read(fd, buf, sizeof buf); + else + continue; + } + + if (output != NULL) { + fwrite(&length, sizeof length, 1, output); + fwrite(buf, 1, length, output); + } + + switch (view) { + case VIEW_TRANSACTION: + handle_packet(buf, length); + break; + case VIEW_PACKET: + print_packet(buf, length); + break; + case VIEW_STATS: + print_stats(buf, length); + break; + } + } + + if (output != NULL) + fclose(output); + + close(fd); + + poptFreeContext(con); + + return 0; +} diff --git a/tools/firewire/nosy-dump.h b/tools/firewire/nosy-dump.h new file mode 100644 index 00000000..3a4b5b33 --- /dev/null +++ b/tools/firewire/nosy-dump.h @@ -0,0 +1,173 @@ +#ifndef __nosy_dump_h__ +#define __nosy_dump_h__ + +#define array_length(array) (sizeof(array) / sizeof(array[0])) + +#define ACK_NO_ACK 0x0 +#define ACK_DONE(a) ((a >> 2) == 0) +#define ACK_BUSY(a) ((a >> 2) == 1) +#define ACK_ERROR(a) ((a >> 2) == 3) + +#include <stdint.h> + +struct phy_packet { + uint32_t timestamp; + union { + struct { + uint32_t zero:24; + uint32_t phy_id:6; + uint32_t identifier:2; + } common, link_on; + + struct { + uint32_t zero:16; + uint32_t gap_count:6; + uint32_t set_gap_count:1; + uint32_t set_root:1; + uint32_t root_id:6; + uint32_t identifier:2; + } phy_config; + + struct { + uint32_t more_packets:1; + uint32_t initiated_reset:1; + uint32_t port2:2; + uint32_t port1:2; + uint32_t port0:2; + uint32_t power_class:3; + uint32_t contender:1; + uint32_t phy_delay:2; + uint32_t phy_speed:2; + uint32_t gap_count:6; + uint32_t link_active:1; + uint32_t extended:1; + uint32_t phy_id:6; + uint32_t identifier:2; + } self_id; + + struct { + uint32_t more_packets:1; + uint32_t reserved1:1; + uint32_t porth:2; + uint32_t portg:2; + uint32_t portf:2; + uint32_t porte:2; + uint32_t portd:2; + uint32_t portc:2; + uint32_t portb:2; + uint32_t porta:2; + uint32_t reserved0:2; + uint32_t sequence:3; + uint32_t extended:1; + uint32_t phy_id:6; + uint32_t identifier:2; + } ext_self_id; + }; + uint32_t inverted; + uint32_t ack; +}; + +#define TCODE_PHY_PACKET 0x10 + +#define PHY_PACKET_CONFIGURATION 0x00 +#define PHY_PACKET_LINK_ON 0x01 +#define PHY_PACKET_SELF_ID 0x02 + +struct link_packet { + uint32_t timestamp; + union { + struct { + uint32_t priority:4; + uint32_t tcode:4; + uint32_t rt:2; + uint32_t tlabel:6; + uint32_t destination:16; + + uint32_t offset_high:16; + uint32_t source:16; + + uint32_t offset_low; + } common; + + struct { + uint32_t common[3]; + uint32_t crc; + } read_quadlet; + + struct { + uint32_t common[3]; + uint32_t data; + uint32_t crc; + } read_quadlet_response; + + struct { + uint32_t common[3]; + uint32_t extended_tcode:16; + uint32_t data_length:16; + uint32_t crc; + } read_block; + + struct { + uint32_t common[3]; + uint32_t extended_tcode:16; + uint32_t data_length:16; + uint32_t crc; + uint32_t data[0]; + /* crc and ack follows. */ + } read_block_response; + + struct { + uint32_t common[3]; + uint32_t data; + uint32_t crc; + } write_quadlet; + + struct { + uint32_t common[3]; + uint32_t extended_tcode:16; + uint32_t data_length:16; + uint32_t crc; + uint32_t data[0]; + /* crc and ack follows. */ + } write_block; + + struct { + uint32_t common[3]; + uint32_t crc; + } write_response; + + struct { + uint32_t common[3]; + uint32_t data; + uint32_t crc; + } cycle_start; + + struct { + uint32_t sy:4; + uint32_t tcode:4; + uint32_t channel:6; + uint32_t tag:2; + uint32_t data_length:16; + + uint32_t crc; + } iso_data; + }; +}; + +struct subaction { + uint32_t ack; + size_t length; + struct list link; + struct link_packet packet; +}; + +struct link_transaction { + int request_node, response_node, tlabel; + struct subaction *request, *response; + struct list request_list, response_list; + struct list link; +}; + +int decode_fcp(struct link_transaction *t); + +#endif /* __nosy_dump_h__ */ diff --git a/tools/perf/.gitignore b/tools/perf/.gitignore new file mode 100644 index 00000000..416684be --- /dev/null +++ b/tools/perf/.gitignore @@ -0,0 +1,21 @@ +PERF-CFLAGS +PERF-GUI-VARS +PERF-VERSION-FILE +perf +perf-help +perf-record +perf-report +perf-stat +perf-top +perf*.1 +perf*.xml +perf*.html +common-cmds.h +perf.data +perf.data.old +perf-archive +tags +TAGS +cscope* +config.mak +config.mak.autogen diff --git a/tools/perf/CREDITS b/tools/perf/CREDITS new file mode 100644 index 00000000..c2ddcb3a --- /dev/null +++ b/tools/perf/CREDITS @@ -0,0 +1,30 @@ +Most of the infrastructure that 'perf' uses here has been reused +from the Git project, as of version: + + 66996ec: Sync with 1.6.2.4 + +Here is an (incomplete!) list of main contributors to those files +in util/* and elsewhere: + + Alex Riesen + Christian Couder + Dmitry Potapov + Jeff King + Johannes Schindelin + Johannes Sixt + Junio C Hamano + Linus Torvalds + Matthias Kestenholz + Michal Ostrowski + Miklos Vajna + Petr Baudis + Pierre Habouzit + René Scharfe + Samuel Tardieu + Shawn O. Pearce + Steffen Prohaska + Steve Haslam + +Thanks guys! + +The full history of the files can be found in the upstream Git commits. diff --git a/tools/perf/Documentation/Makefile b/tools/perf/Documentation/Makefile new file mode 100644 index 00000000..4626a398 --- /dev/null +++ b/tools/perf/Documentation/Makefile @@ -0,0 +1,303 @@ +MAN1_TXT= \ + $(filter-out $(addsuffix .txt, $(ARTICLES) $(SP_ARTICLES)), \ + $(wildcard perf-*.txt)) \ + perf.txt +MAN5_TXT= +MAN7_TXT= + +MAN_TXT = $(MAN1_TXT) $(MAN5_TXT) $(MAN7_TXT) +MAN_XML=$(patsubst %.txt,%.xml,$(MAN_TXT)) +MAN_HTML=$(patsubst %.txt,%.html,$(MAN_TXT)) + +DOC_HTML=$(MAN_HTML) + +ARTICLES = +# with their own formatting rules. +SP_ARTICLES = +API_DOCS = $(patsubst %.txt,%,$(filter-out technical/api-index-skel.txt technical/api-index.txt, $(wildcard technical/api-*.txt))) +SP_ARTICLES += $(API_DOCS) +SP_ARTICLES += technical/api-index + +DOC_HTML += $(patsubst %,%.html,$(ARTICLES) $(SP_ARTICLES)) + +DOC_MAN1=$(patsubst %.txt,%.1,$(MAN1_TXT)) +DOC_MAN5=$(patsubst %.txt,%.5,$(MAN5_TXT)) +DOC_MAN7=$(patsubst %.txt,%.7,$(MAN7_TXT)) + +# Make the path relative to DESTDIR, not prefix +ifndef DESTDIR +prefix?=$(HOME) +endif +bindir?=$(prefix)/bin +htmldir?=$(prefix)/share/doc/perf-doc +pdfdir?=$(prefix)/share/doc/perf-doc +mandir?=$(prefix)/share/man +man1dir=$(mandir)/man1 +man5dir=$(mandir)/man5 +man7dir=$(mandir)/man7 + +ASCIIDOC=asciidoc +ASCIIDOC_EXTRA = --unsafe +MANPAGE_XSL = manpage-normal.xsl +XMLTO_EXTRA = +INSTALL?=install +RM ?= rm -f +DOC_REF = origin/man +HTML_REF = origin/html + +infodir?=$(prefix)/share/info +MAKEINFO=makeinfo +INSTALL_INFO=install-info +DOCBOOK2X_TEXI=docbook2x-texi +DBLATEX=dblatex +ifndef PERL_PATH + PERL_PATH = /usr/bin/perl +endif + +-include ../config.mak.autogen +-include ../config.mak + +# +# For asciidoc ... +# -7.1.2, no extra settings are needed. +# 8.0-, set ASCIIDOC8. +# + +# +# For docbook-xsl ... +# -1.68.1, set ASCIIDOC_NO_ROFF? (based on changelog from 1.73.0) +# 1.69.0, no extra settings are needed? +# 1.69.1-1.71.0, set DOCBOOK_SUPPRESS_SP? +# 1.71.1, no extra settings are needed? +# 1.72.0, set DOCBOOK_XSL_172. +# 1.73.0-, set ASCIIDOC_NO_ROFF +# + +# +# If you had been using DOCBOOK_XSL_172 in an attempt to get rid +# of 'the ".ft C" problem' in your generated manpages, and you +# instead ended up with weird characters around callouts, try +# using ASCIIDOC_NO_ROFF instead (it works fine with ASCIIDOC8). +# + +ifdef ASCIIDOC8 +ASCIIDOC_EXTRA += -a asciidoc7compatible +endif +ifdef DOCBOOK_XSL_172 +ASCIIDOC_EXTRA += -a perf-asciidoc-no-roff +MANPAGE_XSL = manpage-1.72.xsl +else + ifdef ASCIIDOC_NO_ROFF + # docbook-xsl after 1.72 needs the regular XSL, but will not + # pass-thru raw roff codes from asciidoc.conf, so turn them off. + ASCIIDOC_EXTRA += -a perf-asciidoc-no-roff + endif +endif +ifdef MAN_BOLD_LITERAL +XMLTO_EXTRA += -m manpage-bold-literal.xsl +endif +ifdef DOCBOOK_SUPPRESS_SP +XMLTO_EXTRA += -m manpage-suppress-sp.xsl +endif + +SHELL_PATH ?= $(SHELL) +# Shell quote; +SHELL_PATH_SQ = $(subst ','\'',$(SHELL_PATH)) + +# +# Please note that there is a minor bug in asciidoc. +# The version after 6.0.3 _will_ include the patch found here: +# http://marc.theaimsgroup.com/?l=perf&m=111558757202243&w=2 +# +# Until that version is released you may have to apply the patch +# yourself - yes, all 6 characters of it! +# + +QUIET_SUBDIR0 = +$(MAKE) -C # space to separate -C and subdir +QUIET_SUBDIR1 = + +ifneq ($(findstring $(MAKEFLAGS),w),w) +PRINT_DIR = --no-print-directory +else # "make -w" +NO_SUBDIR = : +endif + +ifneq ($(findstring $(MAKEFLAGS),s),s) +ifndef V + QUIET_ASCIIDOC = @echo ' ' ASCIIDOC $@; + QUIET_XMLTO = @echo ' ' XMLTO $@; + QUIET_DB2TEXI = @echo ' ' DB2TEXI $@; + QUIET_MAKEINFO = @echo ' ' MAKEINFO $@; + QUIET_DBLATEX = @echo ' ' DBLATEX $@; + QUIET_XSLTPROC = @echo ' ' XSLTPROC $@; + QUIET_GEN = @echo ' ' GEN $@; + QUIET_STDERR = 2> /dev/null + QUIET_SUBDIR0 = +@subdir= + QUIET_SUBDIR1 = ;$(NO_SUBDIR) echo ' ' SUBDIR $$subdir; \ + $(MAKE) $(PRINT_DIR) -C $$subdir + export V +endif +endif + +all: html man + +html: $(DOC_HTML) + +$(DOC_HTML) $(DOC_MAN1) $(DOC_MAN5) $(DOC_MAN7): asciidoc.conf + +man: man1 man5 man7 +man1: $(DOC_MAN1) +man5: $(DOC_MAN5) +man7: $(DOC_MAN7) + +info: perf.info perfman.info + +pdf: user-manual.pdf + +install: install-man + +install-man: man + $(INSTALL) -d -m 755 $(DESTDIR)$(man1dir) +# $(INSTALL) -d -m 755 $(DESTDIR)$(man5dir) +# $(INSTALL) -d -m 755 $(DESTDIR)$(man7dir) + $(INSTALL) -m 644 $(DOC_MAN1) $(DESTDIR)$(man1dir) +# $(INSTALL) -m 644 $(DOC_MAN5) $(DESTDIR)$(man5dir) +# $(INSTALL) -m 644 $(DOC_MAN7) $(DESTDIR)$(man7dir) + +install-info: info + $(INSTALL) -d -m 755 $(DESTDIR)$(infodir) + $(INSTALL) -m 644 perf.info perfman.info $(DESTDIR)$(infodir) + if test -r $(DESTDIR)$(infodir)/dir; then \ + $(INSTALL_INFO) --info-dir=$(DESTDIR)$(infodir) perf.info ;\ + $(INSTALL_INFO) --info-dir=$(DESTDIR)$(infodir) perfman.info ;\ + else \ + echo "No directory found in $(DESTDIR)$(infodir)" >&2 ; \ + fi + +install-pdf: pdf + $(INSTALL) -d -m 755 $(DESTDIR)$(pdfdir) + $(INSTALL) -m 644 user-manual.pdf $(DESTDIR)$(pdfdir) + +#install-html: html +# '$(SHELL_PATH_SQ)' ./install-webdoc.sh $(DESTDIR)$(htmldir) + +../PERF-VERSION-FILE: .FORCE-PERF-VERSION-FILE + $(QUIET_SUBDIR0)../ $(QUIET_SUBDIR1) PERF-VERSION-FILE + +-include ../PERF-VERSION-FILE + +# +# Determine "include::" file references in asciidoc files. +# +doc.dep : $(wildcard *.txt) build-docdep.perl + $(QUIET_GEN)$(RM) $@+ $@ && \ + $(PERL_PATH) ./build-docdep.perl >$@+ $(QUIET_STDERR) && \ + mv $@+ $@ + +-include doc.dep + +cmds_txt = cmds-ancillaryinterrogators.txt \ + cmds-ancillarymanipulators.txt \ + cmds-mainporcelain.txt \ + cmds-plumbinginterrogators.txt \ + cmds-plumbingmanipulators.txt \ + cmds-synchingrepositories.txt \ + cmds-synchelpers.txt \ + cmds-purehelpers.txt \ + cmds-foreignscminterface.txt + +$(cmds_txt): cmd-list.made + +cmd-list.made: cmd-list.perl ../command-list.txt $(MAN1_TXT) + $(QUIET_GEN)$(RM) $@ && \ + $(PERL_PATH) ./cmd-list.perl ../command-list.txt $(QUIET_STDERR) && \ + date >$@ + +clean: + $(RM) *.xml *.xml+ *.html *.html+ *.1 *.5 *.7 + $(RM) *.texi *.texi+ *.texi++ perf.info perfman.info + $(RM) howto-index.txt howto/*.html doc.dep + $(RM) technical/api-*.html technical/api-index.txt + $(RM) $(cmds_txt) *.made + +$(MAN_HTML): %.html : %.txt + $(QUIET_ASCIIDOC)$(RM) $@+ $@ && \ + $(ASCIIDOC) -b xhtml11 -d manpage -f asciidoc.conf \ + $(ASCIIDOC_EXTRA) -aperf_version=$(PERF_VERSION) -o $@+ $< && \ + mv $@+ $@ + +%.1 %.5 %.7 : %.xml + $(QUIET_XMLTO)$(RM) $@ && \ + xmlto -m $(MANPAGE_XSL) $(XMLTO_EXTRA) man $< + +%.xml : %.txt + $(QUIET_ASCIIDOC)$(RM) $@+ $@ && \ + $(ASCIIDOC) -b docbook -d manpage -f asciidoc.conf \ + $(ASCIIDOC_EXTRA) -aperf_version=$(PERF_VERSION) -o $@+ $< && \ + mv $@+ $@ + +XSLT = docbook.xsl +XSLTOPTS = --xinclude --stringparam html.stylesheet docbook-xsl.css + +user-manual.html: user-manual.xml + $(QUIET_XSLTPROC)xsltproc $(XSLTOPTS) -o $@ $(XSLT) $< + +perf.info: user-manual.texi + $(QUIET_MAKEINFO)$(MAKEINFO) --no-split -o $@ user-manual.texi + +user-manual.texi: user-manual.xml + $(QUIET_DB2TEXI)$(RM) $@+ $@ && \ + $(DOCBOOK2X_TEXI) user-manual.xml --encoding=UTF-8 --to-stdout >$@++ && \ + $(PERL_PATH) fix-texi.perl <$@++ >$@+ && \ + rm $@++ && \ + mv $@+ $@ + +user-manual.pdf: user-manual.xml + $(QUIET_DBLATEX)$(RM) $@+ $@ && \ + $(DBLATEX) -o $@+ -p /etc/asciidoc/dblatex/asciidoc-dblatex.xsl -s /etc/asciidoc/dblatex/asciidoc-dblatex.sty $< && \ + mv $@+ $@ + +perfman.texi: $(MAN_XML) cat-texi.perl + $(QUIET_DB2TEXI)$(RM) $@+ $@ && \ + ($(foreach xml,$(MAN_XML),$(DOCBOOK2X_TEXI) --encoding=UTF-8 \ + --to-stdout $(xml) &&) true) > $@++ && \ + $(PERL_PATH) cat-texi.perl $@ <$@++ >$@+ && \ + rm $@++ && \ + mv $@+ $@ + +perfman.info: perfman.texi + $(QUIET_MAKEINFO)$(MAKEINFO) --no-split --no-validate $*.texi + +$(patsubst %.txt,%.texi,$(MAN_TXT)): %.texi : %.xml + $(QUIET_DB2TEXI)$(RM) $@+ $@ && \ + $(DOCBOOK2X_TEXI) --to-stdout $*.xml >$@+ && \ + mv $@+ $@ + +howto-index.txt: howto-index.sh $(wildcard howto/*.txt) + $(QUIET_GEN)$(RM) $@+ $@ && \ + '$(SHELL_PATH_SQ)' ./howto-index.sh $(wildcard howto/*.txt) >$@+ && \ + mv $@+ $@ + +$(patsubst %,%.html,$(ARTICLES)) : %.html : %.txt + $(QUIET_ASCIIDOC)$(ASCIIDOC) -b xhtml11 $*.txt + +WEBDOC_DEST = /pub/software/tools/perf/docs + +$(patsubst %.txt,%.html,$(wildcard howto/*.txt)): %.html : %.txt + $(QUIET_ASCIIDOC)$(RM) $@+ $@ && \ + sed -e '1,/^$$/d' $< | $(ASCIIDOC) -b xhtml11 - >$@+ && \ + mv $@+ $@ + +# UNIMPLEMENTED +#install-webdoc : html +# '$(SHELL_PATH_SQ)' ./install-webdoc.sh $(WEBDOC_DEST) + +# quick-install: quick-install-man + +# quick-install-man: +# '$(SHELL_PATH_SQ)' ./install-doc-quick.sh $(DOC_REF) $(DESTDIR)$(mandir) + +#quick-install-html: +# '$(SHELL_PATH_SQ)' ./install-doc-quick.sh $(HTML_REF) $(DESTDIR)$(htmldir) + +.PHONY: .FORCE-PERF-VERSION-FILE diff --git a/tools/perf/Documentation/asciidoc.conf b/tools/perf/Documentation/asciidoc.conf new file mode 100644 index 00000000..356b23a4 --- /dev/null +++ b/tools/perf/Documentation/asciidoc.conf @@ -0,0 +1,91 @@ +## linkperf: macro +# +# Usage: linkperf:command[manpage-section] +# +# Note, {0} is the manpage section, while {target} is the command. +# +# Show PERF link as: <command>(<section>); if section is defined, else just show +# the command. + +[macros] +(?su)[\\]?(?P<name>linkperf):(?P<target>\S*?)\[(?P<attrlist>.*?)\]= + +[attributes] +asterisk=* +plus=+ +caret=^ +startsb=[ +endsb=] +tilde=~ + +ifdef::backend-docbook[] +[linkperf-inlinemacro] +{0%{target}} +{0#<citerefentry>} +{0#<refentrytitle>{target}</refentrytitle><manvolnum>{0}</manvolnum>} +{0#</citerefentry>} +endif::backend-docbook[] + +ifdef::backend-docbook[] +ifndef::perf-asciidoc-no-roff[] +# "unbreak" docbook-xsl v1.68 for manpages. v1.69 works with or without this. +# v1.72 breaks with this because it replaces dots not in roff requests. +[listingblock] +<example><title>{title}</title> +<literallayout> +ifdef::doctype-manpage[] + .ft C +endif::doctype-manpage[] +| +ifdef::doctype-manpage[] + .ft +endif::doctype-manpage[] +</literallayout> +{title#}</example> +endif::perf-asciidoc-no-roff[] + +ifdef::perf-asciidoc-no-roff[] +ifdef::doctype-manpage[] +# The following two small workarounds insert a simple paragraph after screen +[listingblock] +<example><title>{title}</title> +<literallayout> +| +</literallayout><simpara></simpara> +{title#}</example> + +[verseblock] +<formalpara{id? id="{id}"}><title>{title}</title><para> +{title%}<literallayout{id? id="{id}"}> +{title#}<literallayout> +| +</literallayout> +{title#}</para></formalpara> +{title%}<simpara></simpara> +endif::doctype-manpage[] +endif::perf-asciidoc-no-roff[] +endif::backend-docbook[] + +ifdef::doctype-manpage[] +ifdef::backend-docbook[] +[header] +template::[header-declarations] +<refentry> +<refmeta> +<refentrytitle>{mantitle}</refentrytitle> +<manvolnum>{manvolnum}</manvolnum> +<refmiscinfo class="source">perf</refmiscinfo> +<refmiscinfo class="version">{perf_version}</refmiscinfo> +<refmiscinfo class="manual">perf Manual</refmiscinfo> +</refmeta> +<refnamediv> + <refname>{manname}</refname> + <refpurpose>{manpurpose}</refpurpose> +</refnamediv> +endif::backend-docbook[] +endif::doctype-manpage[] + +ifdef::backend-xhtml11[] +[linkperf-inlinemacro] +<a href="{target}.html">{target}{0?({0})}</a> +endif::backend-xhtml11[] diff --git a/tools/perf/Documentation/examples.txt b/tools/perf/Documentation/examples.txt new file mode 100644 index 00000000..8eb6c489 --- /dev/null +++ b/tools/perf/Documentation/examples.txt @@ -0,0 +1,225 @@ + + ------------------------------ + ****** perf by examples ****** + ------------------------------ + +[ From an e-mail by Ingo Molnar, http://lkml.org/lkml/2009/8/4/346 ] + + +First, discovery/enumeration of available counters can be done via +'perf list': + +titan:~> perf list + [...] + kmem:kmalloc [Tracepoint event] + kmem:kmem_cache_alloc [Tracepoint event] + kmem:kmalloc_node [Tracepoint event] + kmem:kmem_cache_alloc_node [Tracepoint event] + kmem:kfree [Tracepoint event] + kmem:kmem_cache_free [Tracepoint event] + kmem:mm_page_free_direct [Tracepoint event] + kmem:mm_pagevec_free [Tracepoint event] + kmem:mm_page_alloc [Tracepoint event] + kmem:mm_page_alloc_zone_locked [Tracepoint event] + kmem:mm_page_pcpu_drain [Tracepoint event] + kmem:mm_page_alloc_extfrag [Tracepoint event] + +Then any (or all) of the above event sources can be activated and +measured. For example the page alloc/free properties of a 'hackbench +run' are: + + titan:~> perf stat -e kmem:mm_page_pcpu_drain -e kmem:mm_page_alloc + -e kmem:mm_pagevec_free -e kmem:mm_page_free_direct ./hackbench 10 + Time: 0.575 + + Performance counter stats for './hackbench 10': + + 13857 kmem:mm_page_pcpu_drain + 27576 kmem:mm_page_alloc + 6025 kmem:mm_pagevec_free + 20934 kmem:mm_page_free_direct + + 0.613972165 seconds time elapsed + +You can observe the statistical properties as well, by using the +'repeat the workload N times' feature of perf stat: + + titan:~> perf stat --repeat 5 -e kmem:mm_page_pcpu_drain -e + kmem:mm_page_alloc -e kmem:mm_pagevec_free -e + kmem:mm_page_free_direct ./hackbench 10 + Time: 0.627 + Time: 0.644 + Time: 0.564 + Time: 0.559 + Time: 0.626 + + Performance counter stats for './hackbench 10' (5 runs): + + 12920 kmem:mm_page_pcpu_drain ( +- 3.359% ) + 25035 kmem:mm_page_alloc ( +- 3.783% ) + 6104 kmem:mm_pagevec_free ( +- 0.934% ) + 18376 kmem:mm_page_free_direct ( +- 4.941% ) + + 0.643954516 seconds time elapsed ( +- 2.363% ) + +Furthermore, these tracepoints can be used to sample the workload as +well. For example the page allocations done by a 'git gc' can be +captured the following way: + + titan:~/git> perf record -f -e kmem:mm_page_alloc -c 1 ./git gc + Counting objects: 1148, done. + Delta compression using up to 2 threads. + Compressing objects: 100% (450/450), done. + Writing objects: 100% (1148/1148), done. + Total 1148 (delta 690), reused 1148 (delta 690) + [ perf record: Captured and wrote 0.267 MB perf.data (~11679 samples) ] + +To check which functions generated page allocations: + + titan:~/git> perf report + # Samples: 10646 + # + # Overhead Command Shared Object + # ........ ............... .......................... + # + 23.57% git-repack /lib64/libc-2.5.so + 21.81% git /lib64/libc-2.5.so + 14.59% git ./git + 11.79% git-repack ./git + 7.12% git /lib64/ld-2.5.so + 3.16% git-repack /lib64/libpthread-2.5.so + 2.09% git-repack /bin/bash + 1.97% rm /lib64/libc-2.5.so + 1.39% mv /lib64/ld-2.5.so + 1.37% mv /lib64/libc-2.5.so + 1.12% git-repack /lib64/ld-2.5.so + 0.95% rm /lib64/ld-2.5.so + 0.90% git-update-serv /lib64/libc-2.5.so + 0.73% git-update-serv /lib64/ld-2.5.so + 0.68% perf /lib64/libpthread-2.5.so + 0.64% git-repack /usr/lib64/libz.so.1.2.3 + +Or to see it on a more finegrained level: + +titan:~/git> perf report --sort comm,dso,symbol +# Samples: 10646 +# +# Overhead Command Shared Object Symbol +# ........ ............... .......................... ...... +# + 9.35% git-repack ./git [.] insert_obj_hash + 9.12% git ./git [.] insert_obj_hash + 7.31% git /lib64/libc-2.5.so [.] memcpy + 6.34% git-repack /lib64/libc-2.5.so [.] _int_malloc + 6.24% git-repack /lib64/libc-2.5.so [.] memcpy + 5.82% git-repack /lib64/libc-2.5.so [.] __GI___fork + 5.47% git /lib64/libc-2.5.so [.] _int_malloc + 2.99% git /lib64/libc-2.5.so [.] memset + +Furthermore, call-graph sampling can be done too, of page +allocations - to see precisely what kind of page allocations there +are: + + titan:~/git> perf record -f -g -e kmem:mm_page_alloc -c 1 ./git gc + Counting objects: 1148, done. + Delta compression using up to 2 threads. + Compressing objects: 100% (450/450), done. + Writing objects: 100% (1148/1148), done. + Total 1148 (delta 690), reused 1148 (delta 690) + [ perf record: Captured and wrote 0.963 MB perf.data (~42069 samples) ] + + titan:~/git> perf report -g + # Samples: 10686 + # + # Overhead Command Shared Object + # ........ ............... .......................... + # + 23.25% git-repack /lib64/libc-2.5.so + | + |--50.00%-- _int_free + | + |--37.50%-- __GI___fork + | make_child + | + |--12.50%-- ptmalloc_unlock_all2 + | make_child + | + --6.25%-- __GI_strcpy + 21.61% git /lib64/libc-2.5.so + | + |--30.00%-- __GI_read + | | + | --83.33%-- git_config_from_file + | git_config + | | + [...] + +Or you can observe the whole system's page allocations for 10 +seconds: + +titan:~/git> perf stat -a -e kmem:mm_page_pcpu_drain -e +kmem:mm_page_alloc -e kmem:mm_pagevec_free -e +kmem:mm_page_free_direct sleep 10 + + Performance counter stats for 'sleep 10': + + 171585 kmem:mm_page_pcpu_drain + 322114 kmem:mm_page_alloc + 73623 kmem:mm_pagevec_free + 254115 kmem:mm_page_free_direct + + 10.000591410 seconds time elapsed + +Or observe how fluctuating the page allocations are, via statistical +analysis done over ten 1-second intervals: + + titan:~/git> perf stat --repeat 10 -a -e kmem:mm_page_pcpu_drain -e + kmem:mm_page_alloc -e kmem:mm_pagevec_free -e + kmem:mm_page_free_direct sleep 1 + + Performance counter stats for 'sleep 1' (10 runs): + + 17254 kmem:mm_page_pcpu_drain ( +- 3.709% ) + 34394 kmem:mm_page_alloc ( +- 4.617% ) + 7509 kmem:mm_pagevec_free ( +- 4.820% ) + 25653 kmem:mm_page_free_direct ( +- 3.672% ) + + 1.058135029 seconds time elapsed ( +- 3.089% ) + +Or you can annotate the recorded 'git gc' run on a per symbol basis +and check which instructions/source-code generated page allocations: + + titan:~/git> perf annotate __GI___fork + ------------------------------------------------ + Percent | Source code & Disassembly of libc-2.5.so + ------------------------------------------------ + : + : + : Disassembly of section .plt: + : Disassembly of section .text: + : + : 00000031a2e95560 <__fork>: + [...] + 0.00 : 31a2e95602: b8 38 00 00 00 mov $0x38,%eax + 0.00 : 31a2e95607: 0f 05 syscall + 83.42 : 31a2e95609: 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax + 0.00 : 31a2e9560f: 0f 87 4d 01 00 00 ja 31a2e95762 <__fork+0x202> + 0.00 : 31a2e95615: 85 c0 test %eax,%eax + +( this shows that 83.42% of __GI___fork's page allocations come from + the 0x38 system call it performs. ) + +etc. etc. - a lot more is possible. I could list a dozen of +other different usecases straight away - neither of which is +possible via /proc/vmstat. + +/proc/vmstat is not in the same league really, in terms of +expressive power of system analysis and performance +analysis. + +All that the above results needed were those new tracepoints +in include/tracing/events/kmem.h. + + Ingo + + diff --git a/tools/perf/Documentation/manpage-1.72.xsl b/tools/perf/Documentation/manpage-1.72.xsl new file mode 100644 index 00000000..b4d315cb --- /dev/null +++ b/tools/perf/Documentation/manpage-1.72.xsl @@ -0,0 +1,14 @@ +<!-- manpage-1.72.xsl: + special settings for manpages rendered from asciidoc+docbook + handles peculiarities in docbook-xsl 1.72.0 --> +<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" + version="1.0"> + +<xsl:import href="manpage-base.xsl"/> + +<!-- these are the special values for the roff control characters + needed for docbook-xsl 1.72.0 --> +<xsl:param name="git.docbook.backslash">▓</xsl:param> +<xsl:param name="git.docbook.dot" >⌂</xsl:param> + +</xsl:stylesheet> diff --git a/tools/perf/Documentation/manpage-base.xsl b/tools/perf/Documentation/manpage-base.xsl new file mode 100644 index 00000000..a264fa61 --- /dev/null +++ b/tools/perf/Documentation/manpage-base.xsl @@ -0,0 +1,35 @@ +<!-- manpage-base.xsl: + special formatting for manpages rendered from asciidoc+docbook --> +<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" + version="1.0"> + +<!-- these params silence some output from xmlto --> +<xsl:param name="man.output.quietly" select="1"/> +<xsl:param name="refentry.meta.get.quietly" select="1"/> + +<!-- convert asciidoc callouts to man page format; + git.docbook.backslash and git.docbook.dot params + must be supplied by another XSL file or other means --> +<xsl:template match="co"> + <xsl:value-of select="concat( + $git.docbook.backslash,'fB(', + substring-after(@id,'-'),')', + $git.docbook.backslash,'fR')"/> +</xsl:template> +<xsl:template match="calloutlist"> + <xsl:value-of select="$git.docbook.dot"/> + <xsl:text>sp </xsl:text> + <xsl:apply-templates/> + <xsl:text> </xsl:text> +</xsl:template> +<xsl:template match="callout"> + <xsl:value-of select="concat( + $git.docbook.backslash,'fB', + substring-after(@arearefs,'-'), + '. ',$git.docbook.backslash,'fR')"/> + <xsl:apply-templates/> + <xsl:value-of select="$git.docbook.dot"/> + <xsl:text>br </xsl:text> +</xsl:template> + +</xsl:stylesheet> diff --git a/tools/perf/Documentation/manpage-bold-literal.xsl b/tools/perf/Documentation/manpage-bold-literal.xsl new file mode 100644 index 00000000..608eb5df --- /dev/null +++ b/tools/perf/Documentation/manpage-bold-literal.xsl @@ -0,0 +1,17 @@ +<!-- manpage-bold-literal.xsl: + special formatting for manpages rendered from asciidoc+docbook --> +<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" + version="1.0"> + +<!-- render literal text as bold (instead of plain or monospace); + this makes literal text easier to distinguish in manpages + viewed on a tty --> +<xsl:template match="literal"> + <xsl:value-of select="$git.docbook.backslash"/> + <xsl:text>fB</xsl:text> + <xsl:apply-templates/> + <xsl:value-of select="$git.docbook.backslash"/> + <xsl:text>fR</xsl:text> +</xsl:template> + +</xsl:stylesheet> diff --git a/tools/perf/Documentation/manpage-normal.xsl b/tools/perf/Documentation/manpage-normal.xsl new file mode 100644 index 00000000..a48f5b11 --- /dev/null +++ b/tools/perf/Documentation/manpage-normal.xsl @@ -0,0 +1,13 @@ +<!-- manpage-normal.xsl: + special settings for manpages rendered from asciidoc+docbook + handles anything we want to keep away from docbook-xsl 1.72.0 --> +<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" + version="1.0"> + +<xsl:import href="manpage-base.xsl"/> + +<!-- these are the normal values for the roff control characters --> +<xsl:param name="git.docbook.backslash">\</xsl:param> +<xsl:param name="git.docbook.dot" >.</xsl:param> + +</xsl:stylesheet> diff --git a/tools/perf/Documentation/manpage-suppress-sp.xsl b/tools/perf/Documentation/manpage-suppress-sp.xsl new file mode 100644 index 00000000..a63c7632 --- /dev/null +++ b/tools/perf/Documentation/manpage-suppress-sp.xsl @@ -0,0 +1,21 @@ +<!-- manpage-suppress-sp.xsl: + special settings for manpages rendered from asciidoc+docbook + handles erroneous, inline .sp in manpage output of some + versions of docbook-xsl --> +<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" + version="1.0"> + +<!-- attempt to work around spurious .sp at the tail of the line + that some versions of docbook stylesheets seem to add --> +<xsl:template match="simpara"> + <xsl:variable name="content"> + <xsl:apply-templates/> + </xsl:variable> + <xsl:value-of select="normalize-space($content)"/> + <xsl:if test="not(ancestor::authorblurb) and + not(ancestor::personblurb)"> + <xsl:text> </xsl:text> + </xsl:if> +</xsl:template> + +</xsl:stylesheet> diff --git a/tools/perf/Documentation/perf-annotate.txt b/tools/perf/Documentation/perf-annotate.txt new file mode 100644 index 00000000..6f5a4986 --- /dev/null +++ b/tools/perf/Documentation/perf-annotate.txt @@ -0,0 +1,71 @@ +perf-annotate(1) +================ + +NAME +---- +perf-annotate - Read perf.data (created by perf record) and display annotated code + +SYNOPSIS +-------- +[verse] +'perf annotate' [-i <file> | --input=file] [symbol_name] + +DESCRIPTION +----------- +This command reads the input file and displays an annotated version of the +code. If the object file has debug symbols then the source code will be +displayed alongside assembly code. + +If there is no debug info in the object, then annotated assembly is displayed. + +OPTIONS +------- +-i:: +--input=:: + Input file name. (default: perf.data) + +-d:: +--dsos=<dso[,dso...]>:: + Only consider symbols in these dsos. +-s:: +--symbol=<symbol>:: + Symbol to annotate. + +-f:: +--force:: + Don't complain, do it. + +-v:: +--verbose:: + Be more verbose. (Show symbol address, etc) + +-D:: +--dump-raw-trace:: + Dump raw trace in ASCII. + +-k:: +--vmlinux=<file>:: + vmlinux pathname. + +-m:: +--modules:: + Load module symbols. WARNING: use only with -k and LIVE kernel. + +-l:: +--print-line:: + Print matching source lines (may be slow). + +-P:: +--full-paths:: + Don't shorten the displayed pathnames. + +--stdio:: Use the stdio interface. + +--tui:: Use the TUI interface Use of --tui requires a tty, if one is not + present, as when piping to other commands, the stdio interface is + used. This interfaces starts by centering on the line with more + samples, TAB/UNTAB cycles through the lines with more samples. + +SEE ALSO +-------- +linkperf:perf-record[1], linkperf:perf-report[1] diff --git a/tools/perf/Documentation/perf-archive.txt b/tools/perf/Documentation/perf-archive.txt new file mode 100644 index 00000000..fae174dc --- /dev/null +++ b/tools/perf/Documentation/perf-archive.txt @@ -0,0 +1,22 @@ +perf-archive(1) +=============== + +NAME +---- +perf-archive - Create archive with object files with build-ids found in perf.data file + +SYNOPSIS +-------- +[verse] +'perf archive' [file] + +DESCRIPTION +----------- +This command runs runs perf-buildid-list --with-hits, and collects the files +with the buildids found so that analisys of perf.data contents can be possible +on another machine. + + +SEE ALSO +-------- +linkperf:perf-record[1], linkperf:perf-buildid-list[1], linkperf:perf-report[1] diff --git a/tools/perf/Documentation/perf-bench.txt b/tools/perf/Documentation/perf-bench.txt new file mode 100644 index 00000000..a3dbadb2 --- /dev/null +++ b/tools/perf/Documentation/perf-bench.txt @@ -0,0 +1,120 @@ +perf-bench(1) +============= + +NAME +---- +perf-bench - General framework for benchmark suites + +SYNOPSIS +-------- +[verse] +'perf bench' [<common options>] <subsystem> <suite> [<options>] + +DESCRIPTION +----------- +This 'perf bench' command is general framework for benchmark suites. + +COMMON OPTIONS +-------------- +-f:: +--format=:: +Specify format style. +Current available format styles are: + +'default':: +Default style. This is mainly for human reading. +--------------------- +% perf bench sched pipe # with no style specified +(executing 1000000 pipe operations between two tasks) + Total time:5.855 sec + 5.855061 usecs/op + 170792 ops/sec +--------------------- + +'simple':: +This simple style is friendly for automated +processing by scripts. +--------------------- +% perf bench --format=simple sched pipe # specified simple +5.988 +--------------------- + +SUBSYSTEM +--------- + +'sched':: + Scheduler and IPC mechanisms. + +SUITES FOR 'sched' +~~~~~~~~~~~~~~~~~~ +*messaging*:: +Suite for evaluating performance of scheduler and IPC mechanisms. +Based on hackbench by Rusty Russell. + +Options of *pipe* +^^^^^^^^^^^^^^^^^ +-p:: +--pipe:: +Use pipe() instead of socketpair() + +-t:: +--thread:: +Be multi thread instead of multi process + +-g:: +--group=:: +Specify number of groups + +-l:: +--loop=:: +Specify number of loops + +Example of *messaging* +^^^^^^^^^^^^^^^^^^^^^^ + +--------------------- +% perf bench sched messaging # run with default +options (20 sender and receiver processes per group) +(10 groups == 400 processes run) + + Total time:0.308 sec + +% perf bench sched messaging -t -g 20 # be multi-thread, with 20 groups +(20 sender and receiver threads per group) +(20 groups == 800 threads run) + + Total time:0.582 sec +--------------------- + +*pipe*:: +Suite for pipe() system call. +Based on pipe-test-1m.c by Ingo Molnar. + +Options of *pipe* +^^^^^^^^^^^^^^^^^ +-l:: +--loop=:: +Specify number of loops. + +Example of *pipe* +^^^^^^^^^^^^^^^^^ + +--------------------- +% perf bench sched pipe +(executing 1000000 pipe operations between two tasks) + + Total time:8.091 sec + 8.091833 usecs/op + 123581 ops/sec + +% perf bench sched pipe -l 1000 # loop 1000 +(executing 1000 pipe operations between two tasks) + + Total time:0.016 sec + 16.948000 usecs/op + 59004 ops/sec +--------------------- + +SEE ALSO +-------- +linkperf:perf[1] diff --git a/tools/perf/Documentation/perf-buildid-cache.txt b/tools/perf/Documentation/perf-buildid-cache.txt new file mode 100644 index 00000000..c1057701 --- /dev/null +++ b/tools/perf/Documentation/perf-buildid-cache.txt @@ -0,0 +1,33 @@ +perf-buildid-cache(1) +===================== + +NAME +---- +perf-buildid-cache - Manage build-id cache. + +SYNOPSIS +-------- +[verse] +'perf buildid-cache <options>' + +DESCRIPTION +----------- +This command manages the build-id cache. It can add and remove files to/from +the cache. In the future it should as well purge older entries, set upper +limits for the space used by the cache, etc. + +OPTIONS +------- +-a:: +--add=:: + Add specified file to the cache. +-r:: +--remove=:: + Remove specified file from the cache. +-v:: +--verbose:: + Be more verbose. + +SEE ALSO +-------- +linkperf:perf-record[1], linkperf:perf-report[1], linkperf:perf-buildid-list[1] diff --git a/tools/perf/Documentation/perf-buildid-list.txt b/tools/perf/Documentation/perf-buildid-list.txt new file mode 100644 index 00000000..5eaac6f2 --- /dev/null +++ b/tools/perf/Documentation/perf-buildid-list.txt @@ -0,0 +1,37 @@ +perf-buildid-list(1) +==================== + +NAME +---- +perf-buildid-list - List the buildids in a perf.data file + +SYNOPSIS +-------- +[verse] +'perf buildid-list <options>' + +DESCRIPTION +----------- +This command displays the buildids found in a perf.data file, so that other +tools can be used to fetch packages with matching symbol tables for use by +perf report. + +OPTIONS +------- +-H:: +--with-hits:: + Show only DSOs with hits. +-i:: +--input=:: + Input file name. (default: perf.data) +-f:: +--force:: + Don't do ownership validation. +-v:: +--verbose:: + Be more verbose. + +SEE ALSO +-------- +linkperf:perf-record[1], linkperf:perf-top[1], +linkperf:perf-report[1] diff --git a/tools/perf/Documentation/perf-diff.txt b/tools/perf/Documentation/perf-diff.txt new file mode 100644 index 00000000..74d7481e --- /dev/null +++ b/tools/perf/Documentation/perf-diff.txt @@ -0,0 +1,74 @@ +perf-diff(1) +============ + +NAME +---- +perf-diff - Read two perf.data files and display the differential profile + +SYNOPSIS +-------- +[verse] +'perf diff' [oldfile] [newfile] + +DESCRIPTION +----------- +This command displays the performance difference amongst two perf.data files +captured via perf record. + +If no parameters are passed it will assume perf.data.old and perf.data. + +OPTIONS +------- +-M:: +--displacement:: + Show position displacement relative to baseline. + +-D:: +--dump-raw-trace:: + Dump raw trace in ASCII. + +-m:: +--modules:: + Load module symbols. WARNING: use only with -k and LIVE kernel + +-d:: +--dsos=:: + Only consider symbols in these dsos. CSV that understands + file://filename entries. + +-C:: +--comms=:: + Only consider symbols in these comms. CSV that understands + file://filename entries. + +-S:: +--symbols=:: + Only consider these symbols. CSV that understands + file://filename entries. + +-s:: +--sort=:: + Sort by key(s): pid, comm, dso, symbol. + +-t:: +--field-separator=:: + + Use a special separator character and don't pad with spaces, replacing + all occurrences of this separator in symbol names (and other output) + with a '.' character, that thus it's the only non valid separator. + +-v:: +--verbose:: + Be verbose, for instance, show the raw counts in addition to the + diff. + +-f:: +--force:: + Don't complain, do it. + +--symfs=<directory>:: + Look for files with symbols relative to this directory. + +SEE ALSO +-------- +linkperf:perf-record[1] diff --git a/tools/perf/Documentation/perf-evlist.txt b/tools/perf/Documentation/perf-evlist.txt new file mode 100644 index 00000000..0cada9e0 --- /dev/null +++ b/tools/perf/Documentation/perf-evlist.txt @@ -0,0 +1,26 @@ +perf-evlist(1) +============== + +NAME +---- +perf-evlist - List the event names in a perf.data file + +SYNOPSIS +-------- +[verse] +'perf evlist <options>' + +DESCRIPTION +----------- +This command displays the names of events sampled in a perf.data file. + +OPTIONS +------- +-i:: +--input=:: + Input file name. (default: perf.data) + +SEE ALSO +-------- +linkperf:perf-record[1], linkperf:perf-list[1], +linkperf:perf-report[1] diff --git a/tools/perf/Documentation/perf-help.txt b/tools/perf/Documentation/perf-help.txt new file mode 100644 index 00000000..51439181 --- /dev/null +++ b/tools/perf/Documentation/perf-help.txt @@ -0,0 +1,38 @@ +perf-help(1) +============ + +NAME +---- +perf-help - display help information about perf + +SYNOPSIS +-------- +'perf help' [-a|--all] [COMMAND] + +DESCRIPTION +----------- + +With no options and no COMMAND given, the synopsis of the 'perf' +command and a list of the most commonly used perf commands are printed +on the standard output. + +If the option '--all' or '-a' is given, then all available commands are +printed on the standard output. + +If a perf command is named, a manual page for that command is brought +up. The 'man' program is used by default for this purpose, but this +can be overridden by other options or configuration variables. + +Note that `perf --help ...` is identical to `perf help ...` because the +former is internally converted into the latter. + +OPTIONS +------- +-a:: +--all:: + Prints all the available commands on the standard output. This + option supersedes any other option. + +PERF +---- +Part of the linkperf:perf[1] suite diff --git a/tools/perf/Documentation/perf-inject.txt b/tools/perf/Documentation/perf-inject.txt new file mode 100644 index 00000000..025630d4 --- /dev/null +++ b/tools/perf/Documentation/perf-inject.txt @@ -0,0 +1,35 @@ +perf-inject(1) +============== + +NAME +---- +perf-inject - Filter to augment the events stream with additional information + +SYNOPSIS +-------- +[verse] +'perf inject <options>' + +DESCRIPTION +----------- +perf-inject reads a perf-record event stream and repipes it to stdout. At any +point the processing code can inject other events into the event stream - in +this case build-ids (-b option) are read and injected as needed into the event +stream. + +Build-ids are just the first user of perf-inject - potentially anything that +needs userspace processing to augment the events stream with additional +information could make use of this facility. + +OPTIONS +------- +-b:: +--build-ids=:: + Inject build-ids into the output stream +-v:: +--verbose:: + Be more verbose. + +SEE ALSO +-------- +linkperf:perf-record[1], linkperf:perf-report[1], linkperf:perf-archive[1] diff --git a/tools/perf/Documentation/perf-kmem.txt b/tools/perf/Documentation/perf-kmem.txt new file mode 100644 index 00000000..a52fcde8 --- /dev/null +++ b/tools/perf/Documentation/perf-kmem.txt @@ -0,0 +1,47 @@ +perf-kmem(1) +============ + +NAME +---- +perf-kmem - Tool to trace/measure kernel memory(slab) properties + +SYNOPSIS +-------- +[verse] +'perf kmem' {record|stat} [<options>] + +DESCRIPTION +----------- +There are two variants of perf kmem: + + 'perf kmem record <command>' to record the kmem events + of an arbitrary workload. + + 'perf kmem stat' to report kernel memory statistics. + +OPTIONS +------- +-i <file>:: +--input=<file>:: + Select the input file (default: perf.data) + +--caller:: + Show per-callsite statistics + +--alloc:: + Show per-allocation statistics + +-s <key[,key2...]>:: +--sort=<key[,key2...]>:: + Sort the output (default: frag,hit,bytes) + +-l <num>:: +--line=<num>:: + Print n lines only + +--raw-ip:: + Print raw ip instead of symbol + +SEE ALSO +-------- +linkperf:perf-record[1] diff --git a/tools/perf/Documentation/perf-kvm.txt b/tools/perf/Documentation/perf-kvm.txt new file mode 100644 index 00000000..dd84cb2f --- /dev/null +++ b/tools/perf/Documentation/perf-kvm.txt @@ -0,0 +1,74 @@ +perf-kvm(1) +=========== + +NAME +---- +perf-kvm - Tool to trace/measure kvm guest os + +SYNOPSIS +-------- +[verse] +'perf kvm' [--host] [--guest] [--guestmount=<path> + [--guestkallsyms=<path> --guestmodules=<path> | --guestvmlinux=<path>]] + {top|record|report|diff|buildid-list} +'perf kvm' [--host] [--guest] [--guestkallsyms=<path> --guestmodules=<path> + | --guestvmlinux=<path>] {top|record|report|diff|buildid-list} + +DESCRIPTION +----------- +There are a couple of variants of perf kvm: + + 'perf kvm [options] top <command>' to generates and displays + a performance counter profile of guest os in realtime + of an arbitrary workload. + + 'perf kvm record <command>' to record the performance counter profile + of an arbitrary workload and save it into a perf data file. If both + --host and --guest are input, the perf data file name is perf.data.kvm. + If there is no --host but --guest, the file name is perf.data.guest. + If there is no --guest but --host, the file name is perf.data.host. + + 'perf kvm report' to display the performance counter profile information + recorded via perf kvm record. + + 'perf kvm diff' to displays the performance difference amongst two perf.data + files captured via perf record. + + 'perf kvm buildid-list' to display the buildids found in a perf data file, + so that other tools can be used to fetch packages with matching symbol tables + for use by perf report. + +OPTIONS +------- +-i:: +--input=:: + Input file name. +-o:: +--output:: + Output file name. +--host=:: + Collect host side performance profile. +--guest=:: + Collect guest side performance profile. +--guestmount=<path>:: + Guest os root file system mount directory. Users mounts guest os + root directories under <path> by a specific filesystem access method, + typically, sshfs. For example, start 2 guest os. The one's pid is 8888 + and the other's is 9999. + #mkdir ~/guestmount; cd ~/guestmount + #sshfs -o allow_other,direct_io -p 5551 localhost:/ 8888/ + #sshfs -o allow_other,direct_io -p 5552 localhost:/ 9999/ + #perf kvm --host --guest --guestmount=~/guestmount top +--guestkallsyms=<path>:: + Guest os /proc/kallsyms file copy. 'perf' kvm' reads it to get guest + kernel symbols. Users copy it out from guest os. +--guestmodules=<path>:: + Guest os /proc/modules file copy. 'perf' kvm' reads it to get guest + kernel module information. Users copy it out from guest os. +--guestvmlinux=<path>:: + Guest os kernel vmlinux. + +SEE ALSO +-------- +linkperf:perf-top[1], linkperf:perf-record[1], linkperf:perf-report[1], +linkperf:perf-diff[1], linkperf:perf-buildid-list[1] diff --git a/tools/perf/Documentation/perf-list.txt b/tools/perf/Documentation/perf-list.txt new file mode 100644 index 00000000..7a527f7e --- /dev/null +++ b/tools/perf/Documentation/perf-list.txt @@ -0,0 +1,92 @@ +perf-list(1) +============ + +NAME +---- +perf-list - List all symbolic event types + +SYNOPSIS +-------- +[verse] +'perf list' [hw|sw|cache|tracepoint|event_glob] + +DESCRIPTION +----------- +This command displays the symbolic event types which can be selected in the +various perf commands with the -e option. + +EVENT MODIFIERS +--------------- + +Events can optionally have a modifer by appending a colon and one or +more modifiers. Modifiers allow the user to restrict when events are +counted with 'u' for user-space, 'k' for kernel, 'h' for hypervisor. + +The 'p' modifier can be used for specifying how precise the instruction +address should be. The 'p' modifier is currently only implemented for +Intel PEBS and can be specified multiple times: + 0 - SAMPLE_IP can have arbitrary skid + 1 - SAMPLE_IP must have constant skid + 2 - SAMPLE_IP requested to have 0 skid + 3 - SAMPLE_IP must have 0 skid + +The PEBS implementation now supports up to 2. + +RAW HARDWARE EVENT DESCRIPTOR +----------------------------- +Even when an event is not available in a symbolic form within perf right now, +it can be encoded in a per processor specific way. + +For instance For x86 CPUs NNN represents the raw register encoding with the +layout of IA32_PERFEVTSELx MSRs (see [Intel® 64 and IA-32 Architectures Software Developer's Manual Volume 3B: System Programming Guide] Figure 30-1 Layout +of IA32_PERFEVTSELx MSRs) or AMD's PerfEvtSeln (see [AMD64 Architecture Programmer’s Manual Volume 2: System Programming], Page 344, +Figure 13-7 Performance Event-Select Register (PerfEvtSeln)). + +Example: + +If the Intel docs for a QM720 Core i7 describe an event as: + + Event Umask Event Mask + Num. Value Mnemonic Description Comment + + A8H 01H LSD.UOPS Counts the number of micro-ops Use cmask=1 and + delivered by loop stream detector invert to count + cycles + +raw encoding of 0x1A8 can be used: + + perf stat -e r1a8 -a sleep 1 + perf record -e r1a8 ... + +You should refer to the processor specific documentation for getting these +details. Some of them are referenced in the SEE ALSO section below. + +OPTIONS +------- + +Without options all known events will be listed. + +To limit the list use: + +. 'hw' or 'hardware' to list hardware events such as cache-misses, etc. + +. 'sw' or 'software' to list software events such as context switches, etc. + +. 'cache' or 'hwcache' to list hardware cache events such as L1-dcache-loads, etc. + +. 'tracepoint' to list all tracepoint events, alternatively use + 'subsys_glob:event_glob' to filter by tracepoint subsystems such as sched, + block, etc. + +. If none of the above is matched, it will apply the supplied glob to all + events, printing the ones that match. + +One or more types can be used at the same time, listing the events for the +types specified. + +SEE ALSO +-------- +linkperf:perf-stat[1], linkperf:perf-top[1], +linkperf:perf-record[1], +http://www.intel.com/Assets/PDF/manual/253669.pdf[Intel® 64 and IA-32 Architectures Software Developer's Manual Volume 3B: System Programming Guide], +http://support.amd.com/us/Processor_TechDocs/24593.pdf[AMD64 Architecture Programmer’s Manual Volume 2: System Programming] diff --git a/tools/perf/Documentation/perf-lock.txt b/tools/perf/Documentation/perf-lock.txt new file mode 100644 index 00000000..4a26a2f3 --- /dev/null +++ b/tools/perf/Documentation/perf-lock.txt @@ -0,0 +1,52 @@ +perf-lock(1) +============ + +NAME +---- +perf-lock - Analyze lock events + +SYNOPSIS +-------- +[verse] +'perf lock' {record|report|trace} + +DESCRIPTION +----------- +You can analyze various lock behaviours +and statistics with this 'perf lock' command. + + 'perf lock record <command>' records lock events + between start and end <command>. And this command + produces the file "perf.data" which contains tracing + results of lock events. + + 'perf lock trace' shows raw lock events. + + 'perf lock report' reports statistical data. + +COMMON OPTIONS +-------------- + +-i:: +--input=<file>:: + Input file name. + +-v:: +--verbose:: + Be more verbose (show symbol address, etc). + +-D:: +--dump-raw-trace:: + Dump raw trace in ASCII. + +REPORT OPTIONS +-------------- + +-k:: +--key=<value>:: + Sorting key. Possible values: acquired (default), contended, + wait_total, wait_max, wait_min. + +SEE ALSO +-------- +linkperf:perf[1] diff --git a/tools/perf/Documentation/perf-probe.txt b/tools/perf/Documentation/perf-probe.txt new file mode 100644 index 00000000..02bafce4 --- /dev/null +++ b/tools/perf/Documentation/perf-probe.txt @@ -0,0 +1,186 @@ +perf-probe(1) +============= + +NAME +---- +perf-probe - Define new dynamic tracepoints + +SYNOPSIS +-------- +[verse] +'perf probe' [options] --add='PROBE' [...] +or +'perf probe' [options] PROBE +or +'perf probe' [options] --del='[GROUP:]EVENT' [...] +or +'perf probe' --list +or +'perf probe' [options] --line='LINE' +or +'perf probe' [options] --vars='PROBEPOINT' + +DESCRIPTION +----------- +This command defines dynamic tracepoint events, by symbol and registers +without debuginfo, or by C expressions (C line numbers, C function names, +and C local variables) with debuginfo. + + +OPTIONS +------- +-k:: +--vmlinux=PATH:: + Specify vmlinux path which has debuginfo (Dwarf binary). + +-m:: +--module=MODNAME:: + Specify module name in which perf-probe searches probe points + or lines. + +-s:: +--source=PATH:: + Specify path to kernel source. + +-v:: +--verbose:: + Be more verbose (show parsed arguments, etc). + +-a:: +--add=:: + Define a probe event (see PROBE SYNTAX for detail). + +-d:: +--del=:: + Delete probe events. This accepts glob wildcards('*', '?') and character + classes(e.g. [a-z], [!A-Z]). + +-l:: +--list:: + List up current probe events. + +-L:: +--line=:: + Show source code lines which can be probed. This needs an argument + which specifies a range of the source code. (see LINE SYNTAX for detail) + +-V:: +--vars=:: + Show available local variables at given probe point. The argument + syntax is same as PROBE SYNTAX, but NO ARGs. + +--externs:: + (Only for --vars) Show external defined variables in addition to local + variables. + +-F:: +--funcs:: + Show available functions in given module or kernel. + +--filter=FILTER:: + (Only for --vars and --funcs) Set filter. FILTER is a combination of glob + pattern, see FILTER PATTERN for detail. + Default FILTER is "!__k???tab_* & !__crc_*" for --vars, and "!_*" + for --funcs. + If several filters are specified, only the last filter is used. + +-f:: +--force:: + Forcibly add events with existing name. + +-n:: +--dry-run:: + Dry run. With this option, --add and --del doesn't execute actual + adding and removal operations. + +--max-probes:: + Set the maximum number of probe points for an event. Default is 128. + +PROBE SYNTAX +------------ +Probe points are defined by following syntax. + + 1) Define event based on function name + [EVENT=]FUNC[@SRC][:RLN|+OFFS|%return|;PTN] [ARG ...] + + 2) Define event based on source file with line number + [EVENT=]SRC:ALN [ARG ...] + + 3) Define event based on source file with lazy pattern + [EVENT=]SRC;PTN [ARG ...] + + +'EVENT' specifies the name of new event, if omitted, it will be set the name of the probed function. Currently, event group name is set as 'probe'. +'FUNC' specifies a probed function name, and it may have one of the following options; '+OFFS' is the offset from function entry address in bytes, ':RLN' is the relative-line number from function entry line, and '%return' means that it probes function return. And ';PTN' means lazy matching pattern (see LAZY MATCHING). Note that ';PTN' must be the end of the probe point definition. In addition, '@SRC' specifies a source file which has that function. +It is also possible to specify a probe point by the source line number or lazy matching by using 'SRC:ALN' or 'SRC;PTN' syntax, where 'SRC' is the source file path, ':ALN' is the line number and ';PTN' is the lazy matching pattern. +'ARG' specifies the arguments of this probe point, (see PROBE ARGUMENT). + +PROBE ARGUMENT +-------------- +Each probe argument follows below syntax. + + [NAME=]LOCALVAR|$retval|%REG|@SYMBOL[:TYPE] + +'NAME' specifies the name of this argument (optional). You can use the name of local variable, local data structure member (e.g. var->field, var.field2), local array with fixed index (e.g. array[1], var->array[0], var->pointer[2]), or kprobe-tracer argument format (e.g. $retval, %ax, etc). Note that the name of this argument will be set as the last member name if you specify a local data structure member (e.g. field2 for 'var->field1.field2'.) +'TYPE' casts the type of this argument (optional). If omitted, perf probe automatically set the type based on debuginfo. You can specify 'string' type only for the local variable or structure member which is an array of or a pointer to 'char' or 'unsigned char' type. + +LINE SYNTAX +----------- +Line range is described by following syntax. + + "FUNC[@SRC][:RLN[+NUM|-RLN2]]|SRC[:ALN[+NUM|-ALN2]]" + +FUNC specifies the function name of showing lines. 'RLN' is the start line +number from function entry line, and 'RLN2' is the end line number. As same as +probe syntax, 'SRC' means the source file path, 'ALN' is start line number, +and 'ALN2' is end line number in the file. It is also possible to specify how +many lines to show by using 'NUM'. Moreover, 'FUNC@SRC' combination is good +for searching a specific function when several functions share same name. +So, "source.c:100-120" shows lines between 100th to l20th in source.c file. And "func:10+20" shows 20 lines from 10th line of func function. + +LAZY MATCHING +------------- + The lazy line matching is similar to glob matching but ignoring spaces in both of pattern and target. So this accepts wildcards('*', '?') and character classes(e.g. [a-z], [!A-Z]). + +e.g. + 'a=*' can matches 'a=b', 'a = b', 'a == b' and so on. + +This provides some sort of flexibility and robustness to probe point definitions against minor code changes. For example, actual 10th line of schedule() can be moved easily by modifying schedule(), but the same line matching 'rq=cpu_rq*' may still exist in the function.) + +FILTER PATTERN +-------------- + The filter pattern is a glob matching pattern(s) to filter variables. + In addition, you can use "!" for specifying filter-out rule. You also can give several rules combined with "&" or "|", and fold those rules as one rule by using "(" ")". + +e.g. + With --filter "foo* | bar*", perf probe -V shows variables which start with "foo" or "bar". + With --filter "!foo* & *bar", perf probe -V shows variables which don't start with "foo" and end with "bar", like "fizzbar". But "foobar" is filtered out. + +EXAMPLES +-------- +Display which lines in schedule() can be probed: + + ./perf probe --line schedule + +Add a probe on schedule() function 12th line with recording cpu local variable: + + ./perf probe schedule:12 cpu + or + ./perf probe --add='schedule:12 cpu' + + this will add one or more probes which has the name start with "schedule". + + Add probes on lines in schedule() function which calls update_rq_clock(). + + ./perf probe 'schedule;update_rq_clock*' + or + ./perf probe --add='schedule;update_rq_clock*' + +Delete all probes on schedule(). + + ./perf probe --del='schedule*' + + +SEE ALSO +-------- +linkperf:perf-trace[1], linkperf:perf-record[1] diff --git a/tools/perf/Documentation/perf-record.txt b/tools/perf/Documentation/perf-record.txt new file mode 100644 index 00000000..5a520f82 --- /dev/null +++ b/tools/perf/Documentation/perf-record.txt @@ -0,0 +1,153 @@ +perf-record(1) +============== + +NAME +---- +perf-record - Run a command and record its profile into perf.data + +SYNOPSIS +-------- +[verse] +'perf record' [-e <EVENT> | --event=EVENT] [-l] [-a] <command> +'perf record' [-e <EVENT> | --event=EVENT] [-l] [-a] -- <command> [<options>] + +DESCRIPTION +----------- +This command runs a command and gathers a performance counter profile +from it, into perf.data - without displaying anything. + +This file can then be inspected later on, using 'perf report'. + + +OPTIONS +------- +<command>...:: + Any command you can specify in a shell. + +-e:: +--event=:: + Select the PMU event. Selection can be: + + - a symbolic event name (use 'perf list' to list all events) + + - a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a + hexadecimal event descriptor. + + - a hardware breakpoint event in the form of '\mem:addr[:access]' + where addr is the address in memory you want to break in. + Access is the memory access type (read, write, execute) it can + be passed as follows: '\mem:addr[:[r][w][x]]'. + If you want to profile read-write accesses in 0x1000, just set + 'mem:0x1000:rw'. + +--filter=<filter>:: + Event filter. + +-a:: +--all-cpus:: + System-wide collection from all CPUs. + +-l:: + Scale counter values. + +-p:: +--pid=:: + Record events on existing process ID. + +-t:: +--tid=:: + Record events on existing thread ID. + +-r:: +--realtime=:: + Collect data with this RT SCHED_FIFO priority. +-D:: +--no-delay:: + Collect data without buffering. +-A:: +--append:: + Append to the output file to do incremental profiling. + +-f:: +--force:: + Overwrite existing data file. (deprecated) + +-c:: +--count=:: + Event period to sample. + +-o:: +--output=:: + Output file name. + +-i:: +--no-inherit:: + Child tasks do not inherit counters. +-F:: +--freq=:: + Profile at this frequency. + +-m:: +--mmap-pages=:: + Number of mmap data pages. + +-g:: +--call-graph:: + Do call-graph (stack chain/backtrace) recording. + +-q:: +--quiet:: + Don't print any message, useful for scripting. + +-v:: +--verbose:: + Be more verbose (show counter open errors, etc). + +-s:: +--stat:: + Per thread counts. + +-d:: +--data:: + Sample addresses. + +-T:: +--timestamp:: + Sample timestamps. Use it with 'perf report -D' to see the timestamps, + for instance. + +-n:: +--no-samples:: + Don't sample. + +-R:: +--raw-samples:: +Collect raw sample records from all opened counters (default for tracepoint counters). + +-C:: +--cpu:: +Collect samples only on the list of CPUs provided. Multiple CPUs can be provided as a +comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2. +In per-thread mode with inheritance mode on (default), samples are captured only when +the thread executes on the designated CPUs. Default is to monitor all CPUs. + +-N:: +--no-buildid-cache:: +Do not update the builid cache. This saves some overhead in situations +where the information in the perf.data file (which includes buildids) +is sufficient. + +-G name,...:: +--cgroup name,...:: +monitor only in the container (cgroup) called "name". This option is available only +in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to +container "name" are monitored when they run on the monitored CPUs. Multiple cgroups +can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup +to first event, second cgroup to second event and so on. It is possible to provide +an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have +corresponding events, i.e., they always refer to events defined earlier on the command +line. + +SEE ALSO +-------- +linkperf:perf-stat[1], linkperf:perf-list[1] diff --git a/tools/perf/Documentation/perf-report.txt b/tools/perf/Documentation/perf-report.txt new file mode 100644 index 00000000..8ba03d6e --- /dev/null +++ b/tools/perf/Documentation/perf-report.txt @@ -0,0 +1,124 @@ +perf-report(1) +============== + +NAME +---- +perf-report - Read perf.data (created by perf record) and display the profile + +SYNOPSIS +-------- +[verse] +'perf report' [-i <file> | --input=file] + +DESCRIPTION +----------- +This command displays the performance counter profile information recorded +via perf record. + +OPTIONS +------- +-i:: +--input=:: + Input file name. (default: perf.data) + +-v:: +--verbose:: + Be more verbose. (show symbol address, etc) + +-d:: +--dsos=:: + Only consider symbols in these dsos. CSV that understands + file://filename entries. +-n:: +--show-nr-samples:: + Show the number of samples for each symbol + +--showcpuutilization:: + Show sample percentage for different cpu modes. + +-T:: +--threads:: + Show per-thread event counters +-C:: +--comms=:: + Only consider symbols in these comms. CSV that understands + file://filename entries. +-S:: +--symbols=:: + Only consider these symbols. CSV that understands + file://filename entries. + +-U:: +--hide-unresolved:: + Only display entries resolved to a symbol. + +-s:: +--sort=:: + Sort by key(s): pid, comm, dso, symbol, parent. + +-p:: +--parent=<regex>:: + regex filter to identify parent, see: '--sort parent' + +-x:: +--exclude-other:: + Only display entries with parent-match. + +-w:: +--column-widths=<width[,width...]>:: + Force each column width to the provided list, for large terminal + readability. + +-t:: +--field-separator=:: + + Use a special separator character and don't pad with spaces, replacing + all occurrences of this separator in symbol names (and other output) + with a '.' character, that thus it's the only non valid separator. + +-D:: +--dump-raw-trace:: + Dump raw trace in ASCII. + +-g [type,min]:: +--call-graph:: + Display call chains using type and min percent threshold. + type can be either: + - flat: single column, linear exposure of call chains. + - graph: use a graph tree, displaying absolute overhead rates. + - fractal: like graph, but displays relative rates. Each branch of + the tree is considered as a new profiled object. + + Default: fractal,0.5. + +--pretty=<key>:: + Pretty printing style. key: normal, raw + +--stdio:: Use the stdio interface. + +--tui:: Use the TUI interface, that is integrated with annotate and allows + zooming into DSOs or threads, among other features. Use of --tui + requires a tty, if one is not present, as when piping to other + commands, the stdio interface is used. + +-k:: +--vmlinux=<file>:: + vmlinux pathname + +--kallsyms=<file>:: + kallsyms pathname + +-m:: +--modules:: + Load module symbols. WARNING: This should only be used with -k and + a LIVE kernel. + +-f:: +--force:: + Don't complain, do it. + +--symfs=<directory>:: + Look for files with symbols relative to this directory. + +SEE ALSO +-------- +linkperf:perf-stat[1] diff --git a/tools/perf/Documentation/perf-sched.txt b/tools/perf/Documentation/perf-sched.txt new file mode 100644 index 00000000..46822d5f --- /dev/null +++ b/tools/perf/Documentation/perf-sched.txt @@ -0,0 +1,55 @@ +perf-sched(1) +============== + +NAME +---- +perf-sched - Tool to trace/measure scheduler properties (latencies) + +SYNOPSIS +-------- +[verse] +'perf sched' {record|latency|map|replay|trace} + +DESCRIPTION +----------- +There are five variants of perf sched: + + 'perf sched record <command>' to record the scheduling events + of an arbitrary workload. + + 'perf sched latency' to report the per task scheduling latencies + and other scheduling properties of the workload. + + 'perf sched trace' to see a detailed trace of the workload that + was recorded. + + 'perf sched replay' to simulate the workload that was recorded + via perf sched record. (this is done by starting up mockup threads + that mimic the workload based on the events in the trace. These + threads can then replay the timings (CPU runtime and sleep patterns) + of the workload as it occurred when it was recorded - and can repeat + it a number of times, measuring its performance.) + + 'perf sched map' to print a textual context-switching outline of + workload captured via perf sched record. Columns stand for + individual CPUs, and the two-letter shortcuts stand for tasks that + are running on a CPU. A '*' denotes the CPU that had the event, and + a dot signals an idle CPU. + +OPTIONS +------- +-i:: +--input=<file>:: + Input file name. (default: perf.data) + +-v:: +--verbose:: + Be more verbose. (show symbol address, etc) + +-D:: +--dump-raw-trace=:: + Display verbose dump of the sched data. + +SEE ALSO +-------- +linkperf:perf-record[1] diff --git a/tools/perf/Documentation/perf-script-perl.txt b/tools/perf/Documentation/perf-script-perl.txt new file mode 100644 index 00000000..3152cca1 --- /dev/null +++ b/tools/perf/Documentation/perf-script-perl.txt @@ -0,0 +1,216 @@ +perf-script-perl(1) +================== + +NAME +---- +perf-script-perl - Process trace data with a Perl script + +SYNOPSIS +-------- +[verse] +'perf script' [-s [Perl]:script[.pl] ] + +DESCRIPTION +----------- + +This perf script option is used to process perf script data using perf's +built-in Perl interpreter. It reads and processes the input file and +displays the results of the trace analysis implemented in the given +Perl script, if any. + +STARTER SCRIPTS +--------------- + +You can avoid reading the rest of this document by running 'perf script +-g perl' in the same directory as an existing perf.data trace file. +That will generate a starter script containing a handler for each of +the event types in the trace file; it simply prints every available +field for each event in the trace file. + +You can also look at the existing scripts in +~/libexec/perf-core/scripts/perl for typical examples showing how to +do basic things like aggregate event data, print results, etc. Also, +the check-perf-script.pl script, while not interesting for its results, +attempts to exercise all of the main scripting features. + +EVENT HANDLERS +-------------- + +When perf script is invoked using a trace script, a user-defined +'handler function' is called for each event in the trace. If there's +no handler function defined for a given event type, the event is +ignored (or passed to a 'trace_handled' function, see below) and the +next event is processed. + +Most of the event's field values are passed as arguments to the +handler function; some of the less common ones aren't - those are +available as calls back into the perf executable (see below). + +As an example, the following perf record command can be used to record +all sched_wakeup events in the system: + + # perf record -a -e sched:sched_wakeup + +Traces meant to be processed using a script should be recorded with +the above option: -a to enable system-wide collection. + +The format file for the sched_wakep event defines the following fields +(see /sys/kernel/debug/tracing/events/sched/sched_wakeup/format): + +---- + format: + field:unsigned short common_type; + field:unsigned char common_flags; + field:unsigned char common_preempt_count; + field:int common_pid; + + field:char comm[TASK_COMM_LEN]; + field:pid_t pid; + field:int prio; + field:int success; + field:int target_cpu; +---- + +The handler function for this event would be defined as: + +---- +sub sched::sched_wakeup +{ + my ($event_name, $context, $common_cpu, $common_secs, + $common_nsecs, $common_pid, $common_comm, + $comm, $pid, $prio, $success, $target_cpu) = @_; +} +---- + +The handler function takes the form subsystem::event_name. + +The $common_* arguments in the handler's argument list are the set of +arguments passed to all event handlers; some of the fields correspond +to the common_* fields in the format file, but some are synthesized, +and some of the common_* fields aren't common enough to to be passed +to every event as arguments but are available as library functions. + +Here's a brief description of each of the invariant event args: + + $event_name the name of the event as text + $context an opaque 'cookie' used in calls back into perf + $common_cpu the cpu the event occurred on + $common_secs the secs portion of the event timestamp + $common_nsecs the nsecs portion of the event timestamp + $common_pid the pid of the current task + $common_comm the name of the current process + +All of the remaining fields in the event's format file have +counterparts as handler function arguments of the same name, as can be +seen in the example above. + +The above provides the basics needed to directly access every field of +every event in a trace, which covers 90% of what you need to know to +write a useful trace script. The sections below cover the rest. + +SCRIPT LAYOUT +------------- + +Every perf script Perl script should start by setting up a Perl module +search path and 'use'ing a few support modules (see module +descriptions below): + +---- + use lib "$ENV{'PERF_EXEC_PATH'}/scripts/perl/perf-script-Util/lib"; + use lib "./perf-script-Util/lib"; + use Perf::Trace::Core; + use Perf::Trace::Context; + use Perf::Trace::Util; +---- + +The rest of the script can contain handler functions and support +functions in any order. + +Aside from the event handler functions discussed above, every script +can implement a set of optional functions: + +*trace_begin*, if defined, is called before any event is processed and +gives scripts a chance to do setup tasks: + +---- + sub trace_begin + { + } +---- + +*trace_end*, if defined, is called after all events have been + processed and gives scripts a chance to do end-of-script tasks, such + as display results: + +---- +sub trace_end +{ +} +---- + +*trace_unhandled*, if defined, is called after for any event that + doesn't have a handler explicitly defined for it. The standard set + of common arguments are passed into it: + +---- +sub trace_unhandled +{ + my ($event_name, $context, $common_cpu, $common_secs, + $common_nsecs, $common_pid, $common_comm) = @_; +} +---- + +The remaining sections provide descriptions of each of the available +built-in perf script Perl modules and their associated functions. + +AVAILABLE MODULES AND FUNCTIONS +------------------------------- + +The following sections describe the functions and variables available +via the various Perf::Trace::* Perl modules. To use the functions and +variables from the given module, add the corresponding 'use +Perf::Trace::XXX' line to your perf script script. + +Perf::Trace::Core Module +~~~~~~~~~~~~~~~~~~~~~~~~ + +These functions provide some essential functions to user scripts. + +The *flag_str* and *symbol_str* functions provide human-readable +strings for flag and symbolic fields. These correspond to the strings +and values parsed from the 'print fmt' fields of the event format +files: + + flag_str($event_name, $field_name, $field_value) - returns the string represention corresponding to $field_value for the flag field $field_name of event $event_name + symbol_str($event_name, $field_name, $field_value) - returns the string represention corresponding to $field_value for the symbolic field $field_name of event $event_name + +Perf::Trace::Context Module +~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Some of the 'common' fields in the event format file aren't all that +common, but need to be made accessible to user scripts nonetheless. + +Perf::Trace::Context defines a set of functions that can be used to +access this data in the context of the current event. Each of these +functions expects a $context variable, which is the same as the +$context variable passed into every event handler as the second +argument. + + common_pc($context) - returns common_preempt count for the current event + common_flags($context) - returns common_flags for the current event + common_lock_depth($context) - returns common_lock_depth for the current event + +Perf::Trace::Util Module +~~~~~~~~~~~~~~~~~~~~~~~~ + +Various utility functions for use with perf script: + + nsecs($secs, $nsecs) - returns total nsecs given secs/nsecs pair + nsecs_secs($nsecs) - returns whole secs portion given nsecs + nsecs_nsecs($nsecs) - returns nsecs remainder given nsecs + nsecs_str($nsecs) - returns printable string in the form secs.nsecs + avg($total, $n) - returns average given a sum and a total number of values + +SEE ALSO +-------- +linkperf:perf-script[1] diff --git a/tools/perf/Documentation/perf-script-python.txt b/tools/perf/Documentation/perf-script-python.txt new file mode 100644 index 00000000..47102206 --- /dev/null +++ b/tools/perf/Documentation/perf-script-python.txt @@ -0,0 +1,622 @@ +perf-script-python(1) +==================== + +NAME +---- +perf-script-python - Process trace data with a Python script + +SYNOPSIS +-------- +[verse] +'perf script' [-s [Python]:script[.py] ] + +DESCRIPTION +----------- + +This perf script option is used to process perf script data using perf's +built-in Python interpreter. It reads and processes the input file and +displays the results of the trace analysis implemented in the given +Python script, if any. + +A QUICK EXAMPLE +--------------- + +This section shows the process, start to finish, of creating a working +Python script that aggregates and extracts useful information from a +raw perf script stream. You can avoid reading the rest of this +document if an example is enough for you; the rest of the document +provides more details on each step and lists the library functions +available to script writers. + +This example actually details the steps that were used to create the +'syscall-counts' script you see when you list the available perf script +scripts via 'perf script -l'. As such, this script also shows how to +integrate your script into the list of general-purpose 'perf script' +scripts listed by that command. + +The syscall-counts script is a simple script, but demonstrates all the +basic ideas necessary to create a useful script. Here's an example +of its output (syscall names are not yet supported, they will appear +as numbers): + +---- +syscall events: + +event count +---------------------------------------- ----------- +sys_write 455067 +sys_getdents 4072 +sys_close 3037 +sys_swapoff 1769 +sys_read 923 +sys_sched_setparam 826 +sys_open 331 +sys_newfstat 326 +sys_mmap 217 +sys_munmap 216 +sys_futex 141 +sys_select 102 +sys_poll 84 +sys_setitimer 12 +sys_writev 8 +15 8 +sys_lseek 7 +sys_rt_sigprocmask 6 +sys_wait4 3 +sys_ioctl 3 +sys_set_robust_list 1 +sys_exit 1 +56 1 +sys_access 1 +---- + +Basically our task is to keep a per-syscall tally that gets updated +every time a system call occurs in the system. Our script will do +that, but first we need to record the data that will be processed by +that script. Theoretically, there are a couple of ways we could do +that: + +- we could enable every event under the tracing/events/syscalls + directory, but this is over 600 syscalls, well beyond the number + allowable by perf. These individual syscall events will however be + useful if we want to later use the guidance we get from the + general-purpose scripts to drill down and get more detail about + individual syscalls of interest. + +- we can enable the sys_enter and/or sys_exit syscalls found under + tracing/events/raw_syscalls. These are called for all syscalls; the + 'id' field can be used to distinguish between individual syscall + numbers. + +For this script, we only need to know that a syscall was entered; we +don't care how it exited, so we'll use 'perf record' to record only +the sys_enter events: + +---- +# perf record -a -e raw_syscalls:sys_enter + +^C[ perf record: Woken up 1 times to write data ] +[ perf record: Captured and wrote 56.545 MB perf.data (~2470503 samples) ] +---- + +The options basically say to collect data for every syscall event +system-wide and multiplex the per-cpu output into a single stream. +That single stream will be recorded in a file in the current directory +called perf.data. + +Once we have a perf.data file containing our data, we can use the -g +'perf script' option to generate a Python script that will contain a +callback handler for each event type found in the perf.data trace +stream (for more details, see the STARTER SCRIPTS section). + +---- +# perf script -g python +generated Python script: perf-script.py + +The output file created also in the current directory is named +perf-script.py. Here's the file in its entirety: + +# perf script event handlers, generated by perf script -g python +# Licensed under the terms of the GNU GPL License version 2 + +# The common_* event handler fields are the most useful fields common to +# all events. They don't necessarily correspond to the 'common_*' fields +# in the format files. Those fields not available as handler params can +# be retrieved using Python functions of the form common_*(context). +# See the perf-script-python Documentation for the list of available functions. + +import os +import sys + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/perf-script-Util/lib/Perf/Trace') + +from perf_trace_context import * +from Core import * + +def trace_begin(): + print "in trace_begin" + +def trace_end(): + print "in trace_end" + +def raw_syscalls__sys_enter(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + id, args): + print_header(event_name, common_cpu, common_secs, common_nsecs, + common_pid, common_comm) + + print "id=%d, args=%s\n" % \ + (id, args), + +def trace_unhandled(event_name, context, common_cpu, common_secs, common_nsecs, + common_pid, common_comm): + print_header(event_name, common_cpu, common_secs, common_nsecs, + common_pid, common_comm) + +def print_header(event_name, cpu, secs, nsecs, pid, comm): + print "%-20s %5u %05u.%09u %8u %-20s " % \ + (event_name, cpu, secs, nsecs, pid, comm), +---- + +At the top is a comment block followed by some import statements and a +path append which every perf script script should include. + +Following that are a couple generated functions, trace_begin() and +trace_end(), which are called at the beginning and the end of the +script respectively (for more details, see the SCRIPT_LAYOUT section +below). + +Following those are the 'event handler' functions generated one for +every event in the 'perf record' output. The handler functions take +the form subsystem__event_name, and contain named parameters, one for +each field in the event; in this case, there's only one event, +raw_syscalls__sys_enter(). (see the EVENT HANDLERS section below for +more info on event handlers). + +The final couple of functions are, like the begin and end functions, +generated for every script. The first, trace_unhandled(), is called +every time the script finds an event in the perf.data file that +doesn't correspond to any event handler in the script. This could +mean either that the record step recorded event types that it wasn't +really interested in, or the script was run against a trace file that +doesn't correspond to the script. + +The script generated by -g option simply prints a line for each +event found in the trace stream i.e. it basically just dumps the event +and its parameter values to stdout. The print_header() function is +simply a utility function used for that purpose. Let's rename the +script and run it to see the default output: + +---- +# mv perf-script.py syscall-counts.py +# perf script -s syscall-counts.py + +raw_syscalls__sys_enter 1 00840.847582083 7506 perf id=1, args= +raw_syscalls__sys_enter 1 00840.847595764 7506 perf id=1, args= +raw_syscalls__sys_enter 1 00840.847620860 7506 perf id=1, args= +raw_syscalls__sys_enter 1 00840.847710478 6533 npviewer.bin id=78, args= +raw_syscalls__sys_enter 1 00840.847719204 6533 npviewer.bin id=142, args= +raw_syscalls__sys_enter 1 00840.847755445 6533 npviewer.bin id=3, args= +raw_syscalls__sys_enter 1 00840.847775601 6533 npviewer.bin id=3, args= +raw_syscalls__sys_enter 1 00840.847781820 6533 npviewer.bin id=3, args= +. +. +. +---- + +Of course, for this script, we're not interested in printing every +trace event, but rather aggregating it in a useful way. So we'll get +rid of everything to do with printing as well as the trace_begin() and +trace_unhandled() functions, which we won't be using. That leaves us +with this minimalistic skeleton: + +---- +import os +import sys + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/perf-script-Util/lib/Perf/Trace') + +from perf_trace_context import * +from Core import * + +def trace_end(): + print "in trace_end" + +def raw_syscalls__sys_enter(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + id, args): +---- + +In trace_end(), we'll simply print the results, but first we need to +generate some results to print. To do that we need to have our +sys_enter() handler do the necessary tallying until all events have +been counted. A hash table indexed by syscall id is a good way to +store that information; every time the sys_enter() handler is called, +we simply increment a count associated with that hash entry indexed by +that syscall id: + +---- + syscalls = autodict() + + try: + syscalls[id] += 1 + except TypeError: + syscalls[id] = 1 +---- + +The syscalls 'autodict' object is a special kind of Python dictionary +(implemented in Core.py) that implements Perl's 'autovivifying' hashes +in Python i.e. with autovivifying hashes, you can assign nested hash +values without having to go to the trouble of creating intermediate +levels if they don't exist e.g syscalls[comm][pid][id] = 1 will create +the intermediate hash levels and finally assign the value 1 to the +hash entry for 'id' (because the value being assigned isn't a hash +object itself, the initial value is assigned in the TypeError +exception. Well, there may be a better way to do this in Python but +that's what works for now). + +Putting that code into the raw_syscalls__sys_enter() handler, we +effectively end up with a single-level dictionary keyed on syscall id +and having the counts we've tallied as values. + +The print_syscall_totals() function iterates over the entries in the +dictionary and displays a line for each entry containing the syscall +name (the dictonary keys contain the syscall ids, which are passed to +the Util function syscall_name(), which translates the raw syscall +numbers to the corresponding syscall name strings). The output is +displayed after all the events in the trace have been processed, by +calling the print_syscall_totals() function from the trace_end() +handler called at the end of script processing. + +The final script producing the output shown above is shown in its +entirety below (syscall_name() helper is not yet available, you can +only deal with id's for now): + +---- +import os +import sys + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/perf-script-Util/lib/Perf/Trace') + +from perf_trace_context import * +from Core import * +from Util import * + +syscalls = autodict() + +def trace_end(): + print_syscall_totals() + +def raw_syscalls__sys_enter(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + id, args): + try: + syscalls[id] += 1 + except TypeError: + syscalls[id] = 1 + +def print_syscall_totals(): + if for_comm is not None: + print "\nsyscall events for %s:\n\n" % (for_comm), + else: + print "\nsyscall events:\n\n", + + print "%-40s %10s\n" % ("event", "count"), + print "%-40s %10s\n" % ("----------------------------------------", \ + "-----------"), + + for id, val in sorted(syscalls.iteritems(), key = lambda(k, v): (v, k), \ + reverse = True): + print "%-40s %10d\n" % (syscall_name(id), val), +---- + +The script can be run just as before: + + # perf script -s syscall-counts.py + +So those are the essential steps in writing and running a script. The +process can be generalized to any tracepoint or set of tracepoints +you're interested in - basically find the tracepoint(s) you're +interested in by looking at the list of available events shown by +'perf list' and/or look in /sys/kernel/debug/tracing events for +detailed event and field info, record the corresponding trace data +using 'perf record', passing it the list of interesting events, +generate a skeleton script using 'perf script -g python' and modify the +code to aggregate and display it for your particular needs. + +After you've done that you may end up with a general-purpose script +that you want to keep around and have available for future use. By +writing a couple of very simple shell scripts and putting them in the +right place, you can have your script listed alongside the other +scripts listed by the 'perf script -l' command e.g.: + +---- +root@tropicana:~# perf script -l +List of available trace scripts: + workqueue-stats workqueue stats (ins/exe/create/destroy) + wakeup-latency system-wide min/max/avg wakeup latency + rw-by-file <comm> r/w activity for a program, by file + rw-by-pid system-wide r/w activity +---- + +A nice side effect of doing this is that you also then capture the +probably lengthy 'perf record' command needed to record the events for +the script. + +To have the script appear as a 'built-in' script, you write two simple +scripts, one for recording and one for 'reporting'. + +The 'record' script is a shell script with the same base name as your +script, but with -record appended. The shell script should be put +into the perf/scripts/python/bin directory in the kernel source tree. +In that script, you write the 'perf record' command-line needed for +your script: + +---- +# cat kernel-source/tools/perf/scripts/python/bin/syscall-counts-record + +#!/bin/bash +perf record -a -e raw_syscalls:sys_enter +---- + +The 'report' script is also a shell script with the same base name as +your script, but with -report appended. It should also be located in +the perf/scripts/python/bin directory. In that script, you write the +'perf script -s' command-line needed for running your script: + +---- +# cat kernel-source/tools/perf/scripts/python/bin/syscall-counts-report + +#!/bin/bash +# description: system-wide syscall counts +perf script -s ~/libexec/perf-core/scripts/python/syscall-counts.py +---- + +Note that the location of the Python script given in the shell script +is in the libexec/perf-core/scripts/python directory - this is where +the script will be copied by 'make install' when you install perf. +For the installation to install your script there, your script needs +to be located in the perf/scripts/python directory in the kernel +source tree: + +---- +# ls -al kernel-source/tools/perf/scripts/python + +root@tropicana:/home/trz/src/tip# ls -al tools/perf/scripts/python +total 32 +drwxr-xr-x 4 trz trz 4096 2010-01-26 22:30 . +drwxr-xr-x 4 trz trz 4096 2010-01-26 22:29 .. +drwxr-xr-x 2 trz trz 4096 2010-01-26 22:29 bin +-rw-r--r-- 1 trz trz 2548 2010-01-26 22:29 check-perf-script.py +drwxr-xr-x 3 trz trz 4096 2010-01-26 22:49 perf-script-Util +-rw-r--r-- 1 trz trz 1462 2010-01-26 22:30 syscall-counts.py +---- + +Once you've done that (don't forget to do a new 'make install', +otherwise your script won't show up at run-time), 'perf script -l' +should show a new entry for your script: + +---- +root@tropicana:~# perf script -l +List of available trace scripts: + workqueue-stats workqueue stats (ins/exe/create/destroy) + wakeup-latency system-wide min/max/avg wakeup latency + rw-by-file <comm> r/w activity for a program, by file + rw-by-pid system-wide r/w activity + syscall-counts system-wide syscall counts +---- + +You can now perform the record step via 'perf script record': + + # perf script record syscall-counts + +and display the output using 'perf script report': + + # perf script report syscall-counts + +STARTER SCRIPTS +--------------- + +You can quickly get started writing a script for a particular set of +trace data by generating a skeleton script using 'perf script -g +python' in the same directory as an existing perf.data trace file. +That will generate a starter script containing a handler for each of +the event types in the trace file; it simply prints every available +field for each event in the trace file. + +You can also look at the existing scripts in +~/libexec/perf-core/scripts/python for typical examples showing how to +do basic things like aggregate event data, print results, etc. Also, +the check-perf-script.py script, while not interesting for its results, +attempts to exercise all of the main scripting features. + +EVENT HANDLERS +-------------- + +When perf script is invoked using a trace script, a user-defined +'handler function' is called for each event in the trace. If there's +no handler function defined for a given event type, the event is +ignored (or passed to a 'trace_handled' function, see below) and the +next event is processed. + +Most of the event's field values are passed as arguments to the +handler function; some of the less common ones aren't - those are +available as calls back into the perf executable (see below). + +As an example, the following perf record command can be used to record +all sched_wakeup events in the system: + + # perf record -a -e sched:sched_wakeup + +Traces meant to be processed using a script should be recorded with +the above option: -a to enable system-wide collection. + +The format file for the sched_wakep event defines the following fields +(see /sys/kernel/debug/tracing/events/sched/sched_wakeup/format): + +---- + format: + field:unsigned short common_type; + field:unsigned char common_flags; + field:unsigned char common_preempt_count; + field:int common_pid; + + field:char comm[TASK_COMM_LEN]; + field:pid_t pid; + field:int prio; + field:int success; + field:int target_cpu; +---- + +The handler function for this event would be defined as: + +---- +def sched__sched_wakeup(event_name, context, common_cpu, common_secs, + common_nsecs, common_pid, common_comm, + comm, pid, prio, success, target_cpu): + pass +---- + +The handler function takes the form subsystem__event_name. + +The common_* arguments in the handler's argument list are the set of +arguments passed to all event handlers; some of the fields correspond +to the common_* fields in the format file, but some are synthesized, +and some of the common_* fields aren't common enough to to be passed +to every event as arguments but are available as library functions. + +Here's a brief description of each of the invariant event args: + + event_name the name of the event as text + context an opaque 'cookie' used in calls back into perf + common_cpu the cpu the event occurred on + common_secs the secs portion of the event timestamp + common_nsecs the nsecs portion of the event timestamp + common_pid the pid of the current task + common_comm the name of the current process + +All of the remaining fields in the event's format file have +counterparts as handler function arguments of the same name, as can be +seen in the example above. + +The above provides the basics needed to directly access every field of +every event in a trace, which covers 90% of what you need to know to +write a useful trace script. The sections below cover the rest. + +SCRIPT LAYOUT +------------- + +Every perf script Python script should start by setting up a Python +module search path and 'import'ing a few support modules (see module +descriptions below): + +---- + import os + import sys + + sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/perf-script-Util/lib/Perf/Trace') + + from perf_trace_context import * + from Core import * +---- + +The rest of the script can contain handler functions and support +functions in any order. + +Aside from the event handler functions discussed above, every script +can implement a set of optional functions: + +*trace_begin*, if defined, is called before any event is processed and +gives scripts a chance to do setup tasks: + +---- +def trace_begin: + pass +---- + +*trace_end*, if defined, is called after all events have been + processed and gives scripts a chance to do end-of-script tasks, such + as display results: + +---- +def trace_end: + pass +---- + +*trace_unhandled*, if defined, is called after for any event that + doesn't have a handler explicitly defined for it. The standard set + of common arguments are passed into it: + +---- +def trace_unhandled(event_name, context, common_cpu, common_secs, + common_nsecs, common_pid, common_comm): + pass +---- + +The remaining sections provide descriptions of each of the available +built-in perf script Python modules and their associated functions. + +AVAILABLE MODULES AND FUNCTIONS +------------------------------- + +The following sections describe the functions and variables available +via the various perf script Python modules. To use the functions and +variables from the given module, add the corresponding 'from XXXX +import' line to your perf script script. + +Core.py Module +~~~~~~~~~~~~~~ + +These functions provide some essential functions to user scripts. + +The *flag_str* and *symbol_str* functions provide human-readable +strings for flag and symbolic fields. These correspond to the strings +and values parsed from the 'print fmt' fields of the event format +files: + + flag_str(event_name, field_name, field_value) - returns the string represention corresponding to field_value for the flag field field_name of event event_name + symbol_str(event_name, field_name, field_value) - returns the string represention corresponding to field_value for the symbolic field field_name of event event_name + +The *autodict* function returns a special kind of Python +dictionary that implements Perl's 'autovivifying' hashes in Python +i.e. with autovivifying hashes, you can assign nested hash values +without having to go to the trouble of creating intermediate levels if +they don't exist. + + autodict() - returns an autovivifying dictionary instance + + +perf_trace_context Module +~~~~~~~~~~~~~~~~~~~~~~~~~ + +Some of the 'common' fields in the event format file aren't all that +common, but need to be made accessible to user scripts nonetheless. + +perf_trace_context defines a set of functions that can be used to +access this data in the context of the current event. Each of these +functions expects a context variable, which is the same as the +context variable passed into every event handler as the second +argument. + + common_pc(context) - returns common_preempt count for the current event + common_flags(context) - returns common_flags for the current event + common_lock_depth(context) - returns common_lock_depth for the current event + +Util.py Module +~~~~~~~~~~~~~~ + +Various utility functions for use with perf script: + + nsecs(secs, nsecs) - returns total nsecs given secs/nsecs pair + nsecs_secs(nsecs) - returns whole secs portion given nsecs + nsecs_nsecs(nsecs) - returns nsecs remainder given nsecs + nsecs_str(nsecs) - returns printable string in the form secs.nsecs + avg(total, n) - returns average given a sum and a total number of values + +SEE ALSO +-------- +linkperf:perf-script[1] diff --git a/tools/perf/Documentation/perf-script.txt b/tools/perf/Documentation/perf-script.txt new file mode 100644 index 00000000..86c87e21 --- /dev/null +++ b/tools/perf/Documentation/perf-script.txt @@ -0,0 +1,188 @@ +perf-script(1) +============= + +NAME +---- +perf-script - Read perf.data (created by perf record) and display trace output + +SYNOPSIS +-------- +[verse] +'perf script' [<options>] +'perf script' [<options>] record <script> [<record-options>] <command> +'perf script' [<options>] report <script> [script-args] +'perf script' [<options>] <script> <required-script-args> [<record-options>] <command> +'perf script' [<options>] <top-script> [script-args] + +DESCRIPTION +----------- +This command reads the input file and displays the trace recorded. + +There are several variants of perf script: + + 'perf script' to see a detailed trace of the workload that was + recorded. + + You can also run a set of pre-canned scripts that aggregate and + summarize the raw trace data in various ways (the list of scripts is + available via 'perf script -l'). The following variants allow you to + record and run those scripts: + + 'perf script record <script> <command>' to record the events required + for 'perf script report'. <script> is the name displayed in the + output of 'perf script --list' i.e. the actual script name minus any + language extension. If <command> is not specified, the events are + recorded using the -a (system-wide) 'perf record' option. + + 'perf script report <script> [args]' to run and display the results + of <script>. <script> is the name displayed in the output of 'perf + trace --list' i.e. the actual script name minus any language + extension. The perf.data output from a previous run of 'perf script + record <script>' is used and should be present for this command to + succeed. [args] refers to the (mainly optional) args expected by + the script. + + 'perf script <script> <required-script-args> <command>' to both + record the events required for <script> and to run the <script> + using 'live-mode' i.e. without writing anything to disk. <script> + is the name displayed in the output of 'perf script --list' i.e. the + actual script name minus any language extension. If <command> is + not specified, the events are recorded using the -a (system-wide) + 'perf record' option. If <script> has any required args, they + should be specified before <command>. This mode doesn't allow for + optional script args to be specified; if optional script args are + desired, they can be specified using separate 'perf script record' + and 'perf script report' commands, with the stdout of the record step + piped to the stdin of the report script, using the '-o -' and '-i -' + options of the corresponding commands. + + 'perf script <top-script>' to both record the events required for + <top-script> and to run the <top-script> using 'live-mode' + i.e. without writing anything to disk. <top-script> is the name + displayed in the output of 'perf script --list' i.e. the actual + script name minus any language extension; a <top-script> is defined + as any script name ending with the string 'top'. + + [<record-options>] can be passed to the record steps of 'perf script + record' and 'live-mode' variants; this isn't possible however for + <top-script> 'live-mode' or 'perf script report' variants. + + See the 'SEE ALSO' section for links to language-specific + information on how to write and run your own trace scripts. + +OPTIONS +------- +<command>...:: + Any command you can specify in a shell. + +-D:: +--dump-raw-script=:: + Display verbose dump of the trace data. + +-L:: +--Latency=:: + Show latency attributes (irqs/preemption disabled, etc). + +-l:: +--list=:: + Display a list of available trace scripts. + +-s ['lang']:: +--script=:: + Process trace data with the given script ([lang]:script[.ext]). + If the string 'lang' is specified in place of a script name, a + list of supported languages will be displayed instead. + +-g:: +--gen-script=:: + Generate perf-script.[ext] starter script for given language, + using current perf.data. + +-a:: + Force system-wide collection. Scripts run without a <command> + normally use -a by default, while scripts run with a <command> + normally don't - this option allows the latter to be run in + system-wide mode. + +-i:: +--input=:: + Input file name. + +-d:: +--debug-mode:: + Do various checks like samples ordering and lost events. + +-f:: +--fields:: + Comma separated list of fields to print. Options are: + comm, tid, pid, time, cpu, event, trace, sym. Field + list can be prepended with the type, trace, sw or hw, + to indicate to which event type the field list applies. + e.g., -f sw:comm,tid,time,sym and -f trace:time,cpu,trace + + perf script -f <fields> + + is equivalent to: + + perf script -f trace:<fields> -f sw:<fields> -f hw:<fields> + + i.e., the specified fields apply to all event types if the type string + is not given. + + The arguments are processed in the order received. A later usage can + reset a prior request. e.g.: + + -f trace: -f comm,tid,time,sym + + The first -f suppresses trace events (field list is ""), but then the + second invocation sets the fields to comm,tid,time,sym. In this case a + warning is given to the user: + + "Overriding previous field request for all events." + + Alternativey, consider the order: + + -f comm,tid,time,sym -f trace: + + The first -f sets the fields for all events and the second -f + suppresses trace events. The user is given a warning message about + the override, and the result of the above is that only S/W and H/W + events are displayed with the given fields. + + For the 'wildcard' option if a user selected field is invalid for an + event type, a message is displayed to the user that the option is + ignored for that type. For example: + + $ perf script -f comm,tid,trace + 'trace' not valid for hardware events. Ignoring. + 'trace' not valid for software events. Ignoring. + + Alternatively, if the type is given an invalid field is specified it + is an error. For example: + + perf script -v -f sw:comm,tid,trace + 'trace' not valid for software events. + + At this point usage is displayed, and perf-script exits. + + Finally, a user may not set fields to none for all event types. + i.e., -f "" is not allowed. + +-k:: +--vmlinux=<file>:: + vmlinux pathname + +--kallsyms=<file>:: + kallsyms pathname + +--symfs=<directory>:: + Look for files with symbols relative to this directory. + +-G:: +--hide-call-graph:: + When printing symbols do not display call chain. + +SEE ALSO +-------- +linkperf:perf-record[1], linkperf:perf-script-perl[1], +linkperf:perf-script-python[1] diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt new file mode 100644 index 00000000..918cc38e --- /dev/null +++ b/tools/perf/Documentation/perf-stat.txt @@ -0,0 +1,117 @@ +perf-stat(1) +============ + +NAME +---- +perf-stat - Run a command and gather performance counter statistics + +SYNOPSIS +-------- +[verse] +'perf stat' [-e <EVENT> | --event=EVENT] [-a] <command> +'perf stat' [-e <EVENT> | --event=EVENT] [-a] -- <command> [<options>] + +DESCRIPTION +----------- +This command runs a command and gathers performance counter statistics +from it. + + +OPTIONS +------- +<command>...:: + Any command you can specify in a shell. + + +-e:: +--event=:: + Select the PMU event. Selection can be a symbolic event name + (use 'perf list' to list all events) or a raw PMU + event (eventsel+umask) in the form of rNNN where NNN is a + hexadecimal event descriptor. + +-i:: +--no-inherit:: + child tasks do not inherit counters +-p:: +--pid=<pid>:: + stat events on existing process id + +-t:: +--tid=<tid>:: + stat events on existing thread id + + +-a:: +--all-cpus:: + system-wide collection from all CPUs + +-c:: +--scale:: + scale/normalize counter values + +-r:: +--repeat=<n>:: + repeat command and print average + stddev (max: 100) + +-B:: +--big-num:: + print large numbers with thousands' separators according to locale + +-C:: +--cpu=:: +Count only on the list of CPUs provided. Multiple CPUs can be provided as a +comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2. +In per-thread mode, this option is ignored. The -a option is still necessary +to activate system-wide monitoring. Default is to count on all CPUs. + +-A:: +--no-aggr:: +Do not aggregate counts across all monitored CPUs in system-wide mode (-a). +This option is only valid in system-wide mode. + +-n:: +--null:: + null run - don't start any counters + +-v:: +--verbose:: + be more verbose (show counter open errors, etc) + +-x SEP:: +--field-separator SEP:: +print counts using a CSV-style output to make it easy to import directly into +spreadsheets. Columns are separated by the string specified in SEP. + +-G name:: +--cgroup name:: +monitor only in the container (cgroup) called "name". This option is available only +in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to +container "name" are monitored when they run on the monitored CPUs. Multiple cgroups +can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup +to first event, second cgroup to second event and so on. It is possible to provide +an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have +corresponding events, i.e., they always refer to events defined earlier on the command +line. + +EXAMPLES +-------- + +$ perf stat -- make -j + + Performance counter stats for 'make -j': + + 8117.370256 task clock ticks # 11.281 CPU utilization factor + 678 context switches # 0.000 M/sec + 133 CPU migrations # 0.000 M/sec + 235724 pagefaults # 0.029 M/sec + 24821162526 CPU cycles # 3057.784 M/sec + 18687303457 instructions # 2302.138 M/sec + 172158895 cache references # 21.209 M/sec + 27075259 cache misses # 3.335 M/sec + + Wall-clock time elapsed: 719.554352 msecs + +SEE ALSO +-------- +linkperf:perf-top[1], linkperf:perf-list[1] diff --git a/tools/perf/Documentation/perf-test.txt b/tools/perf/Documentation/perf-test.txt new file mode 100644 index 00000000..2c3b462f --- /dev/null +++ b/tools/perf/Documentation/perf-test.txt @@ -0,0 +1,22 @@ +perf-test(1) +============ + +NAME +---- +perf-test - Runs sanity tests. + +SYNOPSIS +-------- +[verse] +'perf test <options>' + +DESCRIPTION +----------- +This command does assorted sanity tests, initially through linked routines but +also will look for a directory with more tests in the form of scripts. + +OPTIONS +------- +-v:: +--verbose:: + Be more verbose. diff --git a/tools/perf/Documentation/perf-timechart.txt b/tools/perf/Documentation/perf-timechart.txt new file mode 100644 index 00000000..d7b79e2b --- /dev/null +++ b/tools/perf/Documentation/perf-timechart.txt @@ -0,0 +1,46 @@ +perf-timechart(1) +================= + +NAME +---- +perf-timechart - Tool to visualize total system behavior during a workload + +SYNOPSIS +-------- +[verse] +'perf timechart' {record} + +DESCRIPTION +----------- +There are two variants of perf timechart: + + 'perf timechart record <command>' to record the system level events + of an arbitrary workload. + + 'perf timechart' to turn a trace into a Scalable Vector Graphics file, + that can be viewed with popular SVG viewers such as 'Inkscape'. + +OPTIONS +------- +-o:: +--output=:: + Select the output file (default: output.svg) +-i:: +--input=:: + Select the input file (default: perf.data) +-w:: +--width=:: + Select the width of the SVG file (default: 1000) +-P:: +--power-only:: + Only output the CPU power section of the diagram +-p:: +--process:: + Select the processes to display, by name or PID + +--symfs=<directory>:: + Look for files with symbols relative to this directory. + +SEE ALSO +-------- +linkperf:perf-record[1] diff --git a/tools/perf/Documentation/perf-top.txt b/tools/perf/Documentation/perf-top.txt new file mode 100644 index 00000000..f6eb1cda --- /dev/null +++ b/tools/perf/Documentation/perf-top.txt @@ -0,0 +1,147 @@ +perf-top(1) +=========== + +NAME +---- +perf-top - System profiling tool. + +SYNOPSIS +-------- +[verse] +'perf top' [-e <EVENT> | --event=EVENT] [<options>] + +DESCRIPTION +----------- +This command generates and displays a performance counter profile in real time. + + +OPTIONS +------- +-a:: +--all-cpus:: + System-wide collection. (default) + +-c <count>:: +--count=<count>:: + Event period to sample. + +-C <cpu-list>:: +--cpu=<cpu>:: +Monitor only on the list of CPUs provided. Multiple CPUs can be provided as a +comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2. +Default is to monitor all CPUS. + +-d <seconds>:: +--delay=<seconds>:: + Number of seconds to delay between refreshes. + +-e <event>:: +--event=<event>:: + Select the PMU event. Selection can be a symbolic event name + (use 'perf list' to list all events) or a raw PMU + event (eventsel+umask) in the form of rNNN where NNN is a + hexadecimal event descriptor. + +-E <entries>:: +--entries=<entries>:: + Display this many functions. + +-f <count>:: +--count-filter=<count>:: + Only display functions with more events than this. + +-g:: +--group:: + Put the counters into a counter group. + +-F <freq>:: +--freq=<freq>:: + Profile at this frequency. + +-i:: +--inherit:: + Child tasks inherit counters, only makes sens with -p option. + +-k <path>:: +--vmlinux=<path>:: + Path to vmlinux. Required for annotation functionality. + +-m <pages>:: +--mmap-pages=<pages>:: + Number of mmapped data pages. + +-p <pid>:: +--pid=<pid>:: + Profile events on existing Process ID. + +-t <tid>:: +--tid=<tid>:: + Profile events on existing thread ID. + +-r <priority>:: +--realtime=<priority>:: + Collect data with this RT SCHED_FIFO priority. + +-s <symbol>:: +--sym-annotate=<symbol>:: + Annotate this symbol. + +-K:: +--hide_kernel_symbols:: + Hide kernel symbols. + +-U:: +--hide_user_symbols:: + Hide user symbols. + +-D:: +--dump-symtab:: + Dump the symbol table used for profiling. + +-v:: +--verbose:: + Be more verbose (show counter open errors, etc). + +-z:: +--zero:: + Zero history across display updates. + +INTERACTIVE PROMPTING KEYS +-------------------------- + +[d]:: + Display refresh delay. + +[e]:: + Number of entries to display. + +[E]:: + Event to display when multiple counters are active. + +[f]:: + Profile display filter (>= hit count). + +[F]:: + Annotation display filter (>= % of total). + +[s]:: + Annotate symbol. + +[S]:: + Stop annotation, return to full profile display. + +[w]:: + Toggle between weighted sum and individual count[E]r profile. + +[z]:: + Toggle event count zeroing across display updates. + +[qQ]:: + Quit. + +Pressing any unmapped key displays a menu, and prompts for input. + + +SEE ALSO +-------- +linkperf:perf-stat[1], linkperf:perf-list[1] diff --git a/tools/perf/Documentation/perf.txt b/tools/perf/Documentation/perf.txt new file mode 100644 index 00000000..0eeb247d --- /dev/null +++ b/tools/perf/Documentation/perf.txt @@ -0,0 +1,24 @@ +perf(1) +======= + +NAME +---- +perf - Performance analysis tools for Linux + +SYNOPSIS +-------- +[verse] +'perf' [--version] [--help] COMMAND [ARGS] + +DESCRIPTION +----------- +Performance counters for Linux are a new kernel-based subsystem +that provide a framework for all things performance analysis. It +covers hardware level (CPU/PMU, Performance Monitoring Unit) features +and software features (software counters, tracepoints) as well. + +SEE ALSO +-------- +linkperf:perf-stat[1], linkperf:perf-top[1], +linkperf:perf-record[1], linkperf:perf-report[1], +linkperf:perf-list[1] diff --git a/tools/perf/MANIFEST b/tools/perf/MANIFEST new file mode 100644 index 00000000..c12659d8 --- /dev/null +++ b/tools/perf/MANIFEST @@ -0,0 +1,13 @@ +tools/perf +include/linux/perf_event.h +include/linux/rbtree.h +include/linux/list.h +include/linux/hash.h +include/linux/stringify.h +lib/rbtree.c +include/linux/swab.h +arch/*/include/asm/unistd*.h +arch/*/lib/memcpy*.S +include/linux/poison.h +include/linux/magic.h +include/linux/hw_breakpoint.h diff --git a/tools/perf/Makefile b/tools/perf/Makefile new file mode 100644 index 00000000..c1683661 --- /dev/null +++ b/tools/perf/Makefile @@ -0,0 +1,905 @@ +ifeq ("$(origin O)", "command line") + OUTPUT := $(O)/ +endif + +# The default target of this Makefile is... +all: + +include config/utilities.mak + +ifneq ($(OUTPUT),) +# check that the output directory actually exists +OUTDIR := $(shell cd $(OUTPUT) && /bin/pwd) +$(if $(OUTDIR),, $(error output directory "$(OUTPUT)" does not exist)) +endif + +# Define V to have a more verbose compile. +# +# Define PYTHON to point to the python binary if the default +# `python' is not correct; for example: PYTHON=python2 +# +# Define PYTHON_CONFIG to point to the python-config binary if +# the default `$(PYTHON)-config' is not correct. +# +# Define ASCIIDOC8 if you want to format documentation with AsciiDoc 8 +# +# Define DOCBOOK_XSL_172 if you want to format man pages with DocBook XSL v1.72. +# +# Define LDFLAGS=-static to build a static binary. +# +# Define EXTRA_CFLAGS=-m64 or EXTRA_CFLAGS=-m32 as appropriate for cross-builds. +# +# Define NO_DWARF if you do not want debug-info analysis feature at all. + +$(OUTPUT)PERF-VERSION-FILE: .FORCE-PERF-VERSION-FILE + @$(SHELL_PATH) util/PERF-VERSION-GEN $(OUTPUT) +-include $(OUTPUT)PERF-VERSION-FILE + +uname_M := $(shell uname -m 2>/dev/null || echo not) + +ARCH ?= $(shell echo $(uname_M) | sed -e s/i.86/i386/ -e s/sun4u/sparc64/ \ + -e s/arm.*/arm/ -e s/sa110/arm/ \ + -e s/s390x/s390/ -e s/parisc64/parisc/ \ + -e s/ppc.*/powerpc/ -e s/mips.*/mips/ \ + -e s/sh[234].*/sh/ ) + +CC = $(CROSS_COMPILE)gcc +AR = $(CROSS_COMPILE)ar + +# Additional ARCH settings for x86 +ifeq ($(ARCH),i386) + ARCH := x86 +endif +ifeq ($(ARCH),x86_64) + ARCH := x86 + IS_X86_64 := 0 + ifeq (, $(findstring m32,$(EXTRA_CFLAGS))) + IS_X86_64 := $(shell echo __x86_64__ | ${CC} -E -xc - | tail -n 1) + endif + ifeq (${IS_X86_64}, 1) + RAW_ARCH := x86_64 + ARCH_CFLAGS := -DARCH_X86_64 + ARCH_INCLUDE = ../../arch/x86/lib/memcpy_64.S + endif +endif + +# +# Include saner warnings here, which can catch bugs: +# + +EXTRA_WARNINGS := -Wformat +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wformat-security +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wformat-y2k +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wshadow +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Winit-self +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wpacked +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wredundant-decls +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wstrict-aliasing=3 +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wswitch-default +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wswitch-enum +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wno-system-headers +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wundef +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wwrite-strings +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wbad-function-cast +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wmissing-declarations +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wmissing-prototypes +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wnested-externs +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wold-style-definition +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wstrict-prototypes +EXTRA_WARNINGS := $(EXTRA_WARNINGS) -Wdeclaration-after-statement + +ifeq ("$(origin DEBUG)", "command line") + PERF_DEBUG = $(DEBUG) +endif +ifndef PERF_DEBUG + CFLAGS_OPTIMIZE = -O6 +endif + +CFLAGS = -fno-omit-frame-pointer -ggdb3 -Wall -Wextra -std=gnu99 -Werror $(CFLAGS_OPTIMIZE) -D_FORTIFY_SOURCE=2 $(EXTRA_WARNINGS) $(EXTRA_CFLAGS) +EXTLIBS = -lpthread -lrt -lelf -lm +ALL_CFLAGS = $(CFLAGS) -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 +ALL_LDFLAGS = $(LDFLAGS) +STRIP ?= strip + +# Among the variables below, these: +# perfexecdir +# template_dir +# mandir +# infodir +# htmldir +# ETC_PERFCONFIG (but not sysconfdir) +# can be specified as a relative path some/where/else; +# this is interpreted as relative to $(prefix) and "perf" at +# runtime figures out where they are based on the path to the executable. +# This can help installing the suite in a relocatable way. + +# Make the path relative to DESTDIR, not to prefix +ifndef DESTDIR +prefix = $(HOME) +endif +bindir_relative = bin +bindir = $(prefix)/$(bindir_relative) +mandir = share/man +infodir = share/info +perfexecdir = libexec/perf-core +sharedir = $(prefix)/share +template_dir = share/perf-core/templates +htmldir = share/doc/perf-doc +ifeq ($(prefix),/usr) +sysconfdir = /etc +ETC_PERFCONFIG = $(sysconfdir)/perfconfig +else +sysconfdir = $(prefix)/etc +ETC_PERFCONFIG = etc/perfconfig +endif +lib = lib + +export prefix bindir sharedir sysconfdir + +RM = rm -f +MKDIR = mkdir +FIND = find +INSTALL = install + +# sparse is architecture-neutral, which means that we need to tell it +# explicitly what architecture to check for. Fix this up for yours.. +SPARSE_FLAGS = -D__BIG_ENDIAN__ -D__powerpc__ + +-include config/feature-tests.mak + +ifeq ($(call try-cc,$(SOURCE_HELLO),-Werror -fstack-protector-all),y) + CFLAGS := $(CFLAGS) -fstack-protector-all +endif + +ifeq ($(call try-cc,$(SOURCE_HELLO),-Werror -Wstack-protector),y) + CFLAGS := $(CFLAGS) -Wstack-protector +endif + +ifeq ($(call try-cc,$(SOURCE_HELLO),-Werror -Wvolatile-register-var),y) + CFLAGS := $(CFLAGS) -Wvolatile-register-var +endif + +### --- END CONFIGURATION SECTION --- + +# Those must not be GNU-specific; they are shared with perl/ which may +# be built by a different compiler. (Note that this is an artifact now +# but it still might be nice to keep that distinction.) +BASIC_CFLAGS = -Iutil/include -Iarch/$(ARCH)/include +BASIC_LDFLAGS = + +# Guard against environment variables +BUILTIN_OBJS = +LIB_H = +LIB_OBJS = +PYRF_OBJS = +SCRIPT_SH = + +SCRIPT_SH += perf-archive.sh + +grep-libs = $(filter -l%,$(1)) +strip-libs = $(filter-out -l%,$(1)) + +$(OUTPUT)python/perf.so: $(PYRF_OBJS) + $(QUIET_GEN)CFLAGS='$(BASIC_CFLAGS)' $(PYTHON_WORD) util/setup.py \ + --quiet build_ext \ + --build-lib='$(OUTPUT)python' \ + --build-temp='$(OUTPUT)python/temp' +# +# No Perl scripts right now: +# + +SCRIPTS = $(patsubst %.sh,%,$(SCRIPT_SH)) + +# +# Single 'perf' binary right now: +# +PROGRAMS += $(OUTPUT)perf + +LANG_BINDINGS = + +# what 'all' will build and 'install' will install, in perfexecdir +ALL_PROGRAMS = $(PROGRAMS) $(SCRIPTS) + +# what 'all' will build but not install in perfexecdir +OTHER_PROGRAMS = $(OUTPUT)perf + +# Set paths to tools early so that they can be used for version tests. +ifndef SHELL_PATH + SHELL_PATH = /bin/sh +endif +ifndef PERL_PATH + PERL_PATH = /usr/bin/perl +endif + +export PERL_PATH + +LIB_FILE=$(OUTPUT)libperf.a + +LIB_H += ../../include/linux/perf_event.h +LIB_H += ../../include/linux/rbtree.h +LIB_H += ../../include/linux/list.h +LIB_H += ../../include/linux/const.h +LIB_H += ../../include/linux/hash.h +LIB_H += ../../include/linux/stringify.h +LIB_H += util/include/linux/bitmap.h +LIB_H += util/include/linux/bitops.h +LIB_H += util/include/linux/compiler.h +LIB_H += util/include/linux/const.h +LIB_H += util/include/linux/ctype.h +LIB_H += util/include/linux/kernel.h +LIB_H += util/include/linux/list.h +LIB_H += util/include/linux/module.h +LIB_H += util/include/linux/poison.h +LIB_H += util/include/linux/prefetch.h +LIB_H += util/include/linux/rbtree.h +LIB_H += util/include/linux/string.h +LIB_H += util/include/linux/types.h +LIB_H += util/include/linux/linkage.h +LIB_H += util/include/asm/asm-offsets.h +LIB_H += util/include/asm/bug.h +LIB_H += util/include/asm/byteorder.h +LIB_H += util/include/asm/hweight.h +LIB_H += util/include/asm/swab.h +LIB_H += util/include/asm/system.h +LIB_H += util/include/asm/uaccess.h +LIB_H += util/include/dwarf-regs.h +LIB_H += util/include/asm/dwarf2.h +LIB_H += util/include/asm/cpufeature.h +LIB_H += perf.h +LIB_H += util/annotate.h +LIB_H += util/cache.h +LIB_H += util/callchain.h +LIB_H += util/build-id.h +LIB_H += util/debug.h +LIB_H += util/debugfs.h +LIB_H += util/event.h +LIB_H += util/evsel.h +LIB_H += util/evlist.h +LIB_H += util/exec_cmd.h +LIB_H += util/types.h +LIB_H += util/levenshtein.h +LIB_H += util/map.h +LIB_H += util/parse-options.h +LIB_H += util/parse-events.h +LIB_H += util/quote.h +LIB_H += util/util.h +LIB_H += util/xyarray.h +LIB_H += util/header.h +LIB_H += util/help.h +LIB_H += util/session.h +LIB_H += util/strbuf.h +LIB_H += util/strlist.h +LIB_H += util/strfilter.h +LIB_H += util/svghelper.h +LIB_H += util/run-command.h +LIB_H += util/sigchain.h +LIB_H += util/symbol.h +LIB_H += util/color.h +LIB_H += util/values.h +LIB_H += util/sort.h +LIB_H += util/hist.h +LIB_H += util/thread.h +LIB_H += util/thread_map.h +LIB_H += util/trace-event.h +LIB_H += util/probe-finder.h +LIB_H += util/probe-event.h +LIB_H += util/pstack.h +LIB_H += util/cpumap.h +LIB_H += util/top.h +LIB_H += $(ARCH_INCLUDE) +LIB_H += util/cgroup.h + +LIB_OBJS += $(OUTPUT)util/abspath.o +LIB_OBJS += $(OUTPUT)util/alias.o +LIB_OBJS += $(OUTPUT)util/annotate.o +LIB_OBJS += $(OUTPUT)util/build-id.o +LIB_OBJS += $(OUTPUT)util/config.o +LIB_OBJS += $(OUTPUT)util/ctype.o +LIB_OBJS += $(OUTPUT)util/debugfs.o +LIB_OBJS += $(OUTPUT)util/environment.o +LIB_OBJS += $(OUTPUT)util/event.o +LIB_OBJS += $(OUTPUT)util/evlist.o +LIB_OBJS += $(OUTPUT)util/evsel.o +LIB_OBJS += $(OUTPUT)util/exec_cmd.o +LIB_OBJS += $(OUTPUT)util/help.o +LIB_OBJS += $(OUTPUT)util/levenshtein.o +LIB_OBJS += $(OUTPUT)util/parse-options.o +LIB_OBJS += $(OUTPUT)util/parse-events.o +LIB_OBJS += $(OUTPUT)util/path.o +LIB_OBJS += $(OUTPUT)util/rbtree.o +LIB_OBJS += $(OUTPUT)util/bitmap.o +LIB_OBJS += $(OUTPUT)util/hweight.o +LIB_OBJS += $(OUTPUT)util/run-command.o +LIB_OBJS += $(OUTPUT)util/quote.o +LIB_OBJS += $(OUTPUT)util/strbuf.o +LIB_OBJS += $(OUTPUT)util/string.o +LIB_OBJS += $(OUTPUT)util/strlist.o +LIB_OBJS += $(OUTPUT)util/strfilter.o +LIB_OBJS += $(OUTPUT)util/top.o +LIB_OBJS += $(OUTPUT)util/usage.o +LIB_OBJS += $(OUTPUT)util/wrapper.o +LIB_OBJS += $(OUTPUT)util/sigchain.o +LIB_OBJS += $(OUTPUT)util/symbol.o +LIB_OBJS += $(OUTPUT)util/color.o +LIB_OBJS += $(OUTPUT)util/pager.o +LIB_OBJS += $(OUTPUT)util/header.o +LIB_OBJS += $(OUTPUT)util/callchain.o +LIB_OBJS += $(OUTPUT)util/values.o +LIB_OBJS += $(OUTPUT)util/debug.o +LIB_OBJS += $(OUTPUT)util/map.o +LIB_OBJS += $(OUTPUT)util/pstack.o +LIB_OBJS += $(OUTPUT)util/session.o +LIB_OBJS += $(OUTPUT)util/thread.o +LIB_OBJS += $(OUTPUT)util/thread_map.o +LIB_OBJS += $(OUTPUT)util/trace-event-parse.o +LIB_OBJS += $(OUTPUT)util/trace-event-read.o +LIB_OBJS += $(OUTPUT)util/trace-event-info.o +LIB_OBJS += $(OUTPUT)util/trace-event-scripting.o +LIB_OBJS += $(OUTPUT)util/svghelper.o +LIB_OBJS += $(OUTPUT)util/sort.o +LIB_OBJS += $(OUTPUT)util/hist.o +LIB_OBJS += $(OUTPUT)util/probe-event.o +LIB_OBJS += $(OUTPUT)util/util.o +LIB_OBJS += $(OUTPUT)util/xyarray.o +LIB_OBJS += $(OUTPUT)util/cpumap.o +LIB_OBJS += $(OUTPUT)util/cgroup.o + +BUILTIN_OBJS += $(OUTPUT)builtin-annotate.o + +BUILTIN_OBJS += $(OUTPUT)builtin-bench.o + +# Benchmark modules +BUILTIN_OBJS += $(OUTPUT)bench/sched-messaging.o +BUILTIN_OBJS += $(OUTPUT)bench/sched-pipe.o +ifeq ($(RAW_ARCH),x86_64) +BUILTIN_OBJS += $(OUTPUT)bench/mem-memcpy-x86-64-asm.o +endif +BUILTIN_OBJS += $(OUTPUT)bench/mem-memcpy.o + +BUILTIN_OBJS += $(OUTPUT)builtin-diff.o +BUILTIN_OBJS += $(OUTPUT)builtin-evlist.o +BUILTIN_OBJS += $(OUTPUT)builtin-help.o +BUILTIN_OBJS += $(OUTPUT)builtin-sched.o +BUILTIN_OBJS += $(OUTPUT)builtin-buildid-list.o +BUILTIN_OBJS += $(OUTPUT)builtin-buildid-cache.o +BUILTIN_OBJS += $(OUTPUT)builtin-list.o +BUILTIN_OBJS += $(OUTPUT)builtin-record.o +BUILTIN_OBJS += $(OUTPUT)builtin-report.o +BUILTIN_OBJS += $(OUTPUT)builtin-stat.o +BUILTIN_OBJS += $(OUTPUT)builtin-timechart.o +BUILTIN_OBJS += $(OUTPUT)builtin-top.o +BUILTIN_OBJS += $(OUTPUT)builtin-script.o +BUILTIN_OBJS += $(OUTPUT)builtin-probe.o +BUILTIN_OBJS += $(OUTPUT)builtin-kmem.o +BUILTIN_OBJS += $(OUTPUT)builtin-lock.o +BUILTIN_OBJS += $(OUTPUT)builtin-kvm.o +BUILTIN_OBJS += $(OUTPUT)builtin-test.o +BUILTIN_OBJS += $(OUTPUT)builtin-inject.o + +PERFLIBS = $(LIB_FILE) + +# Files needed for the python binding, perf.so +# pyrf is just an internal name needed for all those wrappers. +# This has to be in sync with what is in the 'sources' variable in +# tools/perf/util/setup.py + +PYRF_OBJS += $(OUTPUT)util/cpumap.o +PYRF_OBJS += $(OUTPUT)util/ctype.o +PYRF_OBJS += $(OUTPUT)util/evlist.o +PYRF_OBJS += $(OUTPUT)util/evsel.o +PYRF_OBJS += $(OUTPUT)util/python.o +PYRF_OBJS += $(OUTPUT)util/thread_map.o +PYRF_OBJS += $(OUTPUT)util/util.o +PYRF_OBJS += $(OUTPUT)util/xyarray.o + +# +# Platform specific tweaks +# + +# We choose to avoid "if .. else if .. else .. endif endif" +# because maintaining the nesting to match is a pain. If +# we had "elif" things would have been much nicer... + +-include config.mak.autogen +-include config.mak + +ifndef NO_DWARF +FLAGS_DWARF=$(ALL_CFLAGS) -ldw -lelf $(ALL_LDFLAGS) $(EXTLIBS) +ifneq ($(call try-cc,$(SOURCE_DWARF),$(FLAGS_DWARF)),y) + msg := $(warning No libdw.h found or old libdw.h found or elfutils is older than 0.138, disables dwarf support. Please install new elfutils-devel/libdw-dev); + NO_DWARF := 1 +endif # Dwarf support +endif # NO_DWARF + +-include arch/$(ARCH)/Makefile + +ifneq ($(OUTPUT),) + BASIC_CFLAGS += -I$(OUTPUT) +endif + +FLAGS_LIBELF=$(ALL_CFLAGS) $(ALL_LDFLAGS) $(EXTLIBS) +ifneq ($(call try-cc,$(SOURCE_LIBELF),$(FLAGS_LIBELF)),y) + FLAGS_GLIBC=$(ALL_CFLAGS) $(ALL_LDFLAGS) + ifneq ($(call try-cc,$(SOURCE_GLIBC),$(FLAGS_GLIBC)),y) + msg := $(error No gnu/libc-version.h found, please install glibc-dev[el]/glibc-static); + else + msg := $(error No libelf.h/libelf found, please install libelf-dev/elfutils-libelf-devel); + endif +endif + +ifneq ($(call try-cc,$(SOURCE_ELF_MMAP),$(FLAGS_COMMON)),y) + BASIC_CFLAGS += -DLIBELF_NO_MMAP +endif + +ifndef NO_DWARF +ifeq ($(origin PERF_HAVE_DWARF_REGS), undefined) + msg := $(warning DWARF register mappings have not been defined for architecture $(ARCH), DWARF support disabled); +else + BASIC_CFLAGS += -DDWARF_SUPPORT + EXTLIBS += -lelf -ldw + LIB_OBJS += $(OUTPUT)util/probe-finder.o +endif # PERF_HAVE_DWARF_REGS +endif # NO_DWARF + +ifdef NO_NEWT + BASIC_CFLAGS += -DNO_NEWT_SUPPORT +else + FLAGS_NEWT=$(ALL_CFLAGS) $(ALL_LDFLAGS) $(EXTLIBS) -lnewt + ifneq ($(call try-cc,$(SOURCE_NEWT),$(FLAGS_NEWT)),y) + msg := $(warning newt not found, disables TUI support. Please install newt-devel or libnewt-dev); + BASIC_CFLAGS += -DNO_NEWT_SUPPORT + else + # Fedora has /usr/include/slang/slang.h, but ubuntu /usr/include/slang.h + BASIC_CFLAGS += -I/usr/include/slang + EXTLIBS += -lnewt -lslang + LIB_OBJS += $(OUTPUT)util/ui/setup.o + LIB_OBJS += $(OUTPUT)util/ui/browser.o + LIB_OBJS += $(OUTPUT)util/ui/browsers/annotate.o + LIB_OBJS += $(OUTPUT)util/ui/browsers/hists.o + LIB_OBJS += $(OUTPUT)util/ui/browsers/map.o + LIB_OBJS += $(OUTPUT)util/ui/browsers/top.o + LIB_OBJS += $(OUTPUT)util/ui/helpline.o + LIB_OBJS += $(OUTPUT)util/ui/progress.o + LIB_OBJS += $(OUTPUT)util/ui/util.o + LIB_H += util/ui/browser.h + LIB_H += util/ui/browsers/map.h + LIB_H += util/ui/helpline.h + LIB_H += util/ui/libslang.h + LIB_H += util/ui/progress.h + LIB_H += util/ui/util.h + LIB_H += util/ui/ui.h + endif +endif + +ifdef NO_LIBPERL + BASIC_CFLAGS += -DNO_LIBPERL +else + PERL_EMBED_LDOPTS = $(shell perl -MExtUtils::Embed -e ldopts 2>/dev/null) + PERL_EMBED_LDFLAGS = $(call strip-libs,$(PERL_EMBED_LDOPTS)) + PERL_EMBED_LIBADD = $(call grep-libs,$(PERL_EMBED_LDOPTS)) + PERL_EMBED_CCOPTS = `perl -MExtUtils::Embed -e ccopts 2>/dev/null` + FLAGS_PERL_EMBED=$(PERL_EMBED_CCOPTS) $(PERL_EMBED_LDOPTS) + + ifneq ($(call try-cc,$(SOURCE_PERL_EMBED),$(FLAGS_PERL_EMBED)),y) + BASIC_CFLAGS += -DNO_LIBPERL + else + ALL_LDFLAGS += $(PERL_EMBED_LDFLAGS) + EXTLIBS += $(PERL_EMBED_LIBADD) + LIB_OBJS += $(OUTPUT)util/scripting-engines/trace-event-perl.o + LIB_OBJS += $(OUTPUT)scripts/perl/Perf-Trace-Util/Context.o + endif +endif + +disable-python = $(eval $(disable-python_code)) +define disable-python_code + BASIC_CFLAGS += -DNO_LIBPYTHON + $(if $(1),$(warning No $(1) was found)) + $(warning Python support won't be built) +endef + +override PYTHON := \ + $(call get-executable-or-default,PYTHON,python) + +ifndef PYTHON + $(call disable-python,python interpreter) + python-clean := +else + + PYTHON_WORD := $(call shell-wordify,$(PYTHON)) + + python-clean := $(PYTHON_WORD) util/setup.py clean \ + --build-lib='$(OUTPUT)python' \ + --build-temp='$(OUTPUT)python/temp' + + ifdef NO_LIBPYTHON + $(call disable-python) + else + + override PYTHON_CONFIG := \ + $(call get-executable-or-default,PYTHON_CONFIG,$(PYTHON)-config) + + ifndef PYTHON_CONFIG + $(call disable-python,python-config tool) + else + + PYTHON_CONFIG_SQ := $(call shell-sq,$(PYTHON_CONFIG)) + + PYTHON_EMBED_LDOPTS := $(shell $(PYTHON_CONFIG_SQ) --ldflags 2>/dev/null) + PYTHON_EMBED_LDFLAGS := $(call strip-libs,$(PYTHON_EMBED_LDOPTS)) + PYTHON_EMBED_LIBADD := $(call grep-libs,$(PYTHON_EMBED_LDOPTS)) + PYTHON_EMBED_CCOPTS := $(shell $(PYTHON_CONFIG_SQ) --cflags 2>/dev/null) + FLAGS_PYTHON_EMBED := $(PYTHON_EMBED_CCOPTS) $(PYTHON_EMBED_LDOPTS) + + ifneq ($(call try-cc,$(SOURCE_PYTHON_EMBED),$(FLAGS_PYTHON_EMBED)),y) + $(call disable-python,Python.h (for Python 2.x)) + else + + ifneq ($(call try-cc,$(SOURCE_PYTHON_VERSION),$(FLAGS_PYTHON_EMBED)),y) + $(warning Python 3 is not yet supported; please set) + $(warning PYTHON and/or PYTHON_CONFIG appropriately.) + $(warning If you also have Python 2 installed, then) + $(warning try something like:) + $(warning $(and ,)) + $(warning $(and ,) make PYTHON=python2) + $(warning $(and ,)) + $(warning Otherwise, disable Python support entirely:) + $(warning $(and ,)) + $(warning $(and ,) make NO_LIBPYTHON=1) + $(warning $(and ,)) + $(error $(and ,)) + else + ALL_LDFLAGS += $(PYTHON_EMBED_LDFLAGS) + EXTLIBS += $(PYTHON_EMBED_LIBADD) + LIB_OBJS += $(OUTPUT)util/scripting-engines/trace-event-python.o + LIB_OBJS += $(OUTPUT)scripts/python/Perf-Trace-Util/Context.o + LANG_BINDINGS += $(OUTPUT)python/perf.so + endif + + endif + endif + endif +endif + +ifdef NO_DEMANGLE + BASIC_CFLAGS += -DNO_DEMANGLE +else + ifdef HAVE_CPLUS_DEMANGLE + EXTLIBS += -liberty + BASIC_CFLAGS += -DHAVE_CPLUS_DEMANGLE + else + FLAGS_BFD=$(ALL_CFLAGS) $(ALL_LDFLAGS) $(EXTLIBS) -lbfd + has_bfd := $(call try-cc,$(SOURCE_BFD),$(FLAGS_BFD)) + ifeq ($(has_bfd),y) + EXTLIBS += -lbfd + else + FLAGS_BFD_IBERTY=$(FLAGS_BFD) -liberty + has_bfd_iberty := $(call try-cc,$(SOURCE_BFD),$(FLAGS_BFD_IBERTY)) + ifeq ($(has_bfd_iberty),y) + EXTLIBS += -lbfd -liberty + else + FLAGS_BFD_IBERTY_Z=$(FLAGS_BFD_IBERTY) -lz + has_bfd_iberty_z := $(call try-cc,$(SOURCE_BFD),$(FLAGS_BFD_IBERTY_Z)) + ifeq ($(has_bfd_iberty_z),y) + EXTLIBS += -lbfd -liberty -lz + else + FLAGS_CPLUS_DEMANGLE=$(ALL_CFLAGS) $(ALL_LDFLAGS) $(EXTLIBS) -liberty + has_cplus_demangle := $(call try-cc,$(SOURCE_CPLUS_DEMANGLE),$(FLAGS_CPLUS_DEMANGLE)) + ifeq ($(has_cplus_demangle),y) + EXTLIBS += -liberty + BASIC_CFLAGS += -DHAVE_CPLUS_DEMANGLE + else + msg := $(warning No bfd.h/libbfd found, install binutils-dev[el]/zlib-static to gain symbol demangling) + BASIC_CFLAGS += -DNO_DEMANGLE + endif + endif + endif + endif + endif +endif + + +ifdef NO_STRLCPY + BASIC_CFLAGS += -DNO_STRLCPY +else + ifneq ($(call try-cc,$(SOURCE_STRLCPY),),y) + BASIC_CFLAGS += -DNO_STRLCPY + endif +endif + +ifneq ($(findstring $(MAKEFLAGS),s),s) +ifndef V + QUIET_CC = @echo ' ' CC $@; + QUIET_AR = @echo ' ' AR $@; + QUIET_LINK = @echo ' ' LINK $@; + QUIET_MKDIR = @echo ' ' MKDIR $@; + QUIET_GEN = @echo ' ' GEN $@; +endif +endif + +ifdef ASCIIDOC8 + export ASCIIDOC8 +endif + +# Shell quote (do not use $(call) to accommodate ancient setups); + +ETC_PERFCONFIG_SQ = $(subst ','\'',$(ETC_PERFCONFIG)) + +DESTDIR_SQ = $(subst ','\'',$(DESTDIR)) +bindir_SQ = $(subst ','\'',$(bindir)) +bindir_relative_SQ = $(subst ','\'',$(bindir_relative)) +mandir_SQ = $(subst ','\'',$(mandir)) +infodir_SQ = $(subst ','\'',$(infodir)) +perfexecdir_SQ = $(subst ','\'',$(perfexecdir)) +template_dir_SQ = $(subst ','\'',$(template_dir)) +htmldir_SQ = $(subst ','\'',$(htmldir)) +prefix_SQ = $(subst ','\'',$(prefix)) + +SHELL_PATH_SQ = $(subst ','\'',$(SHELL_PATH)) + +LIBS = -Wl,--whole-archive $(PERFLIBS) -Wl,--no-whole-archive -Wl,--start-group $(EXTLIBS) -Wl,--end-group + +ALL_CFLAGS += $(BASIC_CFLAGS) +ALL_CFLAGS += $(ARCH_CFLAGS) +ALL_LDFLAGS += $(BASIC_LDFLAGS) + +export INSTALL SHELL_PATH + + +### Build rules + +SHELL = $(SHELL_PATH) + +all: shell_compatibility_test $(ALL_PROGRAMS) $(LANG_BINDINGS) $(OTHER_PROGRAMS) + +please_set_SHELL_PATH_to_a_more_modern_shell: + @$$(:) + +shell_compatibility_test: please_set_SHELL_PATH_to_a_more_modern_shell + +strip: $(PROGRAMS) $(OUTPUT)perf + $(STRIP) $(STRIP_OPTS) $(PROGRAMS) $(OUTPUT)perf + +$(OUTPUT)perf.o: perf.c $(OUTPUT)common-cmds.h $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -DPERF_VERSION='"$(PERF_VERSION)"' \ + '-DPERF_HTML_PATH="$(htmldir_SQ)"' \ + $(ALL_CFLAGS) -c $(filter %.c,$^) -o $@ + +$(OUTPUT)perf: $(OUTPUT)perf.o $(BUILTIN_OBJS) $(PERFLIBS) + $(QUIET_LINK)$(CC) $(ALL_CFLAGS) $(ALL_LDFLAGS) $(OUTPUT)perf.o \ + $(BUILTIN_OBJS) $(LIBS) -o $@ + +$(OUTPUT)builtin-help.o: builtin-help.c $(OUTPUT)common-cmds.h $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) \ + '-DPERF_HTML_PATH="$(htmldir_SQ)"' \ + '-DPERF_MAN_PATH="$(mandir_SQ)"' \ + '-DPERF_INFO_PATH="$(infodir_SQ)"' $< + +$(OUTPUT)builtin-timechart.o: builtin-timechart.c $(OUTPUT)common-cmds.h $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) \ + '-DPERF_HTML_PATH="$(htmldir_SQ)"' \ + '-DPERF_MAN_PATH="$(mandir_SQ)"' \ + '-DPERF_INFO_PATH="$(infodir_SQ)"' $< + +$(OUTPUT)common-cmds.h: util/generate-cmdlist.sh command-list.txt + +$(OUTPUT)common-cmds.h: $(wildcard Documentation/perf-*.txt) + $(QUIET_GEN). util/generate-cmdlist.sh > $@+ && mv $@+ $@ + +$(SCRIPTS) : % : %.sh + $(QUIET_GEN)$(INSTALL) '$@.sh' '$(OUTPUT)$@' + +# These can record PERF_VERSION +$(OUTPUT)perf.o perf.spec \ + $(SCRIPTS) \ + : $(OUTPUT)PERF-VERSION-FILE + +$(OUTPUT)%.o: %.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) $< +$(OUTPUT)%.s: %.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -S $(ALL_CFLAGS) $< +$(OUTPUT)%.o: %.S + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) $< + +$(OUTPUT)util/exec_cmd.o: util/exec_cmd.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) \ + '-DPERF_EXEC_PATH="$(perfexecdir_SQ)"' \ + '-DBINDIR="$(bindir_relative_SQ)"' \ + '-DPREFIX="$(prefix_SQ)"' \ + $< + +$(OUTPUT)util/config.o: util/config.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DETC_PERFCONFIG='"$(ETC_PERFCONFIG_SQ)"' $< + +$(OUTPUT)util/ui/browser.o: util/ui/browser.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< + +$(OUTPUT)util/ui/browsers/annotate.o: util/ui/browsers/annotate.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< + +$(OUTPUT)util/ui/browsers/top.o: util/ui/browsers/top.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< + +$(OUTPUT)util/ui/browsers/hists.o: util/ui/browsers/hists.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< + +$(OUTPUT)util/ui/browsers/map.o: util/ui/browsers/map.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< + +$(OUTPUT)util/rbtree.o: ../../lib/rbtree.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DETC_PERFCONFIG='"$(ETC_PERFCONFIG_SQ)"' $< + +$(OUTPUT)util/scripting-engines/trace-event-perl.o: util/scripting-engines/trace-event-perl.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) $(PERL_EMBED_CCOPTS) -Wno-redundant-decls -Wno-strict-prototypes -Wno-unused-parameter -Wno-shadow $< + +$(OUTPUT)scripts/perl/Perf-Trace-Util/Context.o: scripts/perl/Perf-Trace-Util/Context.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) $(PERL_EMBED_CCOPTS) -Wno-redundant-decls -Wno-strict-prototypes -Wno-unused-parameter -Wno-nested-externs $< + +$(OUTPUT)util/scripting-engines/trace-event-python.o: util/scripting-engines/trace-event-python.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) $(PYTHON_EMBED_CCOPTS) -Wno-redundant-decls -Wno-strict-prototypes -Wno-unused-parameter -Wno-shadow $< + +$(OUTPUT)scripts/python/Perf-Trace-Util/Context.o: scripts/python/Perf-Trace-Util/Context.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) $(PYTHON_EMBED_CCOPTS) -Wno-redundant-decls -Wno-strict-prototypes -Wno-unused-parameter -Wno-nested-externs $< + +$(OUTPUT)perf-%: %.o $(PERFLIBS) + $(QUIET_LINK)$(CC) $(ALL_CFLAGS) -o $@ $(ALL_LDFLAGS) $(filter %.o,$^) $(LIBS) + +$(LIB_OBJS) $(BUILTIN_OBJS): $(LIB_H) +$(patsubst perf-%,%.o,$(PROGRAMS)): $(LIB_H) $(wildcard */*.h) + +# we compile into subdirectories. if the target directory is not the source directory, they might not exists. So +# we depend the various files onto their directories. +DIRECTORY_DEPS = $(LIB_OBJS) $(BUILTIN_OBJS) $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)common-cmds.h +$(DIRECTORY_DEPS): | $(sort $(dir $(DIRECTORY_DEPS))) +# In the second step, we make a rule to actually create these directories +$(sort $(dir $(DIRECTORY_DEPS))): + $(QUIET_MKDIR)$(MKDIR) -p $@ 2>/dev/null + +$(LIB_FILE): $(LIB_OBJS) + $(QUIET_AR)$(RM) $@ && $(AR) rcs $@ $(LIB_OBJS) + +help: + @echo 'Perf make targets:' + @echo ' doc - make *all* documentation (see below)' + @echo ' man - make manpage documentation (access with man <foo>)' + @echo ' html - make html documentation' + @echo ' info - make GNU info documentation (access with info <foo>)' + @echo ' pdf - make pdf documentation' + @echo ' TAGS - use etags to make tag information for source browsing' + @echo ' tags - use ctags to make tag information for source browsing' + @echo ' cscope - use cscope to make interactive browsing database' + @echo '' + @echo 'Perf install targets:' + @echo ' NOTE: documentation build requires asciidoc, xmlto packages to be installed' + @echo ' HINT: use "make prefix=<path> <install target>" to install to a particular' + @echo ' path like make prefix=/usr/local install install-doc' + @echo ' install - install compiled binaries' + @echo ' install-doc - install *all* documentation' + @echo ' install-man - install manpage documentation' + @echo ' install-html - install html documentation' + @echo ' install-info - install GNU info documentation' + @echo ' install-pdf - install pdf documentation' + @echo '' + @echo ' quick-install-doc - alias for quick-install-man' + @echo ' quick-install-man - install the documentation quickly' + @echo ' quick-install-html - install the html documentation quickly' + @echo '' + @echo 'Perf maintainer targets:' + @echo ' distclean - alias to clean' + @echo ' clean - clean all binary objects and build output' + +doc: + $(MAKE) -C Documentation all + +man: + $(MAKE) -C Documentation man + +html: + $(MAKE) -C Documentation html + +info: + $(MAKE) -C Documentation info + +pdf: + $(MAKE) -C Documentation pdf + +TAGS: + $(RM) TAGS + $(FIND) . -name '*.[hcS]' -print | xargs etags -a + +tags: + $(RM) tags + $(FIND) . -name '*.[hcS]' -print | xargs ctags -a + +cscope: + $(RM) cscope* + $(FIND) . -name '*.[hcS]' -print | xargs cscope -b + +### Detect prefix changes +TRACK_CFLAGS = $(subst ','\'',$(ALL_CFLAGS)):\ + $(bindir_SQ):$(perfexecdir_SQ):$(template_dir_SQ):$(prefix_SQ) + +$(OUTPUT)PERF-CFLAGS: .FORCE-PERF-CFLAGS + @FLAGS='$(TRACK_CFLAGS)'; \ + if test x"$$FLAGS" != x"`cat $(OUTPUT)PERF-CFLAGS 2>/dev/null`" ; then \ + echo 1>&2 " * new build flags or prefix"; \ + echo "$$FLAGS" >$(OUTPUT)PERF-CFLAGS; \ + fi + +### Testing rules + +# GNU make supports exporting all variables by "export" without parameters. +# However, the environment gets quite big, and some programs have problems +# with that. + +check: $(OUTPUT)common-cmds.h + if sparse; \ + then \ + for i in *.c */*.c; \ + do \ + sparse $(ALL_CFLAGS) $(SPARSE_FLAGS) $$i || exit; \ + done; \ + else \ + exit 1; \ + fi + +### Installation rules + +ifneq ($(filter /%,$(firstword $(perfexecdir))),) +perfexec_instdir = $(perfexecdir) +else +perfexec_instdir = $(prefix)/$(perfexecdir) +endif +perfexec_instdir_SQ = $(subst ','\'',$(perfexec_instdir)) + +install: all + $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(bindir_SQ)' + $(INSTALL) $(OUTPUT)perf '$(DESTDIR_SQ)$(bindir_SQ)' + $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/perl/Perf-Trace-Util/lib/Perf/Trace' + $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/perl/bin' + $(INSTALL) $(OUTPUT)perf-archive -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)' + $(INSTALL) scripts/perl/Perf-Trace-Util/lib/Perf/Trace/* -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/perl/Perf-Trace-Util/lib/Perf/Trace' + $(INSTALL) scripts/perl/*.pl -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/perl' + $(INSTALL) scripts/perl/bin/* -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/perl/bin' + $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/python/Perf-Trace-Util/lib/Perf/Trace' + $(INSTALL) -d -m 755 '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/python/bin' + $(INSTALL) scripts/python/Perf-Trace-Util/lib/Perf/Trace/* -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/python/Perf-Trace-Util/lib/Perf/Trace' + $(INSTALL) scripts/python/*.py -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/python' + $(INSTALL) scripts/python/bin/* -t '$(DESTDIR_SQ)$(perfexec_instdir_SQ)/scripts/python/bin' + +install-doc: + $(MAKE) -C Documentation install + +install-man: + $(MAKE) -C Documentation install-man + +install-html: + $(MAKE) -C Documentation install-html + +install-info: + $(MAKE) -C Documentation install-info + +install-pdf: + $(MAKE) -C Documentation install-pdf + +quick-install-doc: + $(MAKE) -C Documentation quick-install + +quick-install-man: + $(MAKE) -C Documentation quick-install-man + +quick-install-html: + $(MAKE) -C Documentation quick-install-html + +### Cleaning rules + +clean: + $(RM) $(OUTPUT){*.o,*/*.o,*/*/*.o,*/*/*/*.o,$(LIB_FILE),perf-archive} + $(RM) $(ALL_PROGRAMS) perf + $(RM) *.spec *.pyc *.pyo */*.pyc */*.pyo $(OUTPUT)common-cmds.h TAGS tags cscope* + $(MAKE) -C Documentation/ clean + $(RM) $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)PERF-CFLAGS + $(python-clean) + +.PHONY: all install clean strip +.PHONY: shell_compatibility_test please_set_SHELL_PATH_to_a_more_modern_shell +.PHONY: .FORCE-PERF-VERSION-FILE TAGS tags cscope .FORCE-PERF-CFLAGS diff --git a/tools/perf/arch/arm/Makefile b/tools/perf/arch/arm/Makefile new file mode 100644 index 00000000..15130b50 --- /dev/null +++ b/tools/perf/arch/arm/Makefile @@ -0,0 +1,4 @@ +ifndef NO_DWARF +PERF_HAVE_DWARF_REGS := 1 +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o +endif diff --git a/tools/perf/arch/arm/util/dwarf-regs.c b/tools/perf/arch/arm/util/dwarf-regs.c new file mode 100644 index 00000000..fff6450c --- /dev/null +++ b/tools/perf/arch/arm/util/dwarf-regs.c @@ -0,0 +1,64 @@ +/* + * Mapping of DWARF debug register numbers into register names. + * + * Copyright (C) 2010 Will Deacon, ARM Ltd. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include <libio.h> +#include <dwarf-regs.h> + +struct pt_regs_dwarfnum { + const char *name; + unsigned int dwarfnum; +}; + +#define STR(s) #s +#define REG_DWARFNUM_NAME(r, num) {.name = r, .dwarfnum = num} +#define GPR_DWARFNUM_NAME(num) \ + {.name = STR(%r##num), .dwarfnum = num} +#define REG_DWARFNUM_END {.name = NULL, .dwarfnum = 0} + +/* + * Reference: + * http://infocenter.arm.com/help/topic/com.arm.doc.ihi0040a/IHI0040A_aadwarf.pdf + */ +static const struct pt_regs_dwarfnum regdwarfnum_table[] = { + GPR_DWARFNUM_NAME(0), + GPR_DWARFNUM_NAME(1), + GPR_DWARFNUM_NAME(2), + GPR_DWARFNUM_NAME(3), + GPR_DWARFNUM_NAME(4), + GPR_DWARFNUM_NAME(5), + GPR_DWARFNUM_NAME(6), + GPR_DWARFNUM_NAME(7), + GPR_DWARFNUM_NAME(8), + GPR_DWARFNUM_NAME(9), + GPR_DWARFNUM_NAME(10), + REG_DWARFNUM_NAME("%fp", 11), + REG_DWARFNUM_NAME("%ip", 12), + REG_DWARFNUM_NAME("%sp", 13), + REG_DWARFNUM_NAME("%lr", 14), + REG_DWARFNUM_NAME("%pc", 15), + REG_DWARFNUM_END, +}; + +/** + * get_arch_regstr() - lookup register name from it's DWARF register number + * @n: the DWARF register number + * + * get_arch_regstr() returns the name of the register in struct + * regdwarfnum_table from it's DWARF register number. If the register is not + * found in the table, this returns NULL; + */ +const char *get_arch_regstr(unsigned int n) +{ + const struct pt_regs_dwarfnum *roff; + for (roff = regdwarfnum_table; roff->name != NULL; roff++) + if (roff->dwarfnum == n) + return roff->name; + return NULL; +} diff --git a/tools/perf/arch/powerpc/Makefile b/tools/perf/arch/powerpc/Makefile new file mode 100644 index 00000000..15130b50 --- /dev/null +++ b/tools/perf/arch/powerpc/Makefile @@ -0,0 +1,4 @@ +ifndef NO_DWARF +PERF_HAVE_DWARF_REGS := 1 +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o +endif diff --git a/tools/perf/arch/powerpc/util/dwarf-regs.c b/tools/perf/arch/powerpc/util/dwarf-regs.c new file mode 100644 index 00000000..48ae0c5e --- /dev/null +++ b/tools/perf/arch/powerpc/util/dwarf-regs.c @@ -0,0 +1,88 @@ +/* + * Mapping of DWARF debug register numbers into register names. + * + * Copyright (C) 2010 Ian Munsie, IBM Corporation. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + */ + +#include <libio.h> +#include <dwarf-regs.h> + + +struct pt_regs_dwarfnum { + const char *name; + unsigned int dwarfnum; +}; + +#define STR(s) #s +#define REG_DWARFNUM_NAME(r, num) {.name = r, .dwarfnum = num} +#define GPR_DWARFNUM_NAME(num) \ + {.name = STR(%gpr##num), .dwarfnum = num} +#define REG_DWARFNUM_END {.name = NULL, .dwarfnum = 0} + +/* + * Reference: + * http://refspecs.linuxfoundation.org/ELF/ppc64/PPC-elf64abi-1.9.html + */ +static const struct pt_regs_dwarfnum regdwarfnum_table[] = { + GPR_DWARFNUM_NAME(0), + GPR_DWARFNUM_NAME(1), + GPR_DWARFNUM_NAME(2), + GPR_DWARFNUM_NAME(3), + GPR_DWARFNUM_NAME(4), + GPR_DWARFNUM_NAME(5), + GPR_DWARFNUM_NAME(6), + GPR_DWARFNUM_NAME(7), + GPR_DWARFNUM_NAME(8), + GPR_DWARFNUM_NAME(9), + GPR_DWARFNUM_NAME(10), + GPR_DWARFNUM_NAME(11), + GPR_DWARFNUM_NAME(12), + GPR_DWARFNUM_NAME(13), + GPR_DWARFNUM_NAME(14), + GPR_DWARFNUM_NAME(15), + GPR_DWARFNUM_NAME(16), + GPR_DWARFNUM_NAME(17), + GPR_DWARFNUM_NAME(18), + GPR_DWARFNUM_NAME(19), + GPR_DWARFNUM_NAME(20), + GPR_DWARFNUM_NAME(21), + GPR_DWARFNUM_NAME(22), + GPR_DWARFNUM_NAME(23), + GPR_DWARFNUM_NAME(24), + GPR_DWARFNUM_NAME(25), + GPR_DWARFNUM_NAME(26), + GPR_DWARFNUM_NAME(27), + GPR_DWARFNUM_NAME(28), + GPR_DWARFNUM_NAME(29), + GPR_DWARFNUM_NAME(30), + GPR_DWARFNUM_NAME(31), + REG_DWARFNUM_NAME("%msr", 66), + REG_DWARFNUM_NAME("%ctr", 109), + REG_DWARFNUM_NAME("%link", 108), + REG_DWARFNUM_NAME("%xer", 101), + REG_DWARFNUM_NAME("%dar", 119), + REG_DWARFNUM_NAME("%dsisr", 118), + REG_DWARFNUM_END, +}; + +/** + * get_arch_regstr() - lookup register name from it's DWARF register number + * @n: the DWARF register number + * + * get_arch_regstr() returns the name of the register in struct + * regdwarfnum_table from it's DWARF register number. If the register is not + * found in the table, this returns NULL; + */ +const char *get_arch_regstr(unsigned int n) +{ + const struct pt_regs_dwarfnum *roff; + for (roff = regdwarfnum_table; roff->name != NULL; roff++) + if (roff->dwarfnum == n) + return roff->name; + return NULL; +} diff --git a/tools/perf/arch/s390/Makefile b/tools/perf/arch/s390/Makefile new file mode 100644 index 00000000..15130b50 --- /dev/null +++ b/tools/perf/arch/s390/Makefile @@ -0,0 +1,4 @@ +ifndef NO_DWARF +PERF_HAVE_DWARF_REGS := 1 +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o +endif diff --git a/tools/perf/arch/s390/util/dwarf-regs.c b/tools/perf/arch/s390/util/dwarf-regs.c new file mode 100644 index 00000000..e19653e0 --- /dev/null +++ b/tools/perf/arch/s390/util/dwarf-regs.c @@ -0,0 +1,22 @@ +/* + * Mapping of DWARF debug register numbers into register names. + * + * Copyright IBM Corp. 2010 + * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>, + * + */ + +#include <libio.h> +#include <dwarf-regs.h> + +#define NUM_GPRS 16 + +static const char *gpr_names[NUM_GPRS] = { + "%r0", "%r1", "%r2", "%r3", "%r4", "%r5", "%r6", "%r7", + "%r8", "%r9", "%r10", "%r11", "%r12", "%r13", "%r14", "%r15", +}; + +const char *get_arch_regstr(unsigned int n) +{ + return (n >= NUM_GPRS) ? NULL : gpr_names[n]; +} diff --git a/tools/perf/arch/sh/Makefile b/tools/perf/arch/sh/Makefile new file mode 100644 index 00000000..15130b50 --- /dev/null +++ b/tools/perf/arch/sh/Makefile @@ -0,0 +1,4 @@ +ifndef NO_DWARF +PERF_HAVE_DWARF_REGS := 1 +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o +endif diff --git a/tools/perf/arch/sh/util/dwarf-regs.c b/tools/perf/arch/sh/util/dwarf-regs.c new file mode 100644 index 00000000..a11edb00 --- /dev/null +++ b/tools/perf/arch/sh/util/dwarf-regs.c @@ -0,0 +1,55 @@ +/* + * Mapping of DWARF debug register numbers into register names. + * + * Copyright (C) 2010 Matt Fleming <matt@console-pimps.org> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + */ + +#include <libio.h> +#include <dwarf-regs.h> + +/* + * Generic dwarf analysis helpers + */ + +#define SH_MAX_REGS 18 +const char *sh_regs_table[SH_MAX_REGS] = { + "r0", + "r1", + "r2", + "r3", + "r4", + "r5", + "r6", + "r7", + "r8", + "r9", + "r10", + "r11", + "r12", + "r13", + "r14", + "r15", + "pc", + "pr", +}; + +/* Return architecture dependent register string (for kprobe-tracer) */ +const char *get_arch_regstr(unsigned int n) +{ + return (n <= SH_MAX_REGS) ? sh_regs_table[n] : NULL; +} diff --git a/tools/perf/arch/sparc/Makefile b/tools/perf/arch/sparc/Makefile new file mode 100644 index 00000000..15130b50 --- /dev/null +++ b/tools/perf/arch/sparc/Makefile @@ -0,0 +1,4 @@ +ifndef NO_DWARF +PERF_HAVE_DWARF_REGS := 1 +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o +endif diff --git a/tools/perf/arch/sparc/util/dwarf-regs.c b/tools/perf/arch/sparc/util/dwarf-regs.c new file mode 100644 index 00000000..0ab88483 --- /dev/null +++ b/tools/perf/arch/sparc/util/dwarf-regs.c @@ -0,0 +1,43 @@ +/* + * Mapping of DWARF debug register numbers into register names. + * + * Copyright (C) 2010 David S. Miller <davem@davemloft.net> + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + */ + +#include <libio.h> +#include <dwarf-regs.h> + +#define SPARC_MAX_REGS 96 + +const char *sparc_regs_table[SPARC_MAX_REGS] = { + "%g0", "%g1", "%g2", "%g3", "%g4", "%g5", "%g6", "%g7", + "%o0", "%o1", "%o2", "%o3", "%o4", "%o5", "%sp", "%o7", + "%l0", "%l1", "%l2", "%l3", "%l4", "%l5", "%l6", "%l7", + "%i0", "%i1", "%i2", "%i3", "%i4", "%i5", "%fp", "%i7", + "%f0", "%f1", "%f2", "%f3", "%f4", "%f5", "%f6", "%f7", + "%f8", "%f9", "%f10", "%f11", "%f12", "%f13", "%f14", "%f15", + "%f16", "%f17", "%f18", "%f19", "%f20", "%f21", "%f22", "%f23", + "%f24", "%f25", "%f26", "%f27", "%f28", "%f29", "%f30", "%f31", + "%f32", "%f33", "%f34", "%f35", "%f36", "%f37", "%f38", "%f39", + "%f40", "%f41", "%f42", "%f43", "%f44", "%f45", "%f46", "%f47", + "%f48", "%f49", "%f50", "%f51", "%f52", "%f53", "%f54", "%f55", + "%f56", "%f57", "%f58", "%f59", "%f60", "%f61", "%f62", "%f63", +}; + +/** + * get_arch_regstr() - lookup register name from it's DWARF register number + * @n: the DWARF register number + * + * get_arch_regstr() returns the name of the register in struct + * regdwarfnum_table from it's DWARF register number. If the register is not + * found in the table, this returns NULL; + */ +const char *get_arch_regstr(unsigned int n) +{ + return (n <= SPARC_MAX_REGS) ? sparc_regs_table[n] : NULL; +} diff --git a/tools/perf/arch/x86/Makefile b/tools/perf/arch/x86/Makefile new file mode 100644 index 00000000..15130b50 --- /dev/null +++ b/tools/perf/arch/x86/Makefile @@ -0,0 +1,4 @@ +ifndef NO_DWARF +PERF_HAVE_DWARF_REGS := 1 +LIB_OBJS += $(OUTPUT)arch/$(ARCH)/util/dwarf-regs.o +endif diff --git a/tools/perf/arch/x86/util/dwarf-regs.c b/tools/perf/arch/x86/util/dwarf-regs.c new file mode 100644 index 00000000..a794d308 --- /dev/null +++ b/tools/perf/arch/x86/util/dwarf-regs.c @@ -0,0 +1,75 @@ +/* + * dwarf-regs.c : Mapping of DWARF debug register numbers into register names. + * Extracted from probe-finder.c + * + * Written by Masami Hiramatsu <mhiramat@redhat.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + */ + +#include <libio.h> +#include <dwarf-regs.h> + +/* + * Generic dwarf analysis helpers + */ + +#define X86_32_MAX_REGS 8 +const char *x86_32_regs_table[X86_32_MAX_REGS] = { + "%ax", + "%cx", + "%dx", + "%bx", + "$stack", /* Stack address instead of %sp */ + "%bp", + "%si", + "%di", +}; + +#define X86_64_MAX_REGS 16 +const char *x86_64_regs_table[X86_64_MAX_REGS] = { + "%ax", + "%dx", + "%cx", + "%bx", + "%si", + "%di", + "%bp", + "%sp", + "%r8", + "%r9", + "%r10", + "%r11", + "%r12", + "%r13", + "%r14", + "%r15", +}; + +/* TODO: switching by dwarf address size */ +#ifdef __x86_64__ +#define ARCH_MAX_REGS X86_64_MAX_REGS +#define arch_regs_table x86_64_regs_table +#else +#define ARCH_MAX_REGS X86_32_MAX_REGS +#define arch_regs_table x86_32_regs_table +#endif + +/* Return architecture dependent register string (for kprobe-tracer) */ +const char *get_arch_regstr(unsigned int n) +{ + return (n <= ARCH_MAX_REGS) ? arch_regs_table[n] : NULL; +} diff --git a/tools/perf/bench/bench.h b/tools/perf/bench/bench.h new file mode 100644 index 00000000..f7781c62 --- /dev/null +++ b/tools/perf/bench/bench.h @@ -0,0 +1,17 @@ +#ifndef BENCH_H +#define BENCH_H + +extern int bench_sched_messaging(int argc, const char **argv, const char *prefix); +extern int bench_sched_pipe(int argc, const char **argv, const char *prefix); +extern int bench_mem_memcpy(int argc, const char **argv, const char *prefix __used); + +#define BENCH_FORMAT_DEFAULT_STR "default" +#define BENCH_FORMAT_DEFAULT 0 +#define BENCH_FORMAT_SIMPLE_STR "simple" +#define BENCH_FORMAT_SIMPLE 1 + +#define BENCH_FORMAT_UNKNOWN -1 + +extern int bench_format; + +#endif diff --git a/tools/perf/bench/mem-memcpy-arch.h b/tools/perf/bench/mem-memcpy-arch.h new file mode 100644 index 00000000..a72e36cb --- /dev/null +++ b/tools/perf/bench/mem-memcpy-arch.h @@ -0,0 +1,12 @@ + +#ifdef ARCH_X86_64 + +#define MEMCPY_FN(fn, name, desc) \ + extern void *fn(void *, const void *, size_t); + +#include "mem-memcpy-x86-64-asm-def.h" + +#undef MEMCPY_FN + +#endif + diff --git a/tools/perf/bench/mem-memcpy-x86-64-asm-def.h b/tools/perf/bench/mem-memcpy-x86-64-asm-def.h new file mode 100644 index 00000000..d588b876 --- /dev/null +++ b/tools/perf/bench/mem-memcpy-x86-64-asm-def.h @@ -0,0 +1,4 @@ + +MEMCPY_FN(__memcpy, + "x86-64-unrolled", + "unrolled memcpy() in arch/x86/lib/memcpy_64.S") diff --git a/tools/perf/bench/mem-memcpy-x86-64-asm.S b/tools/perf/bench/mem-memcpy-x86-64-asm.S new file mode 100644 index 00000000..185a96d6 --- /dev/null +++ b/tools/perf/bench/mem-memcpy-x86-64-asm.S @@ -0,0 +1,8 @@ + +#include "../../../arch/x86/lib/memcpy_64.S" +/* + * We need to provide note.GNU-stack section, saying that we want + * NOT executable stack. Otherwise the final linking will assume that + * the ELF stack should not be restricted at all and set it RWX. + */ +.section .note.GNU-stack,"",@progbits diff --git a/tools/perf/bench/mem-memcpy.c b/tools/perf/bench/mem-memcpy.c new file mode 100644 index 00000000..db82021f --- /dev/null +++ b/tools/perf/bench/mem-memcpy.c @@ -0,0 +1,297 @@ +/* + * mem-memcpy.c + * + * memcpy: Simple memory copy in various ways + * + * Written by Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> + */ +#include <ctype.h> + +#include "../perf.h" +#include "../util/util.h" +#include "../util/parse-options.h" +#include "../util/header.h" +#include "bench.h" +#include "mem-memcpy-arch.h" + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <sys/time.h> +#include <errno.h> + +#define K 1024 + +static const char *length_str = "1MB"; +static const char *routine = "default"; +static bool use_clock; +static int clock_fd; +static bool only_prefault; +static bool no_prefault; + +static const struct option options[] = { + OPT_STRING('l', "length", &length_str, "1MB", + "Specify length of memory to copy. " + "available unit: B, MB, GB (upper and lower)"), + OPT_STRING('r', "routine", &routine, "default", + "Specify routine to copy"), + OPT_BOOLEAN('c', "clock", &use_clock, + "Use CPU clock for measuring"), + OPT_BOOLEAN('o', "only-prefault", &only_prefault, + "Show only the result with page faults before memcpy()"), + OPT_BOOLEAN('n', "no-prefault", &no_prefault, + "Show only the result without page faults before memcpy()"), + OPT_END() +}; + +typedef void *(*memcpy_t)(void *, const void *, size_t); + +struct routine { + const char *name; + const char *desc; + memcpy_t fn; +}; + +struct routine routines[] = { + { "default", + "Default memcpy() provided by glibc", + memcpy }, +#ifdef ARCH_X86_64 + +#define MEMCPY_FN(fn, name, desc) { name, desc, fn }, +#include "mem-memcpy-x86-64-asm-def.h" +#undef MEMCPY_FN + +#endif + + { NULL, + NULL, + NULL } +}; + +static const char * const bench_mem_memcpy_usage[] = { + "perf bench mem memcpy <options>", + NULL +}; + +static struct perf_event_attr clock_attr = { + .type = PERF_TYPE_HARDWARE, + .config = PERF_COUNT_HW_CPU_CYCLES +}; + +static void init_clock(void) +{ + clock_fd = sys_perf_event_open(&clock_attr, getpid(), -1, -1, 0); + + if (clock_fd < 0 && errno == ENOSYS) + die("No CONFIG_PERF_EVENTS=y kernel support configured?\n"); + else + BUG_ON(clock_fd < 0); +} + +static u64 get_clock(void) +{ + int ret; + u64 clk; + + ret = read(clock_fd, &clk, sizeof(u64)); + BUG_ON(ret != sizeof(u64)); + + return clk; +} + +static double timeval2double(struct timeval *ts) +{ + return (double)ts->tv_sec + + (double)ts->tv_usec / (double)1000000; +} + +static void alloc_mem(void **dst, void **src, size_t length) +{ + *dst = zalloc(length); + if (!dst) + die("memory allocation failed - maybe length is too large?\n"); + + *src = zalloc(length); + if (!src) + die("memory allocation failed - maybe length is too large?\n"); +} + +static u64 do_memcpy_clock(memcpy_t fn, size_t len, bool prefault) +{ + u64 clock_start = 0ULL, clock_end = 0ULL; + void *src = NULL, *dst = NULL; + + alloc_mem(&src, &dst, len); + + if (prefault) + fn(dst, src, len); + + clock_start = get_clock(); + fn(dst, src, len); + clock_end = get_clock(); + + free(src); + free(dst); + return clock_end - clock_start; +} + +static double do_memcpy_gettimeofday(memcpy_t fn, size_t len, bool prefault) +{ + struct timeval tv_start, tv_end, tv_diff; + void *src = NULL, *dst = NULL; + + alloc_mem(&src, &dst, len); + + if (prefault) + fn(dst, src, len); + + BUG_ON(gettimeofday(&tv_start, NULL)); + fn(dst, src, len); + BUG_ON(gettimeofday(&tv_end, NULL)); + + timersub(&tv_end, &tv_start, &tv_diff); + + free(src); + free(dst); + return (double)((double)len / timeval2double(&tv_diff)); +} + +#define pf (no_prefault ? 0 : 1) + +#define print_bps(x) do { \ + if (x < K) \ + printf(" %14lf B/Sec", x); \ + else if (x < K * K) \ + printf(" %14lfd KB/Sec", x / K); \ + else if (x < K * K * K) \ + printf(" %14lf MB/Sec", x / K / K); \ + else \ + printf(" %14lf GB/Sec", x / K / K / K); \ + } while (0) + +int bench_mem_memcpy(int argc, const char **argv, + const char *prefix __used) +{ + int i; + size_t len; + double result_bps[2]; + u64 result_clock[2]; + + argc = parse_options(argc, argv, options, + bench_mem_memcpy_usage, 0); + + if (use_clock) + init_clock(); + + len = (size_t)perf_atoll((char *)length_str); + + result_clock[0] = result_clock[1] = 0ULL; + result_bps[0] = result_bps[1] = 0.0; + + if ((s64)len <= 0) { + fprintf(stderr, "Invalid length:%s\n", length_str); + return 1; + } + + /* same to without specifying either of prefault and no-prefault */ + if (only_prefault && no_prefault) + only_prefault = no_prefault = false; + + for (i = 0; routines[i].name; i++) { + if (!strcmp(routines[i].name, routine)) + break; + } + if (!routines[i].name) { + printf("Unknown routine:%s\n", routine); + printf("Available routines...\n"); + for (i = 0; routines[i].name; i++) { + printf("\t%s ... %s\n", + routines[i].name, routines[i].desc); + } + return 1; + } + + if (bench_format == BENCH_FORMAT_DEFAULT) + printf("# Copying %s Bytes ...\n\n", length_str); + + if (!only_prefault && !no_prefault) { + /* show both of results */ + if (use_clock) { + result_clock[0] = + do_memcpy_clock(routines[i].fn, len, false); + result_clock[1] = + do_memcpy_clock(routines[i].fn, len, true); + } else { + result_bps[0] = + do_memcpy_gettimeofday(routines[i].fn, + len, false); + result_bps[1] = + do_memcpy_gettimeofday(routines[i].fn, + len, true); + } + } else { + if (use_clock) { + result_clock[pf] = + do_memcpy_clock(routines[i].fn, + len, only_prefault); + } else { + result_bps[pf] = + do_memcpy_gettimeofday(routines[i].fn, + len, only_prefault); + } + } + + switch (bench_format) { + case BENCH_FORMAT_DEFAULT: + if (!only_prefault && !no_prefault) { + if (use_clock) { + printf(" %14lf Clock/Byte\n", + (double)result_clock[0] + / (double)len); + printf(" %14lf Clock/Byte (with prefault)\n", + (double)result_clock[1] + / (double)len); + } else { + print_bps(result_bps[0]); + printf("\n"); + print_bps(result_bps[1]); + printf(" (with prefault)\n"); + } + } else { + if (use_clock) { + printf(" %14lf Clock/Byte", + (double)result_clock[pf] + / (double)len); + } else + print_bps(result_bps[pf]); + + printf("%s\n", only_prefault ? " (with prefault)" : ""); + } + break; + case BENCH_FORMAT_SIMPLE: + if (!only_prefault && !no_prefault) { + if (use_clock) { + printf("%lf %lf\n", + (double)result_clock[0] / (double)len, + (double)result_clock[1] / (double)len); + } else { + printf("%lf %lf\n", + result_bps[0], result_bps[1]); + } + } else { + if (use_clock) { + printf("%lf\n", (double)result_clock[pf] + / (double)len); + } else + printf("%lf\n", result_bps[pf]); + } + break; + default: + /* reaching this means there's some disaster: */ + die("unknown format: %d\n", bench_format); + break; + } + + return 0; +} diff --git a/tools/perf/bench/sched-messaging.c b/tools/perf/bench/sched-messaging.c new file mode 100644 index 00000000..d1d1b30f --- /dev/null +++ b/tools/perf/bench/sched-messaging.c @@ -0,0 +1,336 @@ +/* + * + * sched-messaging.c + * + * messaging: Benchmark for scheduler and IPC mechanisms + * + * Based on hackbench by Rusty Russell <rusty@rustcorp.com.au> + * Ported to perf by Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> + * + */ + +#include "../perf.h" +#include "../util/util.h" +#include "../util/parse-options.h" +#include "../builtin.h" +#include "bench.h" + +/* Test groups of 20 processes spraying to 20 receivers */ +#include <pthread.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <errno.h> +#include <unistd.h> +#include <sys/types.h> +#include <sys/socket.h> +#include <sys/wait.h> +#include <sys/time.h> +#include <sys/poll.h> +#include <limits.h> + +#define DATASIZE 100 + +static bool use_pipes = false; +static unsigned int loops = 100; +static bool thread_mode = false; +static unsigned int num_groups = 10; + +struct sender_context { + unsigned int num_fds; + int ready_out; + int wakefd; + int out_fds[0]; +}; + +struct receiver_context { + unsigned int num_packets; + int in_fds[2]; + int ready_out; + int wakefd; +}; + +static void barf(const char *msg) +{ + fprintf(stderr, "%s (error: %s)\n", msg, strerror(errno)); + exit(1); +} + +static void fdpair(int fds[2]) +{ + if (use_pipes) { + if (pipe(fds) == 0) + return; + } else { + if (socketpair(AF_UNIX, SOCK_STREAM, 0, fds) == 0) + return; + } + + barf(use_pipes ? "pipe()" : "socketpair()"); +} + +/* Block until we're ready to go */ +static void ready(int ready_out, int wakefd) +{ + char dummy; + struct pollfd pollfd = { .fd = wakefd, .events = POLLIN }; + + /* Tell them we're ready. */ + if (write(ready_out, &dummy, 1) != 1) + barf("CLIENT: ready write"); + + /* Wait for "GO" signal */ + if (poll(&pollfd, 1, -1) != 1) + barf("poll"); +} + +/* Sender sprays loops messages down each file descriptor */ +static void *sender(struct sender_context *ctx) +{ + char data[DATASIZE]; + unsigned int i, j; + + ready(ctx->ready_out, ctx->wakefd); + + /* Now pump to every receiver. */ + for (i = 0; i < loops; i++) { + for (j = 0; j < ctx->num_fds; j++) { + int ret, done = 0; + +again: + ret = write(ctx->out_fds[j], data + done, + sizeof(data)-done); + if (ret < 0) + barf("SENDER: write"); + done += ret; + if (done < DATASIZE) + goto again; + } + } + + return NULL; +} + + +/* One receiver per fd */ +static void *receiver(struct receiver_context* ctx) +{ + unsigned int i; + + if (!thread_mode) + close(ctx->in_fds[1]); + + /* Wait for start... */ + ready(ctx->ready_out, ctx->wakefd); + + /* Receive them all */ + for (i = 0; i < ctx->num_packets; i++) { + char data[DATASIZE]; + int ret, done = 0; + +again: + ret = read(ctx->in_fds[0], data + done, DATASIZE - done); + if (ret < 0) + barf("SERVER: read"); + done += ret; + if (done < DATASIZE) + goto again; + } + + return NULL; +} + +static pthread_t create_worker(void *ctx, void *(*func)(void *)) +{ + pthread_attr_t attr; + pthread_t childid; + int err; + + if (!thread_mode) { + /* process mode */ + /* Fork the receiver. */ + switch (fork()) { + case -1: + barf("fork()"); + break; + case 0: + (*func) (ctx); + exit(0); + break; + default: + break; + } + + return (pthread_t)0; + } + + if (pthread_attr_init(&attr) != 0) + barf("pthread_attr_init:"); + +#ifndef __ia64__ + if (pthread_attr_setstacksize(&attr, PTHREAD_STACK_MIN) != 0) + barf("pthread_attr_setstacksize"); +#endif + + err = pthread_create(&childid, &attr, func, ctx); + if (err != 0) { + fprintf(stderr, "pthread_create failed: %s (%d)\n", + strerror(err), err); + exit(-1); + } + return childid; +} + +static void reap_worker(pthread_t id) +{ + int proc_status; + void *thread_status; + + if (!thread_mode) { + /* process mode */ + wait(&proc_status); + if (!WIFEXITED(proc_status)) + exit(1); + } else { + pthread_join(id, &thread_status); + } +} + +/* One group of senders and receivers */ +static unsigned int group(pthread_t *pth, + unsigned int num_fds, + int ready_out, + int wakefd) +{ + unsigned int i; + struct sender_context *snd_ctx = malloc(sizeof(struct sender_context) + + num_fds * sizeof(int)); + + if (!snd_ctx) + barf("malloc()"); + + for (i = 0; i < num_fds; i++) { + int fds[2]; + struct receiver_context *ctx = malloc(sizeof(*ctx)); + + if (!ctx) + barf("malloc()"); + + + /* Create the pipe between client and server */ + fdpair(fds); + + ctx->num_packets = num_fds * loops; + ctx->in_fds[0] = fds[0]; + ctx->in_fds[1] = fds[1]; + ctx->ready_out = ready_out; + ctx->wakefd = wakefd; + + pth[i] = create_worker(ctx, (void *)receiver); + + snd_ctx->out_fds[i] = fds[1]; + if (!thread_mode) + close(fds[0]); + } + + /* Now we have all the fds, fork the senders */ + for (i = 0; i < num_fds; i++) { + snd_ctx->ready_out = ready_out; + snd_ctx->wakefd = wakefd; + snd_ctx->num_fds = num_fds; + + pth[num_fds+i] = create_worker(snd_ctx, (void *)sender); + } + + /* Close the fds we have left */ + if (!thread_mode) + for (i = 0; i < num_fds; i++) + close(snd_ctx->out_fds[i]); + + /* Return number of children to reap */ + return num_fds * 2; +} + +static const struct option options[] = { + OPT_BOOLEAN('p', "pipe", &use_pipes, + "Use pipe() instead of socketpair()"), + OPT_BOOLEAN('t', "thread", &thread_mode, + "Be multi thread instead of multi process"), + OPT_UINTEGER('g', "group", &num_groups, "Specify number of groups"), + OPT_UINTEGER('l', "loop", &loops, "Specify number of loops"), + OPT_END() +}; + +static const char * const bench_sched_message_usage[] = { + "perf bench sched messaging <options>", + NULL +}; + +int bench_sched_messaging(int argc, const char **argv, + const char *prefix __used) +{ + unsigned int i, total_children; + struct timeval start, stop, diff; + unsigned int num_fds = 20; + int readyfds[2], wakefds[2]; + char dummy; + pthread_t *pth_tab; + + argc = parse_options(argc, argv, options, + bench_sched_message_usage, 0); + + pth_tab = malloc(num_fds * 2 * num_groups * sizeof(pthread_t)); + if (!pth_tab) + barf("main:malloc()"); + + fdpair(readyfds); + fdpair(wakefds); + + total_children = 0; + for (i = 0; i < num_groups; i++) + total_children += group(pth_tab+total_children, num_fds, + readyfds[1], wakefds[0]); + + /* Wait for everyone to be ready */ + for (i = 0; i < total_children; i++) + if (read(readyfds[0], &dummy, 1) != 1) + barf("Reading for readyfds"); + + gettimeofday(&start, NULL); + + /* Kick them off */ + if (write(wakefds[1], &dummy, 1) != 1) + barf("Writing to start them"); + + /* Reap them all */ + for (i = 0; i < total_children; i++) + reap_worker(pth_tab[i]); + + gettimeofday(&stop, NULL); + + timersub(&stop, &start, &diff); + + switch (bench_format) { + case BENCH_FORMAT_DEFAULT: + printf("# %d sender and receiver %s per group\n", + num_fds, thread_mode ? "threads" : "processes"); + printf("# %d groups == %d %s run\n\n", + num_groups, num_groups * 2 * num_fds, + thread_mode ? "threads" : "processes"); + printf(" %14s: %lu.%03lu [sec]\n", "Total time", + diff.tv_sec, + (unsigned long) (diff.tv_usec/1000)); + break; + case BENCH_FORMAT_SIMPLE: + printf("%lu.%03lu\n", diff.tv_sec, + (unsigned long) (diff.tv_usec/1000)); + break; + default: + /* reaching here is something disaster */ + fprintf(stderr, "Unknown format:%d\n", bench_format); + exit(1); + break; + } + + return 0; +} diff --git a/tools/perf/bench/sched-pipe.c b/tools/perf/bench/sched-pipe.c new file mode 100644 index 00000000..0c7454f8 --- /dev/null +++ b/tools/perf/bench/sched-pipe.c @@ -0,0 +1,127 @@ +/* + * + * sched-pipe.c + * + * pipe: Benchmark for pipe() + * + * Based on pipe-test-1m.c by Ingo Molnar <mingo@redhat.com> + * http://people.redhat.com/mingo/cfs-scheduler/tools/pipe-test-1m.c + * Ported to perf by Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> + * + */ + +#include "../perf.h" +#include "../util/util.h" +#include "../util/parse-options.h" +#include "../builtin.h" +#include "bench.h" + +#include <unistd.h> +#include <stdio.h> +#include <stdlib.h> +#include <signal.h> +#include <sys/wait.h> +#include <linux/unistd.h> +#include <string.h> +#include <errno.h> +#include <assert.h> +#include <sys/time.h> +#include <sys/types.h> + +#define LOOPS_DEFAULT 1000000 +static int loops = LOOPS_DEFAULT; + +static const struct option options[] = { + OPT_INTEGER('l', "loop", &loops, + "Specify number of loops"), + OPT_END() +}; + +static const char * const bench_sched_pipe_usage[] = { + "perf bench sched pipe <options>", + NULL +}; + +int bench_sched_pipe(int argc, const char **argv, + const char *prefix __used) +{ + int pipe_1[2], pipe_2[2]; + int m = 0, i; + struct timeval start, stop, diff; + unsigned long long result_usec = 0; + + /* + * why does "ret" exist? + * discarding returned value of read(), write() + * causes error in building environment for perf + */ + int __used ret, wait_stat; + pid_t pid, retpid; + + argc = parse_options(argc, argv, options, + bench_sched_pipe_usage, 0); + + assert(!pipe(pipe_1)); + assert(!pipe(pipe_2)); + + pid = fork(); + assert(pid >= 0); + + gettimeofday(&start, NULL); + + if (!pid) { + for (i = 0; i < loops; i++) { + ret = read(pipe_1[0], &m, sizeof(int)); + ret = write(pipe_2[1], &m, sizeof(int)); + } + } else { + for (i = 0; i < loops; i++) { + ret = write(pipe_1[1], &m, sizeof(int)); + ret = read(pipe_2[0], &m, sizeof(int)); + } + } + + gettimeofday(&stop, NULL); + timersub(&stop, &start, &diff); + + if (pid) { + retpid = waitpid(pid, &wait_stat, 0); + assert((retpid == pid) && WIFEXITED(wait_stat)); + } else { + exit(0); + } + + switch (bench_format) { + case BENCH_FORMAT_DEFAULT: + printf("# Executed %d pipe operations between two tasks\n\n", + loops); + + result_usec = diff.tv_sec * 1000000; + result_usec += diff.tv_usec; + + printf(" %14s: %lu.%03lu [sec]\n\n", "Total time", + diff.tv_sec, + (unsigned long) (diff.tv_usec/1000)); + + printf(" %14lf usecs/op\n", + (double)result_usec / (double)loops); + printf(" %14d ops/sec\n", + (int)((double)loops / + ((double)result_usec / (double)1000000))); + break; + + case BENCH_FORMAT_SIMPLE: + printf("%lu.%03lu\n", + diff.tv_sec, + (unsigned long) (diff.tv_usec / 1000)); + break; + + default: + /* reaching here is something disaster */ + fprintf(stderr, "Unknown format:%d\n", bench_format); + exit(1); + break; + } + + return 0; +} diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c new file mode 100644 index 00000000..7b139e1e --- /dev/null +++ b/tools/perf/builtin-annotate.c @@ -0,0 +1,294 @@ +/* + * builtin-annotate.c + * + * Builtin annotate command: Analyze the perf.data input file, + * look up and read DSOs and symbol information and display + * a histogram of results, along various sorting keys. + */ +#include "builtin.h" + +#include "util/util.h" +#include "util/color.h" +#include <linux/list.h> +#include "util/cache.h" +#include <linux/rbtree.h> +#include "util/symbol.h" + +#include "perf.h" +#include "util/debug.h" + +#include "util/evlist.h" +#include "util/evsel.h" +#include "util/annotate.h" +#include "util/event.h" +#include "util/parse-options.h" +#include "util/parse-events.h" +#include "util/thread.h" +#include "util/sort.h" +#include "util/hist.h" +#include "util/session.h" + +static char const *input_name = "perf.data"; + +static bool force, use_tui, use_stdio; + +static bool full_paths; + +static bool print_line; + +static const char *sym_hist_filter; + +static int perf_evlist__add_sample(struct perf_evlist *evlist, + struct perf_sample *sample, + struct perf_evsel *evsel, + struct addr_location *al) +{ + struct hist_entry *he; + int ret; + + if (sym_hist_filter != NULL && + (al->sym == NULL || strcmp(sym_hist_filter, al->sym->name) != 0)) { + /* We're only interested in a symbol named sym_hist_filter */ + if (al->sym != NULL) { + rb_erase(&al->sym->rb_node, + &al->map->dso->symbols[al->map->type]); + symbol__delete(al->sym); + } + return 0; + } + + he = __hists__add_entry(&evsel->hists, al, NULL, 1); + if (he == NULL) + return -ENOMEM; + + ret = 0; + if (he->ms.sym != NULL) { + struct annotation *notes = symbol__annotation(he->ms.sym); + if (notes->src == NULL && + symbol__alloc_hist(he->ms.sym, evlist->nr_entries) < 0) + return -ENOMEM; + + ret = hist_entry__inc_addr_samples(he, evsel->idx, al->addr); + } + + evsel->hists.stats.total_period += sample->period; + hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE); + return ret; +} + +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel, + struct perf_session *session) +{ + struct addr_location al; + + if (perf_event__preprocess_sample(event, session, &al, sample, + symbol__annotate_init) < 0) { + pr_warning("problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + if (!al.filtered && + perf_evlist__add_sample(session->evlist, sample, evsel, &al)) { + pr_warning("problem incrementing symbol count, " + "skipping event\n"); + return -1; + } + + return 0; +} + +static int hist_entry__tty_annotate(struct hist_entry *he, int evidx) +{ + return symbol__tty_annotate(he->ms.sym, he->ms.map, evidx, + print_line, full_paths, 0, 0); +} + +static void hists__find_annotations(struct hists *self, int evidx) +{ + struct rb_node *nd = rb_first(&self->entries), *next; + int key = KEY_RIGHT; + + while (nd) { + struct hist_entry *he = rb_entry(nd, struct hist_entry, rb_node); + struct annotation *notes; + + if (he->ms.sym == NULL || he->ms.map->dso->annotate_warned) + goto find_next; + + notes = symbol__annotation(he->ms.sym); + if (notes->src == NULL) { +find_next: + if (key == KEY_LEFT) + nd = rb_prev(nd); + else + nd = rb_next(nd); + continue; + } + + if (use_browser > 0) { + key = hist_entry__tui_annotate(he, evidx); + switch (key) { + case KEY_RIGHT: + next = rb_next(nd); + break; + case KEY_LEFT: + next = rb_prev(nd); + break; + default: + return; + } + + if (next != NULL) + nd = next; + } else { + hist_entry__tty_annotate(he, evidx); + nd = rb_next(nd); + /* + * Since we have a hist_entry per IP for the same + * symbol, free he->ms.sym->src to signal we already + * processed this symbol. + */ + free(notes->src); + notes->src = NULL; + } + } +} + +static struct perf_event_ops event_ops = { + .sample = process_sample_event, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .fork = perf_event__process_task, + .ordered_samples = true, + .ordering_requires_timestamps = true, +}; + +static int __cmd_annotate(void) +{ + int ret; + struct perf_session *session; + struct perf_evsel *pos; + u64 total_nr_samples; + + session = perf_session__new(input_name, O_RDONLY, force, false, &event_ops); + if (session == NULL) + return -ENOMEM; + + ret = perf_session__process_events(session, &event_ops); + if (ret) + goto out_delete; + + if (dump_trace) { + perf_session__fprintf_nr_events(session, stdout); + goto out_delete; + } + + if (verbose > 3) + perf_session__fprintf(session, stdout); + + if (verbose > 2) + perf_session__fprintf_dsos(session, stdout); + + total_nr_samples = 0; + list_for_each_entry(pos, &session->evlist->entries, node) { + struct hists *hists = &pos->hists; + u32 nr_samples = hists->stats.nr_events[PERF_RECORD_SAMPLE]; + + if (nr_samples > 0) { + total_nr_samples += nr_samples; + hists__collapse_resort(hists); + hists__output_resort(hists); + hists__find_annotations(hists, pos->idx); + } + } + + if (total_nr_samples == 0) { + ui__warning("The %s file has no samples!\n", input_name); + goto out_delete; + } +out_delete: + /* + * Speed up the exit process, for large files this can + * take quite a while. + * + * XXX Enable this when using valgrind or if we ever + * librarize this command. + * + * Also experiment with obstacks to see how much speed + * up we'll get here. + * + * perf_session__delete(session); + */ + return ret; +} + +static const char * const annotate_usage[] = { + "perf annotate [<options>] <command>", + NULL +}; + +static const struct option options[] = { + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]", + "only consider symbols in these dsos"), + OPT_STRING('s', "symbol", &sym_hist_filter, "symbol", + "symbol to annotate"), + OPT_BOOLEAN('f', "force", &force, "don't complain, do it"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_BOOLEAN(0, "tui", &use_tui, "Use the TUI interface"), + OPT_BOOLEAN(0, "stdio", &use_stdio, "Use the stdio interface"), + OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name, + "file", "vmlinux pathname"), + OPT_BOOLEAN('m', "modules", &symbol_conf.use_modules, + "load module symbols - WARNING: use only with -k and LIVE kernel"), + OPT_BOOLEAN('l', "print-line", &print_line, + "print matching source lines (may be slow)"), + OPT_BOOLEAN('P', "full-paths", &full_paths, + "Don't shorten the displayed pathnames"), + OPT_END() +}; + +int cmd_annotate(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, options, annotate_usage, 0); + + if (use_stdio) + use_browser = 0; + else if (use_tui) + use_browser = 1; + + setup_browser(true); + + symbol_conf.priv_size = sizeof(struct annotation); + symbol_conf.try_vmlinux_path = true; + + if (symbol__init() < 0) + return -1; + + setup_sorting(annotate_usage, options); + + if (argc) { + /* + * Special case: if there's an argument left then assume tha + * it's a symbol filter: + */ + if (argc > 1) + usage_with_options(annotate_usage, options); + + sym_hist_filter = argv[0]; + } + + if (field_sep && *field_sep == '.') { + pr_err("'.' is the only non valid --field-separator argument\n"); + return -1; + } + + return __cmd_annotate(); +} diff --git a/tools/perf/builtin-bench.c b/tools/perf/builtin-bench.c new file mode 100644 index 00000000..fcb96269 --- /dev/null +++ b/tools/perf/builtin-bench.c @@ -0,0 +1,245 @@ +/* + * + * builtin-bench.c + * + * General benchmarking subsystem provided by perf + * + * Copyright (C) 2009, Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> + * + */ + +/* + * + * Available subsystem list: + * sched ... scheduler and IPC mechanism + * mem ... memory access performance + * + */ + +#include "perf.h" +#include "util/util.h" +#include "util/parse-options.h" +#include "builtin.h" +#include "bench/bench.h" + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +struct bench_suite { + const char *name; + const char *summary; + int (*fn)(int, const char **, const char *); +}; + \ +/* sentinel: easy for help */ +#define suite_all { "all", "test all suite (pseudo suite)", NULL } + +static struct bench_suite sched_suites[] = { + { "messaging", + "Benchmark for scheduler and IPC mechanisms", + bench_sched_messaging }, + { "pipe", + "Flood of communication over pipe() between two processes", + bench_sched_pipe }, + suite_all, + { NULL, + NULL, + NULL } +}; + +static struct bench_suite mem_suites[] = { + { "memcpy", + "Simple memory copy in various ways", + bench_mem_memcpy }, + suite_all, + { NULL, + NULL, + NULL } +}; + +struct bench_subsys { + const char *name; + const char *summary; + struct bench_suite *suites; +}; + +static struct bench_subsys subsystems[] = { + { "sched", + "scheduler and IPC mechanism", + sched_suites }, + { "mem", + "memory access performance", + mem_suites }, + { "all", /* sentinel: easy for help */ + "test all subsystem (pseudo subsystem)", + NULL }, + { NULL, + NULL, + NULL } +}; + +static void dump_suites(int subsys_index) +{ + int i; + + printf("# List of available suites for %s...\n\n", + subsystems[subsys_index].name); + + for (i = 0; subsystems[subsys_index].suites[i].name; i++) + printf("%14s: %s\n", + subsystems[subsys_index].suites[i].name, + subsystems[subsys_index].suites[i].summary); + + printf("\n"); + return; +} + +static const char *bench_format_str; +int bench_format = BENCH_FORMAT_DEFAULT; + +static const struct option bench_options[] = { + OPT_STRING('f', "format", &bench_format_str, "default", + "Specify format style"), + OPT_END() +}; + +static const char * const bench_usage[] = { + "perf bench [<common options>] <subsystem> <suite> [<options>]", + NULL +}; + +static void print_usage(void) +{ + int i; + + printf("Usage: \n"); + for (i = 0; bench_usage[i]; i++) + printf("\t%s\n", bench_usage[i]); + printf("\n"); + + printf("# List of available subsystems...\n\n"); + + for (i = 0; subsystems[i].name; i++) + printf("%14s: %s\n", + subsystems[i].name, subsystems[i].summary); + printf("\n"); +} + +static int bench_str2int(const char *str) +{ + if (!str) + return BENCH_FORMAT_DEFAULT; + + if (!strcmp(str, BENCH_FORMAT_DEFAULT_STR)) + return BENCH_FORMAT_DEFAULT; + else if (!strcmp(str, BENCH_FORMAT_SIMPLE_STR)) + return BENCH_FORMAT_SIMPLE; + + return BENCH_FORMAT_UNKNOWN; +} + +static void all_suite(struct bench_subsys *subsys) /* FROM HERE */ +{ + int i; + const char *argv[2]; + struct bench_suite *suites = subsys->suites; + + argv[1] = NULL; + /* + * TODO: + * preparing preset parameters for + * embedded, ordinary PC, HPC, etc... + * will be helpful + */ + for (i = 0; suites[i].fn; i++) { + printf("# Running %s/%s benchmark...\n", + subsys->name, + suites[i].name); + + argv[1] = suites[i].name; + suites[i].fn(1, argv, NULL); + printf("\n"); + } +} + +static void all_subsystem(void) +{ + int i; + for (i = 0; subsystems[i].suites; i++) + all_suite(&subsystems[i]); +} + +int cmd_bench(int argc, const char **argv, const char *prefix __used) +{ + int i, j, status = 0; + + if (argc < 2) { + /* No subsystem specified. */ + print_usage(); + goto end; + } + + argc = parse_options(argc, argv, bench_options, bench_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + + bench_format = bench_str2int(bench_format_str); + if (bench_format == BENCH_FORMAT_UNKNOWN) { + printf("Unknown format descriptor:%s\n", bench_format_str); + goto end; + } + + if (argc < 1) { + print_usage(); + goto end; + } + + if (!strcmp(argv[0], "all")) { + all_subsystem(); + goto end; + } + + for (i = 0; subsystems[i].name; i++) { + if (strcmp(subsystems[i].name, argv[0])) + continue; + + if (argc < 2) { + /* No suite specified. */ + dump_suites(i); + goto end; + } + + if (!strcmp(argv[1], "all")) { + all_suite(&subsystems[i]); + goto end; + } + + for (j = 0; subsystems[i].suites[j].name; j++) { + if (strcmp(subsystems[i].suites[j].name, argv[1])) + continue; + + if (bench_format == BENCH_FORMAT_DEFAULT) + printf("# Running %s/%s benchmark...\n", + subsystems[i].name, + subsystems[i].suites[j].name); + status = subsystems[i].suites[j].fn(argc - 1, + argv + 1, prefix); + goto end; + } + + if (!strcmp(argv[1], "-h") || !strcmp(argv[1], "--help")) { + dump_suites(i); + goto end; + } + + printf("Unknown suite:%s for %s\n", argv[1], argv[0]); + status = 1; + goto end; + } + + printf("Unknown subsystem:%s\n", argv[0]); + status = 1; + +end: + return status; +} diff --git a/tools/perf/builtin-buildid-cache.c b/tools/perf/builtin-buildid-cache.c new file mode 100644 index 00000000..29ad20e6 --- /dev/null +++ b/tools/perf/builtin-buildid-cache.c @@ -0,0 +1,132 @@ +/* + * builtin-buildid-cache.c + * + * Builtin buildid-cache command: Manages build-id cache + * + * Copyright (C) 2010, Red Hat Inc. + * Copyright (C) 2010, Arnaldo Carvalho de Melo <acme@redhat.com> + */ +#include "builtin.h" +#include "perf.h" +#include "util/cache.h" +#include "util/debug.h" +#include "util/header.h" +#include "util/parse-options.h" +#include "util/strlist.h" +#include "util/symbol.h" + +static char const *add_name_list_str, *remove_name_list_str; + +static const char * const buildid_cache_usage[] = { + "perf buildid-cache [<options>]", + NULL +}; + +static const struct option buildid_cache_options[] = { + OPT_STRING('a', "add", &add_name_list_str, + "file list", "file(s) to add"), + OPT_STRING('r', "remove", &remove_name_list_str, "file list", + "file(s) to remove"), + OPT_INCR('v', "verbose", &verbose, "be more verbose"), + OPT_END() +}; + +static int build_id_cache__add_file(const char *filename, const char *debugdir) +{ + char sbuild_id[BUILD_ID_SIZE * 2 + 1]; + u8 build_id[BUILD_ID_SIZE]; + int err; + + if (filename__read_build_id(filename, &build_id, sizeof(build_id)) < 0) { + pr_debug("Couldn't read a build-id in %s\n", filename); + return -1; + } + + build_id__sprintf(build_id, sizeof(build_id), sbuild_id); + err = build_id_cache__add_s(sbuild_id, debugdir, filename, false); + if (verbose) + pr_info("Adding %s %s: %s\n", sbuild_id, filename, + err ? "FAIL" : "Ok"); + return err; +} + +static int build_id_cache__remove_file(const char *filename __used, + const char *debugdir __used) +{ + u8 build_id[BUILD_ID_SIZE]; + char sbuild_id[BUILD_ID_SIZE * 2 + 1]; + + int err; + + if (filename__read_build_id(filename, &build_id, sizeof(build_id)) < 0) { + pr_debug("Couldn't read a build-id in %s\n", filename); + return -1; + } + + build_id__sprintf(build_id, sizeof(build_id), sbuild_id); + err = build_id_cache__remove_s(sbuild_id, debugdir); + if (verbose) + pr_info("Removing %s %s: %s\n", sbuild_id, filename, + err ? "FAIL" : "Ok"); + + return err; +} + +static int __cmd_buildid_cache(void) +{ + struct strlist *list; + struct str_node *pos; + char debugdir[PATH_MAX]; + + snprintf(debugdir, sizeof(debugdir), "%s", buildid_dir); + + if (add_name_list_str) { + list = strlist__new(true, add_name_list_str); + if (list) { + strlist__for_each(pos, list) + if (build_id_cache__add_file(pos->s, debugdir)) { + if (errno == EEXIST) { + pr_debug("%s already in the cache\n", + pos->s); + continue; + } + pr_warning("Couldn't add %s: %s\n", + pos->s, strerror(errno)); + } + + strlist__delete(list); + } + } + + if (remove_name_list_str) { + list = strlist__new(true, remove_name_list_str); + if (list) { + strlist__for_each(pos, list) + if (build_id_cache__remove_file(pos->s, debugdir)) { + if (errno == ENOENT) { + pr_debug("%s wasn't in the cache\n", + pos->s); + continue; + } + pr_warning("Couldn't remove %s: %s\n", + pos->s, strerror(errno)); + } + + strlist__delete(list); + } + } + + return 0; +} + +int cmd_buildid_cache(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, buildid_cache_options, + buildid_cache_usage, 0); + + if (symbol__init() < 0) + return -1; + + setup_pager(); + return __cmd_buildid_cache(); +} diff --git a/tools/perf/builtin-buildid-list.c b/tools/perf/builtin-buildid-list.c new file mode 100644 index 00000000..5af32ae9 --- /dev/null +++ b/tools/perf/builtin-buildid-list.c @@ -0,0 +1,60 @@ +/* + * builtin-buildid-list.c + * + * Builtin buildid-list command: list buildids in perf.data + * + * Copyright (C) 2009, Red Hat Inc. + * Copyright (C) 2009, Arnaldo Carvalho de Melo <acme@redhat.com> + */ +#include "builtin.h" +#include "perf.h" +#include "util/build-id.h" +#include "util/cache.h" +#include "util/debug.h" +#include "util/parse-options.h" +#include "util/session.h" +#include "util/symbol.h" + +static char const *input_name = "perf.data"; +static bool force; +static bool with_hits; + +static const char * const buildid_list_usage[] = { + "perf buildid-list [<options>]", + NULL +}; + +static const struct option options[] = { + OPT_BOOLEAN('H', "with-hits", &with_hits, "Show only DSOs with hits"), + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_BOOLEAN('f', "force", &force, "don't complain, do it"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose"), + OPT_END() +}; + +static int __cmd_buildid_list(void) +{ + struct perf_session *session; + + session = perf_session__new(input_name, O_RDONLY, force, false, + &build_id__mark_dso_hit_ops); + if (session == NULL) + return -1; + + if (with_hits) + perf_session__process_events(session, &build_id__mark_dso_hit_ops); + + perf_session__fprintf_dsos_buildid(session, stdout, with_hits); + + perf_session__delete(session); + return 0; +} + +int cmd_buildid_list(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, options, buildid_list_usage, 0); + setup_pager(); + return __cmd_buildid_list(); +} diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c new file mode 100644 index 00000000..e8219990 --- /dev/null +++ b/tools/perf/builtin-diff.c @@ -0,0 +1,233 @@ +/* + * builtin-diff.c + * + * Builtin diff command: Analyze two perf.data input files, look up and read + * DSOs and symbol information, sort them and produce a diff. + */ +#include "builtin.h" + +#include "util/debug.h" +#include "util/event.h" +#include "util/hist.h" +#include "util/session.h" +#include "util/sort.h" +#include "util/symbol.h" +#include "util/util.h" + +#include <stdlib.h> + +static char const *input_old = "perf.data.old", + *input_new = "perf.data"; +static char diff__default_sort_order[] = "dso,symbol"; +static bool force; +static bool show_displacement; + +static int hists__add_entry(struct hists *self, + struct addr_location *al, u64 period) +{ + if (__hists__add_entry(self, al, NULL, period) != NULL) + return 0; + return -ENOMEM; +} + +static int diff__process_sample_event(union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel __used, + struct perf_session *session) +{ + struct addr_location al; + + if (perf_event__preprocess_sample(event, session, &al, sample, NULL) < 0) { + pr_warning("problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + if (al.filtered || al.sym == NULL) + return 0; + + if (hists__add_entry(&session->hists, &al, sample->period)) { + pr_warning("problem incrementing symbol period, skipping event\n"); + return -1; + } + + session->hists.stats.total_period += sample->period; + return 0; +} + +static struct perf_event_ops event_ops = { + .sample = diff__process_sample_event, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .exit = perf_event__process_task, + .fork = perf_event__process_task, + .lost = perf_event__process_lost, + .ordered_samples = true, + .ordering_requires_timestamps = true, +}; + +static void perf_session__insert_hist_entry_by_name(struct rb_root *root, + struct hist_entry *he) +{ + struct rb_node **p = &root->rb_node; + struct rb_node *parent = NULL; + struct hist_entry *iter; + + while (*p != NULL) { + parent = *p; + iter = rb_entry(parent, struct hist_entry, rb_node); + if (hist_entry__cmp(he, iter) < 0) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&he->rb_node, parent, p); + rb_insert_color(&he->rb_node, root); +} + +static void hists__resort_entries(struct hists *self) +{ + unsigned long position = 1; + struct rb_root tmp = RB_ROOT; + struct rb_node *next = rb_first(&self->entries); + + while (next != NULL) { + struct hist_entry *n = rb_entry(next, struct hist_entry, rb_node); + + next = rb_next(&n->rb_node); + rb_erase(&n->rb_node, &self->entries); + n->position = position++; + perf_session__insert_hist_entry_by_name(&tmp, n); + } + + self->entries = tmp; +} + +static void hists__set_positions(struct hists *self) +{ + hists__output_resort(self); + hists__resort_entries(self); +} + +static struct hist_entry *hists__find_entry(struct hists *self, + struct hist_entry *he) +{ + struct rb_node *n = self->entries.rb_node; + + while (n) { + struct hist_entry *iter = rb_entry(n, struct hist_entry, rb_node); + int64_t cmp = hist_entry__cmp(he, iter); + + if (cmp < 0) + n = n->rb_left; + else if (cmp > 0) + n = n->rb_right; + else + return iter; + } + + return NULL; +} + +static void hists__match(struct hists *older, struct hists *newer) +{ + struct rb_node *nd; + + for (nd = rb_first(&newer->entries); nd; nd = rb_next(nd)) { + struct hist_entry *pos = rb_entry(nd, struct hist_entry, rb_node); + pos->pair = hists__find_entry(older, pos); + } +} + +static int __cmd_diff(void) +{ + int ret, i; + struct perf_session *session[2]; + + session[0] = perf_session__new(input_old, O_RDONLY, force, false, &event_ops); + session[1] = perf_session__new(input_new, O_RDONLY, force, false, &event_ops); + if (session[0] == NULL || session[1] == NULL) + return -ENOMEM; + + for (i = 0; i < 2; ++i) { + ret = perf_session__process_events(session[i], &event_ops); + if (ret) + goto out_delete; + } + + hists__output_resort(&session[1]->hists); + if (show_displacement) + hists__set_positions(&session[0]->hists); + + hists__match(&session[0]->hists, &session[1]->hists); + hists__fprintf(&session[1]->hists, &session[0]->hists, + show_displacement, stdout); +out_delete: + for (i = 0; i < 2; ++i) + perf_session__delete(session[i]); + return ret; +} + +static const char * const diff_usage[] = { + "perf diff [<options>] [old_file] [new_file]", + NULL, +}; + +static const struct option options[] = { + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('M', "displacement", &show_displacement, + "Show position displacement relative to baseline"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_BOOLEAN('f', "force", &force, "don't complain, do it"), + OPT_BOOLEAN('m', "modules", &symbol_conf.use_modules, + "load module symbols - WARNING: use only with -k and LIVE kernel"), + OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]", + "only consider symbols in these dsos"), + OPT_STRING('C', "comms", &symbol_conf.comm_list_str, "comm[,comm...]", + "only consider symbols in these comms"), + OPT_STRING('S', "symbols", &symbol_conf.sym_list_str, "symbol[,symbol...]", + "only consider these symbols"), + OPT_STRING('s', "sort", &sort_order, "key[,key2...]", + "sort by key(s): pid, comm, dso, symbol, parent"), + OPT_STRING('t', "field-separator", &symbol_conf.field_sep, "separator", + "separator for columns, no spaces will be added between " + "columns '.' is reserved."), + OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory", + "Look for files with symbols relative to this directory"), + OPT_END() +}; + +int cmd_diff(int argc, const char **argv, const char *prefix __used) +{ + sort_order = diff__default_sort_order; + argc = parse_options(argc, argv, options, diff_usage, 0); + if (argc) { + if (argc > 2) + usage_with_options(diff_usage, options); + if (argc == 2) { + input_old = argv[0]; + input_new = argv[1]; + } else + input_new = argv[0]; + } else if (symbol_conf.default_guest_vmlinux_name || + symbol_conf.default_guest_kallsyms) { + input_old = "perf.data.host"; + input_new = "perf.data.guest"; + } + + symbol_conf.exclude_other = false; + if (symbol__init() < 0) + return -1; + + setup_sorting(diff_usage, options); + setup_pager(); + + sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "dso", NULL); + sort_entry__setup_elide(&sort_comm, symbol_conf.comm_list, "comm", NULL); + sort_entry__setup_elide(&sort_sym, symbol_conf.sym_list, "symbol", NULL); + + return __cmd_diff(); +} diff --git a/tools/perf/builtin-evlist.c b/tools/perf/builtin-evlist.c new file mode 100644 index 00000000..4c5e9e04 --- /dev/null +++ b/tools/perf/builtin-evlist.c @@ -0,0 +1,54 @@ +/* + * Builtin evlist command: Show the list of event selectors present + * in a perf.data file. + */ +#include "builtin.h" + +#include "util/util.h" + +#include <linux/list.h> + +#include "perf.h" +#include "util/evlist.h" +#include "util/evsel.h" +#include "util/parse-events.h" +#include "util/parse-options.h" +#include "util/session.h" + +static char const *input_name = "perf.data"; + +static int __cmd_evlist(void) +{ + struct perf_session *session; + struct perf_evsel *pos; + + session = perf_session__new(input_name, O_RDONLY, 0, false, NULL); + if (session == NULL) + return -ENOMEM; + + list_for_each_entry(pos, &session->evlist->entries, node) + printf("%s\n", event_name(pos)); + + perf_session__delete(session); + return 0; +} + +static const char * const evlist_usage[] = { + "perf evlist [<options>]", + NULL +}; + +static const struct option options[] = { + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_END() +}; + +int cmd_evlist(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, options, evlist_usage, 0); + if (argc) + usage_with_options(evlist_usage, options); + + return __cmd_evlist(); +} diff --git a/tools/perf/builtin-help.c b/tools/perf/builtin-help.c new file mode 100644 index 00000000..6d5a8a7f --- /dev/null +++ b/tools/perf/builtin-help.c @@ -0,0 +1,459 @@ +/* + * builtin-help.c + * + * Builtin help command + */ +#include "perf.h" +#include "util/cache.h" +#include "builtin.h" +#include "util/exec_cmd.h" +#include "common-cmds.h" +#include "util/parse-options.h" +#include "util/run-command.h" +#include "util/help.h" + +static struct man_viewer_list { + struct man_viewer_list *next; + char name[FLEX_ARRAY]; +} *man_viewer_list; + +static struct man_viewer_info_list { + struct man_viewer_info_list *next; + const char *info; + char name[FLEX_ARRAY]; +} *man_viewer_info_list; + +enum help_format { + HELP_FORMAT_MAN, + HELP_FORMAT_INFO, + HELP_FORMAT_WEB, +}; + +static bool show_all = false; +static enum help_format help_format = HELP_FORMAT_MAN; +static struct option builtin_help_options[] = { + OPT_BOOLEAN('a', "all", &show_all, "print all available commands"), + OPT_SET_UINT('m', "man", &help_format, "show man page", HELP_FORMAT_MAN), + OPT_SET_UINT('w', "web", &help_format, "show manual in web browser", + HELP_FORMAT_WEB), + OPT_SET_UINT('i', "info", &help_format, "show info page", + HELP_FORMAT_INFO), + OPT_END(), +}; + +static const char * const builtin_help_usage[] = { + "perf help [--all] [--man|--web|--info] [command]", + NULL +}; + +static enum help_format parse_help_format(const char *format) +{ + if (!strcmp(format, "man")) + return HELP_FORMAT_MAN; + if (!strcmp(format, "info")) + return HELP_FORMAT_INFO; + if (!strcmp(format, "web") || !strcmp(format, "html")) + return HELP_FORMAT_WEB; + die("unrecognized help format '%s'", format); +} + +static const char *get_man_viewer_info(const char *name) +{ + struct man_viewer_info_list *viewer; + + for (viewer = man_viewer_info_list; viewer; viewer = viewer->next) { + if (!strcasecmp(name, viewer->name)) + return viewer->info; + } + return NULL; +} + +static int check_emacsclient_version(void) +{ + struct strbuf buffer = STRBUF_INIT; + struct child_process ec_process; + const char *argv_ec[] = { "emacsclient", "--version", NULL }; + int version; + + /* emacsclient prints its version number on stderr */ + memset(&ec_process, 0, sizeof(ec_process)); + ec_process.argv = argv_ec; + ec_process.err = -1; + ec_process.stdout_to_stderr = 1; + if (start_command(&ec_process)) { + fprintf(stderr, "Failed to start emacsclient.\n"); + return -1; + } + strbuf_read(&buffer, ec_process.err, 20); + close(ec_process.err); + + /* + * Don't bother checking return value, because "emacsclient --version" + * seems to always exits with code 1. + */ + finish_command(&ec_process); + + if (prefixcmp(buffer.buf, "emacsclient")) { + fprintf(stderr, "Failed to parse emacsclient version.\n"); + strbuf_release(&buffer); + return -1; + } + + strbuf_remove(&buffer, 0, strlen("emacsclient")); + version = atoi(buffer.buf); + + if (version < 22) { + fprintf(stderr, + "emacsclient version '%d' too old (< 22).\n", + version); + strbuf_release(&buffer); + return -1; + } + + strbuf_release(&buffer); + return 0; +} + +static void exec_woman_emacs(const char *path, const char *page) +{ + if (!check_emacsclient_version()) { + /* This works only with emacsclient version >= 22. */ + struct strbuf man_page = STRBUF_INIT; + + if (!path) + path = "emacsclient"; + strbuf_addf(&man_page, "(woman \"%s\")", page); + execlp(path, "emacsclient", "-e", man_page.buf, NULL); + warning("failed to exec '%s': %s", path, strerror(errno)); + } +} + +static void exec_man_konqueror(const char *path, const char *page) +{ + const char *display = getenv("DISPLAY"); + if (display && *display) { + struct strbuf man_page = STRBUF_INIT; + const char *filename = "kfmclient"; + + /* It's simpler to launch konqueror using kfmclient. */ + if (path) { + const char *file = strrchr(path, '/'); + if (file && !strcmp(file + 1, "konqueror")) { + char *new = strdup(path); + char *dest = strrchr(new, '/'); + + /* strlen("konqueror") == strlen("kfmclient") */ + strcpy(dest + 1, "kfmclient"); + path = new; + } + if (file) + filename = file; + } else + path = "kfmclient"; + strbuf_addf(&man_page, "man:%s(1)", page); + execlp(path, filename, "newTab", man_page.buf, NULL); + warning("failed to exec '%s': %s", path, strerror(errno)); + } +} + +static void exec_man_man(const char *path, const char *page) +{ + if (!path) + path = "man"; + execlp(path, "man", page, NULL); + warning("failed to exec '%s': %s", path, strerror(errno)); +} + +static void exec_man_cmd(const char *cmd, const char *page) +{ + struct strbuf shell_cmd = STRBUF_INIT; + strbuf_addf(&shell_cmd, "%s %s", cmd, page); + execl("/bin/sh", "sh", "-c", shell_cmd.buf, NULL); + warning("failed to exec '%s': %s", cmd, strerror(errno)); +} + +static void add_man_viewer(const char *name) +{ + struct man_viewer_list **p = &man_viewer_list; + size_t len = strlen(name); + + while (*p) + p = &((*p)->next); + *p = zalloc(sizeof(**p) + len + 1); + strncpy((*p)->name, name, len); +} + +static int supported_man_viewer(const char *name, size_t len) +{ + return (!strncasecmp("man", name, len) || + !strncasecmp("woman", name, len) || + !strncasecmp("konqueror", name, len)); +} + +static void do_add_man_viewer_info(const char *name, + size_t len, + const char *value) +{ + struct man_viewer_info_list *new = zalloc(sizeof(*new) + len + 1); + + strncpy(new->name, name, len); + new->info = strdup(value); + new->next = man_viewer_info_list; + man_viewer_info_list = new; +} + +static int add_man_viewer_path(const char *name, + size_t len, + const char *value) +{ + if (supported_man_viewer(name, len)) + do_add_man_viewer_info(name, len, value); + else + warning("'%s': path for unsupported man viewer.\n" + "Please consider using 'man.<tool>.cmd' instead.", + name); + + return 0; +} + +static int add_man_viewer_cmd(const char *name, + size_t len, + const char *value) +{ + if (supported_man_viewer(name, len)) + warning("'%s': cmd for supported man viewer.\n" + "Please consider using 'man.<tool>.path' instead.", + name); + else + do_add_man_viewer_info(name, len, value); + + return 0; +} + +static int add_man_viewer_info(const char *var, const char *value) +{ + const char *name = var + 4; + const char *subkey = strrchr(name, '.'); + + if (!subkey) + return error("Config with no key for man viewer: %s", name); + + if (!strcmp(subkey, ".path")) { + if (!value) + return config_error_nonbool(var); + return add_man_viewer_path(name, subkey - name, value); + } + if (!strcmp(subkey, ".cmd")) { + if (!value) + return config_error_nonbool(var); + return add_man_viewer_cmd(name, subkey - name, value); + } + + warning("'%s': unsupported man viewer sub key.", subkey); + return 0; +} + +static int perf_help_config(const char *var, const char *value, void *cb) +{ + if (!strcmp(var, "help.format")) { + if (!value) + return config_error_nonbool(var); + help_format = parse_help_format(value); + return 0; + } + if (!strcmp(var, "man.viewer")) { + if (!value) + return config_error_nonbool(var); + add_man_viewer(value); + return 0; + } + if (!prefixcmp(var, "man.")) + return add_man_viewer_info(var, value); + + return perf_default_config(var, value, cb); +} + +static struct cmdnames main_cmds, other_cmds; + +void list_common_cmds_help(void) +{ + unsigned int i, longest = 0; + + for (i = 0; i < ARRAY_SIZE(common_cmds); i++) { + if (longest < strlen(common_cmds[i].name)) + longest = strlen(common_cmds[i].name); + } + + puts(" The most commonly used perf commands are:"); + for (i = 0; i < ARRAY_SIZE(common_cmds); i++) { + printf(" %-*s ", longest, common_cmds[i].name); + puts(common_cmds[i].help); + } +} + +static int is_perf_command(const char *s) +{ + return is_in_cmdlist(&main_cmds, s) || + is_in_cmdlist(&other_cmds, s); +} + +static const char *prepend(const char *prefix, const char *cmd) +{ + size_t pre_len = strlen(prefix); + size_t cmd_len = strlen(cmd); + char *p = malloc(pre_len + cmd_len + 1); + memcpy(p, prefix, pre_len); + strcpy(p + pre_len, cmd); + return p; +} + +static const char *cmd_to_page(const char *perf_cmd) +{ + if (!perf_cmd) + return "perf"; + else if (!prefixcmp(perf_cmd, "perf")) + return perf_cmd; + else + return prepend("perf-", perf_cmd); +} + +static void setup_man_path(void) +{ + struct strbuf new_path = STRBUF_INIT; + const char *old_path = getenv("MANPATH"); + + /* We should always put ':' after our path. If there is no + * old_path, the ':' at the end will let 'man' to try + * system-wide paths after ours to find the manual page. If + * there is old_path, we need ':' as delimiter. */ + strbuf_addstr(&new_path, system_path(PERF_MAN_PATH)); + strbuf_addch(&new_path, ':'); + if (old_path) + strbuf_addstr(&new_path, old_path); + + setenv("MANPATH", new_path.buf, 1); + + strbuf_release(&new_path); +} + +static void exec_viewer(const char *name, const char *page) +{ + const char *info = get_man_viewer_info(name); + + if (!strcasecmp(name, "man")) + exec_man_man(info, page); + else if (!strcasecmp(name, "woman")) + exec_woman_emacs(info, page); + else if (!strcasecmp(name, "konqueror")) + exec_man_konqueror(info, page); + else if (info) + exec_man_cmd(info, page); + else + warning("'%s': unknown man viewer.", name); +} + +static void show_man_page(const char *perf_cmd) +{ + struct man_viewer_list *viewer; + const char *page = cmd_to_page(perf_cmd); + const char *fallback = getenv("PERF_MAN_VIEWER"); + + setup_man_path(); + for (viewer = man_viewer_list; viewer; viewer = viewer->next) + exec_viewer(viewer->name, page); /* will return when unable */ + + if (fallback) + exec_viewer(fallback, page); + exec_viewer("man", page); + die("no man viewer handled the request"); +} + +static void show_info_page(const char *perf_cmd) +{ + const char *page = cmd_to_page(perf_cmd); + setenv("INFOPATH", system_path(PERF_INFO_PATH), 1); + execlp("info", "info", "perfman", page, NULL); +} + +static void get_html_page_path(struct strbuf *page_path, const char *page) +{ + struct stat st; + const char *html_path = system_path(PERF_HTML_PATH); + + /* Check that we have a perf documentation directory. */ + if (stat(mkpath("%s/perf.html", html_path), &st) + || !S_ISREG(st.st_mode)) + die("'%s': not a documentation directory.", html_path); + + strbuf_init(page_path, 0); + strbuf_addf(page_path, "%s/%s.html", html_path, page); +} + +/* + * If open_html is not defined in a platform-specific way (see for + * example compat/mingw.h), we use the script web--browse to display + * HTML. + */ +#ifndef open_html +static void open_html(const char *path) +{ + execl_perf_cmd("web--browse", "-c", "help.browser", path, NULL); +} +#endif + +static void show_html_page(const char *perf_cmd) +{ + const char *page = cmd_to_page(perf_cmd); + struct strbuf page_path; /* it leaks but we exec bellow */ + + get_html_page_path(&page_path, page); + + open_html(page_path.buf); +} + +int cmd_help(int argc, const char **argv, const char *prefix __used) +{ + const char *alias; + + load_command_list("perf-", &main_cmds, &other_cmds); + + perf_config(perf_help_config, NULL); + + argc = parse_options(argc, argv, builtin_help_options, + builtin_help_usage, 0); + + if (show_all) { + printf("\n usage: %s\n\n", perf_usage_string); + list_commands("perf commands", &main_cmds, &other_cmds); + printf(" %s\n\n", perf_more_info_string); + return 0; + } + + if (!argv[0]) { + printf("\n usage: %s\n\n", perf_usage_string); + list_common_cmds_help(); + printf("\n %s\n\n", perf_more_info_string); + return 0; + } + + alias = alias_lookup(argv[0]); + if (alias && !is_perf_command(argv[0])) { + printf("`perf %s' is aliased to `%s'\n", argv[0], alias); + return 0; + } + + switch (help_format) { + case HELP_FORMAT_MAN: + show_man_page(argv[0]); + break; + case HELP_FORMAT_INFO: + show_info_page(argv[0]); + break; + case HELP_FORMAT_WEB: + show_html_page(argv[0]); + default: + break; + } + + return 0; +} diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c new file mode 100644 index 00000000..8dfc12bb --- /dev/null +++ b/tools/perf/builtin-inject.c @@ -0,0 +1,250 @@ +/* + * builtin-inject.c + * + * Builtin inject command: Examine the live mode (stdin) event stream + * and repipe it to stdout while optionally injecting additional + * events into it. + */ +#include "builtin.h" + +#include "perf.h" +#include "util/session.h" +#include "util/debug.h" + +#include "util/parse-options.h" + +static char const *input_name = "-"; +static bool inject_build_ids; + +static int perf_event__repipe_synth(union perf_event *event, + struct perf_session *session __used) +{ + uint32_t size; + void *buf = event; + + size = event->header.size; + + while (size) { + int ret = write(STDOUT_FILENO, buf, size); + if (ret < 0) + return -errno; + + size -= ret; + buf += ret; + } + + return 0; +} + +static int perf_event__repipe(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) +{ + return perf_event__repipe_synth(event, session); +} + +static int perf_event__repipe_sample(union perf_event *event, + struct perf_sample *sample __used, + struct perf_evsel *evsel __used, + struct perf_session *session) +{ + return perf_event__repipe_synth(event, session); +} + +static int perf_event__repipe_mmap(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) +{ + int err; + + err = perf_event__process_mmap(event, sample, session); + perf_event__repipe(event, sample, session); + + return err; +} + +static int perf_event__repipe_task(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) +{ + int err; + + err = perf_event__process_task(event, sample, session); + perf_event__repipe(event, sample, session); + + return err; +} + +static int perf_event__repipe_tracing_data(union perf_event *event, + struct perf_session *session) +{ + int err; + + perf_event__repipe_synth(event, session); + err = perf_event__process_tracing_data(event, session); + + return err; +} + +static int dso__read_build_id(struct dso *self) +{ + if (self->has_build_id) + return 0; + + if (filename__read_build_id(self->long_name, self->build_id, + sizeof(self->build_id)) > 0) { + self->has_build_id = true; + return 0; + } + + return -1; +} + +static int dso__inject_build_id(struct dso *self, struct perf_session *session) +{ + u16 misc = PERF_RECORD_MISC_USER; + struct machine *machine; + int err; + + if (dso__read_build_id(self) < 0) { + pr_debug("no build_id found for %s\n", self->long_name); + return -1; + } + + machine = perf_session__find_host_machine(session); + if (machine == NULL) { + pr_err("Can't find machine for session\n"); + return -1; + } + + if (self->kernel) + misc = PERF_RECORD_MISC_KERNEL; + + err = perf_event__synthesize_build_id(self, misc, perf_event__repipe, + machine, session); + if (err) { + pr_err("Can't synthesize build_id event for %s\n", self->long_name); + return -1; + } + + return 0; +} + +static int perf_event__inject_buildid(union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel __used, + struct perf_session *session) +{ + struct addr_location al; + struct thread *thread; + u8 cpumode; + + cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + + thread = perf_session__findnew(session, event->ip.pid); + if (thread == NULL) { + pr_err("problem processing %d event, skipping it.\n", + event->header.type); + goto repipe; + } + + thread__find_addr_map(thread, session, cpumode, MAP__FUNCTION, + event->ip.pid, event->ip.ip, &al); + + if (al.map != NULL) { + if (!al.map->dso->hit) { + al.map->dso->hit = 1; + if (map__load(al.map, NULL) >= 0) { + dso__inject_build_id(al.map->dso, session); + /* + * If this fails, too bad, let the other side + * account this as unresolved. + */ + } else + pr_warning("no symbols found in %s, maybe " + "install a debug package?\n", + al.map->dso->long_name); + } + } + +repipe: + perf_event__repipe(event, sample, session); + return 0; +} + +struct perf_event_ops inject_ops = { + .sample = perf_event__repipe_sample, + .mmap = perf_event__repipe, + .comm = perf_event__repipe, + .fork = perf_event__repipe, + .exit = perf_event__repipe, + .lost = perf_event__repipe, + .read = perf_event__repipe, + .throttle = perf_event__repipe, + .unthrottle = perf_event__repipe, + .attr = perf_event__repipe_synth, + .event_type = perf_event__repipe_synth, + .tracing_data = perf_event__repipe_synth, + .build_id = perf_event__repipe_synth, +}; + +extern volatile int session_done; + +static void sig_handler(int sig __attribute__((__unused__))) +{ + session_done = 1; +} + +static int __cmd_inject(void) +{ + struct perf_session *session; + int ret = -EINVAL; + + signal(SIGINT, sig_handler); + + if (inject_build_ids) { + inject_ops.sample = perf_event__inject_buildid; + inject_ops.mmap = perf_event__repipe_mmap; + inject_ops.fork = perf_event__repipe_task; + inject_ops.tracing_data = perf_event__repipe_tracing_data; + } + + session = perf_session__new(input_name, O_RDONLY, false, true, &inject_ops); + if (session == NULL) + return -ENOMEM; + + ret = perf_session__process_events(session, &inject_ops); + + perf_session__delete(session); + + return ret; +} + +static const char * const report_usage[] = { + "perf inject [<options>]", + NULL +}; + +static const struct option options[] = { + OPT_BOOLEAN('b', "build-ids", &inject_build_ids, + "Inject build-ids into the output stream"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show build ids, etc)"), + OPT_END() +}; + +int cmd_inject(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, options, report_usage, 0); + + /* + * Any (unrecognized) arguments left? + */ + if (argc) + usage_with_options(report_usage, options); + + if (symbol__init() < 0) + return -1; + + return __cmd_inject(); +} diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c new file mode 100644 index 00000000..225e963d --- /dev/null +++ b/tools/perf/builtin-kmem.c @@ -0,0 +1,778 @@ +#include "builtin.h" +#include "perf.h" + +#include "util/util.h" +#include "util/cache.h" +#include "util/symbol.h" +#include "util/thread.h" +#include "util/header.h" +#include "util/session.h" + +#include "util/parse-options.h" +#include "util/trace-event.h" + +#include "util/debug.h" + +#include <linux/rbtree.h> + +struct alloc_stat; +typedef int (*sort_fn_t)(struct alloc_stat *, struct alloc_stat *); + +static char const *input_name = "perf.data"; + +static int alloc_flag; +static int caller_flag; + +static int alloc_lines = -1; +static int caller_lines = -1; + +static bool raw_ip; + +static char default_sort_order[] = "frag,hit,bytes"; + +static int *cpunode_map; +static int max_cpu_num; + +struct alloc_stat { + u64 call_site; + u64 ptr; + u64 bytes_req; + u64 bytes_alloc; + u32 hit; + u32 pingpong; + + short alloc_cpu; + + struct rb_node node; +}; + +static struct rb_root root_alloc_stat; +static struct rb_root root_alloc_sorted; +static struct rb_root root_caller_stat; +static struct rb_root root_caller_sorted; + +static unsigned long total_requested, total_allocated; +static unsigned long nr_allocs, nr_cross_allocs; + +#define PATH_SYS_NODE "/sys/devices/system/node" + +static void init_cpunode_map(void) +{ + FILE *fp; + int i; + + fp = fopen("/sys/devices/system/cpu/kernel_max", "r"); + if (!fp) { + max_cpu_num = 4096; + return; + } + + if (fscanf(fp, "%d", &max_cpu_num) < 1) + die("Failed to read 'kernel_max' from sysfs"); + max_cpu_num++; + + cpunode_map = calloc(max_cpu_num, sizeof(int)); + if (!cpunode_map) + die("calloc"); + for (i = 0; i < max_cpu_num; i++) + cpunode_map[i] = -1; + fclose(fp); +} + +static void setup_cpunode_map(void) +{ + struct dirent *dent1, *dent2; + DIR *dir1, *dir2; + unsigned int cpu, mem; + char buf[PATH_MAX]; + + init_cpunode_map(); + + dir1 = opendir(PATH_SYS_NODE); + if (!dir1) + return; + + while ((dent1 = readdir(dir1)) != NULL) { + if (dent1->d_type != DT_DIR || + sscanf(dent1->d_name, "node%u", &mem) < 1) + continue; + + snprintf(buf, PATH_MAX, "%s/%s", PATH_SYS_NODE, dent1->d_name); + dir2 = opendir(buf); + if (!dir2) + continue; + while ((dent2 = readdir(dir2)) != NULL) { + if (dent2->d_type != DT_LNK || + sscanf(dent2->d_name, "cpu%u", &cpu) < 1) + continue; + cpunode_map[cpu] = mem; + } + } +} + +static void insert_alloc_stat(unsigned long call_site, unsigned long ptr, + int bytes_req, int bytes_alloc, int cpu) +{ + struct rb_node **node = &root_alloc_stat.rb_node; + struct rb_node *parent = NULL; + struct alloc_stat *data = NULL; + + while (*node) { + parent = *node; + data = rb_entry(*node, struct alloc_stat, node); + + if (ptr > data->ptr) + node = &(*node)->rb_right; + else if (ptr < data->ptr) + node = &(*node)->rb_left; + else + break; + } + + if (data && data->ptr == ptr) { + data->hit++; + data->bytes_req += bytes_req; + data->bytes_alloc += bytes_alloc; + } else { + data = malloc(sizeof(*data)); + if (!data) + die("malloc"); + data->ptr = ptr; + data->pingpong = 0; + data->hit = 1; + data->bytes_req = bytes_req; + data->bytes_alloc = bytes_alloc; + + rb_link_node(&data->node, parent, node); + rb_insert_color(&data->node, &root_alloc_stat); + } + data->call_site = call_site; + data->alloc_cpu = cpu; +} + +static void insert_caller_stat(unsigned long call_site, + int bytes_req, int bytes_alloc) +{ + struct rb_node **node = &root_caller_stat.rb_node; + struct rb_node *parent = NULL; + struct alloc_stat *data = NULL; + + while (*node) { + parent = *node; + data = rb_entry(*node, struct alloc_stat, node); + + if (call_site > data->call_site) + node = &(*node)->rb_right; + else if (call_site < data->call_site) + node = &(*node)->rb_left; + else + break; + } + + if (data && data->call_site == call_site) { + data->hit++; + data->bytes_req += bytes_req; + data->bytes_alloc += bytes_alloc; + } else { + data = malloc(sizeof(*data)); + if (!data) + die("malloc"); + data->call_site = call_site; + data->pingpong = 0; + data->hit = 1; + data->bytes_req = bytes_req; + data->bytes_alloc = bytes_alloc; + + rb_link_node(&data->node, parent, node); + rb_insert_color(&data->node, &root_caller_stat); + } +} + +static void process_alloc_event(void *data, + struct event *event, + int cpu, + u64 timestamp __used, + struct thread *thread __used, + int node) +{ + unsigned long call_site; + unsigned long ptr; + int bytes_req; + int bytes_alloc; + int node1, node2; + + ptr = raw_field_value(event, "ptr", data); + call_site = raw_field_value(event, "call_site", data); + bytes_req = raw_field_value(event, "bytes_req", data); + bytes_alloc = raw_field_value(event, "bytes_alloc", data); + + insert_alloc_stat(call_site, ptr, bytes_req, bytes_alloc, cpu); + insert_caller_stat(call_site, bytes_req, bytes_alloc); + + total_requested += bytes_req; + total_allocated += bytes_alloc; + + if (node) { + node1 = cpunode_map[cpu]; + node2 = raw_field_value(event, "node", data); + if (node1 != node2) + nr_cross_allocs++; + } + nr_allocs++; +} + +static int ptr_cmp(struct alloc_stat *, struct alloc_stat *); +static int callsite_cmp(struct alloc_stat *, struct alloc_stat *); + +static struct alloc_stat *search_alloc_stat(unsigned long ptr, + unsigned long call_site, + struct rb_root *root, + sort_fn_t sort_fn) +{ + struct rb_node *node = root->rb_node; + struct alloc_stat key = { .ptr = ptr, .call_site = call_site }; + + while (node) { + struct alloc_stat *data; + int cmp; + + data = rb_entry(node, struct alloc_stat, node); + + cmp = sort_fn(&key, data); + if (cmp < 0) + node = node->rb_left; + else if (cmp > 0) + node = node->rb_right; + else + return data; + } + return NULL; +} + +static void process_free_event(void *data, + struct event *event, + int cpu, + u64 timestamp __used, + struct thread *thread __used) +{ + unsigned long ptr; + struct alloc_stat *s_alloc, *s_caller; + + ptr = raw_field_value(event, "ptr", data); + + s_alloc = search_alloc_stat(ptr, 0, &root_alloc_stat, ptr_cmp); + if (!s_alloc) + return; + + if (cpu != s_alloc->alloc_cpu) { + s_alloc->pingpong++; + + s_caller = search_alloc_stat(0, s_alloc->call_site, + &root_caller_stat, callsite_cmp); + assert(s_caller); + s_caller->pingpong++; + } + s_alloc->alloc_cpu = -1; +} + +static void process_raw_event(union perf_event *raw_event __used, void *data, + int cpu, u64 timestamp, struct thread *thread) +{ + struct event *event; + int type; + + type = trace_parse_common_type(data); + event = trace_find_event(type); + + if (!strcmp(event->name, "kmalloc") || + !strcmp(event->name, "kmem_cache_alloc")) { + process_alloc_event(data, event, cpu, timestamp, thread, 0); + return; + } + + if (!strcmp(event->name, "kmalloc_node") || + !strcmp(event->name, "kmem_cache_alloc_node")) { + process_alloc_event(data, event, cpu, timestamp, thread, 1); + return; + } + + if (!strcmp(event->name, "kfree") || + !strcmp(event->name, "kmem_cache_free")) { + process_free_event(data, event, cpu, timestamp, thread); + return; + } +} + +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel __used, + struct perf_session *session) +{ + struct thread *thread = perf_session__findnew(session, event->ip.pid); + + if (thread == NULL) { + pr_debug("problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid); + + process_raw_event(event, sample->raw_data, sample->cpu, + sample->time, thread); + + return 0; +} + +static struct perf_event_ops event_ops = { + .sample = process_sample_event, + .comm = perf_event__process_comm, + .ordered_samples = true, +}; + +static double fragmentation(unsigned long n_req, unsigned long n_alloc) +{ + if (n_alloc == 0) + return 0.0; + else + return 100.0 - (100.0 * n_req / n_alloc); +} + +static void __print_result(struct rb_root *root, struct perf_session *session, + int n_lines, int is_caller) +{ + struct rb_node *next; + struct machine *machine; + + printf("%.102s\n", graph_dotted_line); + printf(" %-34s |", is_caller ? "Callsite": "Alloc Ptr"); + printf(" Total_alloc/Per | Total_req/Per | Hit | Ping-pong | Frag\n"); + printf("%.102s\n", graph_dotted_line); + + next = rb_first(root); + + machine = perf_session__find_host_machine(session); + if (!machine) { + pr_err("__print_result: couldn't find kernel information\n"); + return; + } + while (next && n_lines--) { + struct alloc_stat *data = rb_entry(next, struct alloc_stat, + node); + struct symbol *sym = NULL; + struct map *map; + char buf[BUFSIZ]; + u64 addr; + + if (is_caller) { + addr = data->call_site; + if (!raw_ip) + sym = machine__find_kernel_function(machine, addr, &map, NULL); + } else + addr = data->ptr; + + if (sym != NULL) + snprintf(buf, sizeof(buf), "%s+%" PRIx64 "", sym->name, + addr - map->unmap_ip(map, sym->start)); + else + snprintf(buf, sizeof(buf), "%#" PRIx64 "", addr); + printf(" %-34s |", buf); + + printf(" %9llu/%-5lu | %9llu/%-5lu | %8lu | %8lu | %6.3f%%\n", + (unsigned long long)data->bytes_alloc, + (unsigned long)data->bytes_alloc / data->hit, + (unsigned long long)data->bytes_req, + (unsigned long)data->bytes_req / data->hit, + (unsigned long)data->hit, + (unsigned long)data->pingpong, + fragmentation(data->bytes_req, data->bytes_alloc)); + + next = rb_next(next); + } + + if (n_lines == -1) + printf(" ... | ... | ... | ... | ... | ... \n"); + + printf("%.102s\n", graph_dotted_line); +} + +static void print_summary(void) +{ + printf("\nSUMMARY\n=======\n"); + printf("Total bytes requested: %lu\n", total_requested); + printf("Total bytes allocated: %lu\n", total_allocated); + printf("Total bytes wasted on internal fragmentation: %lu\n", + total_allocated - total_requested); + printf("Internal fragmentation: %f%%\n", + fragmentation(total_requested, total_allocated)); + printf("Cross CPU allocations: %lu/%lu\n", nr_cross_allocs, nr_allocs); +} + +static void print_result(struct perf_session *session) +{ + if (caller_flag) + __print_result(&root_caller_sorted, session, caller_lines, 1); + if (alloc_flag) + __print_result(&root_alloc_sorted, session, alloc_lines, 0); + print_summary(); +} + +struct sort_dimension { + const char name[20]; + sort_fn_t cmp; + struct list_head list; +}; + +static LIST_HEAD(caller_sort); +static LIST_HEAD(alloc_sort); + +static void sort_insert(struct rb_root *root, struct alloc_stat *data, + struct list_head *sort_list) +{ + struct rb_node **new = &(root->rb_node); + struct rb_node *parent = NULL; + struct sort_dimension *sort; + + while (*new) { + struct alloc_stat *this; + int cmp = 0; + + this = rb_entry(*new, struct alloc_stat, node); + parent = *new; + + list_for_each_entry(sort, sort_list, list) { + cmp = sort->cmp(data, this); + if (cmp) + break; + } + + if (cmp > 0) + new = &((*new)->rb_left); + else + new = &((*new)->rb_right); + } + + rb_link_node(&data->node, parent, new); + rb_insert_color(&data->node, root); +} + +static void __sort_result(struct rb_root *root, struct rb_root *root_sorted, + struct list_head *sort_list) +{ + struct rb_node *node; + struct alloc_stat *data; + + for (;;) { + node = rb_first(root); + if (!node) + break; + + rb_erase(node, root); + data = rb_entry(node, struct alloc_stat, node); + sort_insert(root_sorted, data, sort_list); + } +} + +static void sort_result(void) +{ + __sort_result(&root_alloc_stat, &root_alloc_sorted, &alloc_sort); + __sort_result(&root_caller_stat, &root_caller_sorted, &caller_sort); +} + +static int __cmd_kmem(void) +{ + int err = -EINVAL; + struct perf_session *session = perf_session__new(input_name, O_RDONLY, + 0, false, &event_ops); + if (session == NULL) + return -ENOMEM; + + if (perf_session__create_kernel_maps(session) < 0) + goto out_delete; + + if (!perf_session__has_traces(session, "kmem record")) + goto out_delete; + + setup_pager(); + err = perf_session__process_events(session, &event_ops); + if (err != 0) + goto out_delete; + sort_result(); + print_result(session); +out_delete: + perf_session__delete(session); + return err; +} + +static const char * const kmem_usage[] = { + "perf kmem [<options>] {record|stat}", + NULL +}; + +static int ptr_cmp(struct alloc_stat *l, struct alloc_stat *r) +{ + if (l->ptr < r->ptr) + return -1; + else if (l->ptr > r->ptr) + return 1; + return 0; +} + +static struct sort_dimension ptr_sort_dimension = { + .name = "ptr", + .cmp = ptr_cmp, +}; + +static int callsite_cmp(struct alloc_stat *l, struct alloc_stat *r) +{ + if (l->call_site < r->call_site) + return -1; + else if (l->call_site > r->call_site) + return 1; + return 0; +} + +static struct sort_dimension callsite_sort_dimension = { + .name = "callsite", + .cmp = callsite_cmp, +}; + +static int hit_cmp(struct alloc_stat *l, struct alloc_stat *r) +{ + if (l->hit < r->hit) + return -1; + else if (l->hit > r->hit) + return 1; + return 0; +} + +static struct sort_dimension hit_sort_dimension = { + .name = "hit", + .cmp = hit_cmp, +}; + +static int bytes_cmp(struct alloc_stat *l, struct alloc_stat *r) +{ + if (l->bytes_alloc < r->bytes_alloc) + return -1; + else if (l->bytes_alloc > r->bytes_alloc) + return 1; + return 0; +} + +static struct sort_dimension bytes_sort_dimension = { + .name = "bytes", + .cmp = bytes_cmp, +}; + +static int frag_cmp(struct alloc_stat *l, struct alloc_stat *r) +{ + double x, y; + + x = fragmentation(l->bytes_req, l->bytes_alloc); + y = fragmentation(r->bytes_req, r->bytes_alloc); + + if (x < y) + return -1; + else if (x > y) + return 1; + return 0; +} + +static struct sort_dimension frag_sort_dimension = { + .name = "frag", + .cmp = frag_cmp, +}; + +static int pingpong_cmp(struct alloc_stat *l, struct alloc_stat *r) +{ + if (l->pingpong < r->pingpong) + return -1; + else if (l->pingpong > r->pingpong) + return 1; + return 0; +} + +static struct sort_dimension pingpong_sort_dimension = { + .name = "pingpong", + .cmp = pingpong_cmp, +}; + +static struct sort_dimension *avail_sorts[] = { + &ptr_sort_dimension, + &callsite_sort_dimension, + &hit_sort_dimension, + &bytes_sort_dimension, + &frag_sort_dimension, + &pingpong_sort_dimension, +}; + +#define NUM_AVAIL_SORTS \ + (int)(sizeof(avail_sorts) / sizeof(struct sort_dimension *)) + +static int sort_dimension__add(const char *tok, struct list_head *list) +{ + struct sort_dimension *sort; + int i; + + for (i = 0; i < NUM_AVAIL_SORTS; i++) { + if (!strcmp(avail_sorts[i]->name, tok)) { + sort = malloc(sizeof(*sort)); + if (!sort) + die("malloc"); + memcpy(sort, avail_sorts[i], sizeof(*sort)); + list_add_tail(&sort->list, list); + return 0; + } + } + + return -1; +} + +static int setup_sorting(struct list_head *sort_list, const char *arg) +{ + char *tok; + char *str = strdup(arg); + + if (!str) + die("strdup"); + + while (true) { + tok = strsep(&str, ","); + if (!tok) + break; + if (sort_dimension__add(tok, sort_list) < 0) { + error("Unknown --sort key: '%s'", tok); + return -1; + } + } + + free(str); + return 0; +} + +static int parse_sort_opt(const struct option *opt __used, + const char *arg, int unset __used) +{ + if (!arg) + return -1; + + if (caller_flag > alloc_flag) + return setup_sorting(&caller_sort, arg); + else + return setup_sorting(&alloc_sort, arg); + + return 0; +} + +static int parse_caller_opt(const struct option *opt __used, + const char *arg __used, int unset __used) +{ + caller_flag = (alloc_flag + 1); + return 0; +} + +static int parse_alloc_opt(const struct option *opt __used, + const char *arg __used, int unset __used) +{ + alloc_flag = (caller_flag + 1); + return 0; +} + +static int parse_line_opt(const struct option *opt __used, + const char *arg, int unset __used) +{ + int lines; + + if (!arg) + return -1; + + lines = strtoul(arg, NULL, 10); + + if (caller_flag > alloc_flag) + caller_lines = lines; + else + alloc_lines = lines; + + return 0; +} + +static const struct option kmem_options[] = { + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_CALLBACK_NOOPT(0, "caller", NULL, NULL, + "show per-callsite statistics", + parse_caller_opt), + OPT_CALLBACK_NOOPT(0, "alloc", NULL, NULL, + "show per-allocation statistics", + parse_alloc_opt), + OPT_CALLBACK('s', "sort", NULL, "key[,key2...]", + "sort by keys: ptr, call_site, bytes, hit, pingpong, frag", + parse_sort_opt), + OPT_CALLBACK('l', "line", NULL, "num", + "show n lines", + parse_line_opt), + OPT_BOOLEAN(0, "raw-ip", &raw_ip, "show raw ip instead of symbol"), + OPT_END() +}; + +static const char *record_args[] = { + "record", + "-a", + "-R", + "-f", + "-c", "1", + "-e", "kmem:kmalloc", + "-e", "kmem:kmalloc_node", + "-e", "kmem:kfree", + "-e", "kmem:kmem_cache_alloc", + "-e", "kmem:kmem_cache_alloc_node", + "-e", "kmem:kmem_cache_free", +}; + +static int __cmd_record(int argc, const char **argv) +{ + unsigned int rec_argc, i, j; + const char **rec_argv; + + rec_argc = ARRAY_SIZE(record_args) + argc - 1; + rec_argv = calloc(rec_argc + 1, sizeof(char *)); + + if (rec_argv == NULL) + return -ENOMEM; + + for (i = 0; i < ARRAY_SIZE(record_args); i++) + rec_argv[i] = strdup(record_args[i]); + + for (j = 1; j < (unsigned int)argc; j++, i++) + rec_argv[i] = argv[j]; + + return cmd_record(i, rec_argv, NULL); +} + +int cmd_kmem(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, kmem_options, kmem_usage, 0); + + if (!argc) + usage_with_options(kmem_usage, kmem_options); + + symbol__init(); + + if (!strncmp(argv[0], "rec", 3)) { + return __cmd_record(argc, argv); + } else if (!strcmp(argv[0], "stat")) { + setup_cpunode_map(); + + if (list_empty(&caller_sort)) + setup_sorting(&caller_sort, default_sort_order); + if (list_empty(&alloc_sort)) + setup_sorting(&alloc_sort, default_sort_order); + + return __cmd_kmem(); + } else + usage_with_options(kmem_usage, kmem_options); + + return 0; +} + diff --git a/tools/perf/builtin-kvm.c b/tools/perf/builtin-kvm.c new file mode 100644 index 00000000..34d1e853 --- /dev/null +++ b/tools/perf/builtin-kvm.c @@ -0,0 +1,144 @@ +#include "builtin.h" +#include "perf.h" + +#include "util/util.h" +#include "util/cache.h" +#include "util/symbol.h" +#include "util/thread.h" +#include "util/header.h" +#include "util/session.h" + +#include "util/parse-options.h" +#include "util/trace-event.h" + +#include "util/debug.h" + +#include <sys/prctl.h> + +#include <semaphore.h> +#include <pthread.h> +#include <math.h> + +static const char *file_name; +static char name_buffer[256]; + +bool perf_host = 1; +bool perf_guest; + +static const char * const kvm_usage[] = { + "perf kvm [<options>] {top|record|report|diff|buildid-list}", + NULL +}; + +static const struct option kvm_options[] = { + OPT_STRING('i', "input", &file_name, "file", + "Input file name"), + OPT_STRING('o', "output", &file_name, "file", + "Output file name"), + OPT_BOOLEAN(0, "guest", &perf_guest, + "Collect guest os data"), + OPT_BOOLEAN(0, "host", &perf_host, + "Collect guest os data"), + OPT_STRING(0, "guestmount", &symbol_conf.guestmount, "directory", + "guest mount directory under which every guest os" + " instance has a subdir"), + OPT_STRING(0, "guestvmlinux", &symbol_conf.default_guest_vmlinux_name, + "file", "file saving guest os vmlinux"), + OPT_STRING(0, "guestkallsyms", &symbol_conf.default_guest_kallsyms, + "file", "file saving guest os /proc/kallsyms"), + OPT_STRING(0, "guestmodules", &symbol_conf.default_guest_modules, + "file", "file saving guest os /proc/modules"), + OPT_END() +}; + +static int __cmd_record(int argc, const char **argv) +{ + int rec_argc, i = 0, j; + const char **rec_argv; + + rec_argc = argc + 2; + rec_argv = calloc(rec_argc + 1, sizeof(char *)); + rec_argv[i++] = strdup("record"); + rec_argv[i++] = strdup("-o"); + rec_argv[i++] = strdup(file_name); + for (j = 1; j < argc; j++, i++) + rec_argv[i] = argv[j]; + + BUG_ON(i != rec_argc); + + return cmd_record(i, rec_argv, NULL); +} + +static int __cmd_report(int argc, const char **argv) +{ + int rec_argc, i = 0, j; + const char **rec_argv; + + rec_argc = argc + 2; + rec_argv = calloc(rec_argc + 1, sizeof(char *)); + rec_argv[i++] = strdup("report"); + rec_argv[i++] = strdup("-i"); + rec_argv[i++] = strdup(file_name); + for (j = 1; j < argc; j++, i++) + rec_argv[i] = argv[j]; + + BUG_ON(i != rec_argc); + + return cmd_report(i, rec_argv, NULL); +} + +static int __cmd_buildid_list(int argc, const char **argv) +{ + int rec_argc, i = 0, j; + const char **rec_argv; + + rec_argc = argc + 2; + rec_argv = calloc(rec_argc + 1, sizeof(char *)); + rec_argv[i++] = strdup("buildid-list"); + rec_argv[i++] = strdup("-i"); + rec_argv[i++] = strdup(file_name); + for (j = 1; j < argc; j++, i++) + rec_argv[i] = argv[j]; + + BUG_ON(i != rec_argc); + + return cmd_buildid_list(i, rec_argv, NULL); +} + +int cmd_kvm(int argc, const char **argv, const char *prefix __used) +{ + perf_host = perf_guest = 0; + + argc = parse_options(argc, argv, kvm_options, kvm_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + if (!argc) + usage_with_options(kvm_usage, kvm_options); + + if (!perf_host) + perf_guest = 1; + + if (!file_name) { + if (perf_host && !perf_guest) + sprintf(name_buffer, "perf.data.host"); + else if (!perf_host && perf_guest) + sprintf(name_buffer, "perf.data.guest"); + else + sprintf(name_buffer, "perf.data.kvm"); + file_name = name_buffer; + } + + if (!strncmp(argv[0], "rec", 3)) + return __cmd_record(argc, argv); + else if (!strncmp(argv[0], "rep", 3)) + return __cmd_report(argc, argv); + else if (!strncmp(argv[0], "diff", 4)) + return cmd_diff(argc, argv, NULL); + else if (!strncmp(argv[0], "top", 3)) + return cmd_top(argc, argv, NULL); + else if (!strncmp(argv[0], "buildid-list", 12)) + return __cmd_buildid_list(argc, argv); + else + usage_with_options(kvm_usage, kvm_options); + + return 0; +} diff --git a/tools/perf/builtin-list.c b/tools/perf/builtin-list.c new file mode 100644 index 00000000..6313b6eb --- /dev/null +++ b/tools/perf/builtin-list.c @@ -0,0 +1,60 @@ +/* + * builtin-list.c + * + * Builtin list command: list all event types + * + * Copyright (C) 2009, Thomas Gleixner <tglx@linutronix.de> + * Copyright (C) 2008-2009, Red Hat Inc, Ingo Molnar <mingo@redhat.com> + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + */ +#include "builtin.h" + +#include "perf.h" + +#include "util/parse-events.h" +#include "util/cache.h" + +int cmd_list(int argc, const char **argv, const char *prefix __used) +{ + setup_pager(); + + if (argc == 1) + print_events(NULL); + else { + int i; + + for (i = 1; i < argc; ++i) { + if (i > 1) + putchar('\n'); + if (strncmp(argv[i], "tracepoint", 10) == 0) + print_tracepoint_events(NULL, NULL); + else if (strcmp(argv[i], "hw") == 0 || + strcmp(argv[i], "hardware") == 0) + print_events_type(PERF_TYPE_HARDWARE); + else if (strcmp(argv[i], "sw") == 0 || + strcmp(argv[i], "software") == 0) + print_events_type(PERF_TYPE_SOFTWARE); + else if (strcmp(argv[i], "cache") == 0 || + strcmp(argv[i], "hwcache") == 0) + print_hwcache_events(NULL); + else { + char *sep = strchr(argv[i], ':'), *s; + int sep_idx; + + if (sep == NULL) { + print_events(argv[i]); + continue; + } + sep_idx = sep - argv[i]; + s = strdup(argv[i]); + if (s == NULL) + return -1; + + s[sep_idx] = '\0'; + print_tracepoint_events(s, s + sep_idx + 1); + free(s); + } + } + } + return 0; +} diff --git a/tools/perf/builtin-lock.c b/tools/perf/builtin-lock.c new file mode 100644 index 00000000..9ac05aaf --- /dev/null +++ b/tools/perf/builtin-lock.c @@ -0,0 +1,1017 @@ +#include "builtin.h" +#include "perf.h" + +#include "util/util.h" +#include "util/cache.h" +#include "util/symbol.h" +#include "util/thread.h" +#include "util/header.h" + +#include "util/parse-options.h" +#include "util/trace-event.h" + +#include "util/debug.h" +#include "util/session.h" + +#include <sys/types.h> +#include <sys/prctl.h> +#include <semaphore.h> +#include <pthread.h> +#include <math.h> +#include <limits.h> + +#include <linux/list.h> +#include <linux/hash.h> + +static struct perf_session *session; + +/* based on kernel/lockdep.c */ +#define LOCKHASH_BITS 12 +#define LOCKHASH_SIZE (1UL << LOCKHASH_BITS) + +static struct list_head lockhash_table[LOCKHASH_SIZE]; + +#define __lockhashfn(key) hash_long((unsigned long)key, LOCKHASH_BITS) +#define lockhashentry(key) (lockhash_table + __lockhashfn((key))) + +struct lock_stat { + struct list_head hash_entry; + struct rb_node rb; /* used for sorting */ + + /* + * FIXME: raw_field_value() returns unsigned long long, + * so address of lockdep_map should be dealed as 64bit. + * Is there more better solution? + */ + void *addr; /* address of lockdep_map, used as ID */ + char *name; /* for strcpy(), we cannot use const */ + + unsigned int nr_acquire; + unsigned int nr_acquired; + unsigned int nr_contended; + unsigned int nr_release; + + unsigned int nr_readlock; + unsigned int nr_trylock; + /* these times are in nano sec. */ + u64 wait_time_total; + u64 wait_time_min; + u64 wait_time_max; + + int discard; /* flag of blacklist */ +}; + +/* + * States of lock_seq_stat + * + * UNINITIALIZED is required for detecting first event of acquire. + * As the nature of lock events, there is no guarantee + * that the first event for the locks are acquire, + * it can be acquired, contended or release. + */ +#define SEQ_STATE_UNINITIALIZED 0 /* initial state */ +#define SEQ_STATE_RELEASED 1 +#define SEQ_STATE_ACQUIRING 2 +#define SEQ_STATE_ACQUIRED 3 +#define SEQ_STATE_READ_ACQUIRED 4 +#define SEQ_STATE_CONTENDED 5 + +/* + * MAX_LOCK_DEPTH + * Imported from include/linux/sched.h. + * Should this be synchronized? + */ +#define MAX_LOCK_DEPTH 48 + +/* + * struct lock_seq_stat: + * Place to put on state of one lock sequence + * 1) acquire -> acquired -> release + * 2) acquire -> contended -> acquired -> release + * 3) acquire (with read or try) -> release + * 4) Are there other patterns? + */ +struct lock_seq_stat { + struct list_head list; + int state; + u64 prev_event_time; + void *addr; + + int read_count; +}; + +struct thread_stat { + struct rb_node rb; + + u32 tid; + struct list_head seq_list; +}; + +static struct rb_root thread_stats; + +static struct thread_stat *thread_stat_find(u32 tid) +{ + struct rb_node *node; + struct thread_stat *st; + + node = thread_stats.rb_node; + while (node) { + st = container_of(node, struct thread_stat, rb); + if (st->tid == tid) + return st; + else if (tid < st->tid) + node = node->rb_left; + else + node = node->rb_right; + } + + return NULL; +} + +static void thread_stat_insert(struct thread_stat *new) +{ + struct rb_node **rb = &thread_stats.rb_node; + struct rb_node *parent = NULL; + struct thread_stat *p; + + while (*rb) { + p = container_of(*rb, struct thread_stat, rb); + parent = *rb; + + if (new->tid < p->tid) + rb = &(*rb)->rb_left; + else if (new->tid > p->tid) + rb = &(*rb)->rb_right; + else + BUG_ON("inserting invalid thread_stat\n"); + } + + rb_link_node(&new->rb, parent, rb); + rb_insert_color(&new->rb, &thread_stats); +} + +static struct thread_stat *thread_stat_findnew_after_first(u32 tid) +{ + struct thread_stat *st; + + st = thread_stat_find(tid); + if (st) + return st; + + st = zalloc(sizeof(struct thread_stat)); + if (!st) + die("memory allocation failed\n"); + + st->tid = tid; + INIT_LIST_HEAD(&st->seq_list); + + thread_stat_insert(st); + + return st; +} + +static struct thread_stat *thread_stat_findnew_first(u32 tid); +static struct thread_stat *(*thread_stat_findnew)(u32 tid) = + thread_stat_findnew_first; + +static struct thread_stat *thread_stat_findnew_first(u32 tid) +{ + struct thread_stat *st; + + st = zalloc(sizeof(struct thread_stat)); + if (!st) + die("memory allocation failed\n"); + st->tid = tid; + INIT_LIST_HEAD(&st->seq_list); + + rb_link_node(&st->rb, NULL, &thread_stats.rb_node); + rb_insert_color(&st->rb, &thread_stats); + + thread_stat_findnew = thread_stat_findnew_after_first; + return st; +} + +/* build simple key function one is bigger than two */ +#define SINGLE_KEY(member) \ + static int lock_stat_key_ ## member(struct lock_stat *one, \ + struct lock_stat *two) \ + { \ + return one->member > two->member; \ + } + +SINGLE_KEY(nr_acquired) +SINGLE_KEY(nr_contended) +SINGLE_KEY(wait_time_total) +SINGLE_KEY(wait_time_max) + +static int lock_stat_key_wait_time_min(struct lock_stat *one, + struct lock_stat *two) +{ + u64 s1 = one->wait_time_min; + u64 s2 = two->wait_time_min; + if (s1 == ULLONG_MAX) + s1 = 0; + if (s2 == ULLONG_MAX) + s2 = 0; + return s1 > s2; +} + +struct lock_key { + /* + * name: the value for specify by user + * this should be simpler than raw name of member + * e.g. nr_acquired -> acquired, wait_time_total -> wait_total + */ + const char *name; + int (*key)(struct lock_stat*, struct lock_stat*); +}; + +static const char *sort_key = "acquired"; + +static int (*compare)(struct lock_stat *, struct lock_stat *); + +static struct rb_root result; /* place to store sorted data */ + +#define DEF_KEY_LOCK(name, fn_suffix) \ + { #name, lock_stat_key_ ## fn_suffix } +struct lock_key keys[] = { + DEF_KEY_LOCK(acquired, nr_acquired), + DEF_KEY_LOCK(contended, nr_contended), + DEF_KEY_LOCK(wait_total, wait_time_total), + DEF_KEY_LOCK(wait_min, wait_time_min), + DEF_KEY_LOCK(wait_max, wait_time_max), + + /* extra comparisons much complicated should be here */ + + { NULL, NULL } +}; + +static void select_key(void) +{ + int i; + + for (i = 0; keys[i].name; i++) { + if (!strcmp(keys[i].name, sort_key)) { + compare = keys[i].key; + return; + } + } + + die("Unknown compare key:%s\n", sort_key); +} + +static void insert_to_result(struct lock_stat *st, + int (*bigger)(struct lock_stat *, struct lock_stat *)) +{ + struct rb_node **rb = &result.rb_node; + struct rb_node *parent = NULL; + struct lock_stat *p; + + while (*rb) { + p = container_of(*rb, struct lock_stat, rb); + parent = *rb; + + if (bigger(st, p)) + rb = &(*rb)->rb_left; + else + rb = &(*rb)->rb_right; + } + + rb_link_node(&st->rb, parent, rb); + rb_insert_color(&st->rb, &result); +} + +/* returns left most element of result, and erase it */ +static struct lock_stat *pop_from_result(void) +{ + struct rb_node *node = result.rb_node; + + if (!node) + return NULL; + + while (node->rb_left) + node = node->rb_left; + + rb_erase(node, &result); + return container_of(node, struct lock_stat, rb); +} + +static struct lock_stat *lock_stat_findnew(void *addr, const char *name) +{ + struct list_head *entry = lockhashentry(addr); + struct lock_stat *ret, *new; + + list_for_each_entry(ret, entry, hash_entry) { + if (ret->addr == addr) + return ret; + } + + new = zalloc(sizeof(struct lock_stat)); + if (!new) + goto alloc_failed; + + new->addr = addr; + new->name = zalloc(sizeof(char) * strlen(name) + 1); + if (!new->name) + goto alloc_failed; + strcpy(new->name, name); + + new->wait_time_min = ULLONG_MAX; + + list_add(&new->hash_entry, entry); + return new; + +alloc_failed: + die("memory allocation failed\n"); +} + +static char const *input_name = "perf.data"; + +struct raw_event_sample { + u32 size; + char data[0]; +}; + +struct trace_acquire_event { + void *addr; + const char *name; + int flag; +}; + +struct trace_acquired_event { + void *addr; + const char *name; +}; + +struct trace_contended_event { + void *addr; + const char *name; +}; + +struct trace_release_event { + void *addr; + const char *name; +}; + +struct trace_lock_handler { + void (*acquire_event)(struct trace_acquire_event *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*acquired_event)(struct trace_acquired_event *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*contended_event)(struct trace_contended_event *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*release_event)(struct trace_release_event *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); +}; + +static struct lock_seq_stat *get_seq(struct thread_stat *ts, void *addr) +{ + struct lock_seq_stat *seq; + + list_for_each_entry(seq, &ts->seq_list, list) { + if (seq->addr == addr) + return seq; + } + + seq = zalloc(sizeof(struct lock_seq_stat)); + if (!seq) + die("Not enough memory\n"); + seq->state = SEQ_STATE_UNINITIALIZED; + seq->addr = addr; + + list_add(&seq->list, &ts->seq_list); + return seq; +} + +enum broken_state { + BROKEN_ACQUIRE, + BROKEN_ACQUIRED, + BROKEN_CONTENDED, + BROKEN_RELEASE, + BROKEN_MAX, +}; + +static int bad_hist[BROKEN_MAX]; + +enum acquire_flags { + TRY_LOCK = 1, + READ_LOCK = 2, +}; + +static void +report_lock_acquire_event(struct trace_acquire_event *acquire_event, + struct event *__event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct lock_stat *ls; + struct thread_stat *ts; + struct lock_seq_stat *seq; + + ls = lock_stat_findnew(acquire_event->addr, acquire_event->name); + if (ls->discard) + return; + + ts = thread_stat_findnew(thread->pid); + seq = get_seq(ts, acquire_event->addr); + + switch (seq->state) { + case SEQ_STATE_UNINITIALIZED: + case SEQ_STATE_RELEASED: + if (!acquire_event->flag) { + seq->state = SEQ_STATE_ACQUIRING; + } else { + if (acquire_event->flag & TRY_LOCK) + ls->nr_trylock++; + if (acquire_event->flag & READ_LOCK) + ls->nr_readlock++; + seq->state = SEQ_STATE_READ_ACQUIRED; + seq->read_count = 1; + ls->nr_acquired++; + } + break; + case SEQ_STATE_READ_ACQUIRED: + if (acquire_event->flag & READ_LOCK) { + seq->read_count++; + ls->nr_acquired++; + goto end; + } else { + goto broken; + } + break; + case SEQ_STATE_ACQUIRED: + case SEQ_STATE_ACQUIRING: + case SEQ_STATE_CONTENDED: +broken: + /* broken lock sequence, discard it */ + ls->discard = 1; + bad_hist[BROKEN_ACQUIRE]++; + list_del(&seq->list); + free(seq); + goto end; + break; + default: + BUG_ON("Unknown state of lock sequence found!\n"); + break; + } + + ls->nr_acquire++; + seq->prev_event_time = timestamp; +end: + return; +} + +static void +report_lock_acquired_event(struct trace_acquired_event *acquired_event, + struct event *__event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct lock_stat *ls; + struct thread_stat *ts; + struct lock_seq_stat *seq; + u64 contended_term; + + ls = lock_stat_findnew(acquired_event->addr, acquired_event->name); + if (ls->discard) + return; + + ts = thread_stat_findnew(thread->pid); + seq = get_seq(ts, acquired_event->addr); + + switch (seq->state) { + case SEQ_STATE_UNINITIALIZED: + /* orphan event, do nothing */ + return; + case SEQ_STATE_ACQUIRING: + break; + case SEQ_STATE_CONTENDED: + contended_term = timestamp - seq->prev_event_time; + ls->wait_time_total += contended_term; + if (contended_term < ls->wait_time_min) + ls->wait_time_min = contended_term; + if (ls->wait_time_max < contended_term) + ls->wait_time_max = contended_term; + break; + case SEQ_STATE_RELEASED: + case SEQ_STATE_ACQUIRED: + case SEQ_STATE_READ_ACQUIRED: + /* broken lock sequence, discard it */ + ls->discard = 1; + bad_hist[BROKEN_ACQUIRED]++; + list_del(&seq->list); + free(seq); + goto end; + break; + + default: + BUG_ON("Unknown state of lock sequence found!\n"); + break; + } + + seq->state = SEQ_STATE_ACQUIRED; + ls->nr_acquired++; + seq->prev_event_time = timestamp; +end: + return; +} + +static void +report_lock_contended_event(struct trace_contended_event *contended_event, + struct event *__event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct lock_stat *ls; + struct thread_stat *ts; + struct lock_seq_stat *seq; + + ls = lock_stat_findnew(contended_event->addr, contended_event->name); + if (ls->discard) + return; + + ts = thread_stat_findnew(thread->pid); + seq = get_seq(ts, contended_event->addr); + + switch (seq->state) { + case SEQ_STATE_UNINITIALIZED: + /* orphan event, do nothing */ + return; + case SEQ_STATE_ACQUIRING: + break; + case SEQ_STATE_RELEASED: + case SEQ_STATE_ACQUIRED: + case SEQ_STATE_READ_ACQUIRED: + case SEQ_STATE_CONTENDED: + /* broken lock sequence, discard it */ + ls->discard = 1; + bad_hist[BROKEN_CONTENDED]++; + list_del(&seq->list); + free(seq); + goto end; + break; + default: + BUG_ON("Unknown state of lock sequence found!\n"); + break; + } + + seq->state = SEQ_STATE_CONTENDED; + ls->nr_contended++; + seq->prev_event_time = timestamp; +end: + return; +} + +static void +report_lock_release_event(struct trace_release_event *release_event, + struct event *__event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct lock_stat *ls; + struct thread_stat *ts; + struct lock_seq_stat *seq; + + ls = lock_stat_findnew(release_event->addr, release_event->name); + if (ls->discard) + return; + + ts = thread_stat_findnew(thread->pid); + seq = get_seq(ts, release_event->addr); + + switch (seq->state) { + case SEQ_STATE_UNINITIALIZED: + goto end; + break; + case SEQ_STATE_ACQUIRED: + break; + case SEQ_STATE_READ_ACQUIRED: + seq->read_count--; + BUG_ON(seq->read_count < 0); + if (!seq->read_count) { + ls->nr_release++; + goto end; + } + break; + case SEQ_STATE_ACQUIRING: + case SEQ_STATE_CONTENDED: + case SEQ_STATE_RELEASED: + /* broken lock sequence, discard it */ + ls->discard = 1; + bad_hist[BROKEN_RELEASE]++; + goto free_seq; + break; + default: + BUG_ON("Unknown state of lock sequence found!\n"); + break; + } + + ls->nr_release++; +free_seq: + list_del(&seq->list); + free(seq); +end: + return; +} + +/* lock oriented handlers */ +/* TODO: handlers for CPU oriented, thread oriented */ +static struct trace_lock_handler report_lock_ops = { + .acquire_event = report_lock_acquire_event, + .acquired_event = report_lock_acquired_event, + .contended_event = report_lock_contended_event, + .release_event = report_lock_release_event, +}; + +static struct trace_lock_handler *trace_handler; + +static void +process_lock_acquire_event(void *data, + struct event *event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_acquire_event acquire_event; + u64 tmp; /* this is required for casting... */ + + tmp = raw_field_value(event, "lockdep_addr", data); + memcpy(&acquire_event.addr, &tmp, sizeof(void *)); + acquire_event.name = (char *)raw_field_ptr(event, "name", data); + acquire_event.flag = (int)raw_field_value(event, "flag", data); + + if (trace_handler->acquire_event) + trace_handler->acquire_event(&acquire_event, event, cpu, timestamp, thread); +} + +static void +process_lock_acquired_event(void *data, + struct event *event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_acquired_event acquired_event; + u64 tmp; /* this is required for casting... */ + + tmp = raw_field_value(event, "lockdep_addr", data); + memcpy(&acquired_event.addr, &tmp, sizeof(void *)); + acquired_event.name = (char *)raw_field_ptr(event, "name", data); + + if (trace_handler->acquire_event) + trace_handler->acquired_event(&acquired_event, event, cpu, timestamp, thread); +} + +static void +process_lock_contended_event(void *data, + struct event *event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_contended_event contended_event; + u64 tmp; /* this is required for casting... */ + + tmp = raw_field_value(event, "lockdep_addr", data); + memcpy(&contended_event.addr, &tmp, sizeof(void *)); + contended_event.name = (char *)raw_field_ptr(event, "name", data); + + if (trace_handler->acquire_event) + trace_handler->contended_event(&contended_event, event, cpu, timestamp, thread); +} + +static void +process_lock_release_event(void *data, + struct event *event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_release_event release_event; + u64 tmp; /* this is required for casting... */ + + tmp = raw_field_value(event, "lockdep_addr", data); + memcpy(&release_event.addr, &tmp, sizeof(void *)); + release_event.name = (char *)raw_field_ptr(event, "name", data); + + if (trace_handler->acquire_event) + trace_handler->release_event(&release_event, event, cpu, timestamp, thread); +} + +static void +process_raw_event(void *data, int cpu, u64 timestamp, struct thread *thread) +{ + struct event *event; + int type; + + type = trace_parse_common_type(data); + event = trace_find_event(type); + + if (!strcmp(event->name, "lock_acquire")) + process_lock_acquire_event(data, event, cpu, timestamp, thread); + if (!strcmp(event->name, "lock_acquired")) + process_lock_acquired_event(data, event, cpu, timestamp, thread); + if (!strcmp(event->name, "lock_contended")) + process_lock_contended_event(data, event, cpu, timestamp, thread); + if (!strcmp(event->name, "lock_release")) + process_lock_release_event(data, event, cpu, timestamp, thread); +} + +static void print_bad_events(int bad, int total) +{ + /* Output for debug, this have to be removed */ + int i; + const char *name[4] = + { "acquire", "acquired", "contended", "release" }; + + pr_info("\n=== output for debug===\n\n"); + pr_info("bad: %d, total: %d\n", bad, total); + pr_info("bad rate: %f %%\n", (double)bad / (double)total * 100); + pr_info("histogram of events caused bad sequence\n"); + for (i = 0; i < BROKEN_MAX; i++) + pr_info(" %10s: %d\n", name[i], bad_hist[i]); +} + +/* TODO: various way to print, coloring, nano or milli sec */ +static void print_result(void) +{ + struct lock_stat *st; + char cut_name[20]; + int bad, total; + + pr_info("%20s ", "Name"); + pr_info("%10s ", "acquired"); + pr_info("%10s ", "contended"); + + pr_info("%15s ", "total wait (ns)"); + pr_info("%15s ", "max wait (ns)"); + pr_info("%15s ", "min wait (ns)"); + + pr_info("\n\n"); + + bad = total = 0; + while ((st = pop_from_result())) { + total++; + if (st->discard) { + bad++; + continue; + } + bzero(cut_name, 20); + + if (strlen(st->name) < 16) { + /* output raw name */ + pr_info("%20s ", st->name); + } else { + strncpy(cut_name, st->name, 16); + cut_name[16] = '.'; + cut_name[17] = '.'; + cut_name[18] = '.'; + cut_name[19] = '\0'; + /* cut off name for saving output style */ + pr_info("%20s ", cut_name); + } + + pr_info("%10u ", st->nr_acquired); + pr_info("%10u ", st->nr_contended); + + pr_info("%15" PRIu64 " ", st->wait_time_total); + pr_info("%15" PRIu64 " ", st->wait_time_max); + pr_info("%15" PRIu64 " ", st->wait_time_min == ULLONG_MAX ? + 0 : st->wait_time_min); + pr_info("\n"); + } + + print_bad_events(bad, total); +} + +static bool info_threads, info_map; + +static void dump_threads(void) +{ + struct thread_stat *st; + struct rb_node *node; + struct thread *t; + + pr_info("%10s: comm\n", "Thread ID"); + + node = rb_first(&thread_stats); + while (node) { + st = container_of(node, struct thread_stat, rb); + t = perf_session__findnew(session, st->tid); + pr_info("%10d: %s\n", st->tid, t->comm); + node = rb_next(node); + }; +} + +static void dump_map(void) +{ + unsigned int i; + struct lock_stat *st; + + pr_info("Address of instance: name of class\n"); + for (i = 0; i < LOCKHASH_SIZE; i++) { + list_for_each_entry(st, &lockhash_table[i], hash_entry) { + pr_info(" %p: %s\n", st->addr, st->name); + } + } +} + +static void dump_info(void) +{ + if (info_threads) + dump_threads(); + else if (info_map) + dump_map(); + else + die("Unknown type of information\n"); +} + +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel __used, + struct perf_session *s) +{ + struct thread *thread = perf_session__findnew(s, sample->tid); + + if (thread == NULL) { + pr_debug("problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + process_raw_event(sample->raw_data, sample->cpu, sample->time, thread); + + return 0; +} + +static struct perf_event_ops eops = { + .sample = process_sample_event, + .comm = perf_event__process_comm, + .ordered_samples = true, +}; + +static int read_events(void) +{ + session = perf_session__new(input_name, O_RDONLY, 0, false, &eops); + if (!session) + die("Initializing perf session failed\n"); + + return perf_session__process_events(session, &eops); +} + +static void sort_result(void) +{ + unsigned int i; + struct lock_stat *st; + + for (i = 0; i < LOCKHASH_SIZE; i++) { + list_for_each_entry(st, &lockhash_table[i], hash_entry) { + insert_to_result(st, compare); + } + } +} + +static void __cmd_report(void) +{ + setup_pager(); + select_key(); + read_events(); + sort_result(); + print_result(); +} + +static const char * const report_usage[] = { + "perf lock report [<options>]", + NULL +}; + +static const struct option report_options[] = { + OPT_STRING('k', "key", &sort_key, "acquired", + "key for sorting (acquired / contended / wait_total / wait_max / wait_min)"), + /* TODO: type */ + OPT_END() +}; + +static const char * const info_usage[] = { + "perf lock info [<options>]", + NULL +}; + +static const struct option info_options[] = { + OPT_BOOLEAN('t', "threads", &info_threads, + "dump thread list in perf.data"), + OPT_BOOLEAN('m', "map", &info_map, + "map of lock instances (name:address table)"), + OPT_END() +}; + +static const char * const lock_usage[] = { + "perf lock [<options>] {record|trace|report}", + NULL +}; + +static const struct option lock_options[] = { + OPT_STRING('i', "input", &input_name, "file", "input file name"), + OPT_INCR('v', "verbose", &verbose, "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, "dump raw trace in ASCII"), + OPT_END() +}; + +static const char *record_args[] = { + "record", + "-R", + "-f", + "-m", "1024", + "-c", "1", + "-e", "lock:lock_acquire:r", + "-e", "lock:lock_acquired:r", + "-e", "lock:lock_contended:r", + "-e", "lock:lock_release:r", +}; + +static int __cmd_record(int argc, const char **argv) +{ + unsigned int rec_argc, i, j; + const char **rec_argv; + + rec_argc = ARRAY_SIZE(record_args) + argc - 1; + rec_argv = calloc(rec_argc + 1, sizeof(char *)); + + if (rec_argv == NULL) + return -ENOMEM; + + for (i = 0; i < ARRAY_SIZE(record_args); i++) + rec_argv[i] = strdup(record_args[i]); + + for (j = 1; j < (unsigned int)argc; j++, i++) + rec_argv[i] = argv[j]; + + BUG_ON(i != rec_argc); + + return cmd_record(i, rec_argv, NULL); +} + +int cmd_lock(int argc, const char **argv, const char *prefix __used) +{ + unsigned int i; + + symbol__init(); + for (i = 0; i < LOCKHASH_SIZE; i++) + INIT_LIST_HEAD(lockhash_table + i); + + argc = parse_options(argc, argv, lock_options, lock_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + if (!argc) + usage_with_options(lock_usage, lock_options); + + if (!strncmp(argv[0], "rec", 3)) { + return __cmd_record(argc, argv); + } else if (!strncmp(argv[0], "report", 6)) { + trace_handler = &report_lock_ops; + if (argc) { + argc = parse_options(argc, argv, + report_options, report_usage, 0); + if (argc) + usage_with_options(report_usage, report_options); + } + __cmd_report(); + } else if (!strcmp(argv[0], "script")) { + /* Aliased to 'perf script' */ + return cmd_script(argc, argv, prefix); + } else if (!strcmp(argv[0], "info")) { + if (argc) { + argc = parse_options(argc, argv, + info_options, info_usage, 0); + if (argc) + usage_with_options(info_usage, info_options); + } + /* recycling report_lock_ops */ + trace_handler = &report_lock_ops; + setup_pager(); + read_events(); + dump_info(); + } else { + usage_with_options(lock_usage, lock_options); + } + + return 0; +} diff --git a/tools/perf/builtin-probe.c b/tools/perf/builtin-probe.c new file mode 100644 index 00000000..2c0e64d0 --- /dev/null +++ b/tools/perf/builtin-probe.c @@ -0,0 +1,398 @@ +/* + * builtin-probe.c + * + * Builtin probe command: Set up probe events by C expression + * + * Written by Masami Hiramatsu <mhiramat@redhat.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + */ +#define _GNU_SOURCE +#include <sys/utsname.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <fcntl.h> +#include <errno.h> +#include <stdio.h> +#include <unistd.h> +#include <stdlib.h> +#include <string.h> + +#undef _GNU_SOURCE +#include "perf.h" +#include "builtin.h" +#include "util/util.h" +#include "util/strlist.h" +#include "util/strfilter.h" +#include "util/symbol.h" +#include "util/debug.h" +#include "util/debugfs.h" +#include "util/parse-options.h" +#include "util/probe-finder.h" +#include "util/probe-event.h" + +#define DEFAULT_VAR_FILTER "!__k???tab_* & !__crc_*" +#define DEFAULT_FUNC_FILTER "!_*" +#define MAX_PATH_LEN 256 + +/* Session management structure */ +static struct { + bool list_events; + bool force_add; + bool show_lines; + bool show_vars; + bool show_ext_vars; + bool show_funcs; + bool mod_events; + int nevents; + struct perf_probe_event events[MAX_PROBES]; + struct strlist *dellist; + struct line_range line_range; + const char *target_module; + int max_probe_points; + struct strfilter *filter; +} params; + +/* Parse an event definition. Note that any error must die. */ +static int parse_probe_event(const char *str) +{ + struct perf_probe_event *pev = ¶ms.events[params.nevents]; + int ret; + + pr_debug("probe-definition(%d): %s\n", params.nevents, str); + if (++params.nevents == MAX_PROBES) { + pr_err("Too many probes (> %d) were specified.", MAX_PROBES); + return -1; + } + + /* Parse a perf-probe command into event */ + ret = parse_perf_probe_command(str, pev); + pr_debug("%d arguments\n", pev->nargs); + + return ret; +} + +static int parse_probe_event_argv(int argc, const char **argv) +{ + int i, len, ret; + char *buf; + + /* Bind up rest arguments */ + len = 0; + for (i = 0; i < argc; i++) + len += strlen(argv[i]) + 1; + buf = zalloc(len + 1); + if (buf == NULL) + return -ENOMEM; + len = 0; + for (i = 0; i < argc; i++) + len += sprintf(&buf[len], "%s ", argv[i]); + params.mod_events = true; + ret = parse_probe_event(buf); + free(buf); + return ret; +} + +static int opt_add_probe_event(const struct option *opt __used, + const char *str, int unset __used) +{ + if (str) { + params.mod_events = true; + return parse_probe_event(str); + } else + return 0; +} + +static int opt_del_probe_event(const struct option *opt __used, + const char *str, int unset __used) +{ + if (str) { + params.mod_events = true; + if (!params.dellist) + params.dellist = strlist__new(true, NULL); + strlist__add(params.dellist, str); + } + return 0; +} + +#ifdef DWARF_SUPPORT +static int opt_show_lines(const struct option *opt __used, + const char *str, int unset __used) +{ + int ret = 0; + + if (str) + ret = parse_line_range_desc(str, ¶ms.line_range); + INIT_LIST_HEAD(¶ms.line_range.line_list); + params.show_lines = true; + + return ret; +} + +static int opt_show_vars(const struct option *opt __used, + const char *str, int unset __used) +{ + struct perf_probe_event *pev = ¶ms.events[params.nevents]; + int ret; + + if (!str) + return 0; + + ret = parse_probe_event(str); + if (!ret && pev->nargs != 0) { + pr_err(" Error: '--vars' doesn't accept arguments.\n"); + return -EINVAL; + } + params.show_vars = true; + + return ret; +} +#endif + +static int opt_set_filter(const struct option *opt __used, + const char *str, int unset __used) +{ + const char *err; + + if (str) { + pr_debug2("Set filter: %s\n", str); + if (params.filter) + strfilter__delete(params.filter); + params.filter = strfilter__new(str, &err); + if (!params.filter) { + pr_err("Filter parse error at %td.\n", err - str + 1); + pr_err("Source: \"%s\"\n", str); + pr_err(" %*c\n", (int)(err - str + 1), '^'); + return -EINVAL; + } + } + + return 0; +} + +static const char * const probe_usage[] = { + "perf probe [<options>] 'PROBEDEF' ['PROBEDEF' ...]", + "perf probe [<options>] --add 'PROBEDEF' [--add 'PROBEDEF' ...]", + "perf probe [<options>] --del '[GROUP:]EVENT' ...", + "perf probe --list", +#ifdef DWARF_SUPPORT + "perf probe [<options>] --line 'LINEDESC'", + "perf probe [<options>] --vars 'PROBEPOINT'", +#endif + NULL +}; + +static const struct option options[] = { + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show parsed arguments, etc)"), + OPT_BOOLEAN('l', "list", ¶ms.list_events, + "list up current probe events"), + OPT_CALLBACK('d', "del", NULL, "[GROUP:]EVENT", "delete a probe event.", + opt_del_probe_event), + OPT_CALLBACK('a', "add", NULL, +#ifdef DWARF_SUPPORT + "[EVENT=]FUNC[@SRC][+OFF|%return|:RL|;PT]|SRC:AL|SRC;PT" + " [[NAME=]ARG ...]", +#else + "[EVENT=]FUNC[+OFF|%return] [[NAME=]ARG ...]", +#endif + "probe point definition, where\n" + "\t\tGROUP:\tGroup name (optional)\n" + "\t\tEVENT:\tEvent name\n" + "\t\tFUNC:\tFunction name\n" + "\t\tOFF:\tOffset from function entry (in byte)\n" + "\t\t%return:\tPut the probe at function return\n" +#ifdef DWARF_SUPPORT + "\t\tSRC:\tSource code path\n" + "\t\tRL:\tRelative line number from function entry.\n" + "\t\tAL:\tAbsolute line number in file.\n" + "\t\tPT:\tLazy expression of line code.\n" + "\t\tARG:\tProbe argument (local variable name or\n" + "\t\t\tkprobe-tracer argument format.)\n", +#else + "\t\tARG:\tProbe argument (kprobe-tracer argument format.)\n", +#endif + opt_add_probe_event), + OPT_BOOLEAN('f', "force", ¶ms.force_add, "forcibly add events" + " with existing name"), +#ifdef DWARF_SUPPORT + OPT_CALLBACK('L', "line", NULL, + "FUNC[:RLN[+NUM|-RLN2]]|SRC:ALN[+NUM|-ALN2]", + "Show source code lines.", opt_show_lines), + OPT_CALLBACK('V', "vars", NULL, + "FUNC[@SRC][+OFF|%return|:RL|;PT]|SRC:AL|SRC;PT", + "Show accessible variables on PROBEDEF", opt_show_vars), + OPT_BOOLEAN('\0', "externs", ¶ms.show_ext_vars, + "Show external variables too (with --vars only)"), + OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name, + "file", "vmlinux pathname"), + OPT_STRING('s', "source", &symbol_conf.source_prefix, + "directory", "path to kernel source"), + OPT_STRING('m', "module", ¶ms.target_module, + "modname", "target module name"), +#endif + OPT__DRY_RUN(&probe_event_dry_run), + OPT_INTEGER('\0', "max-probes", ¶ms.max_probe_points, + "Set how many probe points can be found for a probe."), + OPT_BOOLEAN('F', "funcs", ¶ms.show_funcs, + "Show potential probe-able functions."), + OPT_CALLBACK('\0', "filter", NULL, + "[!]FILTER", "Set a filter (with --vars/funcs only)\n" + "\t\t\t(default: \"" DEFAULT_VAR_FILTER "\" for --vars,\n" + "\t\t\t \"" DEFAULT_FUNC_FILTER "\" for --funcs)", + opt_set_filter), + OPT_END() +}; + +int cmd_probe(int argc, const char **argv, const char *prefix __used) +{ + int ret; + + argc = parse_options(argc, argv, options, probe_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + if (argc > 0) { + if (strcmp(argv[0], "-") == 0) { + pr_warning(" Error: '-' is not supported.\n"); + usage_with_options(probe_usage, options); + } + ret = parse_probe_event_argv(argc, argv); + if (ret < 0) { + pr_err(" Error: Parse Error. (%d)\n", ret); + return ret; + } + } + + if (params.max_probe_points == 0) + params.max_probe_points = MAX_PROBES; + + if ((!params.nevents && !params.dellist && !params.list_events && + !params.show_lines && !params.show_funcs)) + usage_with_options(probe_usage, options); + + /* + * Only consider the user's kernel image path if given. + */ + symbol_conf.try_vmlinux_path = (symbol_conf.vmlinux_name == NULL); + + if (params.list_events) { + if (params.mod_events) { + pr_err(" Error: Don't use --list with --add/--del.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_lines) { + pr_err(" Error: Don't use --list with --line.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_vars) { + pr_err(" Error: Don't use --list with --vars.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_funcs) { + pr_err(" Error: Don't use --list with --funcs.\n"); + usage_with_options(probe_usage, options); + } + ret = show_perf_probe_events(); + if (ret < 0) + pr_err(" Error: Failed to show event list. (%d)\n", + ret); + return ret; + } + if (params.show_funcs) { + if (params.nevents != 0 || params.dellist) { + pr_err(" Error: Don't use --funcs with" + " --add/--del.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_lines) { + pr_err(" Error: Don't use --funcs with --line.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_vars) { + pr_err(" Error: Don't use --funcs with --vars.\n"); + usage_with_options(probe_usage, options); + } + if (!params.filter) + params.filter = strfilter__new(DEFAULT_FUNC_FILTER, + NULL); + ret = show_available_funcs(params.target_module, + params.filter); + strfilter__delete(params.filter); + if (ret < 0) + pr_err(" Error: Failed to show functions." + " (%d)\n", ret); + return ret; + } + +#ifdef DWARF_SUPPORT + if (params.show_lines) { + if (params.mod_events) { + pr_err(" Error: Don't use --line with" + " --add/--del.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_vars) { + pr_err(" Error: Don't use --line with --vars.\n"); + usage_with_options(probe_usage, options); + } + + ret = show_line_range(¶ms.line_range, params.target_module); + if (ret < 0) + pr_err(" Error: Failed to show lines. (%d)\n", ret); + return ret; + } + if (params.show_vars) { + if (params.mod_events) { + pr_err(" Error: Don't use --vars with" + " --add/--del.\n"); + usage_with_options(probe_usage, options); + } + if (!params.filter) + params.filter = strfilter__new(DEFAULT_VAR_FILTER, + NULL); + + ret = show_available_vars(params.events, params.nevents, + params.max_probe_points, + params.target_module, + params.filter, + params.show_ext_vars); + strfilter__delete(params.filter); + if (ret < 0) + pr_err(" Error: Failed to show vars. (%d)\n", ret); + return ret; + } +#endif + + if (params.dellist) { + ret = del_perf_probe_events(params.dellist); + strlist__delete(params.dellist); + if (ret < 0) { + pr_err(" Error: Failed to delete events. (%d)\n", ret); + return ret; + } + } + + if (params.nevents) { + ret = add_perf_probe_events(params.events, params.nevents, + params.max_probe_points, + params.target_module, + params.force_add); + if (ret < 0) { + pr_err(" Error: Failed to add events. (%d)\n", ret); + return ret; + } + } + return 0; +} diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c new file mode 100644 index 00000000..8e2c8579 --- /dev/null +++ b/tools/perf/builtin-record.c @@ -0,0 +1,887 @@ +/* + * builtin-record.c + * + * Builtin record command: Record the profile of a workload + * (or a CPU, or a PID) into the perf.data output file - for + * later analysis via perf report. + */ +#define _FILE_OFFSET_BITS 64 + +#include "builtin.h" + +#include "perf.h" + +#include "util/build-id.h" +#include "util/util.h" +#include "util/parse-options.h" +#include "util/parse-events.h" + +#include "util/header.h" +#include "util/event.h" +#include "util/evlist.h" +#include "util/evsel.h" +#include "util/debug.h" +#include "util/session.h" +#include "util/symbol.h" +#include "util/cpumap.h" +#include "util/thread_map.h" + +#include <unistd.h> +#include <sched.h> +#include <sys/mman.h> + +#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) + +enum write_mode_t { + WRITE_FORCE, + WRITE_APPEND +}; + +static u64 user_interval = ULLONG_MAX; +static u64 default_interval = 0; + +static unsigned int page_size; +static unsigned int mmap_pages = UINT_MAX; +static unsigned int user_freq = UINT_MAX; +static int freq = 1000; +static int output; +static int pipe_output = 0; +static const char *output_name = NULL; +static int group = 0; +static int realtime_prio = 0; +static bool nodelay = false; +static bool raw_samples = false; +static bool sample_id_all_avail = true; +static bool system_wide = false; +static pid_t target_pid = -1; +static pid_t target_tid = -1; +static pid_t child_pid = -1; +static bool no_inherit = false; +static enum write_mode_t write_mode = WRITE_FORCE; +static bool call_graph = false; +static bool inherit_stat = false; +static bool no_samples = false; +static bool sample_address = false; +static bool sample_time = false; +static bool no_buildid = false; +static bool no_buildid_cache = false; +static struct perf_evlist *evsel_list; + +static long samples = 0; +static u64 bytes_written = 0; + +static int file_new = 1; +static off_t post_processing_offset; + +static struct perf_session *session; +static const char *cpu_list; + +static void advance_output(size_t size) +{ + bytes_written += size; +} + +static void write_output(void *buf, size_t size) +{ + while (size) { + int ret = write(output, buf, size); + + if (ret < 0) + die("failed to write"); + + size -= ret; + buf += ret; + + bytes_written += ret; + } +} + +static int process_synthesized_event(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *self __used) +{ + write_output(event, event->header.size); + return 0; +} + +static void mmap_read(struct perf_mmap *md) +{ + unsigned int head = perf_mmap__read_head(md); + unsigned int old = md->prev; + unsigned char *data = md->base + page_size; + unsigned long size; + void *buf; + + if (old == head) + return; + + samples++; + + size = head - old; + + if ((old & md->mask) + size != (head & md->mask)) { + buf = &data[old & md->mask]; + size = md->mask + 1 - (old & md->mask); + old += size; + + write_output(buf, size); + } + + buf = &data[old & md->mask]; + size = head - old; + old += size; + + write_output(buf, size); + + md->prev = old; + perf_mmap__write_tail(md, old); +} + +static volatile int done = 0; +static volatile int signr = -1; + +static void sig_handler(int sig) +{ + done = 1; + signr = sig; +} + +static void sig_atexit(void) +{ + if (child_pid > 0) + kill(child_pid, SIGTERM); + + if (signr == -1 || signr == SIGUSR1) + return; + + signal(signr, SIG_DFL); + kill(getpid(), signr); +} + +static void config_attr(struct perf_evsel *evsel, struct perf_evlist *evlist) +{ + struct perf_event_attr *attr = &evsel->attr; + int track = !evsel->idx; /* only the first counter needs these */ + + attr->inherit = !no_inherit; + attr->read_format = PERF_FORMAT_TOTAL_TIME_ENABLED | + PERF_FORMAT_TOTAL_TIME_RUNNING | + PERF_FORMAT_ID; + + attr->sample_type |= PERF_SAMPLE_IP | PERF_SAMPLE_TID; + + if (evlist->nr_entries > 1) + attr->sample_type |= PERF_SAMPLE_ID; + + /* + * We default some events to a 1 default interval. But keep + * it a weak assumption overridable by the user. + */ + if (!attr->sample_period || (user_freq != UINT_MAX && + user_interval != ULLONG_MAX)) { + if (freq) { + attr->sample_type |= PERF_SAMPLE_PERIOD; + attr->freq = 1; + attr->sample_freq = freq; + } else { + attr->sample_period = default_interval; + } + } + + if (no_samples) + attr->sample_freq = 0; + + if (inherit_stat) + attr->inherit_stat = 1; + + if (sample_address) { + attr->sample_type |= PERF_SAMPLE_ADDR; + attr->mmap_data = track; + } + + if (call_graph) + attr->sample_type |= PERF_SAMPLE_CALLCHAIN; + + if (system_wide) + attr->sample_type |= PERF_SAMPLE_CPU; + + if (sample_id_all_avail && + (sample_time || system_wide || !no_inherit || cpu_list)) + attr->sample_type |= PERF_SAMPLE_TIME; + + if (raw_samples) { + attr->sample_type |= PERF_SAMPLE_TIME; + attr->sample_type |= PERF_SAMPLE_RAW; + attr->sample_type |= PERF_SAMPLE_CPU; + } + + if (nodelay) { + attr->watermark = 0; + attr->wakeup_events = 1; + } + + attr->mmap = track; + attr->comm = track; + + if (target_pid == -1 && target_tid == -1 && !system_wide) { + attr->disabled = 1; + attr->enable_on_exec = 1; + } +} + +static bool perf_evlist__equal(struct perf_evlist *evlist, + struct perf_evlist *other) +{ + struct perf_evsel *pos, *pair; + + if (evlist->nr_entries != other->nr_entries) + return false; + + pair = list_entry(other->entries.next, struct perf_evsel, node); + + list_for_each_entry(pos, &evlist->entries, node) { + if (memcmp(&pos->attr, &pair->attr, sizeof(pos->attr) != 0)) + return false; + pair = list_entry(pair->node.next, struct perf_evsel, node); + } + + return true; +} + +static void open_counters(struct perf_evlist *evlist) +{ + struct perf_evsel *pos; + + if (evlist->cpus->map[0] < 0) + no_inherit = true; + + list_for_each_entry(pos, &evlist->entries, node) { + struct perf_event_attr *attr = &pos->attr; + /* + * Check if parse_single_tracepoint_event has already asked for + * PERF_SAMPLE_TIME. + * + * XXX this is kludgy but short term fix for problems introduced by + * eac23d1c that broke 'perf script' by having different sample_types + * when using multiple tracepoint events when we use a perf binary + * that tries to use sample_id_all on an older kernel. + * + * We need to move counter creation to perf_session, support + * different sample_types, etc. + */ + bool time_needed = attr->sample_type & PERF_SAMPLE_TIME; + + config_attr(pos, evlist); +retry_sample_id: + attr->sample_id_all = sample_id_all_avail ? 1 : 0; +try_again: + if (perf_evsel__open(pos, evlist->cpus, evlist->threads, group) < 0) { + int err = errno; + + if (err == EPERM || err == EACCES) { + ui__warning_paranoid(); + exit(EXIT_FAILURE); + } else if (err == ENODEV && cpu_list) { + die("No such device - did you specify" + " an out-of-range profile CPU?\n"); + } else if (err == EINVAL && sample_id_all_avail) { + /* + * Old kernel, no attr->sample_id_type_all field + */ + sample_id_all_avail = false; + if (!sample_time && !raw_samples && !time_needed) + attr->sample_type &= ~PERF_SAMPLE_TIME; + + goto retry_sample_id; + } + + /* + * If it's cycles then fall back to hrtimer + * based cpu-clock-tick sw counter, which + * is always available even if no PMU support: + */ + if (attr->type == PERF_TYPE_HARDWARE + && attr->config == PERF_COUNT_HW_CPU_CYCLES) { + + if (verbose) + ui__warning("The cycles event is not supported, " + "trying to fall back to cpu-clock-ticks\n"); + attr->type = PERF_TYPE_SOFTWARE; + attr->config = PERF_COUNT_SW_CPU_CLOCK; + goto try_again; + } + + if (err == ENOENT) { + ui__warning("The %s event is not supported.\n", + event_name(pos)); + exit(EXIT_FAILURE); + } + + printf("\n"); + error("sys_perf_event_open() syscall returned with %d (%s). /bin/dmesg may provide additional information.\n", + err, strerror(err)); + +#if defined(__i386__) || defined(__x86_64__) + if (attr->type == PERF_TYPE_HARDWARE && err == EOPNOTSUPP) + die("No hardware sampling interrupt available." + " No APIC? If so then you can boot the kernel" + " with the \"lapic\" boot parameter to" + " force-enable it.\n"); +#endif + + die("No CONFIG_PERF_EVENTS=y kernel support configured?\n"); + } + } + + if (perf_evlist__set_filters(evlist)) { + error("failed to set filter with %d (%s)\n", errno, + strerror(errno)); + exit(-1); + } + + if (perf_evlist__mmap(evlist, mmap_pages, false) < 0) + die("failed to mmap with %d (%s)\n", errno, strerror(errno)); + + if (file_new) + session->evlist = evlist; + else { + if (!perf_evlist__equal(session->evlist, evlist)) { + fprintf(stderr, "incompatible append\n"); + exit(-1); + } + } + + perf_session__update_sample_type(session); +} + +static int process_buildids(void) +{ + u64 size = lseek(output, 0, SEEK_CUR); + + if (size == 0) + return 0; + + session->fd = output; + return __perf_session__process_events(session, post_processing_offset, + size - post_processing_offset, + size, &build_id__mark_dso_hit_ops); +} + +static void atexit_header(void) +{ + if (!pipe_output) { + session->header.data_size += bytes_written; + + if (!no_buildid) + process_buildids(); + perf_session__write_header(session, evsel_list, output, true); + perf_session__delete(session); + perf_evlist__delete(evsel_list); + symbol__exit(); + } +} + +static void perf_event__synthesize_guest_os(struct machine *machine, void *data) +{ + int err; + struct perf_session *psession = data; + + if (machine__is_host(machine)) + return; + + /* + *As for guest kernel when processing subcommand record&report, + *we arrange module mmap prior to guest kernel mmap and trigger + *a preload dso because default guest module symbols are loaded + *from guest kallsyms instead of /lib/modules/XXX/XXX. This + *method is used to avoid symbol missing when the first addr is + *in module instead of in guest kernel. + */ + err = perf_event__synthesize_modules(process_synthesized_event, + psession, machine); + if (err < 0) + pr_err("Couldn't record guest kernel [%d]'s reference" + " relocation symbol.\n", machine->pid); + + /* + * We use _stext for guest kernel because guest kernel's /proc/kallsyms + * have no _text sometimes. + */ + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + psession, machine, "_text"); + if (err < 0) + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + psession, machine, + "_stext"); + if (err < 0) + pr_err("Couldn't record guest kernel [%d]'s reference" + " relocation symbol.\n", machine->pid); +} + +static struct perf_event_header finished_round_event = { + .size = sizeof(struct perf_event_header), + .type = PERF_RECORD_FINISHED_ROUND, +}; + +static void mmap_read_all(void) +{ + int i; + + for (i = 0; i < evsel_list->nr_mmaps; i++) { + if (evsel_list->mmap[i].base) + mmap_read(&evsel_list->mmap[i]); + } + + if (perf_header__has_feat(&session->header, HEADER_TRACE_INFO)) + write_output(&finished_round_event, sizeof(finished_round_event)); +} + +static int __cmd_record(int argc, const char **argv) +{ + int i; + struct stat st; + int flags; + int err; + unsigned long waking = 0; + int child_ready_pipe[2], go_pipe[2]; + const bool forks = argc > 0; + char buf; + struct machine *machine; + + page_size = sysconf(_SC_PAGE_SIZE); + + atexit(sig_atexit); + signal(SIGCHLD, sig_handler); + signal(SIGINT, sig_handler); + signal(SIGUSR1, sig_handler); + + if (forks && (pipe(child_ready_pipe) < 0 || pipe(go_pipe) < 0)) { + perror("failed to create pipes"); + exit(-1); + } + + if (!output_name) { + if (!fstat(STDOUT_FILENO, &st) && S_ISFIFO(st.st_mode)) + pipe_output = 1; + else + output_name = "perf.data"; + } + if (output_name) { + if (!strcmp(output_name, "-")) + pipe_output = 1; + else if (!stat(output_name, &st) && st.st_size) { + if (write_mode == WRITE_FORCE) { + char oldname[PATH_MAX]; + snprintf(oldname, sizeof(oldname), "%s.old", + output_name); + unlink(oldname); + rename(output_name, oldname); + } + } else if (write_mode == WRITE_APPEND) { + write_mode = WRITE_FORCE; + } + } + + flags = O_CREAT|O_RDWR; + if (write_mode == WRITE_APPEND) + file_new = 0; + else + flags |= O_TRUNC; + + if (pipe_output) + output = STDOUT_FILENO; + else + output = open(output_name, flags, S_IRUSR | S_IWUSR); + if (output < 0) { + perror("failed to create output file"); + exit(-1); + } + + session = perf_session__new(output_name, O_WRONLY, + write_mode == WRITE_FORCE, false, NULL); + if (session == NULL) { + pr_err("Not enough memory for reading perf file header\n"); + return -1; + } + + if (!no_buildid) + perf_header__set_feat(&session->header, HEADER_BUILD_ID); + + if (!file_new) { + err = perf_session__read_header(session, output); + if (err < 0) + goto out_delete_session; + } + + if (have_tracepoints(&evsel_list->entries)) + perf_header__set_feat(&session->header, HEADER_TRACE_INFO); + + /* 512 kiB: default amount of unprivileged mlocked memory */ + if (mmap_pages == UINT_MAX) + mmap_pages = (512 * 1024) / page_size; + + if (forks) { + child_pid = fork(); + if (child_pid < 0) { + perror("failed to fork"); + exit(-1); + } + + if (!child_pid) { + if (pipe_output) + dup2(2, 1); + close(child_ready_pipe[0]); + close(go_pipe[1]); + fcntl(go_pipe[0], F_SETFD, FD_CLOEXEC); + + /* + * Do a dummy execvp to get the PLT entry resolved, + * so we avoid the resolver overhead on the real + * execvp call. + */ + execvp("", (char **)argv); + + /* + * Tell the parent we're ready to go + */ + close(child_ready_pipe[1]); + + /* + * Wait until the parent tells us to go. + */ + if (read(go_pipe[0], &buf, 1) == -1) + perror("unable to read pipe"); + + execvp(argv[0], (char **)argv); + + perror(argv[0]); + kill(getppid(), SIGUSR1); + exit(-1); + } + + if (!system_wide && target_tid == -1 && target_pid == -1) + evsel_list->threads->map[0] = child_pid; + + close(child_ready_pipe[1]); + close(go_pipe[0]); + /* + * wait for child to settle + */ + if (read(child_ready_pipe[0], &buf, 1) == -1) { + perror("unable to read pipe"); + exit(-1); + } + close(child_ready_pipe[0]); + } + + open_counters(evsel_list); + + /* + * perf_session__delete(session) will be called at atexit_header() + */ + atexit(atexit_header); + + if (pipe_output) { + err = perf_header__write_pipe(output); + if (err < 0) + return err; + } else if (file_new) { + err = perf_session__write_header(session, evsel_list, + output, false); + if (err < 0) + return err; + } + + post_processing_offset = lseek(output, 0, SEEK_CUR); + + if (pipe_output) { + err = perf_session__synthesize_attrs(session, + process_synthesized_event); + if (err < 0) { + pr_err("Couldn't synthesize attrs.\n"); + return err; + } + + err = perf_event__synthesize_event_types(process_synthesized_event, + session); + if (err < 0) { + pr_err("Couldn't synthesize event_types.\n"); + return err; + } + + if (have_tracepoints(&evsel_list->entries)) { + /* + * FIXME err <= 0 here actually means that + * there were no tracepoints so its not really + * an error, just that we don't need to + * synthesize anything. We really have to + * return this more properly and also + * propagate errors that now are calling die() + */ + err = perf_event__synthesize_tracing_data(output, evsel_list, + process_synthesized_event, + session); + if (err <= 0) { + pr_err("Couldn't record tracing data.\n"); + return err; + } + advance_output(err); + } + } + + machine = perf_session__find_host_machine(session); + if (!machine) { + pr_err("Couldn't find native kernel information.\n"); + return -1; + } + + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + session, machine, "_text"); + if (err < 0) + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + session, machine, "_stext"); + if (err < 0) + pr_err("Couldn't record kernel reference relocation symbol\n" + "Symbol resolution may be skewed if relocation was used (e.g. kexec).\n" + "Check /proc/kallsyms permission or run as root.\n"); + + err = perf_event__synthesize_modules(process_synthesized_event, + session, machine); + if (err < 0) + pr_err("Couldn't record kernel module information.\n" + "Symbol resolution may be skewed if relocation was used (e.g. kexec).\n" + "Check /proc/modules permission or run as root.\n"); + + if (perf_guest) + perf_session__process_machines(session, + perf_event__synthesize_guest_os); + + if (!system_wide) + perf_event__synthesize_thread_map(evsel_list->threads, + process_synthesized_event, + session); + else + perf_event__synthesize_threads(process_synthesized_event, + session); + + if (realtime_prio) { + struct sched_param param; + + param.sched_priority = realtime_prio; + if (sched_setscheduler(0, SCHED_FIFO, ¶m)) { + pr_err("Could not set realtime priority.\n"); + exit(-1); + } + } + + /* + * Let the child rip + */ + if (forks) + close(go_pipe[1]); + + for (;;) { + int hits = samples; + int thread; + + mmap_read_all(); + + if (hits == samples) { + if (done) + break; + err = poll(evsel_list->pollfd, evsel_list->nr_fds, -1); + waking++; + } + + if (done) { + for (i = 0; i < evsel_list->cpus->nr; i++) { + struct perf_evsel *pos; + + list_for_each_entry(pos, &evsel_list->entries, node) { + for (thread = 0; + thread < evsel_list->threads->nr; + thread++) + ioctl(FD(pos, i, thread), + PERF_EVENT_IOC_DISABLE); + } + } + } + } + + if (quiet || signr == SIGUSR1) + return 0; + + fprintf(stderr, "[ perf record: Woken up %ld times to write data ]\n", waking); + + /* + * Approximate RIP event size: 24 bytes. + */ + fprintf(stderr, + "[ perf record: Captured and wrote %.3f MB %s (~%" PRIu64 " samples) ]\n", + (double)bytes_written / 1024.0 / 1024.0, + output_name, + bytes_written / 24); + + return 0; + +out_delete_session: + perf_session__delete(session); + return err; +} + +static const char * const record_usage[] = { + "perf record [<options>] [<command>]", + "perf record [<options>] -- <command> [<options>]", + NULL +}; + +static bool force, append_file; + +const struct option record_options[] = { + OPT_CALLBACK('e', "event", &evsel_list, "event", + "event selector. use 'perf list' to list available events", + parse_events), + OPT_CALLBACK(0, "filter", &evsel_list, "filter", + "event filter", parse_filter), + OPT_INTEGER('p', "pid", &target_pid, + "record events on existing process id"), + OPT_INTEGER('t', "tid", &target_tid, + "record events on existing thread id"), + OPT_INTEGER('r', "realtime", &realtime_prio, + "collect data with this RT SCHED_FIFO priority"), + OPT_BOOLEAN('D', "no-delay", &nodelay, + "collect data without buffering"), + OPT_BOOLEAN('R', "raw-samples", &raw_samples, + "collect raw sample records from all opened counters"), + OPT_BOOLEAN('a', "all-cpus", &system_wide, + "system-wide collection from all CPUs"), + OPT_BOOLEAN('A', "append", &append_file, + "append to the output file to do incremental profiling"), + OPT_STRING('C', "cpu", &cpu_list, "cpu", + "list of cpus to monitor"), + OPT_BOOLEAN('f', "force", &force, + "overwrite existing data file (deprecated)"), + OPT_U64('c', "count", &user_interval, "event period to sample"), + OPT_STRING('o', "output", &output_name, "file", + "output file name"), + OPT_BOOLEAN('i', "no-inherit", &no_inherit, + "child tasks do not inherit counters"), + OPT_UINTEGER('F', "freq", &user_freq, "profile at this frequency"), + OPT_UINTEGER('m', "mmap-pages", &mmap_pages, "number of mmap data pages"), + OPT_BOOLEAN('g', "call-graph", &call_graph, + "do call-graph (stack chain/backtrace) recording"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show counter open errors, etc)"), + OPT_BOOLEAN('q', "quiet", &quiet, "don't print any message"), + OPT_BOOLEAN('s', "stat", &inherit_stat, + "per thread counts"), + OPT_BOOLEAN('d', "data", &sample_address, + "Sample addresses"), + OPT_BOOLEAN('T', "timestamp", &sample_time, "Sample timestamps"), + OPT_BOOLEAN('n', "no-samples", &no_samples, + "don't sample"), + OPT_BOOLEAN('N', "no-buildid-cache", &no_buildid_cache, + "do not update the buildid cache"), + OPT_BOOLEAN('B', "no-buildid", &no_buildid, + "do not collect buildids in perf.data"), + OPT_CALLBACK('G', "cgroup", &evsel_list, "name", + "monitor event in cgroup name only", + parse_cgroups), + OPT_END() +}; + +int cmd_record(int argc, const char **argv, const char *prefix __used) +{ + int err = -ENOMEM; + struct perf_evsel *pos; + + evsel_list = perf_evlist__new(NULL, NULL); + if (evsel_list == NULL) + return -ENOMEM; + + argc = parse_options(argc, argv, record_options, record_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + if (!argc && target_pid == -1 && target_tid == -1 && + !system_wide && !cpu_list) + usage_with_options(record_usage, record_options); + + if (force && append_file) { + fprintf(stderr, "Can't overwrite and append at the same time." + " You need to choose between -f and -A"); + usage_with_options(record_usage, record_options); + } else if (append_file) { + write_mode = WRITE_APPEND; + } else { + write_mode = WRITE_FORCE; + } + + if (nr_cgroups && !system_wide) { + fprintf(stderr, "cgroup monitoring only available in" + " system-wide mode\n"); + usage_with_options(record_usage, record_options); + } + + symbol__init(); + + if (symbol_conf.kptr_restrict) + pr_warning( +"WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted,\n" +"check /proc/sys/kernel/kptr_restrict.\n\n" +"Samples in kernel functions may not be resolved if a suitable vmlinux\n" +"file is not found in the buildid cache or in the vmlinux path.\n\n" +"Samples in kernel modules won't be resolved at all.\n\n" +"If some relocation was applied (e.g. kexec) symbols may be misresolved\n" +"even with a suitable vmlinux or kallsyms file.\n\n"); + + if (no_buildid_cache || no_buildid) + disable_buildid_cache(); + + if (evsel_list->nr_entries == 0 && + perf_evlist__add_default(evsel_list) < 0) { + pr_err("Not enough memory for event selector list\n"); + goto out_symbol_exit; + } + + if (target_pid != -1) + target_tid = target_pid; + + if (perf_evlist__create_maps(evsel_list, target_pid, + target_tid, cpu_list) < 0) + usage_with_options(record_usage, record_options); + + list_for_each_entry(pos, &evsel_list->entries, node) { + if (perf_evsel__alloc_fd(pos, evsel_list->cpus->nr, + evsel_list->threads->nr) < 0) + goto out_free_fd; + if (perf_header__push_event(pos->attr.config, event_name(pos))) + goto out_free_fd; + } + + if (perf_evlist__alloc_pollfd(evsel_list) < 0) + goto out_free_fd; + + if (user_interval != ULLONG_MAX) + default_interval = user_interval; + if (user_freq != UINT_MAX) + freq = user_freq; + + /* + * User specified count overrides default frequency. + */ + if (default_interval) + freq = 0; + else if (freq) { + default_interval = freq; + } else { + fprintf(stderr, "frequency and count are zero, aborting\n"); + err = -EINVAL; + goto out_free_fd; + } + + err = __cmd_record(argc, argv); +out_free_fd: + perf_evlist__delete_maps(evsel_list); +out_symbol_exit: + symbol__exit(); + return err; +} diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c new file mode 100644 index 00000000..287a1735 --- /dev/null +++ b/tools/perf/builtin-report.c @@ -0,0 +1,522 @@ +/* + * builtin-report.c + * + * Builtin report command: Analyze the perf.data input file, + * look up and read DSOs and symbol information and display + * a histogram of results, along various sorting keys. + */ +#include "builtin.h" + +#include "util/util.h" + +#include "util/annotate.h" +#include "util/color.h" +#include <linux/list.h> +#include "util/cache.h" +#include <linux/rbtree.h> +#include "util/symbol.h" +#include "util/callchain.h" +#include "util/strlist.h" +#include "util/values.h" + +#include "perf.h" +#include "util/debug.h" +#include "util/evlist.h" +#include "util/evsel.h" +#include "util/header.h" +#include "util/session.h" + +#include "util/parse-options.h" +#include "util/parse-events.h" + +#include "util/thread.h" +#include "util/sort.h" +#include "util/hist.h" + +static char const *input_name = "perf.data"; + +static bool force, use_tui, use_stdio; +static bool hide_unresolved; +static bool dont_use_callchains; + +static bool show_threads; +static struct perf_read_values show_threads_values; + +static const char default_pretty_printing_style[] = "normal"; +static const char *pretty_printing_style = default_pretty_printing_style; + +static char callchain_default_opt[] = "fractal,0.5"; +static symbol_filter_t annotate_init; + +static int perf_session__add_hist_entry(struct perf_session *session, + struct addr_location *al, + struct perf_sample *sample, + struct perf_evsel *evsel) +{ + struct symbol *parent = NULL; + int err = 0; + struct hist_entry *he; + + if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) { + err = perf_session__resolve_callchain(session, al->thread, + sample->callchain, &parent); + if (err) + return err; + } + + he = __hists__add_entry(&evsel->hists, al, parent, sample->period); + if (he == NULL) + return -ENOMEM; + + if (symbol_conf.use_callchain) { + err = callchain_append(he->callchain, &session->callchain_cursor, + sample->period); + if (err) + return err; + } + /* + * Only in the newt browser we are doing integrated annotation, + * so we don't allocated the extra space needed because the stdio + * code will not use it. + */ + if (al->sym != NULL && use_browser > 0) { + struct annotation *notes = symbol__annotation(he->ms.sym); + + assert(evsel != NULL); + + err = -ENOMEM; + if (notes->src == NULL && + symbol__alloc_hist(he->ms.sym, session->evlist->nr_entries) < 0) + goto out; + + err = hist_entry__inc_addr_samples(he, evsel->idx, al->addr); + } + + evsel->hists.stats.total_period += sample->period; + hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE); +out: + return err; +} + + +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel, + struct perf_session *session) +{ + struct addr_location al; + + if (perf_event__preprocess_sample(event, session, &al, sample, + annotate_init) < 0) { + fprintf(stderr, "problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + if (al.filtered || (hide_unresolved && al.sym == NULL)) + return 0; + + if (al.map != NULL) + al.map->dso->hit = 1; + + if (perf_session__add_hist_entry(session, &al, sample, evsel)) { + pr_debug("problem incrementing symbol period, skipping event\n"); + return -1; + } + + return 0; +} + +static int process_read_event(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) +{ + struct perf_evsel *evsel = perf_evlist__id2evsel(session->evlist, + event->read.id); + if (show_threads) { + const char *name = evsel ? event_name(evsel) : "unknown"; + perf_read_values_add_value(&show_threads_values, + event->read.pid, event->read.tid, + event->read.id, + name, + event->read.value); + } + + dump_printf(": %d %d %s %" PRIu64 "\n", event->read.pid, event->read.tid, + evsel ? event_name(evsel) : "FAIL", + event->read.value); + + return 0; +} + +static int perf_session__setup_sample_type(struct perf_session *self) +{ + if (!(self->sample_type & PERF_SAMPLE_CALLCHAIN)) { + if (sort__has_parent) { + fprintf(stderr, "selected --sort parent, but no" + " callchain data. Did you call" + " perf record without -g?\n"); + return -EINVAL; + } + if (symbol_conf.use_callchain) { + fprintf(stderr, "selected -g but no callchain data." + " Did you call perf record without" + " -g?\n"); + return -1; + } + } else if (!dont_use_callchains && callchain_param.mode != CHAIN_NONE && + !symbol_conf.use_callchain) { + symbol_conf.use_callchain = true; + if (callchain_register_param(&callchain_param) < 0) { + fprintf(stderr, "Can't register callchain" + " params\n"); + return -EINVAL; + } + } + + return 0; +} + +static struct perf_event_ops event_ops = { + .sample = process_sample_event, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .exit = perf_event__process_task, + .fork = perf_event__process_task, + .lost = perf_event__process_lost, + .read = process_read_event, + .attr = perf_event__process_attr, + .event_type = perf_event__process_event_type, + .tracing_data = perf_event__process_tracing_data, + .build_id = perf_event__process_build_id, + .ordered_samples = true, + .ordering_requires_timestamps = true, +}; + +extern volatile int session_done; + +static void sig_handler(int sig __used) +{ + session_done = 1; +} + +static size_t hists__fprintf_nr_sample_events(struct hists *self, + const char *evname, FILE *fp) +{ + size_t ret; + char unit; + unsigned long nr_events = self->stats.nr_events[PERF_RECORD_SAMPLE]; + + nr_events = convert_unit(nr_events, &unit); + ret = fprintf(fp, "# Events: %lu%c", nr_events, unit); + if (evname != NULL) + ret += fprintf(fp, " %s", evname); + return ret + fprintf(fp, "\n#\n"); +} + +static int perf_evlist__tty_browse_hists(struct perf_evlist *evlist, + const char *help) +{ + struct perf_evsel *pos; + + list_for_each_entry(pos, &evlist->entries, node) { + struct hists *hists = &pos->hists; + const char *evname = NULL; + + if (rb_first(&hists->entries) != rb_last(&hists->entries)) + evname = event_name(pos); + + hists__fprintf_nr_sample_events(hists, evname, stdout); + hists__fprintf(hists, NULL, false, stdout); + fprintf(stdout, "\n\n"); + } + + if (sort_order == default_sort_order && + parent_pattern == default_parent_pattern) { + fprintf(stdout, "#\n# (%s)\n#\n", help); + + if (show_threads) { + bool style = !strcmp(pretty_printing_style, "raw"); + perf_read_values_display(stdout, &show_threads_values, + style); + perf_read_values_destroy(&show_threads_values); + } + } + + return 0; +} + +static int __cmd_report(void) +{ + int ret = -EINVAL; + u64 nr_samples; + struct perf_session *session; + struct perf_evsel *pos; + struct map *kernel_map; + struct kmap *kernel_kmap; + const char *help = "For a higher level overview, try: perf report --sort comm,dso"; + + signal(SIGINT, sig_handler); + + session = perf_session__new(input_name, O_RDONLY, force, false, &event_ops); + if (session == NULL) + return -ENOMEM; + + if (show_threads) + perf_read_values_init(&show_threads_values); + + ret = perf_session__setup_sample_type(session); + if (ret) + goto out_delete; + + ret = perf_session__process_events(session, &event_ops); + if (ret) + goto out_delete; + + kernel_map = session->host_machine.vmlinux_maps[MAP__FUNCTION]; + kernel_kmap = map__kmap(kernel_map); + if (kernel_map == NULL || + (kernel_map->dso->hit && + (kernel_kmap->ref_reloc_sym == NULL || + kernel_kmap->ref_reloc_sym->addr == 0))) { + const struct dso *kdso = kernel_map->dso; + + ui__warning( +"Kernel address maps (/proc/{kallsyms,modules}) were restricted.\n\n" +"Check /proc/sys/kernel/kptr_restrict before running 'perf record'.\n\n%s\n\n" +"Samples in kernel modules can't be resolved as well.\n\n", + RB_EMPTY_ROOT(&kdso->symbols[MAP__FUNCTION]) ? +"As no suitable kallsyms nor vmlinux was found, kernel samples\n" +"can't be resolved." : +"If some relocation was applied (e.g. kexec) symbols may be misresolved."); + } + + if (dump_trace) { + perf_session__fprintf_nr_events(session, stdout); + goto out_delete; + } + + if (verbose > 3) + perf_session__fprintf(session, stdout); + + if (verbose > 2) + perf_session__fprintf_dsos(session, stdout); + + nr_samples = 0; + list_for_each_entry(pos, &session->evlist->entries, node) { + struct hists *hists = &pos->hists; + + hists__collapse_resort(hists); + hists__output_resort(hists); + nr_samples += hists->stats.nr_events[PERF_RECORD_SAMPLE]; + } + + if (nr_samples == 0) { + ui__warning("The %s file has no samples!\n", input_name); + goto out_delete; + } + + if (use_browser > 0) + perf_evlist__tui_browse_hists(session->evlist, help); + else + perf_evlist__tty_browse_hists(session->evlist, help); + +out_delete: + /* + * Speed up the exit process, for large files this can + * take quite a while. + * + * XXX Enable this when using valgrind or if we ever + * librarize this command. + * + * Also experiment with obstacks to see how much speed + * up we'll get here. + * + * perf_session__delete(session); + */ + return ret; +} + +static int +parse_callchain_opt(const struct option *opt __used, const char *arg, + int unset) +{ + char *tok, *tok2; + char *endptr; + + /* + * --no-call-graph + */ + if (unset) { + dont_use_callchains = true; + return 0; + } + + symbol_conf.use_callchain = true; + + if (!arg) + return 0; + + tok = strtok((char *)arg, ","); + if (!tok) + return -1; + + /* get the output mode */ + if (!strncmp(tok, "graph", strlen(arg))) + callchain_param.mode = CHAIN_GRAPH_ABS; + + else if (!strncmp(tok, "flat", strlen(arg))) + callchain_param.mode = CHAIN_FLAT; + + else if (!strncmp(tok, "fractal", strlen(arg))) + callchain_param.mode = CHAIN_GRAPH_REL; + + else if (!strncmp(tok, "none", strlen(arg))) { + callchain_param.mode = CHAIN_NONE; + symbol_conf.use_callchain = false; + + return 0; + } + + else + return -1; + + /* get the min percentage */ + tok = strtok(NULL, ","); + if (!tok) + goto setup; + + tok2 = strtok(NULL, ","); + callchain_param.min_percent = strtod(tok, &endptr); + if (tok == endptr) + return -1; + + if (tok2) + callchain_param.print_limit = strtod(tok2, &endptr); +setup: + if (callchain_register_param(&callchain_param) < 0) { + fprintf(stderr, "Can't register callchain params\n"); + return -1; + } + return 0; +} + +static const char * const report_usage[] = { + "perf report [<options>] <command>", + NULL +}; + +static const struct option options[] = { + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name, + "file", "vmlinux pathname"), + OPT_STRING(0, "kallsyms", &symbol_conf.kallsyms_name, + "file", "kallsyms pathname"), + OPT_BOOLEAN('f', "force", &force, "don't complain, do it"), + OPT_BOOLEAN('m', "modules", &symbol_conf.use_modules, + "load module symbols - WARNING: use only with -k and LIVE kernel"), + OPT_BOOLEAN('n', "show-nr-samples", &symbol_conf.show_nr_samples, + "Show a column with the number of samples"), + OPT_BOOLEAN('T', "threads", &show_threads, + "Show per-thread event counters"), + OPT_STRING(0, "pretty", &pretty_printing_style, "key", + "pretty printing style key: normal raw"), + OPT_BOOLEAN(0, "tui", &use_tui, "Use the TUI interface"), + OPT_BOOLEAN(0, "stdio", &use_stdio, "Use the stdio interface"), + OPT_STRING('s', "sort", &sort_order, "key[,key2...]", + "sort by key(s): pid, comm, dso, symbol, parent"), + OPT_BOOLEAN(0, "showcpuutilization", &symbol_conf.show_cpu_utilization, + "Show sample percentage for different cpu modes"), + OPT_STRING('p', "parent", &parent_pattern, "regex", + "regex filter to identify parent, see: '--sort parent'"), + OPT_BOOLEAN('x', "exclude-other", &symbol_conf.exclude_other, + "Only display entries with parent-match"), + OPT_CALLBACK_DEFAULT('g', "call-graph", NULL, "output_type,min_percent", + "Display callchains using output_type (graph, flat, fractal, or none) and min percent threshold. " + "Default: fractal,0.5", &parse_callchain_opt, callchain_default_opt), + OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]", + "only consider symbols in these dsos"), + OPT_STRING('C', "comms", &symbol_conf.comm_list_str, "comm[,comm...]", + "only consider symbols in these comms"), + OPT_STRING('S', "symbols", &symbol_conf.sym_list_str, "symbol[,symbol...]", + "only consider these symbols"), + OPT_STRING('w', "column-widths", &symbol_conf.col_width_list_str, + "width[,width...]", + "don't try to adjust column width, use these fixed values"), + OPT_STRING('t', "field-separator", &symbol_conf.field_sep, "separator", + "separator for columns, no spaces will be added between " + "columns '.' is reserved."), + OPT_BOOLEAN('U', "hide-unresolved", &hide_unresolved, + "Only display entries resolved to a symbol"), + OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory", + "Look for files with symbols relative to this directory"), + OPT_END() +}; + +int cmd_report(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, options, report_usage, 0); + + if (use_stdio) + use_browser = 0; + else if (use_tui) + use_browser = 1; + + if (strcmp(input_name, "-") != 0) + setup_browser(true); + else + use_browser = 0; + /* + * Only in the newt browser we are doing integrated annotation, + * so don't allocate extra space that won't be used in the stdio + * implementation. + */ + if (use_browser > 0) { + symbol_conf.priv_size = sizeof(struct annotation); + annotate_init = symbol__annotate_init; + /* + * For searching by name on the "Browse map details". + * providing it only in verbose mode not to bloat too + * much struct symbol. + */ + if (verbose) { + /* + * XXX: Need to provide a less kludgy way to ask for + * more space per symbol, the u32 is for the index on + * the ui browser. + * See symbol__browser_index. + */ + symbol_conf.priv_size += sizeof(u32); + symbol_conf.sort_by_name = true; + } + } + + if (symbol__init() < 0) + return -1; + + setup_sorting(report_usage, options); + + if (parent_pattern != default_parent_pattern) { + if (sort_dimension__add("parent") < 0) + return -1; + sort_parent.elide = 1; + } else + symbol_conf.exclude_other = false; + + /* + * Any (unrecognized) arguments left? + */ + if (argc) + usage_with_options(report_usage, options); + + sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "dso", stdout); + sort_entry__setup_elide(&sort_comm, symbol_conf.comm_list, "comm", stdout); + sort_entry__setup_elide(&sort_sym, symbol_conf.sym_list, "symbol", stdout); + + return __cmd_report(); +} diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c new file mode 100644 index 00000000..dcfe8873 --- /dev/null +++ b/tools/perf/builtin-sched.c @@ -0,0 +1,1917 @@ +#include "builtin.h" +#include "perf.h" + +#include "util/util.h" +#include "util/cache.h" +#include "util/symbol.h" +#include "util/thread.h" +#include "util/header.h" +#include "util/session.h" + +#include "util/parse-options.h" +#include "util/trace-event.h" + +#include "util/debug.h" + +#include <sys/prctl.h> + +#include <semaphore.h> +#include <pthread.h> +#include <math.h> + +static char const *input_name = "perf.data"; + +static char default_sort_order[] = "avg, max, switch, runtime"; +static const char *sort_order = default_sort_order; + +static int profile_cpu = -1; + +#define PR_SET_NAME 15 /* Set process name */ +#define MAX_CPUS 4096 + +static u64 run_measurement_overhead; +static u64 sleep_measurement_overhead; + +#define COMM_LEN 20 +#define SYM_LEN 129 + +#define MAX_PID 65536 + +static unsigned long nr_tasks; + +struct sched_atom; + +struct task_desc { + unsigned long nr; + unsigned long pid; + char comm[COMM_LEN]; + + unsigned long nr_events; + unsigned long curr_event; + struct sched_atom **atoms; + + pthread_t thread; + sem_t sleep_sem; + + sem_t ready_for_work; + sem_t work_done_sem; + + u64 cpu_usage; +}; + +enum sched_event_type { + SCHED_EVENT_RUN, + SCHED_EVENT_SLEEP, + SCHED_EVENT_WAKEUP, + SCHED_EVENT_MIGRATION, +}; + +struct sched_atom { + enum sched_event_type type; + int specific_wait; + u64 timestamp; + u64 duration; + unsigned long nr; + sem_t *wait_sem; + struct task_desc *wakee; +}; + +static struct task_desc *pid_to_task[MAX_PID]; + +static struct task_desc **tasks; + +static pthread_mutex_t start_work_mutex = PTHREAD_MUTEX_INITIALIZER; +static u64 start_time; + +static pthread_mutex_t work_done_wait_mutex = PTHREAD_MUTEX_INITIALIZER; + +static unsigned long nr_run_events; +static unsigned long nr_sleep_events; +static unsigned long nr_wakeup_events; + +static unsigned long nr_sleep_corrections; +static unsigned long nr_run_events_optimized; + +static unsigned long targetless_wakeups; +static unsigned long multitarget_wakeups; + +static u64 cpu_usage; +static u64 runavg_cpu_usage; +static u64 parent_cpu_usage; +static u64 runavg_parent_cpu_usage; + +static unsigned long nr_runs; +static u64 sum_runtime; +static u64 sum_fluct; +static u64 run_avg; + +static unsigned int replay_repeat = 10; +static unsigned long nr_timestamps; +static unsigned long nr_unordered_timestamps; +static unsigned long nr_state_machine_bugs; +static unsigned long nr_context_switch_bugs; +static unsigned long nr_events; +static unsigned long nr_lost_chunks; +static unsigned long nr_lost_events; + +#define TASK_STATE_TO_CHAR_STR "RSDTtZX" + +enum thread_state { + THREAD_SLEEPING = 0, + THREAD_WAIT_CPU, + THREAD_SCHED_IN, + THREAD_IGNORE +}; + +struct work_atom { + struct list_head list; + enum thread_state state; + u64 sched_out_time; + u64 wake_up_time; + u64 sched_in_time; + u64 runtime; +}; + +struct work_atoms { + struct list_head work_list; + struct thread *thread; + struct rb_node node; + u64 max_lat; + u64 max_lat_at; + u64 total_lat; + u64 nb_atoms; + u64 total_runtime; +}; + +typedef int (*sort_fn_t)(struct work_atoms *, struct work_atoms *); + +static struct rb_root atom_root, sorted_atom_root; + +static u64 all_runtime; +static u64 all_count; + + +static u64 get_nsecs(void) +{ + struct timespec ts; + + clock_gettime(CLOCK_MONOTONIC, &ts); + + return ts.tv_sec * 1000000000ULL + ts.tv_nsec; +} + +static void burn_nsecs(u64 nsecs) +{ + u64 T0 = get_nsecs(), T1; + + do { + T1 = get_nsecs(); + } while (T1 + run_measurement_overhead < T0 + nsecs); +} + +static void sleep_nsecs(u64 nsecs) +{ + struct timespec ts; + + ts.tv_nsec = nsecs % 999999999; + ts.tv_sec = nsecs / 999999999; + + nanosleep(&ts, NULL); +} + +static void calibrate_run_measurement_overhead(void) +{ + u64 T0, T1, delta, min_delta = 1000000000ULL; + int i; + + for (i = 0; i < 10; i++) { + T0 = get_nsecs(); + burn_nsecs(0); + T1 = get_nsecs(); + delta = T1-T0; + min_delta = min(min_delta, delta); + } + run_measurement_overhead = min_delta; + + printf("run measurement overhead: %" PRIu64 " nsecs\n", min_delta); +} + +static void calibrate_sleep_measurement_overhead(void) +{ + u64 T0, T1, delta, min_delta = 1000000000ULL; + int i; + + for (i = 0; i < 10; i++) { + T0 = get_nsecs(); + sleep_nsecs(10000); + T1 = get_nsecs(); + delta = T1-T0; + min_delta = min(min_delta, delta); + } + min_delta -= 10000; + sleep_measurement_overhead = min_delta; + + printf("sleep measurement overhead: %" PRIu64 " nsecs\n", min_delta); +} + +static struct sched_atom * +get_new_event(struct task_desc *task, u64 timestamp) +{ + struct sched_atom *event = zalloc(sizeof(*event)); + unsigned long idx = task->nr_events; + size_t size; + + event->timestamp = timestamp; + event->nr = idx; + + task->nr_events++; + size = sizeof(struct sched_atom *) * task->nr_events; + task->atoms = realloc(task->atoms, size); + BUG_ON(!task->atoms); + + task->atoms[idx] = event; + + return event; +} + +static struct sched_atom *last_event(struct task_desc *task) +{ + if (!task->nr_events) + return NULL; + + return task->atoms[task->nr_events - 1]; +} + +static void +add_sched_event_run(struct task_desc *task, u64 timestamp, u64 duration) +{ + struct sched_atom *event, *curr_event = last_event(task); + + /* + * optimize an existing RUN event by merging this one + * to it: + */ + if (curr_event && curr_event->type == SCHED_EVENT_RUN) { + nr_run_events_optimized++; + curr_event->duration += duration; + return; + } + + event = get_new_event(task, timestamp); + + event->type = SCHED_EVENT_RUN; + event->duration = duration; + + nr_run_events++; +} + +static void +add_sched_event_wakeup(struct task_desc *task, u64 timestamp, + struct task_desc *wakee) +{ + struct sched_atom *event, *wakee_event; + + event = get_new_event(task, timestamp); + event->type = SCHED_EVENT_WAKEUP; + event->wakee = wakee; + + wakee_event = last_event(wakee); + if (!wakee_event || wakee_event->type != SCHED_EVENT_SLEEP) { + targetless_wakeups++; + return; + } + if (wakee_event->wait_sem) { + multitarget_wakeups++; + return; + } + + wakee_event->wait_sem = zalloc(sizeof(*wakee_event->wait_sem)); + sem_init(wakee_event->wait_sem, 0, 0); + wakee_event->specific_wait = 1; + event->wait_sem = wakee_event->wait_sem; + + nr_wakeup_events++; +} + +static void +add_sched_event_sleep(struct task_desc *task, u64 timestamp, + u64 task_state __used) +{ + struct sched_atom *event = get_new_event(task, timestamp); + + event->type = SCHED_EVENT_SLEEP; + + nr_sleep_events++; +} + +static struct task_desc *register_pid(unsigned long pid, const char *comm) +{ + struct task_desc *task; + + BUG_ON(pid >= MAX_PID); + + task = pid_to_task[pid]; + + if (task) + return task; + + task = zalloc(sizeof(*task)); + task->pid = pid; + task->nr = nr_tasks; + strcpy(task->comm, comm); + /* + * every task starts in sleeping state - this gets ignored + * if there's no wakeup pointing to this sleep state: + */ + add_sched_event_sleep(task, 0, 0); + + pid_to_task[pid] = task; + nr_tasks++; + tasks = realloc(tasks, nr_tasks*sizeof(struct task_task *)); + BUG_ON(!tasks); + tasks[task->nr] = task; + + if (verbose) + printf("registered task #%ld, PID %ld (%s)\n", nr_tasks, pid, comm); + + return task; +} + + +static void print_task_traces(void) +{ + struct task_desc *task; + unsigned long i; + + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + printf("task %6ld (%20s:%10ld), nr_events: %ld\n", + task->nr, task->comm, task->pid, task->nr_events); + } +} + +static void add_cross_task_wakeups(void) +{ + struct task_desc *task1, *task2; + unsigned long i, j; + + for (i = 0; i < nr_tasks; i++) { + task1 = tasks[i]; + j = i + 1; + if (j == nr_tasks) + j = 0; + task2 = tasks[j]; + add_sched_event_wakeup(task1, 0, task2); + } +} + +static void +process_sched_event(struct task_desc *this_task __used, struct sched_atom *atom) +{ + int ret = 0; + + switch (atom->type) { + case SCHED_EVENT_RUN: + burn_nsecs(atom->duration); + break; + case SCHED_EVENT_SLEEP: + if (atom->wait_sem) + ret = sem_wait(atom->wait_sem); + BUG_ON(ret); + break; + case SCHED_EVENT_WAKEUP: + if (atom->wait_sem) + ret = sem_post(atom->wait_sem); + BUG_ON(ret); + break; + case SCHED_EVENT_MIGRATION: + break; + default: + BUG_ON(1); + } +} + +static u64 get_cpu_usage_nsec_parent(void) +{ + struct rusage ru; + u64 sum; + int err; + + err = getrusage(RUSAGE_SELF, &ru); + BUG_ON(err); + + sum = ru.ru_utime.tv_sec*1e9 + ru.ru_utime.tv_usec*1e3; + sum += ru.ru_stime.tv_sec*1e9 + ru.ru_stime.tv_usec*1e3; + + return sum; +} + +static int self_open_counters(void) +{ + struct perf_event_attr attr; + int fd; + + memset(&attr, 0, sizeof(attr)); + + attr.type = PERF_TYPE_SOFTWARE; + attr.config = PERF_COUNT_SW_TASK_CLOCK; + + fd = sys_perf_event_open(&attr, 0, -1, -1, 0); + + if (fd < 0) + die("Error: sys_perf_event_open() syscall returned" + "with %d (%s)\n", fd, strerror(errno)); + return fd; +} + +static u64 get_cpu_usage_nsec_self(int fd) +{ + u64 runtime; + int ret; + + ret = read(fd, &runtime, sizeof(runtime)); + BUG_ON(ret != sizeof(runtime)); + + return runtime; +} + +static void *thread_func(void *ctx) +{ + struct task_desc *this_task = ctx; + u64 cpu_usage_0, cpu_usage_1; + unsigned long i, ret; + char comm2[22]; + int fd; + + sprintf(comm2, ":%s", this_task->comm); + prctl(PR_SET_NAME, comm2); + fd = self_open_counters(); + +again: + ret = sem_post(&this_task->ready_for_work); + BUG_ON(ret); + ret = pthread_mutex_lock(&start_work_mutex); + BUG_ON(ret); + ret = pthread_mutex_unlock(&start_work_mutex); + BUG_ON(ret); + + cpu_usage_0 = get_cpu_usage_nsec_self(fd); + + for (i = 0; i < this_task->nr_events; i++) { + this_task->curr_event = i; + process_sched_event(this_task, this_task->atoms[i]); + } + + cpu_usage_1 = get_cpu_usage_nsec_self(fd); + this_task->cpu_usage = cpu_usage_1 - cpu_usage_0; + ret = sem_post(&this_task->work_done_sem); + BUG_ON(ret); + + ret = pthread_mutex_lock(&work_done_wait_mutex); + BUG_ON(ret); + ret = pthread_mutex_unlock(&work_done_wait_mutex); + BUG_ON(ret); + + goto again; +} + +static void create_tasks(void) +{ + struct task_desc *task; + pthread_attr_t attr; + unsigned long i; + int err; + + err = pthread_attr_init(&attr); + BUG_ON(err); + err = pthread_attr_setstacksize(&attr, + (size_t) max(16 * 1024, PTHREAD_STACK_MIN)); + BUG_ON(err); + err = pthread_mutex_lock(&start_work_mutex); + BUG_ON(err); + err = pthread_mutex_lock(&work_done_wait_mutex); + BUG_ON(err); + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + sem_init(&task->sleep_sem, 0, 0); + sem_init(&task->ready_for_work, 0, 0); + sem_init(&task->work_done_sem, 0, 0); + task->curr_event = 0; + err = pthread_create(&task->thread, &attr, thread_func, task); + BUG_ON(err); + } +} + +static void wait_for_tasks(void) +{ + u64 cpu_usage_0, cpu_usage_1; + struct task_desc *task; + unsigned long i, ret; + + start_time = get_nsecs(); + cpu_usage = 0; + pthread_mutex_unlock(&work_done_wait_mutex); + + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + ret = sem_wait(&task->ready_for_work); + BUG_ON(ret); + sem_init(&task->ready_for_work, 0, 0); + } + ret = pthread_mutex_lock(&work_done_wait_mutex); + BUG_ON(ret); + + cpu_usage_0 = get_cpu_usage_nsec_parent(); + + pthread_mutex_unlock(&start_work_mutex); + + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + ret = sem_wait(&task->work_done_sem); + BUG_ON(ret); + sem_init(&task->work_done_sem, 0, 0); + cpu_usage += task->cpu_usage; + task->cpu_usage = 0; + } + + cpu_usage_1 = get_cpu_usage_nsec_parent(); + if (!runavg_cpu_usage) + runavg_cpu_usage = cpu_usage; + runavg_cpu_usage = (runavg_cpu_usage*9 + cpu_usage)/10; + + parent_cpu_usage = cpu_usage_1 - cpu_usage_0; + if (!runavg_parent_cpu_usage) + runavg_parent_cpu_usage = parent_cpu_usage; + runavg_parent_cpu_usage = (runavg_parent_cpu_usage*9 + + parent_cpu_usage)/10; + + ret = pthread_mutex_lock(&start_work_mutex); + BUG_ON(ret); + + for (i = 0; i < nr_tasks; i++) { + task = tasks[i]; + sem_init(&task->sleep_sem, 0, 0); + task->curr_event = 0; + } +} + +static void run_one_test(void) +{ + u64 T0, T1, delta, avg_delta, fluct; + + T0 = get_nsecs(); + wait_for_tasks(); + T1 = get_nsecs(); + + delta = T1 - T0; + sum_runtime += delta; + nr_runs++; + + avg_delta = sum_runtime / nr_runs; + if (delta < avg_delta) + fluct = avg_delta - delta; + else + fluct = delta - avg_delta; + sum_fluct += fluct; + if (!run_avg) + run_avg = delta; + run_avg = (run_avg*9 + delta)/10; + + printf("#%-3ld: %0.3f, ", + nr_runs, (double)delta/1000000.0); + + printf("ravg: %0.2f, ", + (double)run_avg/1e6); + + printf("cpu: %0.2f / %0.2f", + (double)cpu_usage/1e6, (double)runavg_cpu_usage/1e6); + +#if 0 + /* + * rusage statistics done by the parent, these are less + * accurate than the sum_exec_runtime based statistics: + */ + printf(" [%0.2f / %0.2f]", + (double)parent_cpu_usage/1e6, + (double)runavg_parent_cpu_usage/1e6); +#endif + + printf("\n"); + + if (nr_sleep_corrections) + printf(" (%ld sleep corrections)\n", nr_sleep_corrections); + nr_sleep_corrections = 0; +} + +static void test_calibrations(void) +{ + u64 T0, T1; + + T0 = get_nsecs(); + burn_nsecs(1e6); + T1 = get_nsecs(); + + printf("the run test took %" PRIu64 " nsecs\n", T1 - T0); + + T0 = get_nsecs(); + sleep_nsecs(1e6); + T1 = get_nsecs(); + + printf("the sleep test took %" PRIu64 " nsecs\n", T1 - T0); +} + +#define FILL_FIELD(ptr, field, event, data) \ + ptr.field = (typeof(ptr.field)) raw_field_value(event, #field, data) + +#define FILL_ARRAY(ptr, array, event, data) \ +do { \ + void *__array = raw_field_ptr(event, #array, data); \ + memcpy(ptr.array, __array, sizeof(ptr.array)); \ +} while(0) + +#define FILL_COMMON_FIELDS(ptr, event, data) \ +do { \ + FILL_FIELD(ptr, common_type, event, data); \ + FILL_FIELD(ptr, common_flags, event, data); \ + FILL_FIELD(ptr, common_preempt_count, event, data); \ + FILL_FIELD(ptr, common_pid, event, data); \ + FILL_FIELD(ptr, common_tgid, event, data); \ +} while (0) + + + +struct trace_switch_event { + u32 size; + + u16 common_type; + u8 common_flags; + u8 common_preempt_count; + u32 common_pid; + u32 common_tgid; + + char prev_comm[16]; + u32 prev_pid; + u32 prev_prio; + u64 prev_state; + char next_comm[16]; + u32 next_pid; + u32 next_prio; +}; + +struct trace_runtime_event { + u32 size; + + u16 common_type; + u8 common_flags; + u8 common_preempt_count; + u32 common_pid; + u32 common_tgid; + + char comm[16]; + u32 pid; + u64 runtime; + u64 vruntime; +}; + +struct trace_wakeup_event { + u32 size; + + u16 common_type; + u8 common_flags; + u8 common_preempt_count; + u32 common_pid; + u32 common_tgid; + + char comm[16]; + u32 pid; + + u32 prio; + u32 success; + u32 cpu; +}; + +struct trace_fork_event { + u32 size; + + u16 common_type; + u8 common_flags; + u8 common_preempt_count; + u32 common_pid; + u32 common_tgid; + + char parent_comm[16]; + u32 parent_pid; + char child_comm[16]; + u32 child_pid; +}; + +struct trace_migrate_task_event { + u32 size; + + u16 common_type; + u8 common_flags; + u8 common_preempt_count; + u32 common_pid; + u32 common_tgid; + + char comm[16]; + u32 pid; + + u32 prio; + u32 cpu; +}; + +struct trace_sched_handler { + void (*switch_event)(struct trace_switch_event *, + struct perf_session *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*runtime_event)(struct trace_runtime_event *, + struct perf_session *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*wakeup_event)(struct trace_wakeup_event *, + struct perf_session *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*fork_event)(struct trace_fork_event *, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); + + void (*migrate_task_event)(struct trace_migrate_task_event *, + struct perf_session *session, + struct event *, + int cpu, + u64 timestamp, + struct thread *thread); +}; + + +static void +replay_wakeup_event(struct trace_wakeup_event *wakeup_event, + struct perf_session *session __used, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct task_desc *waker, *wakee; + + if (verbose) { + printf("sched_wakeup event %p\n", event); + + printf(" ... pid %d woke up %s/%d\n", + wakeup_event->common_pid, + wakeup_event->comm, + wakeup_event->pid); + } + + waker = register_pid(wakeup_event->common_pid, "<unknown>"); + wakee = register_pid(wakeup_event->pid, wakeup_event->comm); + + add_sched_event_wakeup(waker, timestamp, wakee); +} + +static u64 cpu_last_switched[MAX_CPUS]; + +static void +replay_switch_event(struct trace_switch_event *switch_event, + struct perf_session *session __used, + struct event *event, + int cpu, + u64 timestamp, + struct thread *thread __used) +{ + struct task_desc *prev, __used *next; + u64 timestamp0; + s64 delta; + + if (verbose) + printf("sched_switch event %p\n", event); + + if (cpu >= MAX_CPUS || cpu < 0) + return; + + timestamp0 = cpu_last_switched[cpu]; + if (timestamp0) + delta = timestamp - timestamp0; + else + delta = 0; + + if (delta < 0) + die("hm, delta: %" PRIu64 " < 0 ?\n", delta); + + if (verbose) { + printf(" ... switch from %s/%d to %s/%d [ran %" PRIu64 " nsecs]\n", + switch_event->prev_comm, switch_event->prev_pid, + switch_event->next_comm, switch_event->next_pid, + delta); + } + + prev = register_pid(switch_event->prev_pid, switch_event->prev_comm); + next = register_pid(switch_event->next_pid, switch_event->next_comm); + + cpu_last_switched[cpu] = timestamp; + + add_sched_event_run(prev, timestamp, delta); + add_sched_event_sleep(prev, timestamp, switch_event->prev_state); +} + + +static void +replay_fork_event(struct trace_fork_event *fork_event, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + if (verbose) { + printf("sched_fork event %p\n", event); + printf("... parent: %s/%d\n", fork_event->parent_comm, fork_event->parent_pid); + printf("... child: %s/%d\n", fork_event->child_comm, fork_event->child_pid); + } + register_pid(fork_event->parent_pid, fork_event->parent_comm); + register_pid(fork_event->child_pid, fork_event->child_comm); +} + +static struct trace_sched_handler replay_ops = { + .wakeup_event = replay_wakeup_event, + .switch_event = replay_switch_event, + .fork_event = replay_fork_event, +}; + +struct sort_dimension { + const char *name; + sort_fn_t cmp; + struct list_head list; +}; + +static LIST_HEAD(cmp_pid); + +static int +thread_lat_cmp(struct list_head *list, struct work_atoms *l, struct work_atoms *r) +{ + struct sort_dimension *sort; + int ret = 0; + + BUG_ON(list_empty(list)); + + list_for_each_entry(sort, list, list) { + ret = sort->cmp(l, r); + if (ret) + return ret; + } + + return ret; +} + +static struct work_atoms * +thread_atoms_search(struct rb_root *root, struct thread *thread, + struct list_head *sort_list) +{ + struct rb_node *node = root->rb_node; + struct work_atoms key = { .thread = thread }; + + while (node) { + struct work_atoms *atoms; + int cmp; + + atoms = container_of(node, struct work_atoms, node); + + cmp = thread_lat_cmp(sort_list, &key, atoms); + if (cmp > 0) + node = node->rb_left; + else if (cmp < 0) + node = node->rb_right; + else { + BUG_ON(thread != atoms->thread); + return atoms; + } + } + return NULL; +} + +static void +__thread_latency_insert(struct rb_root *root, struct work_atoms *data, + struct list_head *sort_list) +{ + struct rb_node **new = &(root->rb_node), *parent = NULL; + + while (*new) { + struct work_atoms *this; + int cmp; + + this = container_of(*new, struct work_atoms, node); + parent = *new; + + cmp = thread_lat_cmp(sort_list, data, this); + + if (cmp > 0) + new = &((*new)->rb_left); + else + new = &((*new)->rb_right); + } + + rb_link_node(&data->node, parent, new); + rb_insert_color(&data->node, root); +} + +static void thread_atoms_insert(struct thread *thread) +{ + struct work_atoms *atoms = zalloc(sizeof(*atoms)); + if (!atoms) + die("No memory"); + + atoms->thread = thread; + INIT_LIST_HEAD(&atoms->work_list); + __thread_latency_insert(&atom_root, atoms, &cmp_pid); +} + +static void +latency_fork_event(struct trace_fork_event *fork_event __used, + struct event *event __used, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + /* should insert the newcomer */ +} + +__used +static char sched_out_state(struct trace_switch_event *switch_event) +{ + const char *str = TASK_STATE_TO_CHAR_STR; + + return str[switch_event->prev_state]; +} + +static void +add_sched_out_event(struct work_atoms *atoms, + char run_state, + u64 timestamp) +{ + struct work_atom *atom = zalloc(sizeof(*atom)); + if (!atom) + die("Non memory"); + + atom->sched_out_time = timestamp; + + if (run_state == 'R') { + atom->state = THREAD_WAIT_CPU; + atom->wake_up_time = atom->sched_out_time; + } + + list_add_tail(&atom->list, &atoms->work_list); +} + +static void +add_runtime_event(struct work_atoms *atoms, u64 delta, u64 timestamp __used) +{ + struct work_atom *atom; + + BUG_ON(list_empty(&atoms->work_list)); + + atom = list_entry(atoms->work_list.prev, struct work_atom, list); + + atom->runtime += delta; + atoms->total_runtime += delta; +} + +static void +add_sched_in_event(struct work_atoms *atoms, u64 timestamp) +{ + struct work_atom *atom; + u64 delta; + + if (list_empty(&atoms->work_list)) + return; + + atom = list_entry(atoms->work_list.prev, struct work_atom, list); + + if (atom->state != THREAD_WAIT_CPU) + return; + + if (timestamp < atom->wake_up_time) { + atom->state = THREAD_IGNORE; + return; + } + + atom->state = THREAD_SCHED_IN; + atom->sched_in_time = timestamp; + + delta = atom->sched_in_time - atom->wake_up_time; + atoms->total_lat += delta; + if (delta > atoms->max_lat) { + atoms->max_lat = delta; + atoms->max_lat_at = timestamp; + } + atoms->nb_atoms++; +} + +static void +latency_switch_event(struct trace_switch_event *switch_event, + struct perf_session *session, + struct event *event __used, + int cpu, + u64 timestamp, + struct thread *thread __used) +{ + struct work_atoms *out_events, *in_events; + struct thread *sched_out, *sched_in; + u64 timestamp0; + s64 delta; + + BUG_ON(cpu >= MAX_CPUS || cpu < 0); + + timestamp0 = cpu_last_switched[cpu]; + cpu_last_switched[cpu] = timestamp; + if (timestamp0) + delta = timestamp - timestamp0; + else + delta = 0; + + if (delta < 0) + die("hm, delta: %" PRIu64 " < 0 ?\n", delta); + + + sched_out = perf_session__findnew(session, switch_event->prev_pid); + sched_in = perf_session__findnew(session, switch_event->next_pid); + + out_events = thread_atoms_search(&atom_root, sched_out, &cmp_pid); + if (!out_events) { + thread_atoms_insert(sched_out); + out_events = thread_atoms_search(&atom_root, sched_out, &cmp_pid); + if (!out_events) + die("out-event: Internal tree error"); + } + add_sched_out_event(out_events, sched_out_state(switch_event), timestamp); + + in_events = thread_atoms_search(&atom_root, sched_in, &cmp_pid); + if (!in_events) { + thread_atoms_insert(sched_in); + in_events = thread_atoms_search(&atom_root, sched_in, &cmp_pid); + if (!in_events) + die("in-event: Internal tree error"); + /* + * Take came in we have not heard about yet, + * add in an initial atom in runnable state: + */ + add_sched_out_event(in_events, 'R', timestamp); + } + add_sched_in_event(in_events, timestamp); +} + +static void +latency_runtime_event(struct trace_runtime_event *runtime_event, + struct perf_session *session, + struct event *event __used, + int cpu, + u64 timestamp, + struct thread *this_thread __used) +{ + struct thread *thread = perf_session__findnew(session, runtime_event->pid); + struct work_atoms *atoms = thread_atoms_search(&atom_root, thread, &cmp_pid); + + BUG_ON(cpu >= MAX_CPUS || cpu < 0); + if (!atoms) { + thread_atoms_insert(thread); + atoms = thread_atoms_search(&atom_root, thread, &cmp_pid); + if (!atoms) + die("in-event: Internal tree error"); + add_sched_out_event(atoms, 'R', timestamp); + } + + add_runtime_event(atoms, runtime_event->runtime, timestamp); +} + +static void +latency_wakeup_event(struct trace_wakeup_event *wakeup_event, + struct perf_session *session, + struct event *__event __used, + int cpu __used, + u64 timestamp, + struct thread *thread __used) +{ + struct work_atoms *atoms; + struct work_atom *atom; + struct thread *wakee; + + /* Note for later, it may be interesting to observe the failing cases */ + if (!wakeup_event->success) + return; + + wakee = perf_session__findnew(session, wakeup_event->pid); + atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid); + if (!atoms) { + thread_atoms_insert(wakee); + atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid); + if (!atoms) + die("wakeup-event: Internal tree error"); + add_sched_out_event(atoms, 'S', timestamp); + } + + BUG_ON(list_empty(&atoms->work_list)); + + atom = list_entry(atoms->work_list.prev, struct work_atom, list); + + /* + * You WILL be missing events if you've recorded only + * one CPU, or are only looking at only one, so don't + * make useless noise. + */ + if (profile_cpu == -1 && atom->state != THREAD_SLEEPING) + nr_state_machine_bugs++; + + nr_timestamps++; + if (atom->sched_out_time > timestamp) { + nr_unordered_timestamps++; + return; + } + + atom->state = THREAD_WAIT_CPU; + atom->wake_up_time = timestamp; +} + +static void +latency_migrate_task_event(struct trace_migrate_task_event *migrate_task_event, + struct perf_session *session, + struct event *__event __used, + int cpu __used, + u64 timestamp, + struct thread *thread __used) +{ + struct work_atoms *atoms; + struct work_atom *atom; + struct thread *migrant; + + /* + * Only need to worry about migration when profiling one CPU. + */ + if (profile_cpu == -1) + return; + + migrant = perf_session__findnew(session, migrate_task_event->pid); + atoms = thread_atoms_search(&atom_root, migrant, &cmp_pid); + if (!atoms) { + thread_atoms_insert(migrant); + register_pid(migrant->pid, migrant->comm); + atoms = thread_atoms_search(&atom_root, migrant, &cmp_pid); + if (!atoms) + die("migration-event: Internal tree error"); + add_sched_out_event(atoms, 'R', timestamp); + } + + BUG_ON(list_empty(&atoms->work_list)); + + atom = list_entry(atoms->work_list.prev, struct work_atom, list); + atom->sched_in_time = atom->sched_out_time = atom->wake_up_time = timestamp; + + nr_timestamps++; + + if (atom->sched_out_time > timestamp) + nr_unordered_timestamps++; +} + +static struct trace_sched_handler lat_ops = { + .wakeup_event = latency_wakeup_event, + .switch_event = latency_switch_event, + .runtime_event = latency_runtime_event, + .fork_event = latency_fork_event, + .migrate_task_event = latency_migrate_task_event, +}; + +static void output_lat_thread(struct work_atoms *work_list) +{ + int i; + int ret; + u64 avg; + + if (!work_list->nb_atoms) + return; + /* + * Ignore idle threads: + */ + if (!strcmp(work_list->thread->comm, "swapper")) + return; + + all_runtime += work_list->total_runtime; + all_count += work_list->nb_atoms; + + ret = printf(" %s:%d ", work_list->thread->comm, work_list->thread->pid); + + for (i = 0; i < 24 - ret; i++) + printf(" "); + + avg = work_list->total_lat / work_list->nb_atoms; + + printf("|%11.3f ms |%9" PRIu64 " | avg:%9.3f ms | max:%9.3f ms | max at: %9.6f s\n", + (double)work_list->total_runtime / 1e6, + work_list->nb_atoms, (double)avg / 1e6, + (double)work_list->max_lat / 1e6, + (double)work_list->max_lat_at / 1e9); +} + +static int pid_cmp(struct work_atoms *l, struct work_atoms *r) +{ + if (l->thread->pid < r->thread->pid) + return -1; + if (l->thread->pid > r->thread->pid) + return 1; + + return 0; +} + +static struct sort_dimension pid_sort_dimension = { + .name = "pid", + .cmp = pid_cmp, +}; + +static int avg_cmp(struct work_atoms *l, struct work_atoms *r) +{ + u64 avgl, avgr; + + if (!l->nb_atoms) + return -1; + + if (!r->nb_atoms) + return 1; + + avgl = l->total_lat / l->nb_atoms; + avgr = r->total_lat / r->nb_atoms; + + if (avgl < avgr) + return -1; + if (avgl > avgr) + return 1; + + return 0; +} + +static struct sort_dimension avg_sort_dimension = { + .name = "avg", + .cmp = avg_cmp, +}; + +static int max_cmp(struct work_atoms *l, struct work_atoms *r) +{ + if (l->max_lat < r->max_lat) + return -1; + if (l->max_lat > r->max_lat) + return 1; + + return 0; +} + +static struct sort_dimension max_sort_dimension = { + .name = "max", + .cmp = max_cmp, +}; + +static int switch_cmp(struct work_atoms *l, struct work_atoms *r) +{ + if (l->nb_atoms < r->nb_atoms) + return -1; + if (l->nb_atoms > r->nb_atoms) + return 1; + + return 0; +} + +static struct sort_dimension switch_sort_dimension = { + .name = "switch", + .cmp = switch_cmp, +}; + +static int runtime_cmp(struct work_atoms *l, struct work_atoms *r) +{ + if (l->total_runtime < r->total_runtime) + return -1; + if (l->total_runtime > r->total_runtime) + return 1; + + return 0; +} + +static struct sort_dimension runtime_sort_dimension = { + .name = "runtime", + .cmp = runtime_cmp, +}; + +static struct sort_dimension *available_sorts[] = { + &pid_sort_dimension, + &avg_sort_dimension, + &max_sort_dimension, + &switch_sort_dimension, + &runtime_sort_dimension, +}; + +#define NB_AVAILABLE_SORTS (int)(sizeof(available_sorts) / sizeof(struct sort_dimension *)) + +static LIST_HEAD(sort_list); + +static int sort_dimension__add(const char *tok, struct list_head *list) +{ + int i; + + for (i = 0; i < NB_AVAILABLE_SORTS; i++) { + if (!strcmp(available_sorts[i]->name, tok)) { + list_add_tail(&available_sorts[i]->list, list); + + return 0; + } + } + + return -1; +} + +static void setup_sorting(void); + +static void sort_lat(void) +{ + struct rb_node *node; + + for (;;) { + struct work_atoms *data; + node = rb_first(&atom_root); + if (!node) + break; + + rb_erase(node, &atom_root); + data = rb_entry(node, struct work_atoms, node); + __thread_latency_insert(&sorted_atom_root, data, &sort_list); + } +} + +static struct trace_sched_handler *trace_handler; + +static void +process_sched_wakeup_event(void *data, struct perf_session *session, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_wakeup_event wakeup_event; + + FILL_COMMON_FIELDS(wakeup_event, event, data); + + FILL_ARRAY(wakeup_event, comm, event, data); + FILL_FIELD(wakeup_event, pid, event, data); + FILL_FIELD(wakeup_event, prio, event, data); + FILL_FIELD(wakeup_event, success, event, data); + FILL_FIELD(wakeup_event, cpu, event, data); + + if (trace_handler->wakeup_event) + trace_handler->wakeup_event(&wakeup_event, session, event, + cpu, timestamp, thread); +} + +/* + * Track the current task - that way we can know whether there's any + * weird events, such as a task being switched away that is not current. + */ +static int max_cpu; + +static u32 curr_pid[MAX_CPUS] = { [0 ... MAX_CPUS-1] = -1 }; + +static struct thread *curr_thread[MAX_CPUS]; + +static char next_shortname1 = 'A'; +static char next_shortname2 = '0'; + +static void +map_switch_event(struct trace_switch_event *switch_event, + struct perf_session *session, + struct event *event __used, + int this_cpu, + u64 timestamp, + struct thread *thread __used) +{ + struct thread *sched_out __used, *sched_in; + int new_shortname; + u64 timestamp0; + s64 delta; + int cpu; + + BUG_ON(this_cpu >= MAX_CPUS || this_cpu < 0); + + if (this_cpu > max_cpu) + max_cpu = this_cpu; + + timestamp0 = cpu_last_switched[this_cpu]; + cpu_last_switched[this_cpu] = timestamp; + if (timestamp0) + delta = timestamp - timestamp0; + else + delta = 0; + + if (delta < 0) + die("hm, delta: %" PRIu64 " < 0 ?\n", delta); + + + sched_out = perf_session__findnew(session, switch_event->prev_pid); + sched_in = perf_session__findnew(session, switch_event->next_pid); + + curr_thread[this_cpu] = sched_in; + + printf(" "); + + new_shortname = 0; + if (!sched_in->shortname[0]) { + sched_in->shortname[0] = next_shortname1; + sched_in->shortname[1] = next_shortname2; + + if (next_shortname1 < 'Z') { + next_shortname1++; + } else { + next_shortname1='A'; + if (next_shortname2 < '9') { + next_shortname2++; + } else { + next_shortname2='0'; + } + } + new_shortname = 1; + } + + for (cpu = 0; cpu <= max_cpu; cpu++) { + if (cpu != this_cpu) + printf(" "); + else + printf("*"); + + if (curr_thread[cpu]) { + if (curr_thread[cpu]->pid) + printf("%2s ", curr_thread[cpu]->shortname); + else + printf(". "); + } else + printf(" "); + } + + printf(" %12.6f secs ", (double)timestamp/1e9); + if (new_shortname) { + printf("%s => %s:%d\n", + sched_in->shortname, sched_in->comm, sched_in->pid); + } else { + printf("\n"); + } +} + + +static void +process_sched_switch_event(void *data, struct perf_session *session, + struct event *event, + int this_cpu, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_switch_event switch_event; + + FILL_COMMON_FIELDS(switch_event, event, data); + + FILL_ARRAY(switch_event, prev_comm, event, data); + FILL_FIELD(switch_event, prev_pid, event, data); + FILL_FIELD(switch_event, prev_prio, event, data); + FILL_FIELD(switch_event, prev_state, event, data); + FILL_ARRAY(switch_event, next_comm, event, data); + FILL_FIELD(switch_event, next_pid, event, data); + FILL_FIELD(switch_event, next_prio, event, data); + + if (curr_pid[this_cpu] != (u32)-1) { + /* + * Are we trying to switch away a PID that is + * not current? + */ + if (curr_pid[this_cpu] != switch_event.prev_pid) + nr_context_switch_bugs++; + } + if (trace_handler->switch_event) + trace_handler->switch_event(&switch_event, session, event, + this_cpu, timestamp, thread); + + curr_pid[this_cpu] = switch_event.next_pid; +} + +static void +process_sched_runtime_event(void *data, struct perf_session *session, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_runtime_event runtime_event; + + FILL_ARRAY(runtime_event, comm, event, data); + FILL_FIELD(runtime_event, pid, event, data); + FILL_FIELD(runtime_event, runtime, event, data); + FILL_FIELD(runtime_event, vruntime, event, data); + + if (trace_handler->runtime_event) + trace_handler->runtime_event(&runtime_event, session, event, cpu, timestamp, thread); +} + +static void +process_sched_fork_event(void *data, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_fork_event fork_event; + + FILL_COMMON_FIELDS(fork_event, event, data); + + FILL_ARRAY(fork_event, parent_comm, event, data); + FILL_FIELD(fork_event, parent_pid, event, data); + FILL_ARRAY(fork_event, child_comm, event, data); + FILL_FIELD(fork_event, child_pid, event, data); + + if (trace_handler->fork_event) + trace_handler->fork_event(&fork_event, event, + cpu, timestamp, thread); +} + +static void +process_sched_exit_event(struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + if (verbose) + printf("sched_exit event %p\n", event); +} + +static void +process_sched_migrate_task_event(void *data, struct perf_session *session, + struct event *event, + int cpu __used, + u64 timestamp __used, + struct thread *thread __used) +{ + struct trace_migrate_task_event migrate_task_event; + + FILL_COMMON_FIELDS(migrate_task_event, event, data); + + FILL_ARRAY(migrate_task_event, comm, event, data); + FILL_FIELD(migrate_task_event, pid, event, data); + FILL_FIELD(migrate_task_event, prio, event, data); + FILL_FIELD(migrate_task_event, cpu, event, data); + + if (trace_handler->migrate_task_event) + trace_handler->migrate_task_event(&migrate_task_event, session, + event, cpu, timestamp, thread); +} + +static void process_raw_event(union perf_event *raw_event __used, + struct perf_session *session, void *data, int cpu, + u64 timestamp, struct thread *thread) +{ + struct event *event; + int type; + + + type = trace_parse_common_type(data); + event = trace_find_event(type); + + if (!strcmp(event->name, "sched_switch")) + process_sched_switch_event(data, session, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_stat_runtime")) + process_sched_runtime_event(data, session, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_wakeup")) + process_sched_wakeup_event(data, session, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_wakeup_new")) + process_sched_wakeup_event(data, session, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_process_fork")) + process_sched_fork_event(data, event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_process_exit")) + process_sched_exit_event(event, cpu, timestamp, thread); + if (!strcmp(event->name, "sched_migrate_task")) + process_sched_migrate_task_event(data, session, event, cpu, timestamp, thread); +} + +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel __used, + struct perf_session *session) +{ + struct thread *thread; + + if (!(session->sample_type & PERF_SAMPLE_RAW)) + return 0; + + thread = perf_session__findnew(session, sample->pid); + if (thread == NULL) { + pr_debug("problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid); + + if (profile_cpu != -1 && profile_cpu != (int)sample->cpu) + return 0; + + process_raw_event(event, session, sample->raw_data, sample->cpu, + sample->time, thread); + + return 0; +} + +static struct perf_event_ops event_ops = { + .sample = process_sample_event, + .comm = perf_event__process_comm, + .lost = perf_event__process_lost, + .fork = perf_event__process_task, + .ordered_samples = true, +}; + +static int read_events(void) +{ + int err = -EINVAL; + struct perf_session *session = perf_session__new(input_name, O_RDONLY, + 0, false, &event_ops); + if (session == NULL) + return -ENOMEM; + + if (perf_session__has_traces(session, "record -R")) { + err = perf_session__process_events(session, &event_ops); + nr_events = session->hists.stats.nr_events[0]; + nr_lost_events = session->hists.stats.total_lost; + nr_lost_chunks = session->hists.stats.nr_events[PERF_RECORD_LOST]; + } + + perf_session__delete(session); + return err; +} + +static void print_bad_events(void) +{ + if (nr_unordered_timestamps && nr_timestamps) { + printf(" INFO: %.3f%% unordered timestamps (%ld out of %ld)\n", + (double)nr_unordered_timestamps/(double)nr_timestamps*100.0, + nr_unordered_timestamps, nr_timestamps); + } + if (nr_lost_events && nr_events) { + printf(" INFO: %.3f%% lost events (%ld out of %ld, in %ld chunks)\n", + (double)nr_lost_events/(double)nr_events*100.0, + nr_lost_events, nr_events, nr_lost_chunks); + } + if (nr_state_machine_bugs && nr_timestamps) { + printf(" INFO: %.3f%% state machine bugs (%ld out of %ld)", + (double)nr_state_machine_bugs/(double)nr_timestamps*100.0, + nr_state_machine_bugs, nr_timestamps); + if (nr_lost_events) + printf(" (due to lost events?)"); + printf("\n"); + } + if (nr_context_switch_bugs && nr_timestamps) { + printf(" INFO: %.3f%% context switch bugs (%ld out of %ld)", + (double)nr_context_switch_bugs/(double)nr_timestamps*100.0, + nr_context_switch_bugs, nr_timestamps); + if (nr_lost_events) + printf(" (due to lost events?)"); + printf("\n"); + } +} + +static void __cmd_lat(void) +{ + struct rb_node *next; + + setup_pager(); + read_events(); + sort_lat(); + + printf("\n ---------------------------------------------------------------------------------------------------------------\n"); + printf(" Task | Runtime ms | Switches | Average delay ms | Maximum delay ms | Maximum delay at |\n"); + printf(" ---------------------------------------------------------------------------------------------------------------\n"); + + next = rb_first(&sorted_atom_root); + + while (next) { + struct work_atoms *work_list; + + work_list = rb_entry(next, struct work_atoms, node); + output_lat_thread(work_list); + next = rb_next(next); + } + + printf(" -----------------------------------------------------------------------------------------\n"); + printf(" TOTAL: |%11.3f ms |%9" PRIu64 " |\n", + (double)all_runtime/1e6, all_count); + + printf(" ---------------------------------------------------\n"); + + print_bad_events(); + printf("\n"); + +} + +static struct trace_sched_handler map_ops = { + .wakeup_event = NULL, + .switch_event = map_switch_event, + .runtime_event = NULL, + .fork_event = NULL, +}; + +static void __cmd_map(void) +{ + max_cpu = sysconf(_SC_NPROCESSORS_CONF); + + setup_pager(); + read_events(); + print_bad_events(); +} + +static void __cmd_replay(void) +{ + unsigned long i; + + calibrate_run_measurement_overhead(); + calibrate_sleep_measurement_overhead(); + + test_calibrations(); + + read_events(); + + printf("nr_run_events: %ld\n", nr_run_events); + printf("nr_sleep_events: %ld\n", nr_sleep_events); + printf("nr_wakeup_events: %ld\n", nr_wakeup_events); + + if (targetless_wakeups) + printf("target-less wakeups: %ld\n", targetless_wakeups); + if (multitarget_wakeups) + printf("multi-target wakeups: %ld\n", multitarget_wakeups); + if (nr_run_events_optimized) + printf("run atoms optimized: %ld\n", + nr_run_events_optimized); + + print_task_traces(); + add_cross_task_wakeups(); + + create_tasks(); + printf("------------------------------------------------------------\n"); + for (i = 0; i < replay_repeat; i++) + run_one_test(); +} + + +static const char * const sched_usage[] = { + "perf sched [<options>] {record|latency|map|replay|trace}", + NULL +}; + +static const struct option sched_options[] = { + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_END() +}; + +static const char * const latency_usage[] = { + "perf sched latency [<options>]", + NULL +}; + +static const struct option latency_options[] = { + OPT_STRING('s', "sort", &sort_order, "key[,key2...]", + "sort by key(s): runtime, switch, avg, max"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_INTEGER('C', "CPU", &profile_cpu, + "CPU to profile on"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_END() +}; + +static const char * const replay_usage[] = { + "perf sched replay [<options>]", + NULL +}; + +static const struct option replay_options[] = { + OPT_UINTEGER('r', "repeat", &replay_repeat, + "repeat the workload replay N times (-1: infinite)"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_END() +}; + +static void setup_sorting(void) +{ + char *tmp, *tok, *str = strdup(sort_order); + + for (tok = strtok_r(str, ", ", &tmp); + tok; tok = strtok_r(NULL, ", ", &tmp)) { + if (sort_dimension__add(tok, &sort_list) < 0) { + error("Unknown --sort key: `%s'", tok); + usage_with_options(latency_usage, latency_options); + } + } + + free(str); + + sort_dimension__add("pid", &cmp_pid); +} + +static const char *record_args[] = { + "record", + "-a", + "-R", + "-f", + "-m", "1024", + "-c", "1", + "-e", "sched:sched_switch", + "-e", "sched:sched_stat_wait", + "-e", "sched:sched_stat_sleep", + "-e", "sched:sched_stat_iowait", + "-e", "sched:sched_stat_runtime", + "-e", "sched:sched_process_exit", + "-e", "sched:sched_process_fork", + "-e", "sched:sched_wakeup", + "-e", "sched:sched_migrate_task", +}; + +static int __cmd_record(int argc, const char **argv) +{ + unsigned int rec_argc, i, j; + const char **rec_argv; + + rec_argc = ARRAY_SIZE(record_args) + argc - 1; + rec_argv = calloc(rec_argc + 1, sizeof(char *)); + + if (rec_argv == NULL) + return -ENOMEM; + + for (i = 0; i < ARRAY_SIZE(record_args); i++) + rec_argv[i] = strdup(record_args[i]); + + for (j = 1; j < (unsigned int)argc; j++, i++) + rec_argv[i] = argv[j]; + + BUG_ON(i != rec_argc); + + return cmd_record(i, rec_argv, NULL); +} + +int cmd_sched(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, sched_options, sched_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + if (!argc) + usage_with_options(sched_usage, sched_options); + + /* + * Aliased to 'perf script' for now: + */ + if (!strcmp(argv[0], "script")) + return cmd_script(argc, argv, prefix); + + symbol__init(); + if (!strncmp(argv[0], "rec", 3)) { + return __cmd_record(argc, argv); + } else if (!strncmp(argv[0], "lat", 3)) { + trace_handler = &lat_ops; + if (argc > 1) { + argc = parse_options(argc, argv, latency_options, latency_usage, 0); + if (argc) + usage_with_options(latency_usage, latency_options); + } + setup_sorting(); + __cmd_lat(); + } else if (!strcmp(argv[0], "map")) { + trace_handler = &map_ops; + setup_sorting(); + __cmd_map(); + } else if (!strncmp(argv[0], "rep", 3)) { + trace_handler = &replay_ops; + if (argc) { + argc = parse_options(argc, argv, replay_options, replay_usage, 0); + if (argc) + usage_with_options(replay_usage, replay_options); + } + __cmd_replay(); + } else { + usage_with_options(sched_usage, sched_options); + } + + return 0; +} diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c new file mode 100644 index 00000000..22747de7 --- /dev/null +++ b/tools/perf/builtin-script.c @@ -0,0 +1,1230 @@ +#include "builtin.h" + +#include "perf.h" +#include "util/cache.h" +#include "util/debug.h" +#include "util/exec_cmd.h" +#include "util/header.h" +#include "util/parse-options.h" +#include "util/session.h" +#include "util/symbol.h" +#include "util/thread.h" +#include "util/trace-event.h" +#include "util/util.h" +#include "util/evlist.h" +#include "util/evsel.h" + +static char const *script_name; +static char const *generate_script_lang; +static bool debug_mode; +static u64 last_timestamp; +static u64 nr_unordered; +extern const struct option record_options[]; +static bool no_callchain; + +enum perf_output_field { + PERF_OUTPUT_COMM = 1U << 0, + PERF_OUTPUT_TID = 1U << 1, + PERF_OUTPUT_PID = 1U << 2, + PERF_OUTPUT_TIME = 1U << 3, + PERF_OUTPUT_CPU = 1U << 4, + PERF_OUTPUT_EVNAME = 1U << 5, + PERF_OUTPUT_TRACE = 1U << 6, + PERF_OUTPUT_SYM = 1U << 7, +}; + +struct output_option { + const char *str; + enum perf_output_field field; +} all_output_options[] = { + {.str = "comm", .field = PERF_OUTPUT_COMM}, + {.str = "tid", .field = PERF_OUTPUT_TID}, + {.str = "pid", .field = PERF_OUTPUT_PID}, + {.str = "time", .field = PERF_OUTPUT_TIME}, + {.str = "cpu", .field = PERF_OUTPUT_CPU}, + {.str = "event", .field = PERF_OUTPUT_EVNAME}, + {.str = "trace", .field = PERF_OUTPUT_TRACE}, + {.str = "sym", .field = PERF_OUTPUT_SYM}, +}; + +/* default set to maintain compatibility with current format */ +static struct { + bool user_set; + bool wildcard_set; + u64 fields; + u64 invalid_fields; +} output[PERF_TYPE_MAX] = { + + [PERF_TYPE_HARDWARE] = { + .user_set = false, + + .fields = PERF_OUTPUT_COMM | PERF_OUTPUT_TID | + PERF_OUTPUT_CPU | PERF_OUTPUT_TIME | + PERF_OUTPUT_EVNAME | PERF_OUTPUT_SYM, + + .invalid_fields = PERF_OUTPUT_TRACE, + }, + + [PERF_TYPE_SOFTWARE] = { + .user_set = false, + + .fields = PERF_OUTPUT_COMM | PERF_OUTPUT_TID | + PERF_OUTPUT_CPU | PERF_OUTPUT_TIME | + PERF_OUTPUT_EVNAME | PERF_OUTPUT_SYM, + + .invalid_fields = PERF_OUTPUT_TRACE, + }, + + [PERF_TYPE_TRACEPOINT] = { + .user_set = false, + + .fields = PERF_OUTPUT_COMM | PERF_OUTPUT_TID | + PERF_OUTPUT_CPU | PERF_OUTPUT_TIME | + PERF_OUTPUT_EVNAME | PERF_OUTPUT_TRACE, + }, + + [PERF_TYPE_RAW] = { + .user_set = false, + + .fields = PERF_OUTPUT_COMM | PERF_OUTPUT_TID | + PERF_OUTPUT_CPU | PERF_OUTPUT_TIME | + PERF_OUTPUT_EVNAME | PERF_OUTPUT_SYM, + + .invalid_fields = PERF_OUTPUT_TRACE, + }, +}; + +static bool output_set_by_user(void) +{ + int j; + for (j = 0; j < PERF_TYPE_MAX; ++j) { + if (output[j].user_set) + return true; + } + return false; +} + +static const char *output_field2str(enum perf_output_field field) +{ + int i, imax = ARRAY_SIZE(all_output_options); + const char *str = ""; + + for (i = 0; i < imax; ++i) { + if (all_output_options[i].field == field) { + str = all_output_options[i].str; + break; + } + } + return str; +} + +#define PRINT_FIELD(x) (output[attr->type].fields & PERF_OUTPUT_##x) + +static int perf_event_attr__check_stype(struct perf_event_attr *attr, + u64 sample_type, const char *sample_msg, + enum perf_output_field field) +{ + int type = attr->type; + const char *evname; + + if (attr->sample_type & sample_type) + return 0; + + if (output[type].user_set) { + evname = __event_name(attr->type, attr->config); + pr_err("Samples for '%s' event do not have %s attribute set. " + "Cannot print '%s' field.\n", + evname, sample_msg, output_field2str(field)); + return -1; + } + + /* user did not ask for it explicitly so remove from the default list */ + output[type].fields &= ~field; + evname = __event_name(attr->type, attr->config); + pr_debug("Samples for '%s' event do not have %s attribute set. " + "Skipping '%s' field.\n", + evname, sample_msg, output_field2str(field)); + + return 0; +} + +static int perf_evsel__check_attr(struct perf_evsel *evsel, + struct perf_session *session) +{ + struct perf_event_attr *attr = &evsel->attr; + + if (PRINT_FIELD(TRACE) && + !perf_session__has_traces(session, "record -R")) + return -EINVAL; + + if (PRINT_FIELD(SYM)) { + if (perf_event_attr__check_stype(attr, PERF_SAMPLE_IP, "IP", + PERF_OUTPUT_SYM)) + return -EINVAL; + + if (!no_callchain && + !(attr->sample_type & PERF_SAMPLE_CALLCHAIN)) + symbol_conf.use_callchain = false; + } + + if ((PRINT_FIELD(PID) || PRINT_FIELD(TID)) && + perf_event_attr__check_stype(attr, PERF_SAMPLE_TID, "TID", + PERF_OUTPUT_TID|PERF_OUTPUT_PID)) + return -EINVAL; + + if (PRINT_FIELD(TIME) && + perf_event_attr__check_stype(attr, PERF_SAMPLE_TIME, "TIME", + PERF_OUTPUT_TIME)) + return -EINVAL; + + if (PRINT_FIELD(CPU) && + perf_event_attr__check_stype(attr, PERF_SAMPLE_CPU, "CPU", + PERF_OUTPUT_CPU)) + return -EINVAL; + + return 0; +} + +/* + * verify all user requested events exist and the samples + * have the expected data + */ +static int perf_session__check_output_opt(struct perf_session *session) +{ + int j; + struct perf_evsel *evsel; + + for (j = 0; j < PERF_TYPE_MAX; ++j) { + evsel = perf_session__find_first_evtype(session, j); + + /* + * even if fields is set to 0 (ie., show nothing) event must + * exist if user explicitly includes it on the command line + */ + if (!evsel && output[j].user_set && !output[j].wildcard_set) { + pr_err("%s events do not exist. " + "Remove corresponding -f option to proceed.\n", + event_type(j)); + return -1; + } + + if (evsel && output[j].fields && + perf_evsel__check_attr(evsel, session)) + return -1; + } + + return 0; +} + +static void print_sample_start(struct perf_sample *sample, + struct thread *thread, + struct perf_event_attr *attr) +{ + int type; + struct event *event; + const char *evname = NULL; + unsigned long secs; + unsigned long usecs; + unsigned long long nsecs; + + if (PRINT_FIELD(COMM)) { + if (latency_format) + printf("%8.8s ", thread->comm); + else if (PRINT_FIELD(SYM) && symbol_conf.use_callchain) + printf("%s ", thread->comm); + else + printf("%16s ", thread->comm); + } + + if (PRINT_FIELD(PID) && PRINT_FIELD(TID)) + printf("%5d/%-5d ", sample->pid, sample->tid); + else if (PRINT_FIELD(PID)) + printf("%5d ", sample->pid); + else if (PRINT_FIELD(TID)) + printf("%5d ", sample->tid); + + if (PRINT_FIELD(CPU)) { + if (latency_format) + printf("%3d ", sample->cpu); + else + printf("[%03d] ", sample->cpu); + } + + if (PRINT_FIELD(TIME)) { + nsecs = sample->time; + secs = nsecs / NSECS_PER_SEC; + nsecs -= secs * NSECS_PER_SEC; + usecs = nsecs / NSECS_PER_USEC; + printf("%5lu.%06lu: ", secs, usecs); + } + + if (PRINT_FIELD(EVNAME)) { + if (attr->type == PERF_TYPE_TRACEPOINT) { + type = trace_parse_common_type(sample->raw_data); + event = trace_find_event(type); + if (event) + evname = event->name; + } else + evname = __event_name(attr->type, attr->config); + + printf("%s: ", evname ? evname : "(unknown)"); + } +} + +static void process_event(union perf_event *event __unused, + struct perf_sample *sample, + struct perf_evsel *evsel, + struct perf_session *session, + struct thread *thread) +{ + struct perf_event_attr *attr = &evsel->attr; + + if (output[attr->type].fields == 0) + return; + + print_sample_start(sample, thread, attr); + + if (PRINT_FIELD(TRACE)) + print_trace_event(sample->cpu, sample->raw_data, + sample->raw_size); + + if (PRINT_FIELD(SYM)) { + if (!symbol_conf.use_callchain) + printf(" "); + else + printf("\n"); + perf_session__print_symbols(event, sample, session); + } + + printf("\n"); +} + +static int default_start_script(const char *script __unused, + int argc __unused, + const char **argv __unused) +{ + return 0; +} + +static int default_stop_script(void) +{ + return 0; +} + +static int default_generate_script(const char *outfile __unused) +{ + return 0; +} + +static struct scripting_ops default_scripting_ops = { + .start_script = default_start_script, + .stop_script = default_stop_script, + .process_event = process_event, + .generate_script = default_generate_script, +}; + +static struct scripting_ops *scripting_ops; + +static void setup_scripting(void) +{ + setup_perl_scripting(); + setup_python_scripting(); + + scripting_ops = &default_scripting_ops; +} + +static int cleanup_scripting(void) +{ + pr_debug("\nperf script stopped\n"); + + return scripting_ops->stop_script(); +} + +static char const *input_name = "perf.data"; + +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel, + struct perf_session *session) +{ + struct thread *thread = perf_session__findnew(session, event->ip.pid); + + if (thread == NULL) { + pr_debug("problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + if (debug_mode) { + if (sample->time < last_timestamp) { + pr_err("Samples misordered, previous: %" PRIu64 + " this: %" PRIu64 "\n", last_timestamp, + sample->time); + nr_unordered++; + } + last_timestamp = sample->time; + return 0; + } + scripting_ops->process_event(event, sample, evsel, session, thread); + + session->hists.stats.total_period += sample->period; + return 0; +} + +static struct perf_event_ops event_ops = { + .sample = process_sample_event, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .exit = perf_event__process_task, + .fork = perf_event__process_task, + .attr = perf_event__process_attr, + .event_type = perf_event__process_event_type, + .tracing_data = perf_event__process_tracing_data, + .build_id = perf_event__process_build_id, + .ordered_samples = true, + .ordering_requires_timestamps = true, +}; + +extern volatile int session_done; + +static void sig_handler(int sig __unused) +{ + session_done = 1; +} + +static int __cmd_script(struct perf_session *session) +{ + int ret; + + signal(SIGINT, sig_handler); + + ret = perf_session__process_events(session, &event_ops); + + if (debug_mode) + pr_err("Misordered timestamps: %" PRIu64 "\n", nr_unordered); + + return ret; +} + +struct script_spec { + struct list_head node; + struct scripting_ops *ops; + char spec[0]; +}; + +static LIST_HEAD(script_specs); + +static struct script_spec *script_spec__new(const char *spec, + struct scripting_ops *ops) +{ + struct script_spec *s = malloc(sizeof(*s) + strlen(spec) + 1); + + if (s != NULL) { + strcpy(s->spec, spec); + s->ops = ops; + } + + return s; +} + +static void script_spec__delete(struct script_spec *s) +{ + free(s->spec); + free(s); +} + +static void script_spec__add(struct script_spec *s) +{ + list_add_tail(&s->node, &script_specs); +} + +static struct script_spec *script_spec__find(const char *spec) +{ + struct script_spec *s; + + list_for_each_entry(s, &script_specs, node) + if (strcasecmp(s->spec, spec) == 0) + return s; + return NULL; +} + +static struct script_spec *script_spec__findnew(const char *spec, + struct scripting_ops *ops) +{ + struct script_spec *s = script_spec__find(spec); + + if (s) + return s; + + s = script_spec__new(spec, ops); + if (!s) + goto out_delete_spec; + + script_spec__add(s); + + return s; + +out_delete_spec: + script_spec__delete(s); + + return NULL; +} + +int script_spec_register(const char *spec, struct scripting_ops *ops) +{ + struct script_spec *s; + + s = script_spec__find(spec); + if (s) + return -1; + + s = script_spec__findnew(spec, ops); + if (!s) + return -1; + + return 0; +} + +static struct scripting_ops *script_spec__lookup(const char *spec) +{ + struct script_spec *s = script_spec__find(spec); + if (!s) + return NULL; + + return s->ops; +} + +static void list_available_languages(void) +{ + struct script_spec *s; + + fprintf(stderr, "\n"); + fprintf(stderr, "Scripting language extensions (used in " + "perf script -s [spec:]script.[spec]):\n\n"); + + list_for_each_entry(s, &script_specs, node) + fprintf(stderr, " %-42s [%s]\n", s->spec, s->ops->name); + + fprintf(stderr, "\n"); +} + +static int parse_scriptname(const struct option *opt __used, + const char *str, int unset __used) +{ + char spec[PATH_MAX]; + const char *script, *ext; + int len; + + if (strcmp(str, "lang") == 0) { + list_available_languages(); + exit(0); + } + + script = strchr(str, ':'); + if (script) { + len = script - str; + if (len >= PATH_MAX) { + fprintf(stderr, "invalid language specifier"); + return -1; + } + strncpy(spec, str, len); + spec[len] = '\0'; + scripting_ops = script_spec__lookup(spec); + if (!scripting_ops) { + fprintf(stderr, "invalid language specifier"); + return -1; + } + script++; + } else { + script = str; + ext = strrchr(script, '.'); + if (!ext) { + fprintf(stderr, "invalid script extension"); + return -1; + } + scripting_ops = script_spec__lookup(++ext); + if (!scripting_ops) { + fprintf(stderr, "invalid script extension"); + return -1; + } + } + + script_name = strdup(script); + + return 0; +} + +static int parse_output_fields(const struct option *opt __used, + const char *arg, int unset __used) +{ + char *tok; + int i, imax = sizeof(all_output_options) / sizeof(struct output_option); + int j; + int rc = 0; + char *str = strdup(arg); + int type = -1; + + if (!str) + return -ENOMEM; + + /* first word can state for which event type the user is specifying + * the fields. If no type exists, the specified fields apply to all + * event types found in the file minus the invalid fields for a type. + */ + tok = strchr(str, ':'); + if (tok) { + *tok = '\0'; + tok++; + if (!strcmp(str, "hw")) + type = PERF_TYPE_HARDWARE; + else if (!strcmp(str, "sw")) + type = PERF_TYPE_SOFTWARE; + else if (!strcmp(str, "trace")) + type = PERF_TYPE_TRACEPOINT; + else if (!strcmp(str, "raw")) + type = PERF_TYPE_RAW; + else { + fprintf(stderr, "Invalid event type in field string.\n"); + return -EINVAL; + } + + if (output[type].user_set) + pr_warning("Overriding previous field request for %s events.\n", + event_type(type)); + + output[type].fields = 0; + output[type].user_set = true; + output[type].wildcard_set = false; + + } else { + tok = str; + if (strlen(str) == 0) { + fprintf(stderr, + "Cannot set fields to 'none' for all event types.\n"); + rc = -EINVAL; + goto out; + } + + if (output_set_by_user()) + pr_warning("Overriding previous field request for all events.\n"); + + for (j = 0; j < PERF_TYPE_MAX; ++j) { + output[j].fields = 0; + output[j].user_set = true; + output[j].wildcard_set = true; + } + } + + tok = strtok(tok, ","); + while (tok) { + for (i = 0; i < imax; ++i) { + if (strcmp(tok, all_output_options[i].str) == 0) + break; + } + if (i == imax) { + fprintf(stderr, "Invalid field requested.\n"); + rc = -EINVAL; + goto out; + } + + if (type == -1) { + /* add user option to all events types for + * which it is valid + */ + for (j = 0; j < PERF_TYPE_MAX; ++j) { + if (output[j].invalid_fields & all_output_options[i].field) { + pr_warning("\'%s\' not valid for %s events. Ignoring.\n", + all_output_options[i].str, event_type(j)); + } else + output[j].fields |= all_output_options[i].field; + } + } else { + if (output[type].invalid_fields & all_output_options[i].field) { + fprintf(stderr, "\'%s\' not valid for %s events.\n", + all_output_options[i].str, event_type(type)); + + rc = -EINVAL; + goto out; + } + output[type].fields |= all_output_options[i].field; + } + + tok = strtok(NULL, ","); + } + + if (type >= 0) { + if (output[type].fields == 0) { + pr_debug("No fields requested for %s type. " + "Events will not be displayed.\n", event_type(type)); + } + } + +out: + free(str); + return rc; +} + +/* Helper function for filesystems that return a dent->d_type DT_UNKNOWN */ +static int is_directory(const char *base_path, const struct dirent *dent) +{ + char path[PATH_MAX]; + struct stat st; + + sprintf(path, "%s/%s", base_path, dent->d_name); + if (stat(path, &st)) + return 0; + + return S_ISDIR(st.st_mode); +} + +#define for_each_lang(scripts_path, scripts_dir, lang_dirent, lang_next)\ + while (!readdir_r(scripts_dir, &lang_dirent, &lang_next) && \ + lang_next) \ + if ((lang_dirent.d_type == DT_DIR || \ + (lang_dirent.d_type == DT_UNKNOWN && \ + is_directory(scripts_path, &lang_dirent))) && \ + (strcmp(lang_dirent.d_name, ".")) && \ + (strcmp(lang_dirent.d_name, ".."))) + +#define for_each_script(lang_path, lang_dir, script_dirent, script_next)\ + while (!readdir_r(lang_dir, &script_dirent, &script_next) && \ + script_next) \ + if (script_dirent.d_type != DT_DIR && \ + (script_dirent.d_type != DT_UNKNOWN || \ + !is_directory(lang_path, &script_dirent))) + + +#define RECORD_SUFFIX "-record" +#define REPORT_SUFFIX "-report" + +struct script_desc { + struct list_head node; + char *name; + char *half_liner; + char *args; +}; + +static LIST_HEAD(script_descs); + +static struct script_desc *script_desc__new(const char *name) +{ + struct script_desc *s = zalloc(sizeof(*s)); + + if (s != NULL && name) + s->name = strdup(name); + + return s; +} + +static void script_desc__delete(struct script_desc *s) +{ + free(s->name); + free(s->half_liner); + free(s->args); + free(s); +} + +static void script_desc__add(struct script_desc *s) +{ + list_add_tail(&s->node, &script_descs); +} + +static struct script_desc *script_desc__find(const char *name) +{ + struct script_desc *s; + + list_for_each_entry(s, &script_descs, node) + if (strcasecmp(s->name, name) == 0) + return s; + return NULL; +} + +static struct script_desc *script_desc__findnew(const char *name) +{ + struct script_desc *s = script_desc__find(name); + + if (s) + return s; + + s = script_desc__new(name); + if (!s) + goto out_delete_desc; + + script_desc__add(s); + + return s; + +out_delete_desc: + script_desc__delete(s); + + return NULL; +} + +static const char *ends_with(const char *str, const char *suffix) +{ + size_t suffix_len = strlen(suffix); + const char *p = str; + + if (strlen(str) > suffix_len) { + p = str + strlen(str) - suffix_len; + if (!strncmp(p, suffix, suffix_len)) + return p; + } + + return NULL; +} + +static char *ltrim(char *str) +{ + int len = strlen(str); + + while (len && isspace(*str)) { + len--; + str++; + } + + return str; +} + +static int read_script_info(struct script_desc *desc, const char *filename) +{ + char line[BUFSIZ], *p; + FILE *fp; + + fp = fopen(filename, "r"); + if (!fp) + return -1; + + while (fgets(line, sizeof(line), fp)) { + p = ltrim(line); + if (strlen(p) == 0) + continue; + if (*p != '#') + continue; + p++; + if (strlen(p) && *p == '!') + continue; + + p = ltrim(p); + if (strlen(p) && p[strlen(p) - 1] == '\n') + p[strlen(p) - 1] = '\0'; + + if (!strncmp(p, "description:", strlen("description:"))) { + p += strlen("description:"); + desc->half_liner = strdup(ltrim(p)); + continue; + } + + if (!strncmp(p, "args:", strlen("args:"))) { + p += strlen("args:"); + desc->args = strdup(ltrim(p)); + continue; + } + } + + fclose(fp); + + return 0; +} + +static int list_available_scripts(const struct option *opt __used, + const char *s __used, int unset __used) +{ + struct dirent *script_next, *lang_next, script_dirent, lang_dirent; + char scripts_path[MAXPATHLEN]; + DIR *scripts_dir, *lang_dir; + char script_path[MAXPATHLEN]; + char lang_path[MAXPATHLEN]; + struct script_desc *desc; + char first_half[BUFSIZ]; + char *script_root; + char *str; + + snprintf(scripts_path, MAXPATHLEN, "%s/scripts", perf_exec_path()); + + scripts_dir = opendir(scripts_path); + if (!scripts_dir) + return -1; + + for_each_lang(scripts_path, scripts_dir, lang_dirent, lang_next) { + snprintf(lang_path, MAXPATHLEN, "%s/%s/bin", scripts_path, + lang_dirent.d_name); + lang_dir = opendir(lang_path); + if (!lang_dir) + continue; + + for_each_script(lang_path, lang_dir, script_dirent, script_next) { + script_root = strdup(script_dirent.d_name); + str = (char *)ends_with(script_root, REPORT_SUFFIX); + if (str) { + *str = '\0'; + desc = script_desc__findnew(script_root); + snprintf(script_path, MAXPATHLEN, "%s/%s", + lang_path, script_dirent.d_name); + read_script_info(desc, script_path); + } + free(script_root); + } + } + + fprintf(stdout, "List of available trace scripts:\n"); + list_for_each_entry(desc, &script_descs, node) { + sprintf(first_half, "%s %s", desc->name, + desc->args ? desc->args : ""); + fprintf(stdout, " %-36s %s\n", first_half, + desc->half_liner ? desc->half_liner : ""); + } + + exit(0); +} + +static char *get_script_path(const char *script_root, const char *suffix) +{ + struct dirent *script_next, *lang_next, script_dirent, lang_dirent; + char scripts_path[MAXPATHLEN]; + char script_path[MAXPATHLEN]; + DIR *scripts_dir, *lang_dir; + char lang_path[MAXPATHLEN]; + char *str, *__script_root; + char *path = NULL; + + snprintf(scripts_path, MAXPATHLEN, "%s/scripts", perf_exec_path()); + + scripts_dir = opendir(scripts_path); + if (!scripts_dir) + return NULL; + + for_each_lang(scripts_path, scripts_dir, lang_dirent, lang_next) { + snprintf(lang_path, MAXPATHLEN, "%s/%s/bin", scripts_path, + lang_dirent.d_name); + lang_dir = opendir(lang_path); + if (!lang_dir) + continue; + + for_each_script(lang_path, lang_dir, script_dirent, script_next) { + __script_root = strdup(script_dirent.d_name); + str = (char *)ends_with(__script_root, suffix); + if (str) { + *str = '\0'; + if (strcmp(__script_root, script_root)) + continue; + snprintf(script_path, MAXPATHLEN, "%s/%s", + lang_path, script_dirent.d_name); + path = strdup(script_path); + free(__script_root); + break; + } + free(__script_root); + } + } + + return path; +} + +static bool is_top_script(const char *script_path) +{ + return ends_with(script_path, "top") == NULL ? false : true; +} + +static int has_required_arg(char *script_path) +{ + struct script_desc *desc; + int n_args = 0; + char *p; + + desc = script_desc__new(NULL); + + if (read_script_info(desc, script_path)) + goto out; + + if (!desc->args) + goto out; + + for (p = desc->args; *p; p++) + if (*p == '<') + n_args++; +out: + script_desc__delete(desc); + + return n_args; +} + +static const char * const script_usage[] = { + "perf script [<options>]", + "perf script [<options>] record <script> [<record-options>] <command>", + "perf script [<options>] report <script> [script-args]", + "perf script [<options>] <script> [<record-options>] <command>", + "perf script [<options>] <top-script> [script-args]", + NULL +}; + +static const struct option options[] = { + OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, + "dump raw trace in ASCII"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_BOOLEAN('L', "Latency", &latency_format, + "show latency attributes (irqs/preemption disabled, etc)"), + OPT_CALLBACK_NOOPT('l', "list", NULL, NULL, "list available scripts", + list_available_scripts), + OPT_CALLBACK('s', "script", NULL, "name", + "script file name (lang:script name, script name, or *)", + parse_scriptname), + OPT_STRING('g', "gen-script", &generate_script_lang, "lang", + "generate perf-script.xx script in specified language"), + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_BOOLEAN('d', "debug-mode", &debug_mode, + "do various checks like samples ordering and lost events"), + OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name, + "file", "vmlinux pathname"), + OPT_STRING(0, "kallsyms", &symbol_conf.kallsyms_name, + "file", "kallsyms pathname"), + OPT_BOOLEAN('G', "hide-call-graph", &no_callchain, + "When printing symbols do not display call chain"), + OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory", + "Look for files with symbols relative to this directory"), + OPT_CALLBACK('f', "fields", NULL, "str", + "comma separated output fields prepend with 'type:'. Valid types: hw,sw,trace,raw. Fields: comm,tid,pid,time,cpu,event,trace,sym", + parse_output_fields), + + OPT_END() +}; + +static bool have_cmd(int argc, const char **argv) +{ + char **__argv = malloc(sizeof(const char *) * argc); + + if (!__argv) + die("malloc"); + memcpy(__argv, argv, sizeof(const char *) * argc); + argc = parse_options(argc, (const char **)__argv, record_options, + NULL, PARSE_OPT_STOP_AT_NON_OPTION); + free(__argv); + + return argc != 0; +} + +int cmd_script(int argc, const char **argv, const char *prefix __used) +{ + char *rec_script_path = NULL; + char *rep_script_path = NULL; + struct perf_session *session; + char *script_path = NULL; + const char **__argv; + bool system_wide; + int i, j, err; + + setup_scripting(); + + argc = parse_options(argc, argv, options, script_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + + if (argc > 1 && !strncmp(argv[0], "rec", strlen("rec"))) { + rec_script_path = get_script_path(argv[1], RECORD_SUFFIX); + if (!rec_script_path) + return cmd_record(argc, argv, NULL); + } + + if (argc > 1 && !strncmp(argv[0], "rep", strlen("rep"))) { + rep_script_path = get_script_path(argv[1], REPORT_SUFFIX); + if (!rep_script_path) { + fprintf(stderr, + "Please specify a valid report script" + "(see 'perf script -l' for listing)\n"); + return -1; + } + } + + /* make sure PERF_EXEC_PATH is set for scripts */ + perf_set_argv_exec_path(perf_exec_path()); + + if (argc && !script_name && !rec_script_path && !rep_script_path) { + int live_pipe[2]; + int rep_args; + pid_t pid; + + rec_script_path = get_script_path(argv[0], RECORD_SUFFIX); + rep_script_path = get_script_path(argv[0], REPORT_SUFFIX); + + if (!rec_script_path && !rep_script_path) { + fprintf(stderr, " Couldn't find script %s\n\n See perf" + " script -l for available scripts.\n", argv[0]); + usage_with_options(script_usage, options); + } + + if (is_top_script(argv[0])) { + rep_args = argc - 1; + } else { + int rec_args; + + rep_args = has_required_arg(rep_script_path); + rec_args = (argc - 1) - rep_args; + if (rec_args < 0) { + fprintf(stderr, " %s script requires options." + "\n\n See perf script -l for available " + "scripts and options.\n", argv[0]); + usage_with_options(script_usage, options); + } + } + + if (pipe(live_pipe) < 0) { + perror("failed to create pipe"); + exit(-1); + } + + pid = fork(); + if (pid < 0) { + perror("failed to fork"); + exit(-1); + } + + if (!pid) { + system_wide = true; + j = 0; + + dup2(live_pipe[1], 1); + close(live_pipe[0]); + + if (!is_top_script(argv[0])) + system_wide = !have_cmd(argc - rep_args, + &argv[rep_args]); + + __argv = malloc((argc + 6) * sizeof(const char *)); + if (!__argv) + die("malloc"); + + __argv[j++] = "/bin/sh"; + __argv[j++] = rec_script_path; + if (system_wide) + __argv[j++] = "-a"; + __argv[j++] = "-q"; + __argv[j++] = "-o"; + __argv[j++] = "-"; + for (i = rep_args + 1; i < argc; i++) + __argv[j++] = argv[i]; + __argv[j++] = NULL; + + execvp("/bin/sh", (char **)__argv); + free(__argv); + exit(-1); + } + + dup2(live_pipe[0], 0); + close(live_pipe[1]); + + __argv = malloc((argc + 4) * sizeof(const char *)); + if (!__argv) + die("malloc"); + j = 0; + __argv[j++] = "/bin/sh"; + __argv[j++] = rep_script_path; + for (i = 1; i < rep_args + 1; i++) + __argv[j++] = argv[i]; + __argv[j++] = "-i"; + __argv[j++] = "-"; + __argv[j++] = NULL; + + execvp("/bin/sh", (char **)__argv); + free(__argv); + exit(-1); + } + + if (rec_script_path) + script_path = rec_script_path; + if (rep_script_path) + script_path = rep_script_path; + + if (script_path) { + system_wide = false; + j = 0; + + if (rec_script_path) + system_wide = !have_cmd(argc - 1, &argv[1]); + + __argv = malloc((argc + 2) * sizeof(const char *)); + if (!__argv) + die("malloc"); + __argv[j++] = "/bin/sh"; + __argv[j++] = script_path; + if (system_wide) + __argv[j++] = "-a"; + for (i = 2; i < argc; i++) + __argv[j++] = argv[i]; + __argv[j++] = NULL; + + execvp("/bin/sh", (char **)__argv); + free(__argv); + exit(-1); + } + + if (symbol__init() < 0) + return -1; + if (!script_name) + setup_pager(); + + session = perf_session__new(input_name, O_RDONLY, 0, false, &event_ops); + if (session == NULL) + return -ENOMEM; + + if (!no_callchain) + symbol_conf.use_callchain = true; + else + symbol_conf.use_callchain = false; + + if (generate_script_lang) { + struct stat perf_stat; + int input; + + if (output_set_by_user()) { + fprintf(stderr, + "custom fields not supported for generated scripts"); + return -1; + } + + input = open(input_name, O_RDONLY); + if (input < 0) { + perror("failed to open file"); + exit(-1); + } + + err = fstat(input, &perf_stat); + if (err < 0) { + perror("failed to stat file"); + exit(-1); + } + + if (!perf_stat.st_size) { + fprintf(stderr, "zero-sized file, nothing to do!\n"); + exit(0); + } + + scripting_ops = script_spec__lookup(generate_script_lang); + if (!scripting_ops) { + fprintf(stderr, "invalid language specifier"); + return -1; + } + + err = scripting_ops->generate_script("perf-script"); + goto out; + } + + if (script_name) { + err = scripting_ops->start_script(script_name, argc, argv); + if (err) + goto out; + pr_debug("perf script started with script %s\n\n", script_name); + } + + + err = perf_session__check_output_opt(session); + if (err < 0) + goto out; + + err = __cmd_script(session); + + perf_session__delete(session); + cleanup_scripting(); +out: + return err; +} diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c new file mode 100644 index 00000000..a9f06715 --- /dev/null +++ b/tools/perf/builtin-stat.c @@ -0,0 +1,1235 @@ +/* + * builtin-stat.c + * + * Builtin stat command: Give a precise performance counters summary + * overview about any workload, CPU or specific PID. + * + * Sample output: + + $ perf stat ./hackbench 10 + + Time: 0.118 + + Performance counter stats for './hackbench 10': + + 1708.761321 task-clock # 11.037 CPUs utilized + 41,190 context-switches # 0.024 M/sec + 6,735 CPU-migrations # 0.004 M/sec + 17,318 page-faults # 0.010 M/sec + 5,205,202,243 cycles # 3.046 GHz + 3,856,436,920 stalled-cycles-frontend # 74.09% frontend cycles idle + 1,600,790,871 stalled-cycles-backend # 30.75% backend cycles idle + 2,603,501,247 instructions # 0.50 insns per cycle + # 1.48 stalled cycles per insn + 484,357,498 branches # 283.455 M/sec + 6,388,934 branch-misses # 1.32% of all branches + + 0.154822978 seconds time elapsed + + * + * Copyright (C) 2008-2011, Red Hat Inc, Ingo Molnar <mingo@redhat.com> + * + * Improvements and fixes by: + * + * Arjan van de Ven <arjan@linux.intel.com> + * Yanmin Zhang <yanmin.zhang@intel.com> + * Wu Fengguang <fengguang.wu@intel.com> + * Mike Galbraith <efault@gmx.de> + * Paul Mackerras <paulus@samba.org> + * Jaswinder Singh Rajput <jaswinder@kernel.org> + * + * Released under the GPL v2. (and only v2, not any later version) + */ + +#include "perf.h" +#include "builtin.h" +#include "util/util.h" +#include "util/parse-options.h" +#include "util/parse-events.h" +#include "util/event.h" +#include "util/evlist.h" +#include "util/evsel.h" +#include "util/debug.h" +#include "util/color.h" +#include "util/header.h" +#include "util/cpumap.h" +#include "util/thread.h" +#include "util/thread_map.h" + +#include <sys/prctl.h> +#include <math.h> +#include <locale.h> + +#define DEFAULT_SEPARATOR " " + +static struct perf_event_attr default_attrs[] = { + + { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_TASK_CLOCK }, + { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CONTEXT_SWITCHES }, + { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CPU_MIGRATIONS }, + { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_PAGE_FAULTS }, + + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CPU_CYCLES }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_FRONTEND }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_STALLED_CYCLES_BACKEND }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_INSTRUCTIONS }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_INSTRUCTIONS }, + { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_MISSES }, + +}; + +/* + * Detailed stats (-d), covering the L1 and last level data caches: + */ +static struct perf_event_attr detailed_attrs[] = { + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_L1D << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_L1D << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_LL << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_LL << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) }, +}; + +/* + * Very detailed stats (-d -d), covering the instruction cache and the TLB caches: + */ +static struct perf_event_attr very_detailed_attrs[] = { + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_L1I << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_L1I << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_DTLB << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_DTLB << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_ITLB << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_ITLB << 0 | + (PERF_COUNT_HW_CACHE_OP_READ << 8) | + (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) }, + +}; + +/* + * Very, very detailed stats (-d -d -d), adding prefetch events: + */ +static struct perf_event_attr very_very_detailed_attrs[] = { + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_L1D << 0 | + (PERF_COUNT_HW_CACHE_OP_PREFETCH << 8) | + (PERF_COUNT_HW_CACHE_RESULT_ACCESS << 16) }, + + { .type = PERF_TYPE_HW_CACHE, + .config = + PERF_COUNT_HW_CACHE_L1D << 0 | + (PERF_COUNT_HW_CACHE_OP_PREFETCH << 8) | + (PERF_COUNT_HW_CACHE_RESULT_MISS << 16) }, +}; + + + +struct perf_evlist *evsel_list; + +static bool system_wide = false; +static int run_idx = 0; + +static int run_count = 1; +static bool no_inherit = false; +static bool scale = true; +static bool no_aggr = false; +static pid_t target_pid = -1; +static pid_t target_tid = -1; +static pid_t child_pid = -1; +static bool null_run = false; +static int detailed_run = 0; +static bool sync_run = false; +static bool big_num = true; +static int big_num_opt = -1; +static const char *cpu_list; +static const char *csv_sep = NULL; +static bool csv_output = false; + +static volatile int done = 0; + +struct stats +{ + double n, mean, M2; +}; + +struct perf_stat { + struct stats res_stats[3]; +}; + +static int perf_evsel__alloc_stat_priv(struct perf_evsel *evsel) +{ + evsel->priv = zalloc(sizeof(struct perf_stat)); + return evsel->priv == NULL ? -ENOMEM : 0; +} + +static void perf_evsel__free_stat_priv(struct perf_evsel *evsel) +{ + free(evsel->priv); + evsel->priv = NULL; +} + +static void update_stats(struct stats *stats, u64 val) +{ + double delta; + + stats->n++; + delta = val - stats->mean; + stats->mean += delta / stats->n; + stats->M2 += delta*(val - stats->mean); +} + +static double avg_stats(struct stats *stats) +{ + return stats->mean; +} + +/* + * http://en.wikipedia.org/wiki/Algorithms_for_calculating_variance + * + * (\Sum n_i^2) - ((\Sum n_i)^2)/n + * s^2 = ------------------------------- + * n - 1 + * + * http://en.wikipedia.org/wiki/Stddev + * + * The std dev of the mean is related to the std dev by: + * + * s + * s_mean = ------- + * sqrt(n) + * + */ +static double stddev_stats(struct stats *stats) +{ + double variance = stats->M2 / (stats->n - 1); + double variance_mean = variance / stats->n; + + return sqrt(variance_mean); +} + +struct stats runtime_nsecs_stats[MAX_NR_CPUS]; +struct stats runtime_cycles_stats[MAX_NR_CPUS]; +struct stats runtime_stalled_cycles_front_stats[MAX_NR_CPUS]; +struct stats runtime_stalled_cycles_back_stats[MAX_NR_CPUS]; +struct stats runtime_branches_stats[MAX_NR_CPUS]; +struct stats runtime_cacherefs_stats[MAX_NR_CPUS]; +struct stats runtime_l1_dcache_stats[MAX_NR_CPUS]; +struct stats runtime_l1_icache_stats[MAX_NR_CPUS]; +struct stats runtime_ll_cache_stats[MAX_NR_CPUS]; +struct stats runtime_itlb_cache_stats[MAX_NR_CPUS]; +struct stats runtime_dtlb_cache_stats[MAX_NR_CPUS]; +struct stats walltime_nsecs_stats; + +static int create_perf_stat_counter(struct perf_evsel *evsel) +{ + struct perf_event_attr *attr = &evsel->attr; + + if (scale) + attr->read_format = PERF_FORMAT_TOTAL_TIME_ENABLED | + PERF_FORMAT_TOTAL_TIME_RUNNING; + + attr->inherit = !no_inherit; + + if (system_wide) + return perf_evsel__open_per_cpu(evsel, evsel_list->cpus, false); + + if (target_pid == -1 && target_tid == -1) { + attr->disabled = 1; + attr->enable_on_exec = 1; + } + + return perf_evsel__open_per_thread(evsel, evsel_list->threads, false); +} + +/* + * Does the counter have nsecs as a unit? + */ +static inline int nsec_counter(struct perf_evsel *evsel) +{ + if (perf_evsel__match(evsel, SOFTWARE, SW_CPU_CLOCK) || + perf_evsel__match(evsel, SOFTWARE, SW_TASK_CLOCK)) + return 1; + + return 0; +} + +/* + * Update various tracking values we maintain to print + * more semantic information such as miss/hit ratios, + * instruction rates, etc: + */ +static void update_shadow_stats(struct perf_evsel *counter, u64 *count) +{ + if (perf_evsel__match(counter, SOFTWARE, SW_TASK_CLOCK)) + update_stats(&runtime_nsecs_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_CPU_CYCLES)) + update_stats(&runtime_cycles_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_STALLED_CYCLES_FRONTEND)) + update_stats(&runtime_stalled_cycles_front_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_STALLED_CYCLES_BACKEND)) + update_stats(&runtime_stalled_cycles_back_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_BRANCH_INSTRUCTIONS)) + update_stats(&runtime_branches_stats[0], count[0]); + else if (perf_evsel__match(counter, HARDWARE, HW_CACHE_REFERENCES)) + update_stats(&runtime_cacherefs_stats[0], count[0]); + else if (perf_evsel__match(counter, HW_CACHE, HW_CACHE_L1D)) + update_stats(&runtime_l1_dcache_stats[0], count[0]); + else if (perf_evsel__match(counter, HW_CACHE, HW_CACHE_L1I)) + update_stats(&runtime_l1_icache_stats[0], count[0]); + else if (perf_evsel__match(counter, HW_CACHE, HW_CACHE_LL)) + update_stats(&runtime_ll_cache_stats[0], count[0]); + else if (perf_evsel__match(counter, HW_CACHE, HW_CACHE_DTLB)) + update_stats(&runtime_dtlb_cache_stats[0], count[0]); + else if (perf_evsel__match(counter, HW_CACHE, HW_CACHE_ITLB)) + update_stats(&runtime_itlb_cache_stats[0], count[0]); +} + +/* + * Read out the results of a single counter: + * aggregate counts across CPUs in system-wide mode + */ +static int read_counter_aggr(struct perf_evsel *counter) +{ + struct perf_stat *ps = counter->priv; + u64 *count = counter->counts->aggr.values; + int i; + + if (__perf_evsel__read(counter, evsel_list->cpus->nr, + evsel_list->threads->nr, scale) < 0) + return -1; + + for (i = 0; i < 3; i++) + update_stats(&ps->res_stats[i], count[i]); + + if (verbose) { + fprintf(stderr, "%s: %" PRIu64 " %" PRIu64 " %" PRIu64 "\n", + event_name(counter), count[0], count[1], count[2]); + } + + /* + * Save the full runtime - to allow normalization during printout: + */ + update_shadow_stats(counter, count); + + return 0; +} + +/* + * Read out the results of a single counter: + * do not aggregate counts across CPUs in system-wide mode + */ +static int read_counter(struct perf_evsel *counter) +{ + u64 *count; + int cpu; + + for (cpu = 0; cpu < evsel_list->cpus->nr; cpu++) { + if (__perf_evsel__read_on_cpu(counter, cpu, 0, scale) < 0) + return -1; + + count = counter->counts->cpu[cpu].values; + + update_shadow_stats(counter, count); + } + + return 0; +} + +static int run_perf_stat(int argc __used, const char **argv) +{ + unsigned long long t0, t1; + struct perf_evsel *counter; + int status = 0; + int child_ready_pipe[2], go_pipe[2]; + const bool forks = (argc > 0); + char buf; + + if (forks && (pipe(child_ready_pipe) < 0 || pipe(go_pipe) < 0)) { + perror("failed to create pipes"); + exit(1); + } + + if (forks) { + if ((child_pid = fork()) < 0) + perror("failed to fork"); + + if (!child_pid) { + close(child_ready_pipe[0]); + close(go_pipe[1]); + fcntl(go_pipe[0], F_SETFD, FD_CLOEXEC); + + /* + * Do a dummy execvp to get the PLT entry resolved, + * so we avoid the resolver overhead on the real + * execvp call. + */ + execvp("", (char **)argv); + + /* + * Tell the parent we're ready to go + */ + close(child_ready_pipe[1]); + + /* + * Wait until the parent tells us to go. + */ + if (read(go_pipe[0], &buf, 1) == -1) + perror("unable to read pipe"); + + execvp(argv[0], (char **)argv); + + perror(argv[0]); + exit(-1); + } + + if (target_tid == -1 && target_pid == -1 && !system_wide) + evsel_list->threads->map[0] = child_pid; + + /* + * Wait for the child to be ready to exec. + */ + close(child_ready_pipe[1]); + close(go_pipe[0]); + if (read(child_ready_pipe[0], &buf, 1) == -1) + perror("unable to read pipe"); + close(child_ready_pipe[0]); + } + + list_for_each_entry(counter, &evsel_list->entries, node) { + if (create_perf_stat_counter(counter) < 0) { + if (errno == EINVAL || errno == ENOSYS || errno == ENOENT) { + if (verbose) + ui__warning("%s event is not supported by the kernel.\n", + event_name(counter)); + continue; + } + + if (errno == EPERM || errno == EACCES) { + error("You may not have permission to collect %sstats.\n" + "\t Consider tweaking" + " /proc/sys/kernel/perf_event_paranoid or running as root.", + system_wide ? "system-wide " : ""); + } else { + error("open_counter returned with %d (%s). " + "/bin/dmesg may provide additional information.\n", + errno, strerror(errno)); + } + if (child_pid != -1) + kill(child_pid, SIGTERM); + die("Not all events could be opened.\n"); + return -1; + } + } + + if (perf_evlist__set_filters(evsel_list)) { + error("failed to set filter with %d (%s)\n", errno, + strerror(errno)); + return -1; + } + + /* + * Enable counters and exec the command: + */ + t0 = rdclock(); + + if (forks) { + close(go_pipe[1]); + wait(&status); + } else { + while(!done) sleep(1); + } + + t1 = rdclock(); + + update_stats(&walltime_nsecs_stats, t1 - t0); + + if (no_aggr) { + list_for_each_entry(counter, &evsel_list->entries, node) { + read_counter(counter); + perf_evsel__close_fd(counter, evsel_list->cpus->nr, 1); + } + } else { + list_for_each_entry(counter, &evsel_list->entries, node) { + read_counter_aggr(counter); + perf_evsel__close_fd(counter, evsel_list->cpus->nr, + evsel_list->threads->nr); + } + } + + return WEXITSTATUS(status); +} + +static void print_noise_pct(double total, double avg) +{ + double pct = 0.0; + + if (avg) + pct = 100.0*total/avg; + + fprintf(stderr, " ( +-%6.2f%% )", pct); +} + +static void print_noise(struct perf_evsel *evsel, double avg) +{ + struct perf_stat *ps; + + if (run_count == 1) + return; + + ps = evsel->priv; + print_noise_pct(stddev_stats(&ps->res_stats[0]), avg); +} + +static void nsec_printout(int cpu, struct perf_evsel *evsel, double avg) +{ + double msecs = avg / 1e6; + char cpustr[16] = { '\0', }; + const char *fmt = csv_output ? "%s%.6f%s%s" : "%s%18.6f%s%-25s"; + + if (no_aggr) + sprintf(cpustr, "CPU%*d%s", + csv_output ? 0 : -4, + evsel_list->cpus->map[cpu], csv_sep); + + fprintf(stderr, fmt, cpustr, msecs, csv_sep, event_name(evsel)); + + if (evsel->cgrp) + fprintf(stderr, "%s%s", csv_sep, evsel->cgrp->name); + + if (csv_output) + return; + + if (perf_evsel__match(evsel, SOFTWARE, SW_TASK_CLOCK)) + fprintf(stderr, " # %8.3f CPUs utilized ", avg / avg_stats(&walltime_nsecs_stats)); +} + +static void print_stalled_cycles_frontend(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_cycles_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 50.0) + color = PERF_COLOR_RED; + else if (ratio > 30.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 10.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " frontend cycles idle "); +} + +static void print_stalled_cycles_backend(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_cycles_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 75.0) + color = PERF_COLOR_RED; + else if (ratio > 50.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 20.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " backend cycles idle "); +} + +static void print_branch_misses(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_branches_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 20.0) + color = PERF_COLOR_RED; + else if (ratio > 10.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 5.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " of all branches "); +} + +static void print_l1_dcache_misses(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_l1_dcache_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 20.0) + color = PERF_COLOR_RED; + else if (ratio > 10.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 5.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " of all L1-dcache hits "); +} + +static void print_l1_icache_misses(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_l1_icache_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 20.0) + color = PERF_COLOR_RED; + else if (ratio > 10.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 5.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " of all L1-icache hits "); +} + +static void print_dtlb_cache_misses(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_dtlb_cache_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 20.0) + color = PERF_COLOR_RED; + else if (ratio > 10.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 5.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " of all dTLB cache hits "); +} + +static void print_itlb_cache_misses(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_itlb_cache_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 20.0) + color = PERF_COLOR_RED; + else if (ratio > 10.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 5.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " of all iTLB cache hits "); +} + +static void print_ll_cache_misses(int cpu, struct perf_evsel *evsel __used, double avg) +{ + double total, ratio = 0.0; + const char *color; + + total = avg_stats(&runtime_ll_cache_stats[cpu]); + + if (total) + ratio = avg / total * 100.0; + + color = PERF_COLOR_NORMAL; + if (ratio > 20.0) + color = PERF_COLOR_RED; + else if (ratio > 10.0) + color = PERF_COLOR_MAGENTA; + else if (ratio > 5.0) + color = PERF_COLOR_YELLOW; + + fprintf(stderr, " # "); + color_fprintf(stderr, color, "%6.2f%%", ratio); + fprintf(stderr, " of all LL-cache hits "); +} + +static void abs_printout(int cpu, struct perf_evsel *evsel, double avg) +{ + double total, ratio = 0.0; + char cpustr[16] = { '\0', }; + const char *fmt; + + if (csv_output) + fmt = "%s%.0f%s%s"; + else if (big_num) + fmt = "%s%'18.0f%s%-25s"; + else + fmt = "%s%18.0f%s%-25s"; + + if (no_aggr) + sprintf(cpustr, "CPU%*d%s", + csv_output ? 0 : -4, + evsel_list->cpus->map[cpu], csv_sep); + else + cpu = 0; + + fprintf(stderr, fmt, cpustr, avg, csv_sep, event_name(evsel)); + + if (evsel->cgrp) + fprintf(stderr, "%s%s", csv_sep, evsel->cgrp->name); + + if (csv_output) + return; + + if (perf_evsel__match(evsel, HARDWARE, HW_INSTRUCTIONS)) { + total = avg_stats(&runtime_cycles_stats[cpu]); + + if (total) + ratio = avg / total; + + fprintf(stderr, " # %5.2f insns per cycle ", ratio); + + total = avg_stats(&runtime_stalled_cycles_front_stats[cpu]); + total = max(total, avg_stats(&runtime_stalled_cycles_back_stats[cpu])); + + if (total && avg) { + ratio = total / avg; + fprintf(stderr, "\n # %5.2f stalled cycles per insn", ratio); + } + + } else if (perf_evsel__match(evsel, HARDWARE, HW_BRANCH_MISSES) && + runtime_branches_stats[cpu].n != 0) { + print_branch_misses(cpu, evsel, avg); + } else if ( + evsel->attr.type == PERF_TYPE_HW_CACHE && + evsel->attr.config == ( PERF_COUNT_HW_CACHE_L1D | + ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | + ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16)) && + runtime_l1_dcache_stats[cpu].n != 0) { + print_l1_dcache_misses(cpu, evsel, avg); + } else if ( + evsel->attr.type == PERF_TYPE_HW_CACHE && + evsel->attr.config == ( PERF_COUNT_HW_CACHE_L1I | + ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | + ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16)) && + runtime_l1_icache_stats[cpu].n != 0) { + print_l1_icache_misses(cpu, evsel, avg); + } else if ( + evsel->attr.type == PERF_TYPE_HW_CACHE && + evsel->attr.config == ( PERF_COUNT_HW_CACHE_DTLB | + ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | + ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16)) && + runtime_dtlb_cache_stats[cpu].n != 0) { + print_dtlb_cache_misses(cpu, evsel, avg); + } else if ( + evsel->attr.type == PERF_TYPE_HW_CACHE && + evsel->attr.config == ( PERF_COUNT_HW_CACHE_ITLB | + ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | + ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16)) && + runtime_itlb_cache_stats[cpu].n != 0) { + print_itlb_cache_misses(cpu, evsel, avg); + } else if ( + evsel->attr.type == PERF_TYPE_HW_CACHE && + evsel->attr.config == ( PERF_COUNT_HW_CACHE_LL | + ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | + ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16)) && + runtime_ll_cache_stats[cpu].n != 0) { + print_ll_cache_misses(cpu, evsel, avg); + } else if (perf_evsel__match(evsel, HARDWARE, HW_CACHE_MISSES) && + runtime_cacherefs_stats[cpu].n != 0) { + total = avg_stats(&runtime_cacherefs_stats[cpu]); + + if (total) + ratio = avg * 100 / total; + + fprintf(stderr, " # %8.3f %% of all cache refs ", ratio); + + } else if (perf_evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_FRONTEND)) { + print_stalled_cycles_frontend(cpu, evsel, avg); + } else if (perf_evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_BACKEND)) { + print_stalled_cycles_backend(cpu, evsel, avg); + } else if (perf_evsel__match(evsel, HARDWARE, HW_CPU_CYCLES)) { + total = avg_stats(&runtime_nsecs_stats[cpu]); + + if (total) + ratio = 1.0 * avg / total; + + fprintf(stderr, " # %8.3f GHz ", ratio); + } else if (runtime_nsecs_stats[cpu].n != 0) { + total = avg_stats(&runtime_nsecs_stats[cpu]); + + if (total) + ratio = 1000.0 * avg / total; + + fprintf(stderr, " # %8.3f M/sec ", ratio); + } else { + fprintf(stderr, " "); + } +} + +/* + * Print out the results of a single counter: + * aggregated counts in system-wide mode + */ +static void print_counter_aggr(struct perf_evsel *counter) +{ + struct perf_stat *ps = counter->priv; + double avg = avg_stats(&ps->res_stats[0]); + int scaled = counter->counts->scaled; + + if (scaled == -1) { + fprintf(stderr, "%*s%s%*s", + csv_output ? 0 : 18, + "<not counted>", + csv_sep, + csv_output ? 0 : -24, + event_name(counter)); + + if (counter->cgrp) + fprintf(stderr, "%s%s", csv_sep, counter->cgrp->name); + + fputc('\n', stderr); + return; + } + + if (nsec_counter(counter)) + nsec_printout(-1, counter, avg); + else + abs_printout(-1, counter, avg); + + if (csv_output) { + fputc('\n', stderr); + return; + } + + print_noise(counter, avg); + + if (scaled) { + double avg_enabled, avg_running; + + avg_enabled = avg_stats(&ps->res_stats[1]); + avg_running = avg_stats(&ps->res_stats[2]); + + fprintf(stderr, " [%5.2f%%]", 100 * avg_running / avg_enabled); + } + fprintf(stderr, "\n"); +} + +/* + * Print out the results of a single counter: + * does not use aggregated count in system-wide + */ +static void print_counter(struct perf_evsel *counter) +{ + u64 ena, run, val; + int cpu; + + for (cpu = 0; cpu < evsel_list->cpus->nr; cpu++) { + val = counter->counts->cpu[cpu].val; + ena = counter->counts->cpu[cpu].ena; + run = counter->counts->cpu[cpu].run; + if (run == 0 || ena == 0) { + fprintf(stderr, "CPU%*d%s%*s%s%*s", + csv_output ? 0 : -4, + evsel_list->cpus->map[cpu], csv_sep, + csv_output ? 0 : 18, + "<not counted>", csv_sep, + csv_output ? 0 : -24, + event_name(counter)); + + if (counter->cgrp) + fprintf(stderr, "%s%s", csv_sep, counter->cgrp->name); + + fputc('\n', stderr); + continue; + } + + if (nsec_counter(counter)) + nsec_printout(cpu, counter, val); + else + abs_printout(cpu, counter, val); + + if (!csv_output) { + print_noise(counter, 1.0); + + if (run != ena) + fprintf(stderr, " (%.2f%%)", 100.0 * run / ena); + } + fputc('\n', stderr); + } +} + +static void print_stat(int argc, const char **argv) +{ + struct perf_evsel *counter; + int i; + + fflush(stdout); + + if (!csv_output) { + fprintf(stderr, "\n"); + fprintf(stderr, " Performance counter stats for "); + if(target_pid == -1 && target_tid == -1) { + fprintf(stderr, "\'%s", argv[0]); + for (i = 1; i < argc; i++) + fprintf(stderr, " %s", argv[i]); + } else if (target_pid != -1) + fprintf(stderr, "process id \'%d", target_pid); + else + fprintf(stderr, "thread id \'%d", target_tid); + + fprintf(stderr, "\'"); + if (run_count > 1) + fprintf(stderr, " (%d runs)", run_count); + fprintf(stderr, ":\n\n"); + } + + if (no_aggr) { + list_for_each_entry(counter, &evsel_list->entries, node) + print_counter(counter); + } else { + list_for_each_entry(counter, &evsel_list->entries, node) + print_counter_aggr(counter); + } + + if (!csv_output) { + if (!null_run) + fprintf(stderr, "\n"); + fprintf(stderr, " %17.9f seconds time elapsed", + avg_stats(&walltime_nsecs_stats)/1e9); + if (run_count > 1) { + fprintf(stderr, " "); + print_noise_pct(stddev_stats(&walltime_nsecs_stats), + avg_stats(&walltime_nsecs_stats)); + } + fprintf(stderr, "\n\n"); + } +} + +static volatile int signr = -1; + +static void skip_signal(int signo) +{ + if(child_pid == -1) + done = 1; + + signr = signo; +} + +static void sig_atexit(void) +{ + if (child_pid != -1) + kill(child_pid, SIGTERM); + + if (signr == -1) + return; + + signal(signr, SIG_DFL); + kill(getpid(), signr); +} + +static const char * const stat_usage[] = { + "perf stat [<options>] [<command>]", + NULL +}; + +static int stat__set_big_num(const struct option *opt __used, + const char *s __used, int unset) +{ + big_num_opt = unset ? 0 : 1; + return 0; +} + +static const struct option options[] = { + OPT_CALLBACK('e', "event", &evsel_list, "event", + "event selector. use 'perf list' to list available events", + parse_events), + OPT_CALLBACK(0, "filter", &evsel_list, "filter", + "event filter", parse_filter), + OPT_BOOLEAN('i', "no-inherit", &no_inherit, + "child tasks do not inherit counters"), + OPT_INTEGER('p', "pid", &target_pid, + "stat events on existing process id"), + OPT_INTEGER('t', "tid", &target_tid, + "stat events on existing thread id"), + OPT_BOOLEAN('a', "all-cpus", &system_wide, + "system-wide collection from all CPUs"), + OPT_BOOLEAN('c', "scale", &scale, + "scale/normalize counters"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show counter open errors, etc)"), + OPT_INTEGER('r', "repeat", &run_count, + "repeat command and print average + stddev (max: 100)"), + OPT_BOOLEAN('n', "null", &null_run, + "null run - dont start any counters"), + OPT_INCR('d', "detailed", &detailed_run, + "detailed run - start a lot of events"), + OPT_BOOLEAN('S', "sync", &sync_run, + "call sync() before starting a run"), + OPT_CALLBACK_NOOPT('B', "big-num", NULL, NULL, + "print large numbers with thousands\' separators", + stat__set_big_num), + OPT_STRING('C', "cpu", &cpu_list, "cpu", + "list of cpus to monitor in system-wide"), + OPT_BOOLEAN('A', "no-aggr", &no_aggr, + "disable CPU count aggregation"), + OPT_STRING('x', "field-separator", &csv_sep, "separator", + "print counts with custom separator"), + OPT_CALLBACK('G', "cgroup", &evsel_list, "name", + "monitor event in cgroup name only", + parse_cgroups), + OPT_END() +}; + +/* + * Add default attributes, if there were no attributes specified or + * if -d/--detailed, -d -d or -d -d -d is used: + */ +static int add_default_attributes(void) +{ + struct perf_evsel *pos; + size_t attr_nr = 0; + size_t c; + + /* Set attrs if no event is selected and !null_run: */ + if (null_run) + return 0; + + if (!evsel_list->nr_entries) { + for (c = 0; c < ARRAY_SIZE(default_attrs); c++) { + pos = perf_evsel__new(default_attrs + c, c + attr_nr); + if (pos == NULL) + return -1; + perf_evlist__add(evsel_list, pos); + } + attr_nr += c; + } + + /* Detailed events get appended to the event list: */ + + if (detailed_run < 1) + return 0; + + /* Append detailed run extra attributes: */ + for (c = 0; c < ARRAY_SIZE(detailed_attrs); c++) { + pos = perf_evsel__new(detailed_attrs + c, c + attr_nr); + if (pos == NULL) + return -1; + perf_evlist__add(evsel_list, pos); + } + attr_nr += c; + + if (detailed_run < 2) + return 0; + + /* Append very detailed run extra attributes: */ + for (c = 0; c < ARRAY_SIZE(very_detailed_attrs); c++) { + pos = perf_evsel__new(very_detailed_attrs + c, c + attr_nr); + if (pos == NULL) + return -1; + perf_evlist__add(evsel_list, pos); + } + + if (detailed_run < 3) + return 0; + + /* Append very, very detailed run extra attributes: */ + for (c = 0; c < ARRAY_SIZE(very_very_detailed_attrs); c++) { + pos = perf_evsel__new(very_very_detailed_attrs + c, c + attr_nr); + if (pos == NULL) + return -1; + perf_evlist__add(evsel_list, pos); + } + + + return 0; +} + +int cmd_stat(int argc, const char **argv, const char *prefix __used) +{ + struct perf_evsel *pos; + int status = -ENOMEM; + + setlocale(LC_ALL, ""); + + evsel_list = perf_evlist__new(NULL, NULL); + if (evsel_list == NULL) + return -ENOMEM; + + argc = parse_options(argc, argv, options, stat_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + + if (csv_sep) + csv_output = true; + else + csv_sep = DEFAULT_SEPARATOR; + + /* + * let the spreadsheet do the pretty-printing + */ + if (csv_output) { + /* User explicitely passed -B? */ + if (big_num_opt == 1) { + fprintf(stderr, "-B option not supported with -x\n"); + usage_with_options(stat_usage, options); + } else /* Nope, so disable big number formatting */ + big_num = false; + } else if (big_num_opt == 0) /* User passed --no-big-num */ + big_num = false; + + if (!argc && target_pid == -1 && target_tid == -1) + usage_with_options(stat_usage, options); + if (run_count <= 0) + usage_with_options(stat_usage, options); + + /* no_aggr, cgroup are for system-wide only */ + if ((no_aggr || nr_cgroups) && !system_wide) { + fprintf(stderr, "both cgroup and no-aggregation " + "modes only available in system-wide mode\n"); + + usage_with_options(stat_usage, options); + } + + if (add_default_attributes()) + goto out; + + if (target_pid != -1) + target_tid = target_pid; + + evsel_list->threads = thread_map__new(target_pid, target_tid); + if (evsel_list->threads == NULL) { + pr_err("Problems finding threads of monitor\n"); + usage_with_options(stat_usage, options); + } + + if (system_wide) + evsel_list->cpus = cpu_map__new(cpu_list); + else + evsel_list->cpus = cpu_map__dummy_new(); + + if (evsel_list->cpus == NULL) { + perror("failed to parse CPUs map"); + usage_with_options(stat_usage, options); + return -1; + } + + list_for_each_entry(pos, &evsel_list->entries, node) { + if (perf_evsel__alloc_stat_priv(pos) < 0 || + perf_evsel__alloc_counts(pos, evsel_list->cpus->nr) < 0 || + perf_evsel__alloc_fd(pos, evsel_list->cpus->nr, evsel_list->threads->nr) < 0) + goto out_free_fd; + } + + /* + * We dont want to block the signals - that would cause + * child tasks to inherit that and Ctrl-C would not work. + * What we want is for Ctrl-C to work in the exec()-ed + * task, but being ignored by perf stat itself: + */ + atexit(sig_atexit); + signal(SIGINT, skip_signal); + signal(SIGALRM, skip_signal); + signal(SIGABRT, skip_signal); + + status = 0; + for (run_idx = 0; run_idx < run_count; run_idx++) { + if (run_count != 1 && verbose) + fprintf(stderr, "[ perf stat: executing run #%d ... ]\n", run_idx + 1); + + if (sync_run) + sync(); + + status = run_perf_stat(argc, argv); + } + + if (status != -1) + print_stat(argc, argv); +out_free_fd: + list_for_each_entry(pos, &evsel_list->entries, node) + perf_evsel__free_stat_priv(pos); + perf_evlist__delete_maps(evsel_list); +out: + perf_evlist__delete(evsel_list); + return status; +} diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c new file mode 100644 index 00000000..2da91622 --- /dev/null +++ b/tools/perf/builtin-test.c @@ -0,0 +1,674 @@ +/* + * builtin-test.c + * + * Builtin regression testing command: ever growing number of sanity tests + */ +#include "builtin.h" + +#include "util/cache.h" +#include "util/debug.h" +#include "util/evlist.h" +#include "util/parse-options.h" +#include "util/parse-events.h" +#include "util/symbol.h" +#include "util/thread_map.h" + +static long page_size; + +static int vmlinux_matches_kallsyms_filter(struct map *map __used, struct symbol *sym) +{ + bool *visited = symbol__priv(sym); + *visited = true; + return 0; +} + +static int test__vmlinux_matches_kallsyms(void) +{ + int err = -1; + struct rb_node *nd; + struct symbol *sym; + struct map *kallsyms_map, *vmlinux_map; + struct machine kallsyms, vmlinux; + enum map_type type = MAP__FUNCTION; + struct ref_reloc_sym ref_reloc_sym = { .name = "_stext", }; + + /* + * Step 1: + * + * Init the machines that will hold kernel, modules obtained from + * both vmlinux + .ko files and from /proc/kallsyms split by modules. + */ + machine__init(&kallsyms, "", HOST_KERNEL_ID); + machine__init(&vmlinux, "", HOST_KERNEL_ID); + + /* + * Step 2: + * + * Create the kernel maps for kallsyms and the DSO where we will then + * load /proc/kallsyms. Also create the modules maps from /proc/modules + * and find the .ko files that match them in /lib/modules/`uname -r`/. + */ + if (machine__create_kernel_maps(&kallsyms) < 0) { + pr_debug("machine__create_kernel_maps "); + return -1; + } + + /* + * Step 3: + * + * Load and split /proc/kallsyms into multiple maps, one per module. + */ + if (machine__load_kallsyms(&kallsyms, "/proc/kallsyms", type, NULL) <= 0) { + pr_debug("dso__load_kallsyms "); + goto out; + } + + /* + * Step 4: + * + * kallsyms will be internally on demand sorted by name so that we can + * find the reference relocation * symbol, i.e. the symbol we will use + * to see if the running kernel was relocated by checking if it has the + * same value in the vmlinux file we load. + */ + kallsyms_map = machine__kernel_map(&kallsyms, type); + + sym = map__find_symbol_by_name(kallsyms_map, ref_reloc_sym.name, NULL); + if (sym == NULL) { + pr_debug("dso__find_symbol_by_name "); + goto out; + } + + ref_reloc_sym.addr = sym->start; + + /* + * Step 5: + * + * Now repeat step 2, this time for the vmlinux file we'll auto-locate. + */ + if (machine__create_kernel_maps(&vmlinux) < 0) { + pr_debug("machine__create_kernel_maps "); + goto out; + } + + vmlinux_map = machine__kernel_map(&vmlinux, type); + map__kmap(vmlinux_map)->ref_reloc_sym = &ref_reloc_sym; + + /* + * Step 6: + * + * Locate a vmlinux file in the vmlinux path that has a buildid that + * matches the one of the running kernel. + * + * While doing that look if we find the ref reloc symbol, if we find it + * we'll have its ref_reloc_symbol.unrelocated_addr and then + * maps__reloc_vmlinux will notice and set proper ->[un]map_ip routines + * to fixup the symbols. + */ + if (machine__load_vmlinux_path(&vmlinux, type, + vmlinux_matches_kallsyms_filter) <= 0) { + pr_debug("machine__load_vmlinux_path "); + goto out; + } + + err = 0; + /* + * Step 7: + * + * Now look at the symbols in the vmlinux DSO and check if we find all of them + * in the kallsyms dso. For the ones that are in both, check its names and + * end addresses too. + */ + for (nd = rb_first(&vmlinux_map->dso->symbols[type]); nd; nd = rb_next(nd)) { + struct symbol *pair, *first_pair; + bool backwards = true; + + sym = rb_entry(nd, struct symbol, rb_node); + + if (sym->start == sym->end) + continue; + + first_pair = machine__find_kernel_symbol(&kallsyms, type, sym->start, NULL, NULL); + pair = first_pair; + + if (pair && pair->start == sym->start) { +next_pair: + if (strcmp(sym->name, pair->name) == 0) { + /* + * kallsyms don't have the symbol end, so we + * set that by using the next symbol start - 1, + * in some cases we get this up to a page + * wrong, trace_kmalloc when I was developing + * this code was one such example, 2106 bytes + * off the real size. More than that and we + * _really_ have a problem. + */ + s64 skew = sym->end - pair->end; + if (llabs(skew) < page_size) + continue; + + pr_debug("%#" PRIx64 ": diff end addr for %s v: %#" PRIx64 " k: %#" PRIx64 "\n", + sym->start, sym->name, sym->end, pair->end); + } else { + struct rb_node *nnd; +detour: + nnd = backwards ? rb_prev(&pair->rb_node) : + rb_next(&pair->rb_node); + if (nnd) { + struct symbol *next = rb_entry(nnd, struct symbol, rb_node); + + if (next->start == sym->start) { + pair = next; + goto next_pair; + } + } + + if (backwards) { + backwards = false; + pair = first_pair; + goto detour; + } + + pr_debug("%#" PRIx64 ": diff name v: %s k: %s\n", + sym->start, sym->name, pair->name); + } + } else + pr_debug("%#" PRIx64 ": %s not on kallsyms\n", sym->start, sym->name); + + err = -1; + } + + if (!verbose) + goto out; + + pr_info("Maps only in vmlinux:\n"); + + for (nd = rb_first(&vmlinux.kmaps.maps[type]); nd; nd = rb_next(nd)) { + struct map *pos = rb_entry(nd, struct map, rb_node), *pair; + /* + * If it is the kernel, kallsyms is always "[kernel.kallsyms]", while + * the kernel will have the path for the vmlinux file being used, + * so use the short name, less descriptive but the same ("[kernel]" in + * both cases. + */ + pair = map_groups__find_by_name(&kallsyms.kmaps, type, + (pos->dso->kernel ? + pos->dso->short_name : + pos->dso->name)); + if (pair) + pair->priv = 1; + else + map__fprintf(pos, stderr); + } + + pr_info("Maps in vmlinux with a different name in kallsyms:\n"); + + for (nd = rb_first(&vmlinux.kmaps.maps[type]); nd; nd = rb_next(nd)) { + struct map *pos = rb_entry(nd, struct map, rb_node), *pair; + + pair = map_groups__find(&kallsyms.kmaps, type, pos->start); + if (pair == NULL || pair->priv) + continue; + + if (pair->start == pos->start) { + pair->priv = 1; + pr_info(" %" PRIx64 "-%" PRIx64 " %" PRIx64 " %s in kallsyms as", + pos->start, pos->end, pos->pgoff, pos->dso->name); + if (pos->pgoff != pair->pgoff || pos->end != pair->end) + pr_info(": \n*%" PRIx64 "-%" PRIx64 " %" PRIx64 "", + pair->start, pair->end, pair->pgoff); + pr_info(" %s\n", pair->dso->name); + pair->priv = 1; + } + } + + pr_info("Maps only in kallsyms:\n"); + + for (nd = rb_first(&kallsyms.kmaps.maps[type]); + nd; nd = rb_next(nd)) { + struct map *pos = rb_entry(nd, struct map, rb_node); + + if (!pos->priv) + map__fprintf(pos, stderr); + } +out: + return err; +} + +#include "util/cpumap.h" +#include "util/evsel.h" +#include <sys/types.h> + +static int trace_event__id(const char *evname) +{ + char *filename; + int err = -1, fd; + + if (asprintf(&filename, + "/sys/kernel/debug/tracing/events/syscalls/%s/id", + evname) < 0) + return -1; + + fd = open(filename, O_RDONLY); + if (fd >= 0) { + char id[16]; + if (read(fd, id, sizeof(id)) > 0) + err = atoi(id); + close(fd); + } + + free(filename); + return err; +} + +static int test__open_syscall_event(void) +{ + int err = -1, fd; + struct thread_map *threads; + struct perf_evsel *evsel; + struct perf_event_attr attr; + unsigned int nr_open_calls = 111, i; + int id = trace_event__id("sys_enter_open"); + + if (id < 0) { + pr_debug("is debugfs mounted on /sys/kernel/debug?\n"); + return -1; + } + + threads = thread_map__new(-1, getpid()); + if (threads == NULL) { + pr_debug("thread_map__new\n"); + return -1; + } + + memset(&attr, 0, sizeof(attr)); + attr.type = PERF_TYPE_TRACEPOINT; + attr.config = id; + evsel = perf_evsel__new(&attr, 0); + if (evsel == NULL) { + pr_debug("perf_evsel__new\n"); + goto out_thread_map_delete; + } + + if (perf_evsel__open_per_thread(evsel, threads, false) < 0) { + pr_debug("failed to open counter: %s, " + "tweak /proc/sys/kernel/perf_event_paranoid?\n", + strerror(errno)); + goto out_evsel_delete; + } + + for (i = 0; i < nr_open_calls; ++i) { + fd = open("/etc/passwd", O_RDONLY); + close(fd); + } + + if (perf_evsel__read_on_cpu(evsel, 0, 0) < 0) { + pr_debug("perf_evsel__read_on_cpu\n"); + goto out_close_fd; + } + + if (evsel->counts->cpu[0].val != nr_open_calls) { + pr_debug("perf_evsel__read_on_cpu: expected to intercept %d calls, got %" PRIu64 "\n", + nr_open_calls, evsel->counts->cpu[0].val); + goto out_close_fd; + } + + err = 0; +out_close_fd: + perf_evsel__close_fd(evsel, 1, threads->nr); +out_evsel_delete: + perf_evsel__delete(evsel); +out_thread_map_delete: + thread_map__delete(threads); + return err; +} + +#include <sched.h> + +static int test__open_syscall_event_on_all_cpus(void) +{ + int err = -1, fd, cpu; + struct thread_map *threads; + struct cpu_map *cpus; + struct perf_evsel *evsel; + struct perf_event_attr attr; + unsigned int nr_open_calls = 111, i; + cpu_set_t cpu_set; + int id = trace_event__id("sys_enter_open"); + + if (id < 0) { + pr_debug("is debugfs mounted on /sys/kernel/debug?\n"); + return -1; + } + + threads = thread_map__new(-1, getpid()); + if (threads == NULL) { + pr_debug("thread_map__new\n"); + return -1; + } + + cpus = cpu_map__new(NULL); + if (cpus == NULL) { + pr_debug("cpu_map__new\n"); + goto out_thread_map_delete; + } + + + CPU_ZERO(&cpu_set); + + memset(&attr, 0, sizeof(attr)); + attr.type = PERF_TYPE_TRACEPOINT; + attr.config = id; + evsel = perf_evsel__new(&attr, 0); + if (evsel == NULL) { + pr_debug("perf_evsel__new\n"); + goto out_thread_map_delete; + } + + if (perf_evsel__open(evsel, cpus, threads, false) < 0) { + pr_debug("failed to open counter: %s, " + "tweak /proc/sys/kernel/perf_event_paranoid?\n", + strerror(errno)); + goto out_evsel_delete; + } + + for (cpu = 0; cpu < cpus->nr; ++cpu) { + unsigned int ncalls = nr_open_calls + cpu; + /* + * XXX eventually lift this restriction in a way that + * keeps perf building on older glibc installations + * without CPU_ALLOC. 1024 cpus in 2010 still seems + * a reasonable upper limit tho :-) + */ + if (cpus->map[cpu] >= CPU_SETSIZE) { + pr_debug("Ignoring CPU %d\n", cpus->map[cpu]); + continue; + } + + CPU_SET(cpus->map[cpu], &cpu_set); + if (sched_setaffinity(0, sizeof(cpu_set), &cpu_set) < 0) { + pr_debug("sched_setaffinity() failed on CPU %d: %s ", + cpus->map[cpu], + strerror(errno)); + goto out_close_fd; + } + for (i = 0; i < ncalls; ++i) { + fd = open("/etc/passwd", O_RDONLY); + close(fd); + } + CPU_CLR(cpus->map[cpu], &cpu_set); + } + + /* + * Here we need to explicitely preallocate the counts, as if + * we use the auto allocation it will allocate just for 1 cpu, + * as we start by cpu 0. + */ + if (perf_evsel__alloc_counts(evsel, cpus->nr) < 0) { + pr_debug("perf_evsel__alloc_counts(ncpus=%d)\n", cpus->nr); + goto out_close_fd; + } + + err = 0; + + for (cpu = 0; cpu < cpus->nr; ++cpu) { + unsigned int expected; + + if (cpus->map[cpu] >= CPU_SETSIZE) + continue; + + if (perf_evsel__read_on_cpu(evsel, cpu, 0) < 0) { + pr_debug("perf_evsel__read_on_cpu\n"); + err = -1; + break; + } + + expected = nr_open_calls + cpu; + if (evsel->counts->cpu[cpu].val != expected) { + pr_debug("perf_evsel__read_on_cpu: expected to intercept %d calls on cpu %d, got %" PRIu64 "\n", + expected, cpus->map[cpu], evsel->counts->cpu[cpu].val); + err = -1; + } + } + +out_close_fd: + perf_evsel__close_fd(evsel, 1, threads->nr); +out_evsel_delete: + perf_evsel__delete(evsel); +out_thread_map_delete: + thread_map__delete(threads); + return err; +} + +/* + * This test will generate random numbers of calls to some getpid syscalls, + * then establish an mmap for a group of events that are created to monitor + * the syscalls. + * + * It will receive the events, using mmap, use its PERF_SAMPLE_ID generated + * sample.id field to map back to its respective perf_evsel instance. + * + * Then it checks if the number of syscalls reported as perf events by + * the kernel corresponds to the number of syscalls made. + */ +static int test__basic_mmap(void) +{ + int err = -1; + union perf_event *event; + struct thread_map *threads; + struct cpu_map *cpus; + struct perf_evlist *evlist; + struct perf_event_attr attr = { + .type = PERF_TYPE_TRACEPOINT, + .read_format = PERF_FORMAT_ID, + .sample_type = PERF_SAMPLE_ID, + .watermark = 0, + }; + cpu_set_t cpu_set; + const char *syscall_names[] = { "getsid", "getppid", "getpgrp", + "getpgid", }; + pid_t (*syscalls[])(void) = { (void *)getsid, getppid, getpgrp, + (void*)getpgid }; +#define nsyscalls ARRAY_SIZE(syscall_names) + int ids[nsyscalls]; + unsigned int nr_events[nsyscalls], + expected_nr_events[nsyscalls], i, j; + struct perf_evsel *evsels[nsyscalls], *evsel; + int sample_size = __perf_evsel__sample_size(attr.sample_type); + + for (i = 0; i < nsyscalls; ++i) { + char name[64]; + + snprintf(name, sizeof(name), "sys_enter_%s", syscall_names[i]); + ids[i] = trace_event__id(name); + if (ids[i] < 0) { + pr_debug("Is debugfs mounted on /sys/kernel/debug?\n"); + return -1; + } + nr_events[i] = 0; + expected_nr_events[i] = random() % 257; + } + + threads = thread_map__new(-1, getpid()); + if (threads == NULL) { + pr_debug("thread_map__new\n"); + return -1; + } + + cpus = cpu_map__new(NULL); + if (cpus == NULL) { + pr_debug("cpu_map__new\n"); + goto out_free_threads; + } + + CPU_ZERO(&cpu_set); + CPU_SET(cpus->map[0], &cpu_set); + sched_setaffinity(0, sizeof(cpu_set), &cpu_set); + if (sched_setaffinity(0, sizeof(cpu_set), &cpu_set) < 0) { + pr_debug("sched_setaffinity() failed on CPU %d: %s ", + cpus->map[0], strerror(errno)); + goto out_free_cpus; + } + + evlist = perf_evlist__new(cpus, threads); + if (evlist == NULL) { + pr_debug("perf_evlist__new\n"); + goto out_free_cpus; + } + + /* anonymous union fields, can't be initialized above */ + attr.wakeup_events = 1; + attr.sample_period = 1; + + for (i = 0; i < nsyscalls; ++i) { + attr.config = ids[i]; + evsels[i] = perf_evsel__new(&attr, i); + if (evsels[i] == NULL) { + pr_debug("perf_evsel__new\n"); + goto out_free_evlist; + } + + perf_evlist__add(evlist, evsels[i]); + + if (perf_evsel__open(evsels[i], cpus, threads, false) < 0) { + pr_debug("failed to open counter: %s, " + "tweak /proc/sys/kernel/perf_event_paranoid?\n", + strerror(errno)); + goto out_close_fd; + } + } + + if (perf_evlist__mmap(evlist, 128, true) < 0) { + pr_debug("failed to mmap events: %d (%s)\n", errno, + strerror(errno)); + goto out_close_fd; + } + + for (i = 0; i < nsyscalls; ++i) + for (j = 0; j < expected_nr_events[i]; ++j) { + int foo = syscalls[i](); + ++foo; + } + + while ((event = perf_evlist__mmap_read(evlist, 0)) != NULL) { + struct perf_sample sample; + + if (event->header.type != PERF_RECORD_SAMPLE) { + pr_debug("unexpected %s event\n", + perf_event__name(event->header.type)); + goto out_munmap; + } + + err = perf_event__parse_sample(event, attr.sample_type, sample_size, + false, &sample); + if (err) { + pr_err("Can't parse sample, err = %d\n", err); + goto out_munmap; + } + + evsel = perf_evlist__id2evsel(evlist, sample.id); + if (evsel == NULL) { + pr_debug("event with id %" PRIu64 + " doesn't map to an evsel\n", sample.id); + goto out_munmap; + } + nr_events[evsel->idx]++; + } + + list_for_each_entry(evsel, &evlist->entries, node) { + if (nr_events[evsel->idx] != expected_nr_events[evsel->idx]) { + pr_debug("expected %d %s events, got %d\n", + expected_nr_events[evsel->idx], + event_name(evsel), nr_events[evsel->idx]); + goto out_munmap; + } + } + + err = 0; +out_munmap: + perf_evlist__munmap(evlist); +out_close_fd: + for (i = 0; i < nsyscalls; ++i) + perf_evsel__close_fd(evsels[i], 1, threads->nr); +out_free_evlist: + perf_evlist__delete(evlist); +out_free_cpus: + cpu_map__delete(cpus); +out_free_threads: + thread_map__delete(threads); + return err; +#undef nsyscalls +} + +static struct test { + const char *desc; + int (*func)(void); +} tests[] = { + { + .desc = "vmlinux symtab matches kallsyms", + .func = test__vmlinux_matches_kallsyms, + }, + { + .desc = "detect open syscall event", + .func = test__open_syscall_event, + }, + { + .desc = "detect open syscall event on all cpus", + .func = test__open_syscall_event_on_all_cpus, + }, + { + .desc = "read samples using the mmap interface", + .func = test__basic_mmap, + }, + { + .func = NULL, + }, +}; + +static int __cmd_test(void) +{ + int i = 0; + + page_size = sysconf(_SC_PAGE_SIZE); + + while (tests[i].func) { + int err; + pr_info("%2d: %s:", i + 1, tests[i].desc); + pr_debug("\n--- start ---\n"); + err = tests[i].func(); + pr_debug("---- end ----\n%s:", tests[i].desc); + pr_info(" %s\n", err ? "FAILED!\n" : "Ok"); + ++i; + } + + return 0; +} + +static const char * const test_usage[] = { + "perf test [<options>]", + NULL, +}; + +static const struct option test_options[] = { + OPT_INTEGER('v', "verbose", &verbose, + "be more verbose (show symbol address, etc)"), + OPT_END() +}; + +int cmd_test(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, test_options, test_usage, 0); + if (argc) + usage_with_options(test_usage, test_options); + + symbol_conf.priv_size = sizeof(int); + symbol_conf.sort_by_name = true; + symbol_conf.try_vmlinux_path = true; + + if (symbol__init() < 0) + return -1; + + setup_pager(); + + return __cmd_test(); +} diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c new file mode 100644 index 00000000..aa26f4d6 --- /dev/null +++ b/tools/perf/builtin-timechart.c @@ -0,0 +1,1118 @@ +/* + * builtin-timechart.c - make an svg timechart of system activity + * + * (C) Copyright 2009 Intel Corporation + * + * Authors: + * Arjan van de Ven <arjan@linux.intel.com> + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; version 2 + * of the License. + */ + +#include "builtin.h" + +#include "util/util.h" + +#include "util/color.h" +#include <linux/list.h> +#include "util/cache.h" +#include <linux/rbtree.h> +#include "util/symbol.h" +#include "util/callchain.h" +#include "util/strlist.h" + +#include "perf.h" +#include "util/header.h" +#include "util/parse-options.h" +#include "util/parse-events.h" +#include "util/event.h" +#include "util/session.h" +#include "util/svghelper.h" + +#define SUPPORT_OLD_POWER_EVENTS 1 +#define PWR_EVENT_EXIT -1 + + +static char const *input_name = "perf.data"; +static char const *output_name = "output.svg"; + +static unsigned int numcpus; +static u64 min_freq; /* Lowest CPU frequency seen */ +static u64 max_freq; /* Highest CPU frequency seen */ +static u64 turbo_frequency; + +static u64 first_time, last_time; + +static bool power_only; + + +struct per_pid; +struct per_pidcomm; + +struct cpu_sample; +struct power_event; +struct wake_event; + +struct sample_wrapper; + +/* + * Datastructure layout: + * We keep an list of "pid"s, matching the kernels notion of a task struct. + * Each "pid" entry, has a list of "comm"s. + * this is because we want to track different programs different, while + * exec will reuse the original pid (by design). + * Each comm has a list of samples that will be used to draw + * final graph. + */ + +struct per_pid { + struct per_pid *next; + + int pid; + int ppid; + + u64 start_time; + u64 end_time; + u64 total_time; + int display; + + struct per_pidcomm *all; + struct per_pidcomm *current; +}; + + +struct per_pidcomm { + struct per_pidcomm *next; + + u64 start_time; + u64 end_time; + u64 total_time; + + int Y; + int display; + + long state; + u64 state_since; + + char *comm; + + struct cpu_sample *samples; +}; + +struct sample_wrapper { + struct sample_wrapper *next; + + u64 timestamp; + unsigned char data[0]; +}; + +#define TYPE_NONE 0 +#define TYPE_RUNNING 1 +#define TYPE_WAITING 2 +#define TYPE_BLOCKED 3 + +struct cpu_sample { + struct cpu_sample *next; + + u64 start_time; + u64 end_time; + int type; + int cpu; +}; + +static struct per_pid *all_data; + +#define CSTATE 1 +#define PSTATE 2 + +struct power_event { + struct power_event *next; + int type; + int state; + u64 start_time; + u64 end_time; + int cpu; +}; + +struct wake_event { + struct wake_event *next; + int waker; + int wakee; + u64 time; +}; + +static struct power_event *power_events; +static struct wake_event *wake_events; + +struct process_filter; +struct process_filter { + char *name; + int pid; + struct process_filter *next; +}; + +static struct process_filter *process_filter; + + +static struct per_pid *find_create_pid(int pid) +{ + struct per_pid *cursor = all_data; + + while (cursor) { + if (cursor->pid == pid) + return cursor; + cursor = cursor->next; + } + cursor = malloc(sizeof(struct per_pid)); + assert(cursor != NULL); + memset(cursor, 0, sizeof(struct per_pid)); + cursor->pid = pid; + cursor->next = all_data; + all_data = cursor; + return cursor; +} + +static void pid_set_comm(int pid, char *comm) +{ + struct per_pid *p; + struct per_pidcomm *c; + p = find_create_pid(pid); + c = p->all; + while (c) { + if (c->comm && strcmp(c->comm, comm) == 0) { + p->current = c; + return; + } + if (!c->comm) { + c->comm = strdup(comm); + p->current = c; + return; + } + c = c->next; + } + c = malloc(sizeof(struct per_pidcomm)); + assert(c != NULL); + memset(c, 0, sizeof(struct per_pidcomm)); + c->comm = strdup(comm); + p->current = c; + c->next = p->all; + p->all = c; +} + +static void pid_fork(int pid, int ppid, u64 timestamp) +{ + struct per_pid *p, *pp; + p = find_create_pid(pid); + pp = find_create_pid(ppid); + p->ppid = ppid; + if (pp->current && pp->current->comm && !p->current) + pid_set_comm(pid, pp->current->comm); + + p->start_time = timestamp; + if (p->current) { + p->current->start_time = timestamp; + p->current->state_since = timestamp; + } +} + +static void pid_exit(int pid, u64 timestamp) +{ + struct per_pid *p; + p = find_create_pid(pid); + p->end_time = timestamp; + if (p->current) + p->current->end_time = timestamp; +} + +static void +pid_put_sample(int pid, int type, unsigned int cpu, u64 start, u64 end) +{ + struct per_pid *p; + struct per_pidcomm *c; + struct cpu_sample *sample; + + p = find_create_pid(pid); + c = p->current; + if (!c) { + c = malloc(sizeof(struct per_pidcomm)); + assert(c != NULL); + memset(c, 0, sizeof(struct per_pidcomm)); + p->current = c; + c->next = p->all; + p->all = c; + } + + sample = malloc(sizeof(struct cpu_sample)); + assert(sample != NULL); + memset(sample, 0, sizeof(struct cpu_sample)); + sample->start_time = start; + sample->end_time = end; + sample->type = type; + sample->next = c->samples; + sample->cpu = cpu; + c->samples = sample; + + if (sample->type == TYPE_RUNNING && end > start && start > 0) { + c->total_time += (end-start); + p->total_time += (end-start); + } + + if (c->start_time == 0 || c->start_time > start) + c->start_time = start; + if (p->start_time == 0 || p->start_time > start) + p->start_time = start; +} + +#define MAX_CPUS 4096 + +static u64 cpus_cstate_start_times[MAX_CPUS]; +static int cpus_cstate_state[MAX_CPUS]; +static u64 cpus_pstate_start_times[MAX_CPUS]; +static u64 cpus_pstate_state[MAX_CPUS]; + +static int process_comm_event(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session __used) +{ + pid_set_comm(event->comm.tid, event->comm.comm); + return 0; +} + +static int process_fork_event(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session __used) +{ + pid_fork(event->fork.pid, event->fork.ppid, event->fork.time); + return 0; +} + +static int process_exit_event(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session __used) +{ + pid_exit(event->fork.pid, event->fork.time); + return 0; +} + +struct trace_entry { + unsigned short type; + unsigned char flags; + unsigned char preempt_count; + int pid; + int lock_depth; +}; + +#ifdef SUPPORT_OLD_POWER_EVENTS +static int use_old_power_events; +struct power_entry_old { + struct trace_entry te; + u64 type; + u64 value; + u64 cpu_id; +}; +#endif + +struct power_processor_entry { + struct trace_entry te; + u32 state; + u32 cpu_id; +}; + +#define TASK_COMM_LEN 16 +struct wakeup_entry { + struct trace_entry te; + char comm[TASK_COMM_LEN]; + int pid; + int prio; + int success; +}; + +/* + * trace_flag_type is an enumeration that holds different + * states when a trace occurs. These are: + * IRQS_OFF - interrupts were disabled + * IRQS_NOSUPPORT - arch does not support irqs_disabled_flags + * NEED_RESCED - reschedule is requested + * HARDIRQ - inside an interrupt handler + * SOFTIRQ - inside a softirq handler + */ +enum trace_flag_type { + TRACE_FLAG_IRQS_OFF = 0x01, + TRACE_FLAG_IRQS_NOSUPPORT = 0x02, + TRACE_FLAG_NEED_RESCHED = 0x04, + TRACE_FLAG_HARDIRQ = 0x08, + TRACE_FLAG_SOFTIRQ = 0x10, +}; + + + +struct sched_switch { + struct trace_entry te; + char prev_comm[TASK_COMM_LEN]; + int prev_pid; + int prev_prio; + long prev_state; /* Arjan weeps. */ + char next_comm[TASK_COMM_LEN]; + int next_pid; + int next_prio; +}; + +static void c_state_start(int cpu, u64 timestamp, int state) +{ + cpus_cstate_start_times[cpu] = timestamp; + cpus_cstate_state[cpu] = state; +} + +static void c_state_end(int cpu, u64 timestamp) +{ + struct power_event *pwr; + pwr = malloc(sizeof(struct power_event)); + if (!pwr) + return; + memset(pwr, 0, sizeof(struct power_event)); + + pwr->state = cpus_cstate_state[cpu]; + pwr->start_time = cpus_cstate_start_times[cpu]; + pwr->end_time = timestamp; + pwr->cpu = cpu; + pwr->type = CSTATE; + pwr->next = power_events; + + power_events = pwr; +} + +static void p_state_change(int cpu, u64 timestamp, u64 new_freq) +{ + struct power_event *pwr; + pwr = malloc(sizeof(struct power_event)); + + if (new_freq > 8000000) /* detect invalid data */ + return; + + if (!pwr) + return; + memset(pwr, 0, sizeof(struct power_event)); + + pwr->state = cpus_pstate_state[cpu]; + pwr->start_time = cpus_pstate_start_times[cpu]; + pwr->end_time = timestamp; + pwr->cpu = cpu; + pwr->type = PSTATE; + pwr->next = power_events; + + if (!pwr->start_time) + pwr->start_time = first_time; + + power_events = pwr; + + cpus_pstate_state[cpu] = new_freq; + cpus_pstate_start_times[cpu] = timestamp; + + if ((u64)new_freq > max_freq) + max_freq = new_freq; + + if (new_freq < min_freq || min_freq == 0) + min_freq = new_freq; + + if (new_freq == max_freq - 1000) + turbo_frequency = max_freq; +} + +static void +sched_wakeup(int cpu, u64 timestamp, int pid, struct trace_entry *te) +{ + struct wake_event *we; + struct per_pid *p; + struct wakeup_entry *wake = (void *)te; + + we = malloc(sizeof(struct wake_event)); + if (!we) + return; + + memset(we, 0, sizeof(struct wake_event)); + we->time = timestamp; + we->waker = pid; + + if ((te->flags & TRACE_FLAG_HARDIRQ) || (te->flags & TRACE_FLAG_SOFTIRQ)) + we->waker = -1; + + we->wakee = wake->pid; + we->next = wake_events; + wake_events = we; + p = find_create_pid(we->wakee); + + if (p && p->current && p->current->state == TYPE_NONE) { + p->current->state_since = timestamp; + p->current->state = TYPE_WAITING; + } + if (p && p->current && p->current->state == TYPE_BLOCKED) { + pid_put_sample(p->pid, p->current->state, cpu, p->current->state_since, timestamp); + p->current->state_since = timestamp; + p->current->state = TYPE_WAITING; + } +} + +static void sched_switch(int cpu, u64 timestamp, struct trace_entry *te) +{ + struct per_pid *p = NULL, *prev_p; + struct sched_switch *sw = (void *)te; + + + prev_p = find_create_pid(sw->prev_pid); + + p = find_create_pid(sw->next_pid); + + if (prev_p->current && prev_p->current->state != TYPE_NONE) + pid_put_sample(sw->prev_pid, TYPE_RUNNING, cpu, prev_p->current->state_since, timestamp); + if (p && p->current) { + if (p->current->state != TYPE_NONE) + pid_put_sample(sw->next_pid, p->current->state, cpu, p->current->state_since, timestamp); + + p->current->state_since = timestamp; + p->current->state = TYPE_RUNNING; + } + + if (prev_p->current) { + prev_p->current->state = TYPE_NONE; + prev_p->current->state_since = timestamp; + if (sw->prev_state & 2) + prev_p->current->state = TYPE_BLOCKED; + if (sw->prev_state == 0) + prev_p->current->state = TYPE_WAITING; + } +} + + +static int process_sample_event(union perf_event *event __used, + struct perf_sample *sample, + struct perf_evsel *evsel __used, + struct perf_session *session) +{ + struct trace_entry *te; + + if (session->sample_type & PERF_SAMPLE_TIME) { + if (!first_time || first_time > sample->time) + first_time = sample->time; + if (last_time < sample->time) + last_time = sample->time; + } + + te = (void *)sample->raw_data; + if (session->sample_type & PERF_SAMPLE_RAW && sample->raw_size > 0) { + char *event_str; +#ifdef SUPPORT_OLD_POWER_EVENTS + struct power_entry_old *peo; + peo = (void *)te; +#endif + /* + * FIXME: use evsel, its already mapped from id to perf_evsel, + * remove perf_header__find_event infrastructure bits. + * Mapping all these "power:cpu_idle" strings to the tracepoint + * ID and then just comparing against evsel->attr.config. + * + * e.g.: + * + * if (evsel->attr.config == power_cpu_idle_id) + */ + event_str = perf_header__find_event(te->type); + + if (!event_str) + return 0; + + if (sample->cpu > numcpus) + numcpus = sample->cpu; + + if (strcmp(event_str, "power:cpu_idle") == 0) { + struct power_processor_entry *ppe = (void *)te; + if (ppe->state == (u32)PWR_EVENT_EXIT) + c_state_end(ppe->cpu_id, sample->time); + else + c_state_start(ppe->cpu_id, sample->time, + ppe->state); + } + else if (strcmp(event_str, "power:cpu_frequency") == 0) { + struct power_processor_entry *ppe = (void *)te; + p_state_change(ppe->cpu_id, sample->time, ppe->state); + } + + else if (strcmp(event_str, "sched:sched_wakeup") == 0) + sched_wakeup(sample->cpu, sample->time, sample->pid, te); + + else if (strcmp(event_str, "sched:sched_switch") == 0) + sched_switch(sample->cpu, sample->time, te); + +#ifdef SUPPORT_OLD_POWER_EVENTS + if (use_old_power_events) { + if (strcmp(event_str, "power:power_start") == 0) + c_state_start(peo->cpu_id, sample->time, + peo->value); + + else if (strcmp(event_str, "power:power_end") == 0) + c_state_end(sample->cpu, sample->time); + + else if (strcmp(event_str, + "power:power_frequency") == 0) + p_state_change(peo->cpu_id, sample->time, + peo->value); + } +#endif + } + return 0; +} + +/* + * After the last sample we need to wrap up the current C/P state + * and close out each CPU for these. + */ +static void end_sample_processing(void) +{ + u64 cpu; + struct power_event *pwr; + + for (cpu = 0; cpu <= numcpus; cpu++) { + pwr = malloc(sizeof(struct power_event)); + if (!pwr) + return; + memset(pwr, 0, sizeof(struct power_event)); + + /* C state */ +#if 0 + pwr->state = cpus_cstate_state[cpu]; + pwr->start_time = cpus_cstate_start_times[cpu]; + pwr->end_time = last_time; + pwr->cpu = cpu; + pwr->type = CSTATE; + pwr->next = power_events; + + power_events = pwr; +#endif + /* P state */ + + pwr = malloc(sizeof(struct power_event)); + if (!pwr) + return; + memset(pwr, 0, sizeof(struct power_event)); + + pwr->state = cpus_pstate_state[cpu]; + pwr->start_time = cpus_pstate_start_times[cpu]; + pwr->end_time = last_time; + pwr->cpu = cpu; + pwr->type = PSTATE; + pwr->next = power_events; + + if (!pwr->start_time) + pwr->start_time = first_time; + if (!pwr->state) + pwr->state = min_freq; + power_events = pwr; + } +} + +/* + * Sort the pid datastructure + */ +static void sort_pids(void) +{ + struct per_pid *new_list, *p, *cursor, *prev; + /* sort by ppid first, then by pid, lowest to highest */ + + new_list = NULL; + + while (all_data) { + p = all_data; + all_data = p->next; + p->next = NULL; + + if (new_list == NULL) { + new_list = p; + p->next = NULL; + continue; + } + prev = NULL; + cursor = new_list; + while (cursor) { + if (cursor->ppid > p->ppid || + (cursor->ppid == p->ppid && cursor->pid > p->pid)) { + /* must insert before */ + if (prev) { + p->next = prev->next; + prev->next = p; + cursor = NULL; + continue; + } else { + p->next = new_list; + new_list = p; + cursor = NULL; + continue; + } + } + + prev = cursor; + cursor = cursor->next; + if (!cursor) + prev->next = p; + } + } + all_data = new_list; +} + + +static void draw_c_p_states(void) +{ + struct power_event *pwr; + pwr = power_events; + + /* + * two pass drawing so that the P state bars are on top of the C state blocks + */ + while (pwr) { + if (pwr->type == CSTATE) + svg_cstate(pwr->cpu, pwr->start_time, pwr->end_time, pwr->state); + pwr = pwr->next; + } + + pwr = power_events; + while (pwr) { + if (pwr->type == PSTATE) { + if (!pwr->state) + pwr->state = min_freq; + svg_pstate(pwr->cpu, pwr->start_time, pwr->end_time, pwr->state); + } + pwr = pwr->next; + } +} + +static void draw_wakeups(void) +{ + struct wake_event *we; + struct per_pid *p; + struct per_pidcomm *c; + + we = wake_events; + while (we) { + int from = 0, to = 0; + char *task_from = NULL, *task_to = NULL; + + /* locate the column of the waker and wakee */ + p = all_data; + while (p) { + if (p->pid == we->waker || p->pid == we->wakee) { + c = p->all; + while (c) { + if (c->Y && c->start_time <= we->time && c->end_time >= we->time) { + if (p->pid == we->waker && !from) { + from = c->Y; + task_from = strdup(c->comm); + } + if (p->pid == we->wakee && !to) { + to = c->Y; + task_to = strdup(c->comm); + } + } + c = c->next; + } + c = p->all; + while (c) { + if (p->pid == we->waker && !from) { + from = c->Y; + task_from = strdup(c->comm); + } + if (p->pid == we->wakee && !to) { + to = c->Y; + task_to = strdup(c->comm); + } + c = c->next; + } + } + p = p->next; + } + + if (!task_from) { + task_from = malloc(40); + sprintf(task_from, "[%i]", we->waker); + } + if (!task_to) { + task_to = malloc(40); + sprintf(task_to, "[%i]", we->wakee); + } + + if (we->waker == -1) + svg_interrupt(we->time, to); + else if (from && to && abs(from - to) == 1) + svg_wakeline(we->time, from, to); + else + svg_partial_wakeline(we->time, from, task_from, to, task_to); + we = we->next; + + free(task_from); + free(task_to); + } +} + +static void draw_cpu_usage(void) +{ + struct per_pid *p; + struct per_pidcomm *c; + struct cpu_sample *sample; + p = all_data; + while (p) { + c = p->all; + while (c) { + sample = c->samples; + while (sample) { + if (sample->type == TYPE_RUNNING) + svg_process(sample->cpu, sample->start_time, sample->end_time, "sample", c->comm); + + sample = sample->next; + } + c = c->next; + } + p = p->next; + } +} + +static void draw_process_bars(void) +{ + struct per_pid *p; + struct per_pidcomm *c; + struct cpu_sample *sample; + int Y = 0; + + Y = 2 * numcpus + 2; + + p = all_data; + while (p) { + c = p->all; + while (c) { + if (!c->display) { + c->Y = 0; + c = c->next; + continue; + } + + svg_box(Y, c->start_time, c->end_time, "process"); + sample = c->samples; + while (sample) { + if (sample->type == TYPE_RUNNING) + svg_sample(Y, sample->cpu, sample->start_time, sample->end_time); + if (sample->type == TYPE_BLOCKED) + svg_box(Y, sample->start_time, sample->end_time, "blocked"); + if (sample->type == TYPE_WAITING) + svg_waiting(Y, sample->start_time, sample->end_time); + sample = sample->next; + } + + if (c->comm) { + char comm[256]; + if (c->total_time > 5000000000) /* 5 seconds */ + sprintf(comm, "%s:%i (%2.2fs)", c->comm, p->pid, c->total_time / 1000000000.0); + else + sprintf(comm, "%s:%i (%3.1fms)", c->comm, p->pid, c->total_time / 1000000.0); + + svg_text(Y, c->start_time, comm); + } + c->Y = Y; + Y++; + c = c->next; + } + p = p->next; + } +} + +static void add_process_filter(const char *string) +{ + struct process_filter *filt; + int pid; + + pid = strtoull(string, NULL, 10); + filt = malloc(sizeof(struct process_filter)); + if (!filt) + return; + + filt->name = strdup(string); + filt->pid = pid; + filt->next = process_filter; + + process_filter = filt; +} + +static int passes_filter(struct per_pid *p, struct per_pidcomm *c) +{ + struct process_filter *filt; + if (!process_filter) + return 1; + + filt = process_filter; + while (filt) { + if (filt->pid && p->pid == filt->pid) + return 1; + if (strcmp(filt->name, c->comm) == 0) + return 1; + filt = filt->next; + } + return 0; +} + +static int determine_display_tasks_filtered(void) +{ + struct per_pid *p; + struct per_pidcomm *c; + int count = 0; + + p = all_data; + while (p) { + p->display = 0; + if (p->start_time == 1) + p->start_time = first_time; + + /* no exit marker, task kept running to the end */ + if (p->end_time == 0) + p->end_time = last_time; + + c = p->all; + + while (c) { + c->display = 0; + + if (c->start_time == 1) + c->start_time = first_time; + + if (passes_filter(p, c)) { + c->display = 1; + p->display = 1; + count++; + } + + if (c->end_time == 0) + c->end_time = last_time; + + c = c->next; + } + p = p->next; + } + return count; +} + +static int determine_display_tasks(u64 threshold) +{ + struct per_pid *p; + struct per_pidcomm *c; + int count = 0; + + if (process_filter) + return determine_display_tasks_filtered(); + + p = all_data; + while (p) { + p->display = 0; + if (p->start_time == 1) + p->start_time = first_time; + + /* no exit marker, task kept running to the end */ + if (p->end_time == 0) + p->end_time = last_time; + if (p->total_time >= threshold && !power_only) + p->display = 1; + + c = p->all; + + while (c) { + c->display = 0; + + if (c->start_time == 1) + c->start_time = first_time; + + if (c->total_time >= threshold && !power_only) { + c->display = 1; + count++; + } + + if (c->end_time == 0) + c->end_time = last_time; + + c = c->next; + } + p = p->next; + } + return count; +} + + + +#define TIME_THRESH 10000000 + +static void write_svg_file(const char *filename) +{ + u64 i; + int count; + + numcpus++; + + + count = determine_display_tasks(TIME_THRESH); + + /* We'd like to show at least 15 tasks; be less picky if we have fewer */ + if (count < 15) + count = determine_display_tasks(TIME_THRESH / 10); + + open_svg(filename, numcpus, count, first_time, last_time); + + svg_time_grid(); + svg_legenda(); + + for (i = 0; i < numcpus; i++) + svg_cpu_box(i, max_freq, turbo_frequency); + + draw_cpu_usage(); + draw_process_bars(); + draw_c_p_states(); + draw_wakeups(); + + svg_close(); +} + +static struct perf_event_ops event_ops = { + .comm = process_comm_event, + .fork = process_fork_event, + .exit = process_exit_event, + .sample = process_sample_event, + .ordered_samples = true, +}; + +static int __cmd_timechart(void) +{ + struct perf_session *session = perf_session__new(input_name, O_RDONLY, + 0, false, &event_ops); + int ret = -EINVAL; + + if (session == NULL) + return -ENOMEM; + + if (!perf_session__has_traces(session, "timechart record")) + goto out_delete; + + ret = perf_session__process_events(session, &event_ops); + if (ret) + goto out_delete; + + end_sample_processing(); + + sort_pids(); + + write_svg_file(output_name); + + pr_info("Written %2.1f seconds of trace to %s.\n", + (last_time - first_time) / 1000000000.0, output_name); +out_delete: + perf_session__delete(session); + return ret; +} + +static const char * const timechart_usage[] = { + "perf timechart [<options>] {record}", + NULL +}; + +#ifdef SUPPORT_OLD_POWER_EVENTS +static const char * const record_old_args[] = { + "record", + "-a", + "-R", + "-f", + "-c", "1", + "-e", "power:power_start", + "-e", "power:power_end", + "-e", "power:power_frequency", + "-e", "sched:sched_wakeup", + "-e", "sched:sched_switch", +}; +#endif + +static const char * const record_new_args[] = { + "record", + "-a", + "-R", + "-f", + "-c", "1", + "-e", "power:cpu_frequency", + "-e", "power:cpu_idle", + "-e", "sched:sched_wakeup", + "-e", "sched:sched_switch", +}; + +static int __cmd_record(int argc, const char **argv) +{ + unsigned int rec_argc, i, j; + const char **rec_argv; + const char * const *record_args = record_new_args; + unsigned int record_elems = ARRAY_SIZE(record_new_args); + +#ifdef SUPPORT_OLD_POWER_EVENTS + if (!is_valid_tracepoint("power:cpu_idle") && + is_valid_tracepoint("power:power_start")) { + use_old_power_events = 1; + record_args = record_old_args; + record_elems = ARRAY_SIZE(record_old_args); + } +#endif + + rec_argc = record_elems + argc - 1; + rec_argv = calloc(rec_argc + 1, sizeof(char *)); + + if (rec_argv == NULL) + return -ENOMEM; + + for (i = 0; i < record_elems; i++) + rec_argv[i] = strdup(record_args[i]); + + for (j = 1; j < (unsigned int)argc; j++, i++) + rec_argv[i] = argv[j]; + + return cmd_record(i, rec_argv, NULL); +} + +static int +parse_process(const struct option *opt __used, const char *arg, int __used unset) +{ + if (arg) + add_process_filter(arg); + return 0; +} + +static const struct option options[] = { + OPT_STRING('i', "input", &input_name, "file", + "input file name"), + OPT_STRING('o', "output", &output_name, "file", + "output file name"), + OPT_INTEGER('w', "width", &svg_page_width, + "page width"), + OPT_BOOLEAN('P', "power-only", &power_only, + "output power data only"), + OPT_CALLBACK('p', "process", NULL, "process", + "process selector. Pass a pid or process name.", + parse_process), + OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory", + "Look for files with symbols relative to this directory"), + OPT_END() +}; + + +int cmd_timechart(int argc, const char **argv, const char *prefix __used) +{ + argc = parse_options(argc, argv, options, timechart_usage, + PARSE_OPT_STOP_AT_NON_OPTION); + + symbol__init(); + + if (argc && !strncmp(argv[0], "rec", 3)) + return __cmd_record(argc, argv); + else if (argc) + usage_with_options(timechart_usage, options); + + setup_pager(); + + return __cmd_timechart(); +} diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c new file mode 100644 index 00000000..f2f3f493 --- /dev/null +++ b/tools/perf/builtin-top.c @@ -0,0 +1,1139 @@ +/* + * builtin-top.c + * + * Builtin top command: Display a continuously updated profile of + * any workload, CPU or specific PID. + * + * Copyright (C) 2008, Red Hat Inc, Ingo Molnar <mingo@redhat.com> + * + * Improvements and fixes by: + * + * Arjan van de Ven <arjan@linux.intel.com> + * Yanmin Zhang <yanmin.zhang@intel.com> + * Wu Fengguang <fengguang.wu@intel.com> + * Mike Galbraith <efault@gmx.de> + * Paul Mackerras <paulus@samba.org> + * + * Released under the GPL v2. (and only v2, not any later version) + */ +#include "builtin.h" + +#include "perf.h" + +#include "util/annotate.h" +#include "util/cache.h" +#include "util/color.h" +#include "util/evlist.h" +#include "util/evsel.h" +#include "util/session.h" +#include "util/symbol.h" +#include "util/thread.h" +#include "util/thread_map.h" +#include "util/top.h" +#include "util/util.h" +#include <linux/rbtree.h> +#include "util/parse-options.h" +#include "util/parse-events.h" +#include "util/cpumap.h" +#include "util/xyarray.h" + +#include "util/debug.h" + +#include <assert.h> +#include <fcntl.h> + +#include <stdio.h> +#include <termios.h> +#include <unistd.h> +#include <inttypes.h> + +#include <errno.h> +#include <time.h> +#include <sched.h> + +#include <sys/syscall.h> +#include <sys/ioctl.h> +#include <sys/poll.h> +#include <sys/prctl.h> +#include <sys/wait.h> +#include <sys/uio.h> +#include <sys/mman.h> + +#include <linux/unistd.h> +#include <linux/types.h> + +static struct perf_top top = { + .count_filter = 5, + .delay_secs = 2, + .display_weighted = -1, + .target_pid = -1, + .target_tid = -1, + .active_symbols = LIST_HEAD_INIT(top.active_symbols), + .active_symbols_lock = PTHREAD_MUTEX_INITIALIZER, + .active_symbols_cond = PTHREAD_COND_INITIALIZER, + .freq = 1000, /* 1 KHz */ +}; + +static bool system_wide = false; + +static bool use_tui, use_stdio; + +static int default_interval = 0; + +static bool kptr_restrict_warned; +static bool vmlinux_warned; +static bool inherit = false; +static int realtime_prio = 0; +static bool group = false; +static unsigned int page_size; +static unsigned int mmap_pages = 128; + +static bool dump_symtab = false; + +static struct winsize winsize; + +static const char *sym_filter = NULL; +struct sym_entry *sym_filter_entry_sched = NULL; +static int sym_pcnt_filter = 5; + +/* + * Source functions + */ + +void get_term_dimensions(struct winsize *ws) +{ + char *s = getenv("LINES"); + + if (s != NULL) { + ws->ws_row = atoi(s); + s = getenv("COLUMNS"); + if (s != NULL) { + ws->ws_col = atoi(s); + if (ws->ws_row && ws->ws_col) + return; + } + } +#ifdef TIOCGWINSZ + if (ioctl(1, TIOCGWINSZ, ws) == 0 && + ws->ws_row && ws->ws_col) + return; +#endif + ws->ws_row = 25; + ws->ws_col = 80; +} + +static void update_print_entries(struct winsize *ws) +{ + top.print_entries = ws->ws_row; + + if (top.print_entries > 9) + top.print_entries -= 9; +} + +static void sig_winch_handler(int sig __used) +{ + get_term_dimensions(&winsize); + update_print_entries(&winsize); +} + +static int parse_source(struct sym_entry *syme) +{ + struct symbol *sym; + struct annotation *notes; + struct map *map; + int err = -1; + + if (!syme) + return -1; + + sym = sym_entry__symbol(syme); + map = syme->map; + + /* + * We can't annotate with just /proc/kallsyms + */ + if (map->dso->symtab_type == SYMTAB__KALLSYMS) { + pr_err("Can't annotate %s: No vmlinux file was found in the " + "path\n", sym->name); + sleep(1); + return -1; + } + + notes = symbol__annotation(sym); + if (notes->src != NULL) { + pthread_mutex_lock(¬es->lock); + goto out_assign; + } + + pthread_mutex_lock(¬es->lock); + + if (symbol__alloc_hist(sym, top.evlist->nr_entries) < 0) { + pthread_mutex_unlock(¬es->lock); + pr_err("Not enough memory for annotating '%s' symbol!\n", + sym->name); + sleep(1); + return err; + } + + err = symbol__annotate(sym, syme->map, 0); + if (err == 0) { +out_assign: + top.sym_filter_entry = syme; + } + + pthread_mutex_unlock(¬es->lock); + return err; +} + +static void __zero_source_counters(struct sym_entry *syme) +{ + struct symbol *sym = sym_entry__symbol(syme); + symbol__annotate_zero_histograms(sym); +} + +static void record_precise_ip(struct sym_entry *syme, int counter, u64 ip) +{ + struct annotation *notes; + struct symbol *sym; + + if (syme != top.sym_filter_entry) + return; + + sym = sym_entry__symbol(syme); + notes = symbol__annotation(sym); + + if (pthread_mutex_trylock(¬es->lock)) + return; + + ip = syme->map->map_ip(syme->map, ip); + symbol__inc_addr_samples(sym, syme->map, counter, ip); + + pthread_mutex_unlock(¬es->lock); +} + +static void show_details(struct sym_entry *syme) +{ + struct annotation *notes; + struct symbol *symbol; + int more; + + if (!syme) + return; + + symbol = sym_entry__symbol(syme); + notes = symbol__annotation(symbol); + + pthread_mutex_lock(¬es->lock); + + if (notes->src == NULL) + goto out_unlock; + + printf("Showing %s for %s\n", event_name(top.sym_evsel), symbol->name); + printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter); + + more = symbol__annotate_printf(symbol, syme->map, top.sym_evsel->idx, + 0, sym_pcnt_filter, top.print_entries, 4); + if (top.zero) + symbol__annotate_zero_histogram(symbol, top.sym_evsel->idx); + else + symbol__annotate_decay_histogram(symbol, top.sym_evsel->idx); + if (more != 0) + printf("%d lines not displayed, maybe increase display entries [e]\n", more); +out_unlock: + pthread_mutex_unlock(¬es->lock); +} + +static const char CONSOLE_CLEAR[] = "[H[2J"; + +static void __list_insert_active_sym(struct sym_entry *syme) +{ + list_add(&syme->node, &top.active_symbols); +} + +static void print_sym_table(struct perf_session *session) +{ + char bf[160]; + int printed = 0; + struct rb_node *nd; + struct sym_entry *syme; + struct rb_root tmp = RB_ROOT; + const int win_width = winsize.ws_col - 1; + int sym_width, dso_width, dso_short_width; + float sum_ksamples = perf_top__decay_samples(&top, &tmp); + + puts(CONSOLE_CLEAR); + + perf_top__header_snprintf(&top, bf, sizeof(bf)); + printf("%s\n", bf); + + perf_top__reset_sample_counters(&top); + + printf("%-*.*s\n", win_width, win_width, graph_dotted_line); + + if (session->hists.stats.total_lost != 0) { + color_fprintf(stdout, PERF_COLOR_RED, "WARNING:"); + printf(" LOST %" PRIu64 " events, Check IO/CPU overload\n", + session->hists.stats.total_lost); + } + + if (top.sym_filter_entry) { + show_details(top.sym_filter_entry); + return; + } + + perf_top__find_widths(&top, &tmp, &dso_width, &dso_short_width, + &sym_width); + + if (sym_width + dso_width > winsize.ws_col - 29) { + dso_width = dso_short_width; + if (sym_width + dso_width > winsize.ws_col - 29) + sym_width = winsize.ws_col - dso_width - 29; + } + putchar('\n'); + if (top.evlist->nr_entries == 1) + printf(" samples pcnt"); + else + printf(" weight samples pcnt"); + + if (verbose) + printf(" RIP "); + printf(" %-*.*s DSO\n", sym_width, sym_width, "function"); + printf(" %s _______ _____", + top.evlist->nr_entries == 1 ? " " : "______"); + if (verbose) + printf(" ________________"); + printf(" %-*.*s", sym_width, sym_width, graph_line); + printf(" %-*.*s", dso_width, dso_width, graph_line); + puts("\n"); + + for (nd = rb_first(&tmp); nd; nd = rb_next(nd)) { + struct symbol *sym; + double pcnt; + + syme = rb_entry(nd, struct sym_entry, rb_node); + sym = sym_entry__symbol(syme); + if (++printed > top.print_entries || + (int)syme->snap_count < top.count_filter) + continue; + + pcnt = 100.0 - (100.0 * ((sum_ksamples - syme->snap_count) / + sum_ksamples)); + + if (top.evlist->nr_entries == 1 || !top.display_weighted) + printf("%20.2f ", syme->weight); + else + printf("%9.1f %10ld ", syme->weight, syme->snap_count); + + percent_color_fprintf(stdout, "%4.1f%%", pcnt); + if (verbose) + printf(" %016" PRIx64, sym->start); + printf(" %-*.*s", sym_width, sym_width, sym->name); + printf(" %-*.*s\n", dso_width, dso_width, + dso_width >= syme->map->dso->long_name_len ? + syme->map->dso->long_name : + syme->map->dso->short_name); + } +} + +static void prompt_integer(int *target, const char *msg) +{ + char *buf = malloc(0), *p; + size_t dummy = 0; + int tmp; + + fprintf(stdout, "\n%s: ", msg); + if (getline(&buf, &dummy, stdin) < 0) + return; + + p = strchr(buf, '\n'); + if (p) + *p = 0; + + p = buf; + while(*p) { + if (!isdigit(*p)) + goto out_free; + p++; + } + tmp = strtoul(buf, NULL, 10); + *target = tmp; +out_free: + free(buf); +} + +static void prompt_percent(int *target, const char *msg) +{ + int tmp = 0; + + prompt_integer(&tmp, msg); + if (tmp >= 0 && tmp <= 100) + *target = tmp; +} + +static void prompt_symbol(struct sym_entry **target, const char *msg) +{ + char *buf = malloc(0), *p; + struct sym_entry *syme = *target, *n, *found = NULL; + size_t dummy = 0; + + /* zero counters of active symbol */ + if (syme) { + __zero_source_counters(syme); + *target = NULL; + } + + fprintf(stdout, "\n%s: ", msg); + if (getline(&buf, &dummy, stdin) < 0) + goto out_free; + + p = strchr(buf, '\n'); + if (p) + *p = 0; + + pthread_mutex_lock(&top.active_symbols_lock); + syme = list_entry(top.active_symbols.next, struct sym_entry, node); + pthread_mutex_unlock(&top.active_symbols_lock); + + list_for_each_entry_safe_from(syme, n, &top.active_symbols, node) { + struct symbol *sym = sym_entry__symbol(syme); + + if (!strcmp(buf, sym->name)) { + found = syme; + break; + } + } + + if (!found) { + fprintf(stderr, "Sorry, %s is not active.\n", buf); + sleep(1); + return; + } else + parse_source(found); + +out_free: + free(buf); +} + +static void print_mapped_keys(void) +{ + char *name = NULL; + + if (top.sym_filter_entry) { + struct symbol *sym = sym_entry__symbol(top.sym_filter_entry); + name = sym->name; + } + + fprintf(stdout, "\nMapped keys:\n"); + fprintf(stdout, "\t[d] display refresh delay. \t(%d)\n", top.delay_secs); + fprintf(stdout, "\t[e] display entries (lines). \t(%d)\n", top.print_entries); + + if (top.evlist->nr_entries > 1) + fprintf(stdout, "\t[E] active event counter. \t(%s)\n", event_name(top.sym_evsel)); + + fprintf(stdout, "\t[f] profile display filter (count). \t(%d)\n", top.count_filter); + + fprintf(stdout, "\t[F] annotate display filter (percent). \t(%d%%)\n", sym_pcnt_filter); + fprintf(stdout, "\t[s] annotate symbol. \t(%s)\n", name?: "NULL"); + fprintf(stdout, "\t[S] stop annotation.\n"); + + if (top.evlist->nr_entries > 1) + fprintf(stdout, "\t[w] toggle display weighted/count[E]r. \t(%d)\n", top.display_weighted ? 1 : 0); + + fprintf(stdout, + "\t[K] hide kernel_symbols symbols. \t(%s)\n", + top.hide_kernel_symbols ? "yes" : "no"); + fprintf(stdout, + "\t[U] hide user symbols. \t(%s)\n", + top.hide_user_symbols ? "yes" : "no"); + fprintf(stdout, "\t[z] toggle sample zeroing. \t(%d)\n", top.zero ? 1 : 0); + fprintf(stdout, "\t[qQ] quit.\n"); +} + +static int key_mapped(int c) +{ + switch (c) { + case 'd': + case 'e': + case 'f': + case 'z': + case 'q': + case 'Q': + case 'K': + case 'U': + case 'F': + case 's': + case 'S': + return 1; + case 'E': + case 'w': + return top.evlist->nr_entries > 1 ? 1 : 0; + default: + break; + } + + return 0; +} + +static void handle_keypress(struct perf_session *session, int c) +{ + if (!key_mapped(c)) { + struct pollfd stdin_poll = { .fd = 0, .events = POLLIN }; + struct termios tc, save; + + print_mapped_keys(); + fprintf(stdout, "\nEnter selection, or unmapped key to continue: "); + fflush(stdout); + + tcgetattr(0, &save); + tc = save; + tc.c_lflag &= ~(ICANON | ECHO); + tc.c_cc[VMIN] = 0; + tc.c_cc[VTIME] = 0; + tcsetattr(0, TCSANOW, &tc); + + poll(&stdin_poll, 1, -1); + c = getc(stdin); + + tcsetattr(0, TCSAFLUSH, &save); + if (!key_mapped(c)) + return; + } + + switch (c) { + case 'd': + prompt_integer(&top.delay_secs, "Enter display delay"); + if (top.delay_secs < 1) + top.delay_secs = 1; + break; + case 'e': + prompt_integer(&top.print_entries, "Enter display entries (lines)"); + if (top.print_entries == 0) { + sig_winch_handler(SIGWINCH); + signal(SIGWINCH, sig_winch_handler); + } else + signal(SIGWINCH, SIG_DFL); + break; + case 'E': + if (top.evlist->nr_entries > 1) { + /* Select 0 as the default event: */ + int counter = 0; + + fprintf(stderr, "\nAvailable events:"); + + list_for_each_entry(top.sym_evsel, &top.evlist->entries, node) + fprintf(stderr, "\n\t%d %s", top.sym_evsel->idx, event_name(top.sym_evsel)); + + prompt_integer(&counter, "Enter details event counter"); + + if (counter >= top.evlist->nr_entries) { + top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node); + fprintf(stderr, "Sorry, no such event, using %s.\n", event_name(top.sym_evsel)); + sleep(1); + break; + } + list_for_each_entry(top.sym_evsel, &top.evlist->entries, node) + if (top.sym_evsel->idx == counter) + break; + } else + top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node); + break; + case 'f': + prompt_integer(&top.count_filter, "Enter display event count filter"); + break; + case 'F': + prompt_percent(&sym_pcnt_filter, "Enter details display event filter (percent)"); + break; + case 'K': + top.hide_kernel_symbols = !top.hide_kernel_symbols; + break; + case 'q': + case 'Q': + printf("exiting.\n"); + if (dump_symtab) + perf_session__fprintf_dsos(session, stderr); + exit(0); + case 's': + prompt_symbol(&top.sym_filter_entry, "Enter details symbol"); + break; + case 'S': + if (!top.sym_filter_entry) + break; + else { + struct sym_entry *syme = top.sym_filter_entry; + + top.sym_filter_entry = NULL; + __zero_source_counters(syme); + } + break; + case 'U': + top.hide_user_symbols = !top.hide_user_symbols; + break; + case 'w': + top.display_weighted = ~top.display_weighted; + break; + case 'z': + top.zero = !top.zero; + break; + default: + break; + } +} + +static void *display_thread_tui(void *arg __used) +{ + int err = 0; + pthread_mutex_lock(&top.active_symbols_lock); + while (list_empty(&top.active_symbols)) { + err = pthread_cond_wait(&top.active_symbols_cond, + &top.active_symbols_lock); + if (err) + break; + } + pthread_mutex_unlock(&top.active_symbols_lock); + if (!err) + perf_top__tui_browser(&top); + exit_browser(0); + exit(0); + return NULL; +} + +static void *display_thread(void *arg __used) +{ + struct pollfd stdin_poll = { .fd = 0, .events = POLLIN }; + struct termios tc, save; + int delay_msecs, c; + struct perf_session *session = (struct perf_session *) arg; + + tcgetattr(0, &save); + tc = save; + tc.c_lflag &= ~(ICANON | ECHO); + tc.c_cc[VMIN] = 0; + tc.c_cc[VTIME] = 0; + +repeat: + delay_msecs = top.delay_secs * 1000; + tcsetattr(0, TCSANOW, &tc); + /* trash return*/ + getc(stdin); + + do { + print_sym_table(session); + } while (!poll(&stdin_poll, 1, delay_msecs) == 1); + + c = getc(stdin); + tcsetattr(0, TCSAFLUSH, &save); + + handle_keypress(session, c); + goto repeat; + + return NULL; +} + +/* Tag samples to be skipped. */ +static const char *skip_symbols[] = { + "default_idle", + "native_safe_halt", + "cpu_idle", + "enter_idle", + "exit_idle", + "mwait_idle", + "mwait_idle_with_hints", + "poll_idle", + "ppc64_runlatch_off", + "pseries_dedicated_idle_sleep", + NULL +}; + +static int symbol_filter(struct map *map, struct symbol *sym) +{ + struct sym_entry *syme; + const char *name = sym->name; + int i; + + /* + * ppc64 uses function descriptors and appends a '.' to the + * start of every instruction address. Remove it. + */ + if (name[0] == '.') + name++; + + if (!strcmp(name, "_text") || + !strcmp(name, "_etext") || + !strcmp(name, "_sinittext") || + !strncmp("init_module", name, 11) || + !strncmp("cleanup_module", name, 14) || + strstr(name, "_text_start") || + strstr(name, "_text_end")) + return 1; + + syme = symbol__priv(sym); + syme->map = map; + symbol__annotate_init(map, sym); + + if (!top.sym_filter_entry && sym_filter && !strcmp(name, sym_filter)) { + /* schedule initial sym_filter_entry setup */ + sym_filter_entry_sched = syme; + sym_filter = NULL; + } + + for (i = 0; skip_symbols[i]; i++) { + if (!strcmp(skip_symbols[i], name)) { + sym->ignore = true; + break; + } + } + + return 0; +} + +static void perf_event__process_sample(const union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) +{ + u64 ip = event->ip.ip; + struct sym_entry *syme; + struct addr_location al; + struct machine *machine; + u8 origin = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + + ++top.samples; + + switch (origin) { + case PERF_RECORD_MISC_USER: + ++top.us_samples; + if (top.hide_user_symbols) + return; + machine = perf_session__find_host_machine(session); + break; + case PERF_RECORD_MISC_KERNEL: + ++top.kernel_samples; + if (top.hide_kernel_symbols) + return; + machine = perf_session__find_host_machine(session); + break; + case PERF_RECORD_MISC_GUEST_KERNEL: + ++top.guest_kernel_samples; + machine = perf_session__find_machine(session, event->ip.pid); + break; + case PERF_RECORD_MISC_GUEST_USER: + ++top.guest_us_samples; + /* + * TODO: we don't process guest user from host side + * except simple counting. + */ + return; + default: + return; + } + + if (!machine && perf_guest) { + pr_err("Can't find guest [%d]'s kernel information\n", + event->ip.pid); + return; + } + + if (event->header.misc & PERF_RECORD_MISC_EXACT_IP) + top.exact_samples++; + + if (perf_event__preprocess_sample(event, session, &al, sample, + symbol_filter) < 0 || + al.filtered) + return; + + if (!kptr_restrict_warned && + symbol_conf.kptr_restrict && + al.cpumode == PERF_RECORD_MISC_KERNEL) { + ui__warning( +"Kernel address maps (/proc/{kallsyms,modules}) are restricted.\n\n" +"Check /proc/sys/kernel/kptr_restrict.\n\n" +"Kernel%s samples will not be resolved.\n", + !RB_EMPTY_ROOT(&al.map->dso->symbols[MAP__FUNCTION]) ? + " modules" : ""); + if (use_browser <= 0) + sleep(5); + kptr_restrict_warned = true; + } + + if (al.sym == NULL) { + const char *msg = "Kernel samples will not be resolved.\n"; + /* + * As we do lazy loading of symtabs we only will know if the + * specified vmlinux file is invalid when we actually have a + * hit in kernel space and then try to load it. So if we get + * here and there are _no_ symbols in the DSO backing the + * kernel map, bail out. + * + * We may never get here, for instance, if we use -K/ + * --hide-kernel-symbols, even if the user specifies an + * invalid --vmlinux ;-) + */ + if (!kptr_restrict_warned && !vmlinux_warned && + al.map == machine->vmlinux_maps[MAP__FUNCTION] && + RB_EMPTY_ROOT(&al.map->dso->symbols[MAP__FUNCTION])) { + if (symbol_conf.vmlinux_name) { + ui__warning("The %s file can't be used.\n%s", + symbol_conf.vmlinux_name, msg); + } else { + ui__warning("A vmlinux file was not found.\n%s", + msg); + } + + if (use_browser <= 0) + sleep(5); + vmlinux_warned = true; + } + + return; + } + + /* let's see, whether we need to install initial sym_filter_entry */ + if (sym_filter_entry_sched) { + top.sym_filter_entry = sym_filter_entry_sched; + sym_filter_entry_sched = NULL; + if (parse_source(top.sym_filter_entry) < 0) { + struct symbol *sym = sym_entry__symbol(top.sym_filter_entry); + + pr_err("Can't annotate %s", sym->name); + if (top.sym_filter_entry->map->dso->symtab_type == SYMTAB__KALLSYMS) { + pr_err(": No vmlinux file was found in the path:\n"); + machine__fprintf_vmlinux_path(machine, stderr); + } else + pr_err(".\n"); + exit(1); + } + } + + syme = symbol__priv(al.sym); + if (!al.sym->ignore) { + struct perf_evsel *evsel; + + evsel = perf_evlist__id2evsel(top.evlist, sample->id); + assert(evsel != NULL); + syme->count[evsel->idx]++; + record_precise_ip(syme, evsel->idx, ip); + pthread_mutex_lock(&top.active_symbols_lock); + if (list_empty(&syme->node) || !syme->node.next) { + static bool first = true; + __list_insert_active_sym(syme); + if (first) { + pthread_cond_broadcast(&top.active_symbols_cond); + first = false; + } + } + pthread_mutex_unlock(&top.active_symbols_lock); + } +} + +static void perf_session__mmap_read_idx(struct perf_session *self, int idx) +{ + struct perf_sample sample; + union perf_event *event; + int ret; + + while ((event = perf_evlist__mmap_read(top.evlist, idx)) != NULL) { + ret = perf_session__parse_sample(self, event, &sample); + if (ret) { + pr_err("Can't parse sample, err = %d\n", ret); + continue; + } + + if (event->header.type == PERF_RECORD_SAMPLE) + perf_event__process_sample(event, &sample, self); + else + perf_event__process(event, &sample, self); + } +} + +static void perf_session__mmap_read(struct perf_session *self) +{ + int i; + + for (i = 0; i < top.evlist->nr_mmaps; i++) + perf_session__mmap_read_idx(self, i); +} + +static void start_counters(struct perf_evlist *evlist) +{ + struct perf_evsel *counter; + + list_for_each_entry(counter, &evlist->entries, node) { + struct perf_event_attr *attr = &counter->attr; + + attr->sample_type = PERF_SAMPLE_IP | PERF_SAMPLE_TID; + + if (top.freq) { + attr->sample_type |= PERF_SAMPLE_PERIOD; + attr->freq = 1; + attr->sample_freq = top.freq; + } + + if (evlist->nr_entries > 1) { + attr->sample_type |= PERF_SAMPLE_ID; + attr->read_format |= PERF_FORMAT_ID; + } + + attr->mmap = 1; + attr->inherit = inherit; +try_again: + if (perf_evsel__open(counter, top.evlist->cpus, + top.evlist->threads, group) < 0) { + int err = errno; + + if (err == EPERM || err == EACCES) { + ui__warning_paranoid(); + goto out_err; + } + /* + * If it's cycles then fall back to hrtimer + * based cpu-clock-tick sw counter, which + * is always available even if no PMU support: + */ + if (attr->type == PERF_TYPE_HARDWARE && + attr->config == PERF_COUNT_HW_CPU_CYCLES) { + if (verbose) + ui__warning("Cycles event not supported,\n" + "trying to fall back to cpu-clock-ticks\n"); + + attr->type = PERF_TYPE_SOFTWARE; + attr->config = PERF_COUNT_SW_CPU_CLOCK; + goto try_again; + } + + if (err == ENOENT) { + ui__warning("The %s event is not supported.\n", + event_name(counter)); + goto out_err; + } + + ui__warning("The sys_perf_event_open() syscall " + "returned with %d (%s). /bin/dmesg " + "may provide additional information.\n" + "No CONFIG_PERF_EVENTS=y kernel support " + "configured?\n", err, strerror(err)); + goto out_err; + } + } + + if (perf_evlist__mmap(evlist, mmap_pages, false) < 0) { + ui__warning("Failed to mmap with %d (%s)\n", + errno, strerror(errno)); + goto out_err; + } + + return; + +out_err: + exit_browser(0); + exit(0); +} + +static int __cmd_top(void) +{ + pthread_t thread; + int ret __used; + /* + * FIXME: perf_session__new should allow passing a O_MMAP, so that all this + * mmap reading, etc is encapsulated in it. Use O_WRONLY for now. + */ + struct perf_session *session = perf_session__new(NULL, O_WRONLY, false, false, NULL); + if (session == NULL) + return -ENOMEM; + + if (top.target_tid != -1) + perf_event__synthesize_thread_map(top.evlist->threads, + perf_event__process, session); + else + perf_event__synthesize_threads(perf_event__process, session); + + start_counters(top.evlist); + session->evlist = top.evlist; + perf_session__update_sample_type(session); + + /* Wait for a minimal set of events before starting the snapshot */ + poll(top.evlist->pollfd, top.evlist->nr_fds, 100); + + perf_session__mmap_read(session); + + if (pthread_create(&thread, NULL, (use_browser > 0 ? display_thread_tui : + display_thread), session)) { + printf("Could not create display thread.\n"); + exit(-1); + } + + if (realtime_prio) { + struct sched_param param; + + param.sched_priority = realtime_prio; + if (sched_setscheduler(0, SCHED_FIFO, ¶m)) { + printf("Could not set realtime priority.\n"); + exit(-1); + } + } + + while (1) { + u64 hits = top.samples; + + perf_session__mmap_read(session); + + if (hits == top.samples) + ret = poll(top.evlist->pollfd, top.evlist->nr_fds, 100); + } + + return 0; +} + +static const char * const top_usage[] = { + "perf top [<options>]", + NULL +}; + +static const struct option options[] = { + OPT_CALLBACK('e', "event", &top.evlist, "event", + "event selector. use 'perf list' to list available events", + parse_events), + OPT_INTEGER('c', "count", &default_interval, + "event period to sample"), + OPT_INTEGER('p', "pid", &top.target_pid, + "profile events on existing process id"), + OPT_INTEGER('t', "tid", &top.target_tid, + "profile events on existing thread id"), + OPT_BOOLEAN('a', "all-cpus", &system_wide, + "system-wide collection from all CPUs"), + OPT_STRING('C', "cpu", &top.cpu_list, "cpu", + "list of cpus to monitor"), + OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name, + "file", "vmlinux pathname"), + OPT_BOOLEAN('K', "hide_kernel_symbols", &top.hide_kernel_symbols, + "hide kernel symbols"), + OPT_UINTEGER('m', "mmap-pages", &mmap_pages, "number of mmap data pages"), + OPT_INTEGER('r', "realtime", &realtime_prio, + "collect data with this RT SCHED_FIFO priority"), + OPT_INTEGER('d', "delay", &top.delay_secs, + "number of seconds to delay between refreshes"), + OPT_BOOLEAN('D', "dump-symtab", &dump_symtab, + "dump the symbol table used for profiling"), + OPT_INTEGER('f', "count-filter", &top.count_filter, + "only display functions with more events than this"), + OPT_BOOLEAN('g', "group", &group, + "put the counters into a counter group"), + OPT_BOOLEAN('i', "inherit", &inherit, + "child tasks inherit counters"), + OPT_STRING('s', "sym-annotate", &sym_filter, "symbol name", + "symbol to annotate"), + OPT_BOOLEAN('z', "zero", &top.zero, + "zero history across updates"), + OPT_INTEGER('F', "freq", &top.freq, + "profile at this frequency"), + OPT_INTEGER('E', "entries", &top.print_entries, + "display this many functions"), + OPT_BOOLEAN('U', "hide_user_symbols", &top.hide_user_symbols, + "hide user symbols"), + OPT_BOOLEAN(0, "tui", &use_tui, "Use the TUI interface"), + OPT_BOOLEAN(0, "stdio", &use_stdio, "Use the stdio interface"), + OPT_INCR('v', "verbose", &verbose, + "be more verbose (show counter open errors, etc)"), + OPT_END() +}; + +int cmd_top(int argc, const char **argv, const char *prefix __used) +{ + struct perf_evsel *pos; + int status = -ENOMEM; + + top.evlist = perf_evlist__new(NULL, NULL); + if (top.evlist == NULL) + return -ENOMEM; + + page_size = sysconf(_SC_PAGE_SIZE); + + argc = parse_options(argc, argv, options, top_usage, 0); + if (argc) + usage_with_options(top_usage, options); + + /* + * XXX For now start disabled, only using TUI if explicitely asked for. + * Change that when handle_keys equivalent gets written, live annotation + * done, etc. + */ + use_browser = 0; + + if (use_stdio) + use_browser = 0; + else if (use_tui) + use_browser = 1; + + setup_browser(false); + + /* CPU and PID are mutually exclusive */ + if (top.target_tid > 0 && top.cpu_list) { + printf("WARNING: PID switch overriding CPU\n"); + sleep(1); + top.cpu_list = NULL; + } + + if (top.target_pid != -1) + top.target_tid = top.target_pid; + + if (perf_evlist__create_maps(top.evlist, top.target_pid, + top.target_tid, top.cpu_list) < 0) + usage_with_options(top_usage, options); + + if (!top.evlist->nr_entries && + perf_evlist__add_default(top.evlist) < 0) { + pr_err("Not enough memory for event selector list\n"); + return -ENOMEM; + } + + if (top.delay_secs < 1) + top.delay_secs = 1; + + /* + * User specified count overrides default frequency. + */ + if (default_interval) + top.freq = 0; + else if (top.freq) { + default_interval = top.freq; + } else { + fprintf(stderr, "frequency and count are zero, aborting\n"); + exit(EXIT_FAILURE); + } + + list_for_each_entry(pos, &top.evlist->entries, node) { + if (perf_evsel__alloc_fd(pos, top.evlist->cpus->nr, + top.evlist->threads->nr) < 0) + goto out_free_fd; + /* + * Fill in the ones not specifically initialized via -c: + */ + if (pos->attr.sample_period) + continue; + + pos->attr.sample_period = default_interval; + } + + if (perf_evlist__alloc_pollfd(top.evlist) < 0 || + perf_evlist__alloc_mmap(top.evlist) < 0) + goto out_free_fd; + + top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node); + + symbol_conf.priv_size = (sizeof(struct sym_entry) + sizeof(struct annotation) + + (top.evlist->nr_entries + 1) * sizeof(unsigned long)); + + symbol_conf.try_vmlinux_path = (symbol_conf.vmlinux_name == NULL); + if (symbol__init() < 0) + return -1; + + get_term_dimensions(&winsize); + if (top.print_entries == 0) { + update_print_entries(&winsize); + signal(SIGWINCH, sig_winch_handler); + } + + status = __cmd_top(); +out_free_fd: + perf_evlist__delete(top.evlist); + + return status; +} diff --git a/tools/perf/builtin.h b/tools/perf/builtin.h new file mode 100644 index 00000000..4702e244 --- /dev/null +++ b/tools/perf/builtin.h @@ -0,0 +1,40 @@ +#ifndef BUILTIN_H +#define BUILTIN_H + +#include "util/util.h" +#include "util/strbuf.h" + +extern const char perf_version_string[]; +extern const char perf_usage_string[]; +extern const char perf_more_info_string[]; + +extern void list_common_cmds_help(void); +extern const char *help_unknown_cmd(const char *cmd); +extern void prune_packed_objects(int); +extern int read_line_with_nul(char *buf, int size, FILE *file); +extern int check_pager_config(const char *cmd); + +extern int cmd_annotate(int argc, const char **argv, const char *prefix); +extern int cmd_bench(int argc, const char **argv, const char *prefix); +extern int cmd_buildid_cache(int argc, const char **argv, const char *prefix); +extern int cmd_buildid_list(int argc, const char **argv, const char *prefix); +extern int cmd_diff(int argc, const char **argv, const char *prefix); +extern int cmd_evlist(int argc, const char **argv, const char *prefix); +extern int cmd_help(int argc, const char **argv, const char *prefix); +extern int cmd_sched(int argc, const char **argv, const char *prefix); +extern int cmd_list(int argc, const char **argv, const char *prefix); +extern int cmd_record(int argc, const char **argv, const char *prefix); +extern int cmd_report(int argc, const char **argv, const char *prefix); +extern int cmd_stat(int argc, const char **argv, const char *prefix); +extern int cmd_timechart(int argc, const char **argv, const char *prefix); +extern int cmd_top(int argc, const char **argv, const char *prefix); +extern int cmd_script(int argc, const char **argv, const char *prefix); +extern int cmd_version(int argc, const char **argv, const char *prefix); +extern int cmd_probe(int argc, const char **argv, const char *prefix); +extern int cmd_kmem(int argc, const char **argv, const char *prefix); +extern int cmd_lock(int argc, const char **argv, const char *prefix); +extern int cmd_kvm(int argc, const char **argv, const char *prefix); +extern int cmd_test(int argc, const char **argv, const char *prefix); +extern int cmd_inject(int argc, const char **argv, const char *prefix); + +#endif diff --git a/tools/perf/command-list.txt b/tools/perf/command-list.txt new file mode 100644 index 00000000..d695fe40 --- /dev/null +++ b/tools/perf/command-list.txt @@ -0,0 +1,25 @@ +# +# List of known perf commands. +# command name category [deprecated] [common] +# +perf-annotate mainporcelain common +perf-archive mainporcelain common +perf-bench mainporcelain common +perf-buildid-cache mainporcelain common +perf-buildid-list mainporcelain common +perf-diff mainporcelain common +perf-evlist mainporcelain common +perf-inject mainporcelain common +perf-list mainporcelain common +perf-sched mainporcelain common +perf-record mainporcelain common +perf-report mainporcelain common +perf-stat mainporcelain common +perf-timechart mainporcelain common +perf-top mainporcelain common +perf-script mainporcelain common +perf-probe mainporcelain common +perf-kmem mainporcelain common +perf-lock mainporcelain common +perf-kvm mainporcelain common +perf-test mainporcelain common diff --git a/tools/perf/config/feature-tests.mak b/tools/perf/config/feature-tests.mak new file mode 100644 index 00000000..6170fd25 --- /dev/null +++ b/tools/perf/config/feature-tests.mak @@ -0,0 +1,128 @@ +define SOURCE_HELLO +#include <stdio.h> +int main(void) +{ + return puts(\"hi\"); +} +endef + +ifndef NO_DWARF +define SOURCE_DWARF +#include <dwarf.h> +#include <elfutils/libdw.h> +#include <elfutils/version.h> +#ifndef _ELFUTILS_PREREQ +#error +#endif + +int main(void) +{ + Dwarf *dbg = dwarf_begin(0, DWARF_C_READ); + return (long)dbg; +} +endef +endif + +define SOURCE_LIBELF +#include <libelf.h> + +int main(void) +{ + Elf *elf = elf_begin(0, ELF_C_READ, 0); + return (long)elf; +} +endef + +define SOURCE_GLIBC +#include <gnu/libc-version.h> + +int main(void) +{ + const char *version = gnu_get_libc_version(); + return (long)version; +} +endef + +define SOURCE_ELF_MMAP +#include <libelf.h> +int main(void) +{ + Elf *elf = elf_begin(0, ELF_C_READ_MMAP, 0); + return (long)elf; +} +endef + +ifndef NO_NEWT +define SOURCE_NEWT +#include <newt.h> + +int main(void) +{ + newtInit(); + newtCls(); + return newtFinished(); +} +endef +endif + +ifndef NO_LIBPERL +define SOURCE_PERL_EMBED +#include <EXTERN.h> +#include <perl.h> + +int main(void) +{ +perl_alloc(); +return 0; +} +endef +endif + +ifndef NO_LIBPYTHON +define SOURCE_PYTHON_VERSION +#include <Python.h> +#if PY_VERSION_HEX >= 0x03000000 + #error +#endif +int main(void){} +endef +define SOURCE_PYTHON_EMBED +#include <Python.h> +int main(void) +{ + Py_Initialize(); + return 0; +} +endef +endif + +define SOURCE_BFD +#include <bfd.h> + +int main(void) +{ + bfd_demangle(0, 0, 0); + return 0; +} +endef + +define SOURCE_CPLUS_DEMANGLE +extern char *cplus_demangle(const char *, int); + +int main(void) +{ + cplus_demangle(0, 0); + return 0; +} +endef + +define SOURCE_STRLCPY +#include <stdlib.h> +extern size_t strlcpy(char *dest, const char *src, size_t size); + +int main(void) +{ + strlcpy(NULL, NULL, 0); + return 0; +} +endef diff --git a/tools/perf/config/utilities.mak b/tools/perf/config/utilities.mak new file mode 100644 index 00000000..8046182a --- /dev/null +++ b/tools/perf/config/utilities.mak @@ -0,0 +1,188 @@ +# This allows us to work with the newline character: +define newline + + +endef +newline := $(newline) + +# nl-escape +# +# Usage: escape = $(call nl-escape[,escape]) +# +# This is used as the common way to specify +# what should replace a newline when escaping +# newlines; the default is a bizarre string. +# +nl-escape = $(or $(1),m822df3020w6a44id34bt574ctac44eb9f4n) + +# escape-nl +# +# Usage: escaped-text = $(call escape-nl,text[,escape]) +# +# GNU make's $(shell ...) function converts to a +# single space each newline character in the output +# produced during the expansion; this may not be +# desirable. +# +# The only solution is to change each newline into +# something that won't be converted, so that the +# information can be recovered later with +# $(call unescape-nl...) +# +escape-nl = $(subst $(newline),$(call nl-escape,$(2)),$(1)) + +# unescape-nl +# +# Usage: text = $(call unescape-nl,escaped-text[,escape]) +# +# See escape-nl. +# +unescape-nl = $(subst $(call nl-escape,$(2)),$(newline),$(1)) + +# shell-escape-nl +# +# Usage: $(shell some-command | $(call shell-escape-nl[,escape])) +# +# Use this to escape newlines from within a shell call; +# the default escape is a bizarre string. +# +# NOTE: The escape is used directly as a string constant +# in an `awk' program that is delimited by shell +# single-quotes, so be wary of the characters +# that are chosen. +# +define shell-escape-nl +awk 'NR==1 {t=$$0} NR>1 {t=t "$(nl-escape)" $$0} END {printf t}' +endef + +# shell-unescape-nl +# +# Usage: $(shell some-command | $(call shell-unescape-nl[,escape])) +# +# Use this to unescape newlines from within a shell call; +# the default escape is a bizarre string. +# +# NOTE: The escape is used directly as an extended regular +# expression constant in an `awk' program that is +# delimited by shell single-quotes, so be wary +# of the characters that are chosen. +# +# (The bash shell has a bug where `{gsub(...),...}' is +# misinterpreted as a brace expansion; this can be +# overcome by putting a space between `{' and `gsub'). +# +define shell-unescape-nl +awk 'NR==1 {t=$$0} NR>1 {t=t "\n" $$0} END { gsub(/$(nl-escape)/,"\n",t); printf t }' +endef + +# escape-for-shell-sq +# +# Usage: embeddable-text = $(call escape-for-shell-sq,text) +# +# This function produces text that is suitable for +# embedding in a shell string that is delimited by +# single-quotes. +# +escape-for-shell-sq = $(subst ','\'',$(1)) + +# shell-sq +# +# Usage: single-quoted-and-escaped-text = $(call shell-sq,text) +# +shell-sq = '$(escape-for-shell-sq)' + +# shell-wordify +# +# Usage: wordified-text = $(call shell-wordify,text) +# +# For instance: +# +# |define text +# |hello +# |world +# |endef +# | +# |target: +# | echo $(call shell-wordify,$(text)) +# +# At least GNU make gets confused by expanding a newline +# within the context of a command line of a makefile rule +# (this is in constrast to a `$(shell ...)' function call, +# which can handle it just fine). +# +# This function avoids the problem by producing a string +# that works as a shell word, regardless of whether or +# not it contains a newline. +# +# If the text to be wordified contains a newline, then +# an intrictate shell command substitution is constructed +# to render the text as a single line; when the shell +# processes the resulting escaped text, it transforms +# it into the original unescaped text. +# +# If the text does not contain a newline, then this function +# produces the same results as the `$(shell-sq)' function. +# +shell-wordify = $(if $(findstring $(newline),$(1)),$(_sw-esc-nl),$(shell-sq)) +define _sw-esc-nl +"$$(echo $(call escape-nl,$(shell-sq),$(2)) | $(call shell-unescape-nl,$(2)))" +endef + +# is-absolute +# +# Usage: bool-value = $(call is-absolute,path) +# +is-absolute = $(shell echo $(shell-sq) | grep ^/ -q && echo y) + +# lookup +# +# Usage: absolute-executable-path-or-empty = $(call lookup,path) +# +# (It's necessary to use `sh -c' because GNU make messes up by +# trying too hard and getting things wrong). +# +lookup = $(call unescape-nl,$(shell sh -c $(_l-sh))) +_l-sh = $(call shell-sq,command -v $(shell-sq) | $(call shell-escape-nl,)) + +# is-executable +# +# Usage: bool-value = $(call is-executable,path) +# +# (It's necessary to use `sh -c' because GNU make messes up by +# trying too hard and getting things wrong). +# +is-executable = $(call _is-executable-helper,$(shell-sq)) +_is-executable-helper = $(shell sh -c $(_is-executable-sh)) +_is-executable-sh = $(call shell-sq,test -f $(1) -a -x $(1) && echo y) + +# get-executable +# +# Usage: absolute-executable-path-or-empty = $(call get-executable,path) +# +# The goal is to get an absolute path for an executable; +# the `command -v' is defined by POSIX, but it's not +# necessarily very portable, so it's only used if +# relative path resolution is requested, as determined +# by the presence of a leading `/'. +# +get-executable = $(if $(1),$(if $(is-absolute),$(_ge-abspath),$(lookup))) +_ge-abspath = $(if $(is-executable),$(1)) + +# get-supplied-or-default-executable +# +# Usage: absolute-executable-path-or-empty = $(call get-executable-or-default,variable,default) +# +define get-executable-or-default +$(if $($(1)),$(call _ge_attempt,$($(1)),$(1)),$(call _ge_attempt,$(2))) +endef +_ge_attempt = $(or $(get-executable),$(_gea_warn),$(call _gea_err,$(2))) +_gea_warn = $(warning The path '$(1)' is not executable.) +_gea_err = $(if $(1),$(error Please set '$(1)' appropriately)) + +# try-cc +# Usage: option = $(call try-cc, source-to-build, cc-options) +try-cc = $(shell sh -c \ + 'TMP="$(OUTPUT)$(TMPOUT).$$$$"; \ + echo "$(1)" | \ + $(CC) -x c - $(2) -o "$$TMP" > /dev/null 2>&1 && echo y; \ + rm -f "$$TMP"') diff --git a/tools/perf/design.txt b/tools/perf/design.txt new file mode 100644 index 00000000..bd0bb1b1 --- /dev/null +++ b/tools/perf/design.txt @@ -0,0 +1,462 @@ + +Performance Counters for Linux +------------------------------ + +Performance counters are special hardware registers available on most modern +CPUs. These registers count the number of certain types of hw events: such +as instructions executed, cachemisses suffered, or branches mis-predicted - +without slowing down the kernel or applications. These registers can also +trigger interrupts when a threshold number of events have passed - and can +thus be used to profile the code that runs on that CPU. + +The Linux Performance Counter subsystem provides an abstraction of these +hardware capabilities. It provides per task and per CPU counters, counter +groups, and it provides event capabilities on top of those. It +provides "virtual" 64-bit counters, regardless of the width of the +underlying hardware counters. + +Performance counters are accessed via special file descriptors. +There's one file descriptor per virtual counter used. + +The special file descriptor is opened via the perf_event_open() +system call: + + int sys_perf_event_open(struct perf_event_attr *hw_event_uptr, + pid_t pid, int cpu, int group_fd, + unsigned long flags); + +The syscall returns the new fd. The fd can be used via the normal +VFS system calls: read() can be used to read the counter, fcntl() +can be used to set the blocking mode, etc. + +Multiple counters can be kept open at a time, and the counters +can be poll()ed. + +When creating a new counter fd, 'perf_event_attr' is: + +struct perf_event_attr { + /* + * The MSB of the config word signifies if the rest contains cpu + * specific (raw) counter configuration data, if unset, the next + * 7 bits are an event type and the rest of the bits are the event + * identifier. + */ + __u64 config; + + __u64 irq_period; + __u32 record_type; + __u32 read_format; + + __u64 disabled : 1, /* off by default */ + inherit : 1, /* children inherit it */ + pinned : 1, /* must always be on PMU */ + exclusive : 1, /* only group on PMU */ + exclude_user : 1, /* don't count user */ + exclude_kernel : 1, /* ditto kernel */ + exclude_hv : 1, /* ditto hypervisor */ + exclude_idle : 1, /* don't count when idle */ + mmap : 1, /* include mmap data */ + munmap : 1, /* include munmap data */ + comm : 1, /* include comm data */ + + __reserved_1 : 52; + + __u32 extra_config_len; + __u32 wakeup_events; /* wakeup every n events */ + + __u64 __reserved_2; + __u64 __reserved_3; +}; + +The 'config' field specifies what the counter should count. It +is divided into 3 bit-fields: + +raw_type: 1 bit (most significant bit) 0x8000_0000_0000_0000 +type: 7 bits (next most significant) 0x7f00_0000_0000_0000 +event_id: 56 bits (least significant) 0x00ff_ffff_ffff_ffff + +If 'raw_type' is 1, then the counter will count a hardware event +specified by the remaining 63 bits of event_config. The encoding is +machine-specific. + +If 'raw_type' is 0, then the 'type' field says what kind of counter +this is, with the following encoding: + +enum perf_event_types { + PERF_TYPE_HARDWARE = 0, + PERF_TYPE_SOFTWARE = 1, + PERF_TYPE_TRACEPOINT = 2, +}; + +A counter of PERF_TYPE_HARDWARE will count the hardware event +specified by 'event_id': + +/* + * Generalized performance counter event types, used by the hw_event.event_id + * parameter of the sys_perf_event_open() syscall: + */ +enum hw_event_ids { + /* + * Common hardware events, generalized by the kernel: + */ + PERF_COUNT_HW_CPU_CYCLES = 0, + PERF_COUNT_HW_INSTRUCTIONS = 1, + PERF_COUNT_HW_CACHE_REFERENCES = 2, + PERF_COUNT_HW_CACHE_MISSES = 3, + PERF_COUNT_HW_BRANCH_INSTRUCTIONS = 4, + PERF_COUNT_HW_BRANCH_MISSES = 5, + PERF_COUNT_HW_BUS_CYCLES = 6, +}; + +These are standardized types of events that work relatively uniformly +on all CPUs that implement Performance Counters support under Linux, +although there may be variations (e.g., different CPUs might count +cache references and misses at different levels of the cache hierarchy). +If a CPU is not able to count the selected event, then the system call +will return -EINVAL. + +More hw_event_types are supported as well, but they are CPU-specific +and accessed as raw events. For example, to count "External bus +cycles while bus lock signal asserted" events on Intel Core CPUs, pass +in a 0x4064 event_id value and set hw_event.raw_type to 1. + +A counter of type PERF_TYPE_SOFTWARE will count one of the available +software events, selected by 'event_id': + +/* + * Special "software" counters provided by the kernel, even if the hardware + * does not support performance counters. These counters measure various + * physical and sw events of the kernel (and allow the profiling of them as + * well): + */ +enum sw_event_ids { + PERF_COUNT_SW_CPU_CLOCK = 0, + PERF_COUNT_SW_TASK_CLOCK = 1, + PERF_COUNT_SW_PAGE_FAULTS = 2, + PERF_COUNT_SW_CONTEXT_SWITCHES = 3, + PERF_COUNT_SW_CPU_MIGRATIONS = 4, + PERF_COUNT_SW_PAGE_FAULTS_MIN = 5, + PERF_COUNT_SW_PAGE_FAULTS_MAJ = 6, + PERF_COUNT_SW_ALIGNMENT_FAULTS = 7, + PERF_COUNT_SW_EMULATION_FAULTS = 8, +}; + +Counters of the type PERF_TYPE_TRACEPOINT are available when the ftrace event +tracer is available, and event_id values can be obtained from +/debug/tracing/events/*/*/id + + +Counters come in two flavours: counting counters and sampling +counters. A "counting" counter is one that is used for counting the +number of events that occur, and is characterised by having +irq_period = 0. + + +A read() on a counter returns the current value of the counter and possible +additional values as specified by 'read_format', each value is a u64 (8 bytes) +in size. + +/* + * Bits that can be set in hw_event.read_format to request that + * reads on the counter should return the indicated quantities, + * in increasing order of bit value, after the counter value. + */ +enum perf_event_read_format { + PERF_FORMAT_TOTAL_TIME_ENABLED = 1, + PERF_FORMAT_TOTAL_TIME_RUNNING = 2, +}; + +Using these additional values one can establish the overcommit ratio for a +particular counter allowing one to take the round-robin scheduling effect +into account. + + +A "sampling" counter is one that is set up to generate an interrupt +every N events, where N is given by 'irq_period'. A sampling counter +has irq_period > 0. The record_type controls what data is recorded on each +interrupt: + +/* + * Bits that can be set in hw_event.record_type to request information + * in the overflow packets. + */ +enum perf_event_record_format { + PERF_RECORD_IP = 1U << 0, + PERF_RECORD_TID = 1U << 1, + PERF_RECORD_TIME = 1U << 2, + PERF_RECORD_ADDR = 1U << 3, + PERF_RECORD_GROUP = 1U << 4, + PERF_RECORD_CALLCHAIN = 1U << 5, +}; + +Such (and other) events will be recorded in a ring-buffer, which is +available to user-space using mmap() (see below). + +The 'disabled' bit specifies whether the counter starts out disabled +or enabled. If it is initially disabled, it can be enabled by ioctl +or prctl (see below). + +The 'inherit' bit, if set, specifies that this counter should count +events on descendant tasks as well as the task specified. This only +applies to new descendents, not to any existing descendents at the +time the counter is created (nor to any new descendents of existing +descendents). + +The 'pinned' bit, if set, specifies that the counter should always be +on the CPU if at all possible. It only applies to hardware counters +and only to group leaders. If a pinned counter cannot be put onto the +CPU (e.g. because there are not enough hardware counters or because of +a conflict with some other event), then the counter goes into an +'error' state, where reads return end-of-file (i.e. read() returns 0) +until the counter is subsequently enabled or disabled. + +The 'exclusive' bit, if set, specifies that when this counter's group +is on the CPU, it should be the only group using the CPU's counters. +In future, this will allow sophisticated monitoring programs to supply +extra configuration information via 'extra_config_len' to exploit +advanced features of the CPU's Performance Monitor Unit (PMU) that are +not otherwise accessible and that might disrupt other hardware +counters. + +The 'exclude_user', 'exclude_kernel' and 'exclude_hv' bits provide a +way to request that counting of events be restricted to times when the +CPU is in user, kernel and/or hypervisor mode. + +The 'mmap' and 'munmap' bits allow recording of PROT_EXEC mmap/munmap +operations, these can be used to relate userspace IP addresses to actual +code, even after the mapping (or even the whole process) is gone, +these events are recorded in the ring-buffer (see below). + +The 'comm' bit allows tracking of process comm data on process creation. +This too is recorded in the ring-buffer (see below). + +The 'pid' parameter to the perf_event_open() system call allows the +counter to be specific to a task: + + pid == 0: if the pid parameter is zero, the counter is attached to the + current task. + + pid > 0: the counter is attached to a specific task (if the current task + has sufficient privilege to do so) + + pid < 0: all tasks are counted (per cpu counters) + +The 'cpu' parameter allows a counter to be made specific to a CPU: + + cpu >= 0: the counter is restricted to a specific CPU + cpu == -1: the counter counts on all CPUs + +(Note: the combination of 'pid == -1' and 'cpu == -1' is not valid.) + +A 'pid > 0' and 'cpu == -1' counter is a per task counter that counts +events of that task and 'follows' that task to whatever CPU the task +gets schedule to. Per task counters can be created by any user, for +their own tasks. + +A 'pid == -1' and 'cpu == x' counter is a per CPU counter that counts +all events on CPU-x. Per CPU counters need CAP_SYS_ADMIN privilege. + +The 'flags' parameter is currently unused and must be zero. + +The 'group_fd' parameter allows counter "groups" to be set up. A +counter group has one counter which is the group "leader". The leader +is created first, with group_fd = -1 in the perf_event_open call +that creates it. The rest of the group members are created +subsequently, with group_fd giving the fd of the group leader. +(A single counter on its own is created with group_fd = -1 and is +considered to be a group with only 1 member.) + +A counter group is scheduled onto the CPU as a unit, that is, it will +only be put onto the CPU if all of the counters in the group can be +put onto the CPU. This means that the values of the member counters +can be meaningfully compared, added, divided (to get ratios), etc., +with each other, since they have counted events for the same set of +executed instructions. + + +Like stated, asynchronous events, like counter overflow or PROT_EXEC mmap +tracking are logged into a ring-buffer. This ring-buffer is created and +accessed through mmap(). + +The mmap size should be 1+2^n pages, where the first page is a meta-data page +(struct perf_event_mmap_page) that contains various bits of information such +as where the ring-buffer head is. + +/* + * Structure of the page that can be mapped via mmap + */ +struct perf_event_mmap_page { + __u32 version; /* version number of this structure */ + __u32 compat_version; /* lowest version this is compat with */ + + /* + * Bits needed to read the hw counters in user-space. + * + * u32 seq; + * s64 count; + * + * do { + * seq = pc->lock; + * + * barrier() + * if (pc->index) { + * count = pmc_read(pc->index - 1); + * count += pc->offset; + * } else + * goto regular_read; + * + * barrier(); + * } while (pc->lock != seq); + * + * NOTE: for obvious reason this only works on self-monitoring + * processes. + */ + __u32 lock; /* seqlock for synchronization */ + __u32 index; /* hardware counter identifier */ + __s64 offset; /* add to hardware counter value */ + + /* + * Control data for the mmap() data buffer. + * + * User-space reading this value should issue an rmb(), on SMP capable + * platforms, after reading this value -- see perf_event_wakeup(). + */ + __u32 data_head; /* head in the data section */ +}; + +NOTE: the hw-counter userspace bits are arch specific and are currently only + implemented on powerpc. + +The following 2^n pages are the ring-buffer which contains events of the form: + +#define PERF_RECORD_MISC_KERNEL (1 << 0) +#define PERF_RECORD_MISC_USER (1 << 1) +#define PERF_RECORD_MISC_OVERFLOW (1 << 2) + +struct perf_event_header { + __u32 type; + __u16 misc; + __u16 size; +}; + +enum perf_event_type { + + /* + * The MMAP events record the PROT_EXEC mappings so that we can + * correlate userspace IPs to code. They have the following structure: + * + * struct { + * struct perf_event_header header; + * + * u32 pid, tid; + * u64 addr; + * u64 len; + * u64 pgoff; + * char filename[]; + * }; + */ + PERF_RECORD_MMAP = 1, + PERF_RECORD_MUNMAP = 2, + + /* + * struct { + * struct perf_event_header header; + * + * u32 pid, tid; + * char comm[]; + * }; + */ + PERF_RECORD_COMM = 3, + + /* + * When header.misc & PERF_RECORD_MISC_OVERFLOW the event_type field + * will be PERF_RECORD_* + * + * struct { + * struct perf_event_header header; + * + * { u64 ip; } && PERF_RECORD_IP + * { u32 pid, tid; } && PERF_RECORD_TID + * { u64 time; } && PERF_RECORD_TIME + * { u64 addr; } && PERF_RECORD_ADDR + * + * { u64 nr; + * { u64 event, val; } cnt[nr]; } && PERF_RECORD_GROUP + * + * { u16 nr, + * hv, + * kernel, + * user; + * u64 ips[nr]; } && PERF_RECORD_CALLCHAIN + * }; + */ +}; + +NOTE: PERF_RECORD_CALLCHAIN is arch specific and currently only implemented + on x86. + +Notification of new events is possible through poll()/select()/epoll() and +fcntl() managing signals. + +Normally a notification is generated for every page filled, however one can +additionally set perf_event_attr.wakeup_events to generate one every +so many counter overflow events. + +Future work will include a splice() interface to the ring-buffer. + + +Counters can be enabled and disabled in two ways: via ioctl and via +prctl. When a counter is disabled, it doesn't count or generate +events but does continue to exist and maintain its count value. + +An individual counter or counter group can be enabled with + + ioctl(fd, PERF_EVENT_IOC_ENABLE); + +or disabled with + + ioctl(fd, PERF_EVENT_IOC_DISABLE); + +Enabling or disabling the leader of a group enables or disables the +whole group; that is, while the group leader is disabled, none of the +counters in the group will count. Enabling or disabling a member of a +group other than the leader only affects that counter - disabling an +non-leader stops that counter from counting but doesn't affect any +other counter. + +Additionally, non-inherited overflow counters can use + + ioctl(fd, PERF_EVENT_IOC_REFRESH, nr); + +to enable a counter for 'nr' events, after which it gets disabled again. + +A process can enable or disable all the counter groups that are +attached to it, using prctl: + + prctl(PR_TASK_PERF_EVENTS_ENABLE); + + prctl(PR_TASK_PERF_EVENTS_DISABLE); + +This applies to all counters on the current process, whether created +by this process or by another, and doesn't affect any counters that +this process has created on other processes. It only enables or +disables the group leaders, not any other members in the groups. + + +Arch requirements +----------------- + +If your architecture does not have hardware performance metrics, you can +still use the generic software counters based on hrtimers for sampling. + +So to start with, in order to add HAVE_PERF_EVENTS to your Kconfig, you +will need at least this: + - asm/perf_event.h - a basic stub will suffice at first + - support for atomic64 types (and associated helper functions) + - set_perf_event_pending() implemented + +If your architecture does have hardware capabilities, you can override the +weak stub hw_perf_event_init() to register hardware counters. + +Architectures that have d-cache aliassing issues, such as Sparc and ARM, +should select PERF_USE_VMALLOC in order to avoid these for perf mmap(). diff --git a/tools/perf/perf-archive.sh b/tools/perf/perf-archive.sh new file mode 100644 index 00000000..677e59d6 --- /dev/null +++ b/tools/perf/perf-archive.sh @@ -0,0 +1,46 @@ +#!/bin/bash +# perf archive +# Arnaldo Carvalho de Melo <acme@redhat.com> + +PERF_DATA=perf.data +if [ $# -ne 0 ] ; then + PERF_DATA=$1 +fi + +# +# PERF_BUILDID_DIR environment variable set by perf +# path to buildid directory, default to $HOME/.debug +# +if [ -z $PERF_BUILDID_DIR ]; then + PERF_BUILDID_DIR=~/.debug/ +else + # append / to make substitutions work + PERF_BUILDID_DIR=$PERF_BUILDID_DIR/ +fi + +BUILDIDS=$(mktemp /tmp/perf-archive-buildids.XXXXXX) +NOBUILDID=0000000000000000000000000000000000000000 + +perf buildid-list -i $PERF_DATA --with-hits | grep -v "^$NOBUILDID " > $BUILDIDS +if [ ! -s $BUILDIDS ] ; then + echo "perf archive: no build-ids found" + rm -f $BUILDIDS + exit 1 +fi + +MANIFEST=$(mktemp /tmp/perf-archive-manifest.XXXXXX) + +cut -d ' ' -f 1 $BUILDIDS | \ +while read build_id ; do + linkname=$PERF_BUILDID_DIR.build-id/${build_id:0:2}/${build_id:2} + filename=$(readlink -f $linkname) + echo ${linkname#$PERF_BUILDID_DIR} >> $MANIFEST + echo ${filename#$PERF_BUILDID_DIR} >> $MANIFEST +done + +tar cfj $PERF_DATA.tar.bz2 -C $PERF_BUILDID_DIR -T $MANIFEST +rm -f $MANIFEST $BUILDIDS +echo -e "Now please run:\n" +echo -e "$ tar xvf $PERF_DATA.tar.bz2 -C ~/.debug\n" +echo "wherever you need to run 'perf report' on." +exit 0 diff --git a/tools/perf/perf.c b/tools/perf/perf.c new file mode 100644 index 00000000..ec635b7c --- /dev/null +++ b/tools/perf/perf.c @@ -0,0 +1,509 @@ +/* + * perf.c + * + * Performance analysis utility. + * + * This is the main hub from which the sub-commands (perf stat, + * perf top, perf record, perf report, etc.) are started. + */ +#include "builtin.h" + +#include "util/exec_cmd.h" +#include "util/cache.h" +#include "util/quote.h" +#include "util/run-command.h" +#include "util/parse-events.h" +#include "util/debugfs.h" + +const char perf_usage_string[] = + "perf [--version] [--help] COMMAND [ARGS]"; + +const char perf_more_info_string[] = + "See 'perf help COMMAND' for more information on a specific command."; + +int use_browser = -1; +static int use_pager = -1; + +struct pager_config { + const char *cmd; + int val; +}; + +static char debugfs_mntpt[MAXPATHLEN]; + +static int pager_command_config(const char *var, const char *value, void *data) +{ + struct pager_config *c = data; + if (!prefixcmp(var, "pager.") && !strcmp(var + 6, c->cmd)) + c->val = perf_config_bool(var, value); + return 0; +} + +/* returns 0 for "no pager", 1 for "use pager", and -1 for "not specified" */ +int check_pager_config(const char *cmd) +{ + struct pager_config c; + c.cmd = cmd; + c.val = -1; + perf_config(pager_command_config, &c); + return c.val; +} + +static int tui_command_config(const char *var, const char *value, void *data) +{ + struct pager_config *c = data; + if (!prefixcmp(var, "tui.") && !strcmp(var + 4, c->cmd)) + c->val = perf_config_bool(var, value); + return 0; +} + +/* returns 0 for "no tui", 1 for "use tui", and -1 for "not specified" */ +static int check_tui_config(const char *cmd) +{ + struct pager_config c; + c.cmd = cmd; + c.val = -1; + perf_config(tui_command_config, &c); + return c.val; +} + +static void commit_pager_choice(void) +{ + switch (use_pager) { + case 0: + setenv("PERF_PAGER", "cat", 1); + break; + case 1: + /* setup_pager(); */ + break; + default: + break; + } +} + +static void set_debugfs_path(void) +{ + char *path; + + path = getenv(PERF_DEBUGFS_ENVIRONMENT); + snprintf(debugfs_path, MAXPATHLEN, "%s/%s", path ?: debugfs_mntpt, + "tracing/events"); +} + +static int handle_options(const char ***argv, int *argc, int *envchanged) +{ + int handled = 0; + + while (*argc > 0) { + const char *cmd = (*argv)[0]; + if (cmd[0] != '-') + break; + + /* + * For legacy reasons, the "version" and "help" + * commands can be written with "--" prepended + * to make them look like flags. + */ + if (!strcmp(cmd, "--help") || !strcmp(cmd, "--version")) + break; + + /* + * Check remaining flags. + */ + if (!prefixcmp(cmd, CMD_EXEC_PATH)) { + cmd += strlen(CMD_EXEC_PATH); + if (*cmd == '=') + perf_set_argv_exec_path(cmd + 1); + else { + puts(perf_exec_path()); + exit(0); + } + } else if (!strcmp(cmd, "--html-path")) { + puts(system_path(PERF_HTML_PATH)); + exit(0); + } else if (!strcmp(cmd, "-p") || !strcmp(cmd, "--paginate")) { + use_pager = 1; + } else if (!strcmp(cmd, "--no-pager")) { + use_pager = 0; + if (envchanged) + *envchanged = 1; + } else if (!strcmp(cmd, "--perf-dir")) { + if (*argc < 2) { + fprintf(stderr, "No directory given for --perf-dir.\n"); + usage(perf_usage_string); + } + setenv(PERF_DIR_ENVIRONMENT, (*argv)[1], 1); + if (envchanged) + *envchanged = 1; + (*argv)++; + (*argc)--; + handled++; + } else if (!prefixcmp(cmd, CMD_PERF_DIR)) { + setenv(PERF_DIR_ENVIRONMENT, cmd + strlen(CMD_PERF_DIR), 1); + if (envchanged) + *envchanged = 1; + } else if (!strcmp(cmd, "--work-tree")) { + if (*argc < 2) { + fprintf(stderr, "No directory given for --work-tree.\n"); + usage(perf_usage_string); + } + setenv(PERF_WORK_TREE_ENVIRONMENT, (*argv)[1], 1); + if (envchanged) + *envchanged = 1; + (*argv)++; + (*argc)--; + } else if (!prefixcmp(cmd, CMD_WORK_TREE)) { + setenv(PERF_WORK_TREE_ENVIRONMENT, cmd + strlen(CMD_WORK_TREE), 1); + if (envchanged) + *envchanged = 1; + } else if (!strcmp(cmd, "--debugfs-dir")) { + if (*argc < 2) { + fprintf(stderr, "No directory given for --debugfs-dir.\n"); + usage(perf_usage_string); + } + strncpy(debugfs_mntpt, (*argv)[1], MAXPATHLEN); + debugfs_mntpt[MAXPATHLEN - 1] = '\0'; + if (envchanged) + *envchanged = 1; + (*argv)++; + (*argc)--; + } else if (!prefixcmp(cmd, CMD_DEBUGFS_DIR)) { + strncpy(debugfs_mntpt, cmd + strlen(CMD_DEBUGFS_DIR), MAXPATHLEN); + debugfs_mntpt[MAXPATHLEN - 1] = '\0'; + if (envchanged) + *envchanged = 1; + } else { + fprintf(stderr, "Unknown option: %s\n", cmd); + usage(perf_usage_string); + } + + (*argv)++; + (*argc)--; + handled++; + } + return handled; +} + +static int handle_alias(int *argcp, const char ***argv) +{ + int envchanged = 0, ret = 0, saved_errno = errno; + int count, option_count; + const char **new_argv; + const char *alias_command; + char *alias_string; + + alias_command = (*argv)[0]; + alias_string = alias_lookup(alias_command); + if (alias_string) { + if (alias_string[0] == '!') { + if (*argcp > 1) { + struct strbuf buf; + + strbuf_init(&buf, PATH_MAX); + strbuf_addstr(&buf, alias_string); + sq_quote_argv(&buf, (*argv) + 1, PATH_MAX); + free(alias_string); + alias_string = buf.buf; + } + ret = system(alias_string + 1); + if (ret >= 0 && WIFEXITED(ret) && + WEXITSTATUS(ret) != 127) + exit(WEXITSTATUS(ret)); + die("Failed to run '%s' when expanding alias '%s'", + alias_string + 1, alias_command); + } + count = split_cmdline(alias_string, &new_argv); + if (count < 0) + die("Bad alias.%s string", alias_command); + option_count = handle_options(&new_argv, &count, &envchanged); + if (envchanged) + die("alias '%s' changes environment variables\n" + "You can use '!perf' in the alias to do this.", + alias_command); + memmove(new_argv - option_count, new_argv, + count * sizeof(char *)); + new_argv -= option_count; + + if (count < 1) + die("empty alias for %s", alias_command); + + if (!strcmp(alias_command, new_argv[0])) + die("recursive alias: %s", alias_command); + + new_argv = realloc(new_argv, sizeof(char *) * + (count + *argcp + 1)); + /* insert after command name */ + memcpy(new_argv + count, *argv + 1, sizeof(char *) * *argcp); + new_argv[count + *argcp] = NULL; + + *argv = new_argv; + *argcp += count - 1; + + ret = 1; + } + + errno = saved_errno; + + return ret; +} + +const char perf_version_string[] = PERF_VERSION; + +#define RUN_SETUP (1<<0) +#define USE_PAGER (1<<1) +/* + * require working tree to be present -- anything uses this needs + * RUN_SETUP for reading from the configuration file. + */ +#define NEED_WORK_TREE (1<<2) + +struct cmd_struct { + const char *cmd; + int (*fn)(int, const char **, const char *); + int option; +}; + +static int run_builtin(struct cmd_struct *p, int argc, const char **argv) +{ + int status; + struct stat st; + const char *prefix; + + prefix = NULL; + if (p->option & RUN_SETUP) + prefix = NULL; /* setup_perf_directory(); */ + + if (use_browser == -1) + use_browser = check_tui_config(p->cmd); + + if (use_pager == -1 && p->option & RUN_SETUP) + use_pager = check_pager_config(p->cmd); + if (use_pager == -1 && p->option & USE_PAGER) + use_pager = 1; + commit_pager_choice(); + set_debugfs_path(); + + status = p->fn(argc, argv, prefix); + exit_browser(status); + + if (status) + return status & 0xff; + + /* Somebody closed stdout? */ + if (fstat(fileno(stdout), &st)) + return 0; + /* Ignore write errors for pipes and sockets.. */ + if (S_ISFIFO(st.st_mode) || S_ISSOCK(st.st_mode)) + return 0; + + /* Check for ENOSPC and EIO errors.. */ + if (fflush(stdout)) + die("write failure on standard output: %s", strerror(errno)); + if (ferror(stdout)) + die("unknown write failure on standard output"); + if (fclose(stdout)) + die("close failed on standard output: %s", strerror(errno)); + return 0; +} + +static void handle_internal_command(int argc, const char **argv) +{ + const char *cmd = argv[0]; + static struct cmd_struct commands[] = { + { "buildid-cache", cmd_buildid_cache, 0 }, + { "buildid-list", cmd_buildid_list, 0 }, + { "diff", cmd_diff, 0 }, + { "evlist", cmd_evlist, 0 }, + { "help", cmd_help, 0 }, + { "list", cmd_list, 0 }, + { "record", cmd_record, 0 }, + { "report", cmd_report, 0 }, + { "bench", cmd_bench, 0 }, + { "stat", cmd_stat, 0 }, + { "timechart", cmd_timechart, 0 }, + { "top", cmd_top, 0 }, + { "annotate", cmd_annotate, 0 }, + { "version", cmd_version, 0 }, + { "script", cmd_script, 0 }, + { "sched", cmd_sched, 0 }, + { "probe", cmd_probe, 0 }, + { "kmem", cmd_kmem, 0 }, + { "lock", cmd_lock, 0 }, + { "kvm", cmd_kvm, 0 }, + { "test", cmd_test, 0 }, + { "inject", cmd_inject, 0 }, + }; + unsigned int i; + static const char ext[] = STRIP_EXTENSION; + + if (sizeof(ext) > 1) { + i = strlen(argv[0]) - strlen(ext); + if (i > 0 && !strcmp(argv[0] + i, ext)) { + char *argv0 = strdup(argv[0]); + argv[0] = cmd = argv0; + argv0[i] = '\0'; + } + } + + /* Turn "perf cmd --help" into "perf help cmd" */ + if (argc > 1 && !strcmp(argv[1], "--help")) { + argv[1] = argv[0]; + argv[0] = cmd = "help"; + } + + for (i = 0; i < ARRAY_SIZE(commands); i++) { + struct cmd_struct *p = commands+i; + if (strcmp(p->cmd, cmd)) + continue; + exit(run_builtin(p, argc, argv)); + } +} + +static void execv_dashed_external(const char **argv) +{ + struct strbuf cmd = STRBUF_INIT; + const char *tmp; + int status; + + strbuf_addf(&cmd, "perf-%s", argv[0]); + + /* + * argv[0] must be the perf command, but the argv array + * belongs to the caller, and may be reused in + * subsequent loop iterations. Save argv[0] and + * restore it on error. + */ + tmp = argv[0]; + argv[0] = cmd.buf; + + /* + * if we fail because the command is not found, it is + * OK to return. Otherwise, we just pass along the status code. + */ + status = run_command_v_opt(argv, 0); + if (status != -ERR_RUN_COMMAND_EXEC) { + if (IS_RUN_COMMAND_ERR(status)) + die("unable to run '%s'", argv[0]); + exit(-status); + } + errno = ENOENT; /* as if we called execvp */ + + argv[0] = tmp; + + strbuf_release(&cmd); +} + +static int run_argv(int *argcp, const char ***argv) +{ + int done_alias = 0; + + while (1) { + /* See if it's an internal command */ + handle_internal_command(*argcp, *argv); + + /* .. then try the external ones */ + execv_dashed_external(*argv); + + /* It could be an alias -- this works around the insanity + * of overriding "perf log" with "perf show" by having + * alias.log = show + */ + if (done_alias || !handle_alias(argcp, argv)) + break; + done_alias = 1; + } + + return done_alias; +} + +/* mini /proc/mounts parser: searching for "^blah /mount/point debugfs" */ +static void get_debugfs_mntpt(void) +{ + const char *path = debugfs_mount(NULL); + + if (path) + strncpy(debugfs_mntpt, path, sizeof(debugfs_mntpt)); + else + debugfs_mntpt[0] = '\0'; +} + +int main(int argc, const char **argv) +{ + const char *cmd; + + cmd = perf_extract_argv0_path(argv[0]); + if (!cmd) + cmd = "perf-help"; + /* get debugfs mount point from /proc/mounts */ + get_debugfs_mntpt(); + /* + * "perf-xxxx" is the same as "perf xxxx", but we obviously: + * + * - cannot take flags in between the "perf" and the "xxxx". + * - cannot execute it externally (since it would just do + * the same thing over again) + * + * So we just directly call the internal command handler, and + * die if that one cannot handle it. + */ + if (!prefixcmp(cmd, "perf-")) { + cmd += 5; + argv[0] = cmd; + handle_internal_command(argc, argv); + die("cannot handle %s internally", cmd); + } + + /* Look for flags.. */ + argv++; + argc--; + handle_options(&argv, &argc, NULL); + commit_pager_choice(); + set_debugfs_path(); + set_buildid_dir(); + + if (argc > 0) { + if (!prefixcmp(argv[0], "--")) + argv[0] += 2; + } else { + /* The user didn't specify a command; give them help */ + printf("\n usage: %s\n\n", perf_usage_string); + list_common_cmds_help(); + printf("\n %s\n\n", perf_more_info_string); + exit(1); + } + cmd = argv[0]; + + /* + * We use PATH to find perf commands, but we prepend some higher + * precedence paths: the "--exec-path" option, the PERF_EXEC_PATH + * environment, and the $(perfexecdir) from the Makefile at build + * time. + */ + setup_path(); + + while (1) { + static int done_help; + static int was_alias; + + was_alias = run_argv(&argc, &argv); + if (errno != ENOENT) + break; + + if (was_alias) { + fprintf(stderr, "Expansion of alias '%s' failed; " + "'%s' is not a perf-command\n", + cmd, argv[0]); + exit(1); + } + if (!done_help) { + cmd = argv[0] = help_unknown_cmd(cmd); + done_help = 1; + } else + break; + } + + fprintf(stderr, "Failed to run command '%s': %s\n", + cmd, strerror(errno)); + + return 1; +} diff --git a/tools/perf/perf.h b/tools/perf/perf.h new file mode 100644 index 00000000..a5fc660c --- /dev/null +++ b/tools/perf/perf.h @@ -0,0 +1,175 @@ +#ifndef _PERF_PERF_H +#define _PERF_PERF_H + +struct winsize; + +void get_term_dimensions(struct winsize *ws); + +#if defined(__i386__) +#include "../../arch/x86/include/asm/unistd.h" +#define rmb() asm volatile("lock; addl $0,0(%%esp)" ::: "memory") +#define cpu_relax() asm volatile("rep; nop" ::: "memory"); +#endif + +#if defined(__x86_64__) +#include "../../arch/x86/include/asm/unistd.h" +#define rmb() asm volatile("lfence" ::: "memory") +#define cpu_relax() asm volatile("rep; nop" ::: "memory"); +#endif + +#ifdef __powerpc__ +#include "../../arch/powerpc/include/asm/unistd.h" +#define rmb() asm volatile ("sync" ::: "memory") +#define cpu_relax() asm volatile ("" ::: "memory"); +#endif + +#ifdef __s390__ +#include "../../arch/s390/include/asm/unistd.h" +#define rmb() asm volatile("bcr 15,0" ::: "memory") +#define cpu_relax() asm volatile("" ::: "memory"); +#endif + +#ifdef __sh__ +#include "../../arch/sh/include/asm/unistd.h" +#if defined(__SH4A__) || defined(__SH5__) +# define rmb() asm volatile("synco" ::: "memory") +#else +# define rmb() asm volatile("" ::: "memory") +#endif +#define cpu_relax() asm volatile("" ::: "memory") +#endif + +#ifdef __hppa__ +#include "../../arch/parisc/include/asm/unistd.h" +#define rmb() asm volatile("" ::: "memory") +#define cpu_relax() asm volatile("" ::: "memory"); +#endif + +#ifdef __sparc__ +#include "../../arch/sparc/include/asm/unistd.h" +#define rmb() asm volatile("":::"memory") +#define cpu_relax() asm volatile("":::"memory") +#endif + +#ifdef __alpha__ +#include "../../arch/alpha/include/asm/unistd.h" +#define rmb() asm volatile("mb" ::: "memory") +#define cpu_relax() asm volatile("" ::: "memory") +#endif + +#ifdef __ia64__ +#include "../../arch/ia64/include/asm/unistd.h" +#define rmb() asm volatile ("mf" ::: "memory") +#define cpu_relax() asm volatile ("hint @pause" ::: "memory") +#endif + +#ifdef __arm__ +#include "../../arch/arm/include/asm/unistd.h" +/* + * Use the __kuser_memory_barrier helper in the CPU helper page. See + * arch/arm/kernel/entry-armv.S in the kernel source for details. + */ +#define rmb() ((void(*)(void))0xffff0fa0)() +#define cpu_relax() asm volatile("":::"memory") +#endif + +#ifdef __mips__ +#include "../../arch/mips/include/asm/unistd.h" +#define rmb() asm volatile( \ + ".set mips2\n\t" \ + "sync\n\t" \ + ".set mips0" \ + : /* no output */ \ + : /* no input */ \ + : "memory") +#define cpu_relax() asm volatile("" ::: "memory") +#endif + +#include <time.h> +#include <unistd.h> +#include <sys/types.h> +#include <sys/syscall.h> + +#include "../../include/linux/perf_event.h" +#include "util/types.h" +#include <stdbool.h> + +struct perf_mmap { + void *base; + int mask; + unsigned int prev; +}; + +static inline unsigned int perf_mmap__read_head(struct perf_mmap *mm) +{ + struct perf_event_mmap_page *pc = mm->base; + int head = pc->data_head; + rmb(); + return head; +} + +static inline void perf_mmap__write_tail(struct perf_mmap *md, + unsigned long tail) +{ + struct perf_event_mmap_page *pc = md->base; + + /* + * ensure all reads are done before we write the tail out. + */ + /* mb(); */ + pc->data_tail = tail; +} + +/* + * prctl(PR_TASK_PERF_EVENTS_DISABLE) will (cheaply) disable all + * counters in the current task. + */ +#define PR_TASK_PERF_EVENTS_DISABLE 31 +#define PR_TASK_PERF_EVENTS_ENABLE 32 + +#ifndef NSEC_PER_SEC +# define NSEC_PER_SEC 1000000000ULL +#endif + +static inline unsigned long long rdclock(void) +{ + struct timespec ts; + + clock_gettime(CLOCK_MONOTONIC, &ts); + return ts.tv_sec * 1000000000ULL + ts.tv_nsec; +} + +/* + * Pick up some kernel type conventions: + */ +#define __user +#define asmlinkage + +#define unlikely(x) __builtin_expect(!!(x), 0) +#define min(x, y) ({ \ + typeof(x) _min1 = (x); \ + typeof(y) _min2 = (y); \ + (void) (&_min1 == &_min2); \ + _min1 < _min2 ? _min1 : _min2; }) + +static inline int +sys_perf_event_open(struct perf_event_attr *attr, + pid_t pid, int cpu, int group_fd, + unsigned long flags) +{ + attr->size = sizeof(*attr); + return syscall(__NR_perf_event_open, attr, pid, cpu, + group_fd, flags); +} + +#define MAX_COUNTERS 256 +#define MAX_NR_CPUS 256 + +struct ip_callchain { + u64 nr; + u64 ips[0]; +}; + +extern bool perf_host, perf_guest; + +#endif diff --git a/tools/perf/python/twatch.py b/tools/perf/python/twatch.py new file mode 100755 index 00000000..df638c43 --- /dev/null +++ b/tools/perf/python/twatch.py @@ -0,0 +1,41 @@ +#! /usr/bin/python +# -*- python -*- +# -*- coding: utf-8 -*- +# twatch - Experimental use of the perf python interface +# Copyright (C) 2011 Arnaldo Carvalho de Melo <acme@redhat.com> +# +# This application is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License +# as published by the Free Software Foundation; version 2. +# +# This application is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# General Public License for more details. + +import perf + +def main(): + cpus = perf.cpu_map() + threads = perf.thread_map() + evsel = perf.evsel(task = 1, comm = 1, mmap = 0, + wakeup_events = 1, sample_period = 1, + sample_id_all = 1, + sample_type = perf.SAMPLE_PERIOD | perf.SAMPLE_TID | perf.SAMPLE_CPU | perf.SAMPLE_TID) + evsel.open(cpus = cpus, threads = threads); + evlist = perf.evlist(cpus, threads) + evlist.add(evsel) + evlist.mmap() + while True: + evlist.poll(timeout = -1) + for cpu in cpus: + event = evlist.read_on_cpu(cpu) + if not event: + continue + print "cpu: %2d, pid: %4d, tid: %4d" % (event.sample_cpu, + event.sample_pid, + event.sample_tid), + print event + +if __name__ == '__main__': + main() diff --git a/tools/perf/scripts/perl/Perf-Trace-Util/Context.c b/tools/perf/scripts/perl/Perf-Trace-Util/Context.c new file mode 100644 index 00000000..790ceba6 --- /dev/null +++ b/tools/perf/scripts/perl/Perf-Trace-Util/Context.c @@ -0,0 +1,135 @@ +/* + * This file was generated automatically by ExtUtils::ParseXS version 2.18_02 from the + * contents of Context.xs. Do not edit this file, edit Context.xs instead. + * + * ANY CHANGES MADE HERE WILL BE LOST! + * + */ + +#line 1 "Context.xs" +/* + * Context.xs. XS interfaces for perf script. + * + * Copyright (C) 2009 Tom Zanussi <tzanussi@gmail.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#include "EXTERN.h" +#include "perl.h" +#include "XSUB.h" +#include "../../../perf.h" +#include "../../../util/trace-event.h" + +#ifndef PERL_UNUSED_VAR +# define PERL_UNUSED_VAR(var) if (0) var = var +#endif + +#line 42 "Context.c" + +XS(XS_Perf__Trace__Context_common_pc); /* prototype to pass -Wmissing-prototypes */ +XS(XS_Perf__Trace__Context_common_pc) +{ +#ifdef dVAR + dVAR; dXSARGS; +#else + dXSARGS; +#endif + if (items != 1) + Perl_croak(aTHX_ "Usage: %s(%s)", "Perf::Trace::Context::common_pc", "context"); + PERL_UNUSED_VAR(cv); /* -W */ + { + struct scripting_context * context = INT2PTR(struct scripting_context *,SvIV(ST(0))); + int RETVAL; + dXSTARG; + + RETVAL = common_pc(context); + XSprePUSH; PUSHi((IV)RETVAL); + } + XSRETURN(1); +} + + +XS(XS_Perf__Trace__Context_common_flags); /* prototype to pass -Wmissing-prototypes */ +XS(XS_Perf__Trace__Context_common_flags) +{ +#ifdef dVAR + dVAR; dXSARGS; +#else + dXSARGS; +#endif + if (items != 1) + Perl_croak(aTHX_ "Usage: %s(%s)", "Perf::Trace::Context::common_flags", "context"); + PERL_UNUSED_VAR(cv); /* -W */ + { + struct scripting_context * context = INT2PTR(struct scripting_context *,SvIV(ST(0))); + int RETVAL; + dXSTARG; + + RETVAL = common_flags(context); + XSprePUSH; PUSHi((IV)RETVAL); + } + XSRETURN(1); +} + + +XS(XS_Perf__Trace__Context_common_lock_depth); /* prototype to pass -Wmissing-prototypes */ +XS(XS_Perf__Trace__Context_common_lock_depth) +{ +#ifdef dVAR + dVAR; dXSARGS; +#else + dXSARGS; +#endif + if (items != 1) + Perl_croak(aTHX_ "Usage: %s(%s)", "Perf::Trace::Context::common_lock_depth", "context"); + PERL_UNUSED_VAR(cv); /* -W */ + { + struct scripting_context * context = INT2PTR(struct scripting_context *,SvIV(ST(0))); + int RETVAL; + dXSTARG; + + RETVAL = common_lock_depth(context); + XSprePUSH; PUSHi((IV)RETVAL); + } + XSRETURN(1); +} + +#ifdef __cplusplus +extern "C" +#endif +XS(boot_Perf__Trace__Context); /* prototype to pass -Wmissing-prototypes */ +XS(boot_Perf__Trace__Context) +{ +#ifdef dVAR + dVAR; dXSARGS; +#else + dXSARGS; +#endif + const char* file = __FILE__; + + PERL_UNUSED_VAR(cv); /* -W */ + PERL_UNUSED_VAR(items); /* -W */ + XS_VERSION_BOOTCHECK ; + + newXSproto("Perf::Trace::Context::common_pc", XS_Perf__Trace__Context_common_pc, file, "$"); + newXSproto("Perf::Trace::Context::common_flags", XS_Perf__Trace__Context_common_flags, file, "$"); + newXSproto("Perf::Trace::Context::common_lock_depth", XS_Perf__Trace__Context_common_lock_depth, file, "$"); + if (PL_unitcheckav) + call_list(PL_scopestack_ix, PL_unitcheckav); + XSRETURN_YES; +} + diff --git a/tools/perf/scripts/perl/Perf-Trace-Util/Context.xs b/tools/perf/scripts/perl/Perf-Trace-Util/Context.xs new file mode 100644 index 00000000..c1e2ed1e --- /dev/null +++ b/tools/perf/scripts/perl/Perf-Trace-Util/Context.xs @@ -0,0 +1,42 @@ +/* + * Context.xs. XS interfaces for perf script. + * + * Copyright (C) 2009 Tom Zanussi <tzanussi@gmail.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#include "EXTERN.h" +#include "perl.h" +#include "XSUB.h" +#include "../../../perf.h" +#include "../../../util/script-event.h" + +MODULE = Perf::Trace::Context PACKAGE = Perf::Trace::Context +PROTOTYPES: ENABLE + +int +common_pc(context) + struct scripting_context * context + +int +common_flags(context) + struct scripting_context * context + +int +common_lock_depth(context) + struct scripting_context * context + diff --git a/tools/perf/scripts/perl/Perf-Trace-Util/Makefile.PL b/tools/perf/scripts/perl/Perf-Trace-Util/Makefile.PL new file mode 100644 index 00000000..decdeb0f --- /dev/null +++ b/tools/perf/scripts/perl/Perf-Trace-Util/Makefile.PL @@ -0,0 +1,17 @@ +use 5.010000; +use ExtUtils::MakeMaker; +# See lib/ExtUtils/MakeMaker.pm for details of how to influence +# the contents of the Makefile that is written. +WriteMakefile( + NAME => 'Perf::Trace::Context', + VERSION_FROM => 'lib/Perf/Trace/Context.pm', # finds $VERSION + PREREQ_PM => {}, # e.g., Module::Name => 1.1 + ($] >= 5.005 ? ## Add these new keywords supported since 5.005 + (ABSTRACT_FROM => 'lib/Perf/Trace/Context.pm', # retrieve abstract from module + AUTHOR => 'Tom Zanussi <tzanussi@gmail.com>') : ()), + LIBS => [''], # e.g., '-lm' + DEFINE => '-I ../..', # e.g., '-DHAVE_SOMETHING' + INC => '-I.', # e.g., '-I. -I/usr/include/other' + # Un-comment this if you add C files to link with later: + OBJECT => 'Context.o', # link all the C files too +); diff --git a/tools/perf/scripts/perl/Perf-Trace-Util/README b/tools/perf/scripts/perl/Perf-Trace-Util/README new file mode 100644 index 00000000..2f0c7f30 --- /dev/null +++ b/tools/perf/scripts/perl/Perf-Trace-Util/README @@ -0,0 +1,59 @@ +Perf-Trace-Util version 0.01 +============================ + +This module contains utility functions for use with perf script. + +Core.pm and Util.pm are pure Perl modules; Core.pm contains routines +that the core perf support for Perl calls on and should always be +'used', while Util.pm contains useful but optional utility functions +that scripts may want to use. Context.pm contains the Perl->C +interface that allows scripts to access data in the embedding perf +executable; scripts wishing to do that should 'use Context.pm'. + +The Perl->C perf interface is completely driven by Context.xs. If you +want to add new Perl functions that end up accessing C data in the +perf executable, you add desciptions of the new functions here. +scripting_context is a pointer to the perf data in the perf executable +that you want to access - it's passed as the second parameter, +$context, to all handler functions. + +After you do that: + + perl Makefile.PL # to create a Makefile for the next step + make # to create Context.c + + edit Context.c to add const to the char* file = __FILE__ line in + XS(boot_Perf__Trace__Context) to silence a warning/error. + + You can delete the Makefile, object files and anything else that was + generated e.g. blib and shared library, etc, except for of course + Context.c + + You should then be able to run the normal perf make as usual. + +INSTALLATION + +Building perf with perf script Perl scripting should install this +module in the right place. + +You should make sure libperl and ExtUtils/Embed.pm are installed first +e.g. apt-get install libperl-dev or yum install perl-ExtUtils-Embed. + +DEPENDENCIES + +This module requires these other modules and libraries: + + None + +COPYRIGHT AND LICENCE + +Copyright (C) 2009 by Tom Zanussi <tzanussi@gmail.com> + +This library is free software; you can redistribute it and/or modify +it under the same terms as Perl itself, either Perl version 5.10.0 or, +at your option, any later version of Perl 5 you may have available. + +Alternatively, this software may be distributed under the terms of the +GNU General Public License ("GPL") version 2 as published by the Free +Software Foundation. + diff --git a/tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Context.pm b/tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Context.pm new file mode 100644 index 00000000..4e2f6039 --- /dev/null +++ b/tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Context.pm @@ -0,0 +1,55 @@ +package Perf::Trace::Context; + +use 5.010000; +use strict; +use warnings; + +require Exporter; + +our @ISA = qw(Exporter); + +our %EXPORT_TAGS = ( 'all' => [ qw( +) ] ); + +our @EXPORT_OK = ( @{ $EXPORT_TAGS{'all'} } ); + +our @EXPORT = qw( + common_pc common_flags common_lock_depth +); + +our $VERSION = '0.01'; + +require XSLoader; +XSLoader::load('Perf::Trace::Context', $VERSION); + +1; +__END__ +=head1 NAME + +Perf::Trace::Context - Perl extension for accessing functions in perf. + +=head1 SYNOPSIS + + use Perf::Trace::Context; + +=head1 SEE ALSO + +Perf (script) documentation + +=head1 AUTHOR + +Tom Zanussi, E<lt>tzanussi@gmail.com<gt> + +=head1 COPYRIGHT AND LICENSE + +Copyright (C) 2009 by Tom Zanussi + +This library is free software; you can redistribute it and/or modify +it under the same terms as Perl itself, either Perl version 5.10.0 or, +at your option, any later version of Perl 5 you may have available. + +Alternatively, this software may be distributed under the terms of the +GNU General Public License ("GPL") version 2 as published by the Free +Software Foundation. + +=cut diff --git a/tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Core.pm b/tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Core.pm new file mode 100644 index 00000000..9158458d --- /dev/null +++ b/tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Core.pm @@ -0,0 +1,192 @@ +package Perf::Trace::Core; + +use 5.010000; +use strict; +use warnings; + +require Exporter; + +our @ISA = qw(Exporter); + +our %EXPORT_TAGS = ( 'all' => [ qw( +) ] ); + +our @EXPORT_OK = ( @{ $EXPORT_TAGS{'all'} } ); + +our @EXPORT = qw( +define_flag_field define_flag_value flag_str dump_flag_fields +define_symbolic_field define_symbolic_value symbol_str dump_symbolic_fields +trace_flag_str +); + +our $VERSION = '0.01'; + +my %trace_flags = (0x00 => "NONE", + 0x01 => "IRQS_OFF", + 0x02 => "IRQS_NOSUPPORT", + 0x04 => "NEED_RESCHED", + 0x08 => "HARDIRQ", + 0x10 => "SOFTIRQ"); + +sub trace_flag_str +{ + my ($value) = @_; + + my $string; + + my $print_delim = 0; + + foreach my $idx (sort {$a <=> $b} keys %trace_flags) { + if (!$value && !$idx) { + $string .= "NONE"; + last; + } + + if ($idx && ($value & $idx) == $idx) { + if ($print_delim) { + $string .= " | "; + } + $string .= "$trace_flags{$idx}"; + $print_delim = 1; + $value &= ~$idx; + } + } + + return $string; +} + +my %flag_fields; +my %symbolic_fields; + +sub flag_str +{ + my ($event_name, $field_name, $value) = @_; + + my $string; + + if ($flag_fields{$event_name}{$field_name}) { + my $print_delim = 0; + foreach my $idx (sort {$a <=> $b} keys %{$flag_fields{$event_name}{$field_name}{"values"}}) { + if (!$value && !$idx) { + $string .= "$flag_fields{$event_name}{$field_name}{'values'}{$idx}"; + last; + } + if ($idx && ($value & $idx) == $idx) { + if ($print_delim && $flag_fields{$event_name}{$field_name}{'delim'}) { + $string .= " $flag_fields{$event_name}{$field_name}{'delim'} "; + } + $string .= "$flag_fields{$event_name}{$field_name}{'values'}{$idx}"; + $print_delim = 1; + $value &= ~$idx; + } + } + } + + return $string; +} + +sub define_flag_field +{ + my ($event_name, $field_name, $delim) = @_; + + $flag_fields{$event_name}{$field_name}{"delim"} = $delim; +} + +sub define_flag_value +{ + my ($event_name, $field_name, $value, $field_str) = @_; + + $flag_fields{$event_name}{$field_name}{"values"}{$value} = $field_str; +} + +sub dump_flag_fields +{ + for my $event (keys %flag_fields) { + print "event $event:\n"; + for my $field (keys %{$flag_fields{$event}}) { + print " field: $field:\n"; + print " delim: $flag_fields{$event}{$field}{'delim'}\n"; + foreach my $idx (sort {$a <=> $b} keys %{$flag_fields{$event}{$field}{"values"}}) { + print " value $idx: $flag_fields{$event}{$field}{'values'}{$idx}\n"; + } + } + } +} + +sub symbol_str +{ + my ($event_name, $field_name, $value) = @_; + + if ($symbolic_fields{$event_name}{$field_name}) { + foreach my $idx (sort {$a <=> $b} keys %{$symbolic_fields{$event_name}{$field_name}{"values"}}) { + if (!$value && !$idx) { + return "$symbolic_fields{$event_name}{$field_name}{'values'}{$idx}"; + last; + } + if ($value == $idx) { + return "$symbolic_fields{$event_name}{$field_name}{'values'}{$idx}"; + } + } + } + + return undef; +} + +sub define_symbolic_field +{ + my ($event_name, $field_name) = @_; + + # nothing to do, really +} + +sub define_symbolic_value +{ + my ($event_name, $field_name, $value, $field_str) = @_; + + $symbolic_fields{$event_name}{$field_name}{"values"}{$value} = $field_str; +} + +sub dump_symbolic_fields +{ + for my $event (keys %symbolic_fields) { + print "event $event:\n"; + for my $field (keys %{$symbolic_fields{$event}}) { + print " field: $field:\n"; + foreach my $idx (sort {$a <=> $b} keys %{$symbolic_fields{$event}{$field}{"values"}}) { + print " value $idx: $symbolic_fields{$event}{$field}{'values'}{$idx}\n"; + } + } + } +} + +1; +__END__ +=head1 NAME + +Perf::Trace::Core - Perl extension for perf script + +=head1 SYNOPSIS + + use Perf::Trace::Core + +=head1 SEE ALSO + +Perf (script) documentation + +=head1 AUTHOR + +Tom Zanussi, E<lt>tzanussi@gmail.com<gt> + +=head1 COPYRIGHT AND LICENSE + +Copyright (C) 2009 by Tom Zanussi + +This library is free software; you can redistribute it and/or modify +it under the same terms as Perl itself, either Perl version 5.10.0 or, +at your option, any later version of Perl 5 you may have available. + +Alternatively, this software may be distributed under the terms of the +GNU General Public License ("GPL") version 2 as published by the Free +Software Foundation. + +=cut diff --git a/tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Util.pm b/tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Util.pm new file mode 100644 index 00000000..05350011 --- /dev/null +++ b/tools/perf/scripts/perl/Perf-Trace-Util/lib/Perf/Trace/Util.pm @@ -0,0 +1,94 @@ +package Perf::Trace::Util; + +use 5.010000; +use strict; +use warnings; + +require Exporter; + +our @ISA = qw(Exporter); + +our %EXPORT_TAGS = ( 'all' => [ qw( +) ] ); + +our @EXPORT_OK = ( @{ $EXPORT_TAGS{'all'} } ); + +our @EXPORT = qw( +avg nsecs nsecs_secs nsecs_nsecs nsecs_usecs print_nsecs +clear_term +); + +our $VERSION = '0.01'; + +sub avg +{ + my ($total, $n) = @_; + + return $total / $n; +} + +my $NSECS_PER_SEC = 1000000000; + +sub nsecs +{ + my ($secs, $nsecs) = @_; + + return $secs * $NSECS_PER_SEC + $nsecs; +} + +sub nsecs_secs { + my ($nsecs) = @_; + + return $nsecs / $NSECS_PER_SEC; +} + +sub nsecs_nsecs { + my ($nsecs) = @_; + + return $nsecs % $NSECS_PER_SEC; +} + +sub nsecs_str { + my ($nsecs) = @_; + + my $str = sprintf("%5u.%09u", nsecs_secs($nsecs), nsecs_nsecs($nsecs)); + + return $str; +} + +sub clear_term +{ + print "\x1b[H\x1b[2J"; +} + +1; +__END__ +=head1 NAME + +Perf::Trace::Util - Perl extension for perf script + +=head1 SYNOPSIS + + use Perf::Trace::Util; + +=head1 SEE ALSO + +Perf (script) documentation + +=head1 AUTHOR + +Tom Zanussi, E<lt>tzanussi@gmail.com<gt> + +=head1 COPYRIGHT AND LICENSE + +Copyright (C) 2009 by Tom Zanussi + +This library is free software; you can redistribute it and/or modify +it under the same terms as Perl itself, either Perl version 5.10.0 or, +at your option, any later version of Perl 5 you may have available. + +Alternatively, this software may be distributed under the terms of the +GNU General Public License ("GPL") version 2 as published by the Free +Software Foundation. + +=cut diff --git a/tools/perf/scripts/perl/Perf-Trace-Util/typemap b/tools/perf/scripts/perl/Perf-Trace-Util/typemap new file mode 100644 index 00000000..84083680 --- /dev/null +++ b/tools/perf/scripts/perl/Perf-Trace-Util/typemap @@ -0,0 +1 @@ +struct scripting_context * T_PTR diff --git a/tools/perf/scripts/perl/bin/check-perf-trace-record b/tools/perf/scripts/perl/bin/check-perf-trace-record new file mode 100644 index 00000000..423ad6ae --- /dev/null +++ b/tools/perf/scripts/perl/bin/check-perf-trace-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -a -e kmem:kmalloc -e irq:softirq_entry -e kmem:kfree diff --git a/tools/perf/scripts/perl/bin/failed-syscalls-record b/tools/perf/scripts/perl/bin/failed-syscalls-record new file mode 100644 index 00000000..8104895a --- /dev/null +++ b/tools/perf/scripts/perl/bin/failed-syscalls-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -e raw_syscalls:sys_exit $@ diff --git a/tools/perf/scripts/perl/bin/failed-syscalls-report b/tools/perf/scripts/perl/bin/failed-syscalls-report new file mode 100644 index 00000000..9f83cc1a --- /dev/null +++ b/tools/perf/scripts/perl/bin/failed-syscalls-report @@ -0,0 +1,10 @@ +#!/bin/bash +# description: system-wide failed syscalls +# args: [comm] +if [ $# -gt 0 ] ; then + if ! expr match "$1" "-" > /dev/null ; then + comm=$1 + shift + fi +fi +perf script $@ -s "$PERF_EXEC_PATH"/scripts/perl/failed-syscalls.pl $comm diff --git a/tools/perf/scripts/perl/bin/rw-by-file-record b/tools/perf/scripts/perl/bin/rw-by-file-record new file mode 100644 index 00000000..33efc867 --- /dev/null +++ b/tools/perf/scripts/perl/bin/rw-by-file-record @@ -0,0 +1,3 @@ +#!/bin/bash +perf record -e syscalls:sys_enter_read -e syscalls:sys_enter_write $@ + diff --git a/tools/perf/scripts/perl/bin/rw-by-file-report b/tools/perf/scripts/perl/bin/rw-by-file-report new file mode 100644 index 00000000..77200b3f --- /dev/null +++ b/tools/perf/scripts/perl/bin/rw-by-file-report @@ -0,0 +1,10 @@ +#!/bin/bash +# description: r/w activity for a program, by file +# args: <comm> +if [ $# -lt 1 ] ; then + echo "usage: rw-by-file <comm>" + exit +fi +comm=$1 +shift +perf script $@ -s "$PERF_EXEC_PATH"/scripts/perl/rw-by-file.pl $comm diff --git a/tools/perf/scripts/perl/bin/rw-by-pid-record b/tools/perf/scripts/perl/bin/rw-by-pid-record new file mode 100644 index 00000000..7cb9db23 --- /dev/null +++ b/tools/perf/scripts/perl/bin/rw-by-pid-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -e syscalls:sys_enter_read -e syscalls:sys_exit_read -e syscalls:sys_enter_write -e syscalls:sys_exit_write $@ diff --git a/tools/perf/scripts/perl/bin/rw-by-pid-report b/tools/perf/scripts/perl/bin/rw-by-pid-report new file mode 100644 index 00000000..a27b9f31 --- /dev/null +++ b/tools/perf/scripts/perl/bin/rw-by-pid-report @@ -0,0 +1,3 @@ +#!/bin/bash +# description: system-wide r/w activity +perf script $@ -s "$PERF_EXEC_PATH"/scripts/perl/rw-by-pid.pl diff --git a/tools/perf/scripts/perl/bin/rwtop-record b/tools/perf/scripts/perl/bin/rwtop-record new file mode 100644 index 00000000..7cb9db23 --- /dev/null +++ b/tools/perf/scripts/perl/bin/rwtop-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -e syscalls:sys_enter_read -e syscalls:sys_exit_read -e syscalls:sys_enter_write -e syscalls:sys_exit_write $@ diff --git a/tools/perf/scripts/perl/bin/rwtop-report b/tools/perf/scripts/perl/bin/rwtop-report new file mode 100644 index 00000000..83e11ec2 --- /dev/null +++ b/tools/perf/scripts/perl/bin/rwtop-report @@ -0,0 +1,20 @@ +#!/bin/bash +# description: system-wide r/w top +# args: [interval] +n_args=0 +for i in "$@" +do + if expr match "$i" "-" > /dev/null ; then + break + fi + n_args=$(( $n_args + 1 )) +done +if [ "$n_args" -gt 1 ] ; then + echo "usage: rwtop-report [interval]" + exit +fi +if [ "$n_args" -gt 0 ] ; then + interval=$1 + shift +fi +perf script $@ -s "$PERF_EXEC_PATH"/scripts/perl/rwtop.pl $interval diff --git a/tools/perf/scripts/perl/bin/wakeup-latency-record b/tools/perf/scripts/perl/bin/wakeup-latency-record new file mode 100644 index 00000000..464251a1 --- /dev/null +++ b/tools/perf/scripts/perl/bin/wakeup-latency-record @@ -0,0 +1,6 @@ +#!/bin/bash +perf record -e sched:sched_switch -e sched:sched_wakeup $@ + + + + diff --git a/tools/perf/scripts/perl/bin/wakeup-latency-report b/tools/perf/scripts/perl/bin/wakeup-latency-report new file mode 100644 index 00000000..889e8130 --- /dev/null +++ b/tools/perf/scripts/perl/bin/wakeup-latency-report @@ -0,0 +1,3 @@ +#!/bin/bash +# description: system-wide min/max/avg wakeup latency +perf script $@ -s "$PERF_EXEC_PATH"/scripts/perl/wakeup-latency.pl diff --git a/tools/perf/scripts/perl/bin/workqueue-stats-record b/tools/perf/scripts/perl/bin/workqueue-stats-record new file mode 100644 index 00000000..8edda907 --- /dev/null +++ b/tools/perf/scripts/perl/bin/workqueue-stats-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -e workqueue:workqueue_creation -e workqueue:workqueue_destruction -e workqueue:workqueue_execution -e workqueue:workqueue_insertion $@ diff --git a/tools/perf/scripts/perl/bin/workqueue-stats-report b/tools/perf/scripts/perl/bin/workqueue-stats-report new file mode 100644 index 00000000..6d91411d --- /dev/null +++ b/tools/perf/scripts/perl/bin/workqueue-stats-report @@ -0,0 +1,3 @@ +#!/bin/bash +# description: workqueue stats (ins/exe/create/destroy) +perf script $@ -s "$PERF_EXEC_PATH"/scripts/perl/workqueue-stats.pl diff --git a/tools/perf/scripts/perl/check-perf-trace.pl b/tools/perf/scripts/perl/check-perf-trace.pl new file mode 100644 index 00000000..4e7076c2 --- /dev/null +++ b/tools/perf/scripts/perl/check-perf-trace.pl @@ -0,0 +1,106 @@ +# perf script event handlers, generated by perf script -g perl +# (c) 2009, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 + +# This script tests basic functionality such as flag and symbol +# strings, common_xxx() calls back into perf, begin, end, unhandled +# events, etc. Basically, if this script runs successfully and +# displays expected results, perl scripting support should be ok. + +use lib "$ENV{'PERF_EXEC_PATH'}/scripts/perl/Perf-Trace-Util/lib"; +use lib "./Perf-Trace-Util/lib"; +use Perf::Trace::Core; +use Perf::Trace::Context; +use Perf::Trace::Util; + +sub trace_begin +{ + print "trace_begin\n"; +} + +sub trace_end +{ + print "trace_end\n"; + + print_unhandled(); +} + +sub irq::softirq_entry +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $vec) = @_; + + print_header($event_name, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm); + + print_uncommon($context); + + printf("vec=%s\n", + symbol_str("irq::softirq_entry", "vec", $vec)); +} + +sub kmem::kmalloc +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $call_site, $ptr, $bytes_req, $bytes_alloc, + $gfp_flags) = @_; + + print_header($event_name, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm); + + print_uncommon($context); + + printf("call_site=%p, ptr=%p, bytes_req=%u, bytes_alloc=%u, ". + "gfp_flags=%s\n", + $call_site, $ptr, $bytes_req, $bytes_alloc, + + flag_str("kmem::kmalloc", "gfp_flags", $gfp_flags)); +} + +# print trace fields not included in handler args +sub print_uncommon +{ + my ($context) = @_; + + printf("common_preempt_count=%d, common_flags=%s, common_lock_depth=%d, ", + common_pc($context), trace_flag_str(common_flags($context)), + common_lock_depth($context)); + +} + +my %unhandled; + +sub print_unhandled +{ + if ((scalar keys %unhandled) == 0) { + return; + } + + print "\nunhandled events:\n\n"; + + printf("%-40s %10s\n", "event", "count"); + printf("%-40s %10s\n", "----------------------------------------", + "-----------"); + + foreach my $event_name (keys %unhandled) { + printf("%-40s %10d\n", $event_name, $unhandled{$event_name}); + } +} + +sub trace_unhandled +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm) = @_; + + $unhandled{$event_name}++; +} + +sub print_header +{ + my ($event_name, $cpu, $secs, $nsecs, $pid, $comm) = @_; + + printf("%-20s %5u %05u.%09u %8u %-20s ", + $event_name, $cpu, $secs, $nsecs, $pid, $comm); +} diff --git a/tools/perf/scripts/perl/failed-syscalls.pl b/tools/perf/scripts/perl/failed-syscalls.pl new file mode 100644 index 00000000..94bc25a3 --- /dev/null +++ b/tools/perf/scripts/perl/failed-syscalls.pl @@ -0,0 +1,42 @@ +# failed system call counts +# (c) 2010, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 +# +# Displays system-wide failed system call totals +# If a [comm] arg is specified, only syscalls called by [comm] are displayed. + +use lib "$ENV{'PERF_EXEC_PATH'}/scripts/perl/Perf-Trace-Util/lib"; +use lib "./Perf-Trace-Util/lib"; +use Perf::Trace::Core; +use Perf::Trace::Context; +use Perf::Trace::Util; + +my $for_comm = shift; + +my %failed_syscalls; + +sub raw_syscalls::sys_exit +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $id, $ret) = @_; + + if ($ret < 0) { + $failed_syscalls{$common_comm}++; + } +} + +sub trace_end +{ + printf("\nfailed syscalls by comm:\n\n"); + + printf("%-20s %10s\n", "comm", "# errors"); + printf("%-20s %6s %10s\n", "--------------------", "----------"); + + foreach my $comm (sort {$failed_syscalls{$b} <=> $failed_syscalls{$a}} + keys %failed_syscalls) { + next if ($for_comm && $comm ne $for_comm); + + printf("%-20s %10s\n", $comm, $failed_syscalls{$comm}); + } +} diff --git a/tools/perf/scripts/perl/rw-by-file.pl b/tools/perf/scripts/perl/rw-by-file.pl new file mode 100644 index 00000000..74844ee2 --- /dev/null +++ b/tools/perf/scripts/perl/rw-by-file.pl @@ -0,0 +1,106 @@ +#!/usr/bin/perl -w +# (c) 2009, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 + +# Display r/w activity for files read/written to for a given program + +# The common_* event handler fields are the most useful fields common to +# all events. They don't necessarily correspond to the 'common_*' fields +# in the status files. Those fields not available as handler params can +# be retrieved via script functions of the form get_common_*(). + +use 5.010000; +use strict; +use warnings; + +use lib "$ENV{'PERF_EXEC_PATH'}/scripts/perl/Perf-Trace-Util/lib"; +use lib "./Perf-Trace-Util/lib"; +use Perf::Trace::Core; +use Perf::Trace::Util; + +my $usage = "perf script -s rw-by-file.pl <comm>\n"; + +my $for_comm = shift or die $usage; + +my %reads; +my %writes; + +sub syscalls::sys_enter_read +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, $nr, $fd, $buf, $count) = @_; + + if ($common_comm eq $for_comm) { + $reads{$fd}{bytes_requested} += $count; + $reads{$fd}{total_reads}++; + } +} + +sub syscalls::sys_enter_write +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, $nr, $fd, $buf, $count) = @_; + + if ($common_comm eq $for_comm) { + $writes{$fd}{bytes_written} += $count; + $writes{$fd}{total_writes}++; + } +} + +sub trace_end +{ + printf("file read counts for $for_comm:\n\n"); + + printf("%6s %10s %10s\n", "fd", "# reads", "bytes_requested"); + printf("%6s %10s %10s\n", "------", "----------", "-----------"); + + foreach my $fd (sort {$reads{$b}{bytes_requested} <=> + $reads{$a}{bytes_requested}} keys %reads) { + my $total_reads = $reads{$fd}{total_reads}; + my $bytes_requested = $reads{$fd}{bytes_requested}; + printf("%6u %10u %10u\n", $fd, $total_reads, $bytes_requested); + } + + printf("\nfile write counts for $for_comm:\n\n"); + + printf("%6s %10s %10s\n", "fd", "# writes", "bytes_written"); + printf("%6s %10s %10s\n", "------", "----------", "-----------"); + + foreach my $fd (sort {$writes{$b}{bytes_written} <=> + $writes{$a}{bytes_written}} keys %writes) { + my $total_writes = $writes{$fd}{total_writes}; + my $bytes_written = $writes{$fd}{bytes_written}; + printf("%6u %10u %10u\n", $fd, $total_writes, $bytes_written); + } + + print_unhandled(); +} + +my %unhandled; + +sub print_unhandled +{ + if ((scalar keys %unhandled) == 0) { + return; + } + + print "\nunhandled events:\n\n"; + + printf("%-40s %10s\n", "event", "count"); + printf("%-40s %10s\n", "----------------------------------------", + "-----------"); + + foreach my $event_name (keys %unhandled) { + printf("%-40s %10d\n", $event_name, $unhandled{$event_name}); + } +} + +sub trace_unhandled +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm) = @_; + + $unhandled{$event_name}++; +} + + diff --git a/tools/perf/scripts/perl/rw-by-pid.pl b/tools/perf/scripts/perl/rw-by-pid.pl new file mode 100644 index 00000000..9db23c9d --- /dev/null +++ b/tools/perf/scripts/perl/rw-by-pid.pl @@ -0,0 +1,184 @@ +#!/usr/bin/perl -w +# (c) 2009, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 + +# Display r/w activity for all processes + +# The common_* event handler fields are the most useful fields common to +# all events. They don't necessarily correspond to the 'common_*' fields +# in the status files. Those fields not available as handler params can +# be retrieved via script functions of the form get_common_*(). + +use 5.010000; +use strict; +use warnings; + +use lib "$ENV{'PERF_EXEC_PATH'}/scripts/perl/Perf-Trace-Util/lib"; +use lib "./Perf-Trace-Util/lib"; +use Perf::Trace::Core; +use Perf::Trace::Util; + +my %reads; +my %writes; + +sub syscalls::sys_exit_read +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $nr, $ret) = @_; + + if ($ret > 0) { + $reads{$common_pid}{bytes_read} += $ret; + } else { + if (!defined ($reads{$common_pid}{bytes_read})) { + $reads{$common_pid}{bytes_read} = 0; + } + $reads{$common_pid}{errors}{$ret}++; + } +} + +sub syscalls::sys_enter_read +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $nr, $fd, $buf, $count) = @_; + + $reads{$common_pid}{bytes_requested} += $count; + $reads{$common_pid}{total_reads}++; + $reads{$common_pid}{comm} = $common_comm; +} + +sub syscalls::sys_exit_write +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $nr, $ret) = @_; + + if ($ret <= 0) { + $writes{$common_pid}{errors}{$ret}++; + } +} + +sub syscalls::sys_enter_write +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $nr, $fd, $buf, $count) = @_; + + $writes{$common_pid}{bytes_written} += $count; + $writes{$common_pid}{total_writes}++; + $writes{$common_pid}{comm} = $common_comm; +} + +sub trace_end +{ + printf("read counts by pid:\n\n"); + + printf("%6s %20s %10s %10s %10s\n", "pid", "comm", + "# reads", "bytes_requested", "bytes_read"); + printf("%6s %-20s %10s %10s %10s\n", "------", "--------------------", + "-----------", "----------", "----------"); + + foreach my $pid (sort { ($reads{$b}{bytes_read} || 0) <=> + ($reads{$a}{bytes_read} || 0) } keys %reads) { + my $comm = $reads{$pid}{comm} || ""; + my $total_reads = $reads{$pid}{total_reads} || 0; + my $bytes_requested = $reads{$pid}{bytes_requested} || 0; + my $bytes_read = $reads{$pid}{bytes_read} || 0; + + printf("%6s %-20s %10s %10s %10s\n", $pid, $comm, + $total_reads, $bytes_requested, $bytes_read); + } + + printf("\nfailed reads by pid:\n\n"); + + printf("%6s %20s %6s %10s\n", "pid", "comm", "error #", "# errors"); + printf("%6s %20s %6s %10s\n", "------", "--------------------", + "------", "----------"); + + my @errcounts = (); + + foreach my $pid (keys %reads) { + foreach my $error (keys %{$reads{$pid}{errors}}) { + my $comm = $reads{$pid}{comm} || ""; + my $errcount = $reads{$pid}{errors}{$error} || 0; + push @errcounts, [$pid, $comm, $error, $errcount]; + } + } + + @errcounts = sort { $b->[3] <=> $a->[3] } @errcounts; + + for my $i (0 .. $#errcounts) { + printf("%6d %-20s %6d %10s\n", $errcounts[$i][0], + $errcounts[$i][1], $errcounts[$i][2], $errcounts[$i][3]); + } + + printf("\nwrite counts by pid:\n\n"); + + printf("%6s %20s %10s %10s\n", "pid", "comm", + "# writes", "bytes_written"); + printf("%6s %-20s %10s %10s\n", "------", "--------------------", + "-----------", "----------"); + + foreach my $pid (sort { ($writes{$b}{bytes_written} || 0) <=> + ($writes{$a}{bytes_written} || 0)} keys %writes) { + my $comm = $writes{$pid}{comm} || ""; + my $total_writes = $writes{$pid}{total_writes} || 0; + my $bytes_written = $writes{$pid}{bytes_written} || 0; + + printf("%6s %-20s %10s %10s\n", $pid, $comm, + $total_writes, $bytes_written); + } + + printf("\nfailed writes by pid:\n\n"); + + printf("%6s %20s %6s %10s\n", "pid", "comm", "error #", "# errors"); + printf("%6s %20s %6s %10s\n", "------", "--------------------", + "------", "----------"); + + @errcounts = (); + + foreach my $pid (keys %writes) { + foreach my $error (keys %{$writes{$pid}{errors}}) { + my $comm = $writes{$pid}{comm} || ""; + my $errcount = $writes{$pid}{errors}{$error} || 0; + push @errcounts, [$pid, $comm, $error, $errcount]; + } + } + + @errcounts = sort { $b->[3] <=> $a->[3] } @errcounts; + + for my $i (0 .. $#errcounts) { + printf("%6d %-20s %6d %10s\n", $errcounts[$i][0], + $errcounts[$i][1], $errcounts[$i][2], $errcounts[$i][3]); + } + + print_unhandled(); +} + +my %unhandled; + +sub print_unhandled +{ + if ((scalar keys %unhandled) == 0) { + return; + } + + print "\nunhandled events:\n\n"; + + printf("%-40s %10s\n", "event", "count"); + printf("%-40s %10s\n", "----------------------------------------", + "-----------"); + + foreach my $event_name (keys %unhandled) { + printf("%-40s %10d\n", $event_name, $unhandled{$event_name}); + } +} + +sub trace_unhandled +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm) = @_; + + $unhandled{$event_name}++; +} diff --git a/tools/perf/scripts/perl/rwtop.pl b/tools/perf/scripts/perl/rwtop.pl new file mode 100644 index 00000000..4bb3ecd3 --- /dev/null +++ b/tools/perf/scripts/perl/rwtop.pl @@ -0,0 +1,199 @@ +#!/usr/bin/perl -w +# (c) 2010, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 + +# read/write top +# +# Periodically displays system-wide r/w call activity, broken down by +# pid. If an [interval] arg is specified, the display will be +# refreshed every [interval] seconds. The default interval is 3 +# seconds. + +use 5.010000; +use strict; +use warnings; + +use lib "$ENV{'PERF_EXEC_PATH'}/scripts/perl/Perf-Trace-Util/lib"; +use lib "./Perf-Trace-Util/lib"; +use Perf::Trace::Core; +use Perf::Trace::Util; + +my $default_interval = 3; +my $nlines = 20; +my $print_thread; +my $print_pending = 0; + +my %reads; +my %writes; + +my $interval = shift; +if (!$interval) { + $interval = $default_interval; +} + +sub syscalls::sys_exit_read +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $nr, $ret) = @_; + + print_check(); + + if ($ret > 0) { + $reads{$common_pid}{bytes_read} += $ret; + } else { + if (!defined ($reads{$common_pid}{bytes_read})) { + $reads{$common_pid}{bytes_read} = 0; + } + $reads{$common_pid}{errors}{$ret}++; + } +} + +sub syscalls::sys_enter_read +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $nr, $fd, $buf, $count) = @_; + + print_check(); + + $reads{$common_pid}{bytes_requested} += $count; + $reads{$common_pid}{total_reads}++; + $reads{$common_pid}{comm} = $common_comm; +} + +sub syscalls::sys_exit_write +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $nr, $ret) = @_; + + print_check(); + + if ($ret <= 0) { + $writes{$common_pid}{errors}{$ret}++; + } +} + +sub syscalls::sys_enter_write +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $nr, $fd, $buf, $count) = @_; + + print_check(); + + $writes{$common_pid}{bytes_written} += $count; + $writes{$common_pid}{total_writes}++; + $writes{$common_pid}{comm} = $common_comm; +} + +sub trace_begin +{ + $SIG{ALRM} = \&set_print_pending; + alarm 1; +} + +sub trace_end +{ + print_unhandled(); + print_totals(); +} + +sub print_check() +{ + if ($print_pending == 1) { + $print_pending = 0; + print_totals(); + } +} + +sub set_print_pending() +{ + $print_pending = 1; + alarm $interval; +} + +sub print_totals +{ + my $count; + + $count = 0; + + clear_term(); + + printf("\nread counts by pid:\n\n"); + + printf("%6s %20s %10s %10s %10s\n", "pid", "comm", + "# reads", "bytes_req", "bytes_read"); + printf("%6s %-20s %10s %10s %10s\n", "------", "--------------------", + "----------", "----------", "----------"); + + foreach my $pid (sort { ($reads{$b}{bytes_read} || 0) <=> + ($reads{$a}{bytes_read} || 0) } keys %reads) { + my $comm = $reads{$pid}{comm} || ""; + my $total_reads = $reads{$pid}{total_reads} || 0; + my $bytes_requested = $reads{$pid}{bytes_requested} || 0; + my $bytes_read = $reads{$pid}{bytes_read} || 0; + + printf("%6s %-20s %10s %10s %10s\n", $pid, $comm, + $total_reads, $bytes_requested, $bytes_read); + + if (++$count == $nlines) { + last; + } + } + + $count = 0; + + printf("\nwrite counts by pid:\n\n"); + + printf("%6s %20s %10s %13s\n", "pid", "comm", + "# writes", "bytes_written"); + printf("%6s %-20s %10s %13s\n", "------", "--------------------", + "----------", "-------------"); + + foreach my $pid (sort { ($writes{$b}{bytes_written} || 0) <=> + ($writes{$a}{bytes_written} || 0)} keys %writes) { + my $comm = $writes{$pid}{comm} || ""; + my $total_writes = $writes{$pid}{total_writes} || 0; + my $bytes_written = $writes{$pid}{bytes_written} || 0; + + printf("%6s %-20s %10s %13s\n", $pid, $comm, + $total_writes, $bytes_written); + + if (++$count == $nlines) { + last; + } + } + + %reads = (); + %writes = (); +} + +my %unhandled; + +sub print_unhandled +{ + if ((scalar keys %unhandled) == 0) { + return; + } + + print "\nunhandled events:\n\n"; + + printf("%-40s %10s\n", "event", "count"); + printf("%-40s %10s\n", "----------------------------------------", + "-----------"); + + foreach my $event_name (keys %unhandled) { + printf("%-40s %10d\n", $event_name, $unhandled{$event_name}); + } +} + +sub trace_unhandled +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm) = @_; + + $unhandled{$event_name}++; +} diff --git a/tools/perf/scripts/perl/wakeup-latency.pl b/tools/perf/scripts/perl/wakeup-latency.pl new file mode 100644 index 00000000..d9143dce --- /dev/null +++ b/tools/perf/scripts/perl/wakeup-latency.pl @@ -0,0 +1,107 @@ +#!/usr/bin/perl -w +# (c) 2009, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 + +# Display avg/min/max wakeup latency + +# The common_* event handler fields are the most useful fields common to +# all events. They don't necessarily correspond to the 'common_*' fields +# in the status files. Those fields not available as handler params can +# be retrieved via script functions of the form get_common_*(). + +use 5.010000; +use strict; +use warnings; + +use lib "$ENV{'PERF_EXEC_PATH'}/scripts/perl/Perf-Trace-Util/lib"; +use lib "./Perf-Trace-Util/lib"; +use Perf::Trace::Core; +use Perf::Trace::Util; + +my %last_wakeup; + +my $max_wakeup_latency; +my $min_wakeup_latency; +my $total_wakeup_latency = 0; +my $total_wakeups = 0; + +sub sched::sched_switch +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $prev_comm, $prev_pid, $prev_prio, $prev_state, $next_comm, $next_pid, + $next_prio) = @_; + + my $wakeup_ts = $last_wakeup{$common_cpu}{ts}; + if ($wakeup_ts) { + my $switch_ts = nsecs($common_secs, $common_nsecs); + my $wakeup_latency = $switch_ts - $wakeup_ts; + if ($wakeup_latency > $max_wakeup_latency) { + $max_wakeup_latency = $wakeup_latency; + } + if ($wakeup_latency < $min_wakeup_latency) { + $min_wakeup_latency = $wakeup_latency; + } + $total_wakeup_latency += $wakeup_latency; + $total_wakeups++; + } + $last_wakeup{$common_cpu}{ts} = 0; +} + +sub sched::sched_wakeup +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $comm, $pid, $prio, $success, $target_cpu) = @_; + + $last_wakeup{$target_cpu}{ts} = nsecs($common_secs, $common_nsecs); +} + +sub trace_begin +{ + $min_wakeup_latency = 1000000000; + $max_wakeup_latency = 0; +} + +sub trace_end +{ + printf("wakeup_latency stats:\n\n"); + print "total_wakeups: $total_wakeups\n"; + if ($total_wakeups) { + printf("avg_wakeup_latency (ns): %u\n", + avg($total_wakeup_latency, $total_wakeups)); + } else { + printf("avg_wakeup_latency (ns): N/A\n"); + } + printf("min_wakeup_latency (ns): %u\n", $min_wakeup_latency); + printf("max_wakeup_latency (ns): %u\n", $max_wakeup_latency); + + print_unhandled(); +} + +my %unhandled; + +sub print_unhandled +{ + if ((scalar keys %unhandled) == 0) { + return; + } + + print "\nunhandled events:\n\n"; + + printf("%-40s %10s\n", "event", "count"); + printf("%-40s %10s\n", "----------------------------------------", + "-----------"); + + foreach my $event_name (keys %unhandled) { + printf("%-40s %10d\n", $event_name, $unhandled{$event_name}); + } +} + +sub trace_unhandled +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm) = @_; + + $unhandled{$event_name}++; +} diff --git a/tools/perf/scripts/perl/workqueue-stats.pl b/tools/perf/scripts/perl/workqueue-stats.pl new file mode 100644 index 00000000..a8eaff51 --- /dev/null +++ b/tools/perf/scripts/perl/workqueue-stats.pl @@ -0,0 +1,129 @@ +#!/usr/bin/perl -w +# (c) 2009, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 + +# Displays workqueue stats +# +# Usage: +# +# perf record -c 1 -f -a -R -e workqueue:workqueue_creation -e +# workqueue:workqueue_destruction -e workqueue:workqueue_execution +# -e workqueue:workqueue_insertion +# +# perf script -p -s tools/perf/scripts/perl/workqueue-stats.pl + +use 5.010000; +use strict; +use warnings; + +use lib "$ENV{'PERF_EXEC_PATH'}/scripts/perl/Perf-Trace-Util/lib"; +use lib "./Perf-Trace-Util/lib"; +use Perf::Trace::Core; +use Perf::Trace::Util; + +my @cpus; + +sub workqueue::workqueue_destruction +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $thread_comm, $thread_pid) = @_; + + $cpus[$common_cpu]{$thread_pid}{destroyed}++; + $cpus[$common_cpu]{$thread_pid}{comm} = $thread_comm; +} + +sub workqueue::workqueue_creation +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $thread_comm, $thread_pid, $cpu) = @_; + + $cpus[$common_cpu]{$thread_pid}{created}++; + $cpus[$common_cpu]{$thread_pid}{comm} = $thread_comm; +} + +sub workqueue::workqueue_execution +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $thread_comm, $thread_pid, $func) = @_; + + $cpus[$common_cpu]{$thread_pid}{executed}++; + $cpus[$common_cpu]{$thread_pid}{comm} = $thread_comm; +} + +sub workqueue::workqueue_insertion +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm, + $thread_comm, $thread_pid, $func) = @_; + + $cpus[$common_cpu]{$thread_pid}{inserted}++; + $cpus[$common_cpu]{$thread_pid}{comm} = $thread_comm; +} + +sub trace_end +{ + print "workqueue work stats:\n\n"; + my $cpu = 0; + printf("%3s %6s %6s\t%-20s\n", "cpu", "ins", "exec", "name"); + printf("%3s %6s %6s\t%-20s\n", "---", "---", "----", "----"); + foreach my $pidhash (@cpus) { + while ((my $pid, my $wqhash) = each %$pidhash) { + my $ins = $$wqhash{'inserted'} || 0; + my $exe = $$wqhash{'executed'} || 0; + my $comm = $$wqhash{'comm'} || ""; + if ($ins || $exe) { + printf("%3u %6u %6u\t%-20s\n", $cpu, $ins, $exe, $comm); + } + } + $cpu++; + } + + $cpu = 0; + print "\nworkqueue lifecycle stats:\n\n"; + printf("%3s %6s %6s\t%-20s\n", "cpu", "created", "destroyed", "name"); + printf("%3s %6s %6s\t%-20s\n", "---", "-------", "---------", "----"); + foreach my $pidhash (@cpus) { + while ((my $pid, my $wqhash) = each %$pidhash) { + my $created = $$wqhash{'created'} || 0; + my $destroyed = $$wqhash{'destroyed'} || 0; + my $comm = $$wqhash{'comm'} || ""; + if ($created || $destroyed) { + printf("%3u %6u %6u\t%-20s\n", $cpu, $created, $destroyed, + $comm); + } + } + $cpu++; + } + + print_unhandled(); +} + +my %unhandled; + +sub print_unhandled +{ + if ((scalar keys %unhandled) == 0) { + return; + } + + print "\nunhandled events:\n\n"; + + printf("%-40s %10s\n", "event", "count"); + printf("%-40s %10s\n", "----------------------------------------", + "-----------"); + + foreach my $event_name (keys %unhandled) { + printf("%-40s %10d\n", $event_name, $unhandled{$event_name}); + } +} + +sub trace_unhandled +{ + my ($event_name, $context, $common_cpu, $common_secs, $common_nsecs, + $common_pid, $common_comm) = @_; + + $unhandled{$event_name}++; +} diff --git a/tools/perf/scripts/python/Perf-Trace-Util/Context.c b/tools/perf/scripts/python/Perf-Trace-Util/Context.c new file mode 100644 index 00000000..315067b8 --- /dev/null +++ b/tools/perf/scripts/python/Perf-Trace-Util/Context.c @@ -0,0 +1,88 @@ +/* + * Context.c. Python interfaces for perf script. + * + * Copyright (C) 2010 Tom Zanussi <tzanussi@gmail.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#include <Python.h> +#include "../../../perf.h" +#include "../../../util/trace-event.h" + +PyMODINIT_FUNC initperf_trace_context(void); + +static PyObject *perf_trace_context_common_pc(PyObject *self, PyObject *args) +{ + static struct scripting_context *scripting_context; + PyObject *context; + int retval; + + if (!PyArg_ParseTuple(args, "O", &context)) + return NULL; + + scripting_context = PyCObject_AsVoidPtr(context); + retval = common_pc(scripting_context); + + return Py_BuildValue("i", retval); +} + +static PyObject *perf_trace_context_common_flags(PyObject *self, + PyObject *args) +{ + static struct scripting_context *scripting_context; + PyObject *context; + int retval; + + if (!PyArg_ParseTuple(args, "O", &context)) + return NULL; + + scripting_context = PyCObject_AsVoidPtr(context); + retval = common_flags(scripting_context); + + return Py_BuildValue("i", retval); +} + +static PyObject *perf_trace_context_common_lock_depth(PyObject *self, + PyObject *args) +{ + static struct scripting_context *scripting_context; + PyObject *context; + int retval; + + if (!PyArg_ParseTuple(args, "O", &context)) + return NULL; + + scripting_context = PyCObject_AsVoidPtr(context); + retval = common_lock_depth(scripting_context); + + return Py_BuildValue("i", retval); +} + +static PyMethodDef ContextMethods[] = { + { "common_pc", perf_trace_context_common_pc, METH_VARARGS, + "Get the common preempt count event field value."}, + { "common_flags", perf_trace_context_common_flags, METH_VARARGS, + "Get the common flags event field value."}, + { "common_lock_depth", perf_trace_context_common_lock_depth, + METH_VARARGS, "Get the common lock depth event field value."}, + { NULL, NULL, 0, NULL} +}; + +PyMODINIT_FUNC initperf_trace_context(void) +{ + (void) Py_InitModule("perf_trace_context", ContextMethods); +} diff --git a/tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Core.py b/tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Core.py new file mode 100644 index 00000000..de7211e4 --- /dev/null +++ b/tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Core.py @@ -0,0 +1,121 @@ +# Core.py - Python extension for perf script, core functions +# +# Copyright (C) 2010 by Tom Zanussi <tzanussi@gmail.com> +# +# This software may be distributed under the terms of the GNU General +# Public License ("GPL") version 2 as published by the Free Software +# Foundation. + +from collections import defaultdict + +def autodict(): + return defaultdict(autodict) + +flag_fields = autodict() +symbolic_fields = autodict() + +def define_flag_field(event_name, field_name, delim): + flag_fields[event_name][field_name]['delim'] = delim + +def define_flag_value(event_name, field_name, value, field_str): + flag_fields[event_name][field_name]['values'][value] = field_str + +def define_symbolic_field(event_name, field_name): + # nothing to do, really + pass + +def define_symbolic_value(event_name, field_name, value, field_str): + symbolic_fields[event_name][field_name]['values'][value] = field_str + +def flag_str(event_name, field_name, value): + string = "" + + if flag_fields[event_name][field_name]: + print_delim = 0 + keys = flag_fields[event_name][field_name]['values'].keys() + keys.sort() + for idx in keys: + if not value and not idx: + string += flag_fields[event_name][field_name]['values'][idx] + break + if idx and (value & idx) == idx: + if print_delim and flag_fields[event_name][field_name]['delim']: + string += " " + flag_fields[event_name][field_name]['delim'] + " " + string += flag_fields[event_name][field_name]['values'][idx] + print_delim = 1 + value &= ~idx + + return string + +def symbol_str(event_name, field_name, value): + string = "" + + if symbolic_fields[event_name][field_name]: + keys = symbolic_fields[event_name][field_name]['values'].keys() + keys.sort() + for idx in keys: + if not value and not idx: + string = symbolic_fields[event_name][field_name]['values'][idx] + break + if (value == idx): + string = symbolic_fields[event_name][field_name]['values'][idx] + break + + return string + +trace_flags = { 0x00: "NONE", \ + 0x01: "IRQS_OFF", \ + 0x02: "IRQS_NOSUPPORT", \ + 0x04: "NEED_RESCHED", \ + 0x08: "HARDIRQ", \ + 0x10: "SOFTIRQ" } + +def trace_flag_str(value): + string = "" + print_delim = 0 + + keys = trace_flags.keys() + + for idx in keys: + if not value and not idx: + string += "NONE" + break + + if idx and (value & idx) == idx: + if print_delim: + string += " | "; + string += trace_flags[idx] + print_delim = 1 + value &= ~idx + + return string + + +def taskState(state): + states = { + 0 : "R", + 1 : "S", + 2 : "D", + 64: "DEAD" + } + + if state not in states: + return "Unknown" + + return states[state] + + +class EventHeaders: + def __init__(self, common_cpu, common_secs, common_nsecs, + common_pid, common_comm): + self.cpu = common_cpu + self.secs = common_secs + self.nsecs = common_nsecs + self.pid = common_pid + self.comm = common_comm + + def ts(self): + return (self.secs * (10 ** 9)) + self.nsecs + + def ts_format(self): + return "%d.%d" % (self.secs, int(self.nsecs / 1000)) diff --git a/tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/SchedGui.py b/tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/SchedGui.py new file mode 100644 index 00000000..fdd92f69 --- /dev/null +++ b/tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/SchedGui.py @@ -0,0 +1,184 @@ +# SchedGui.py - Python extension for perf script, basic GUI code for +# traces drawing and overview. +# +# Copyright (C) 2010 by Frederic Weisbecker <fweisbec@gmail.com> +# +# This software is distributed under the terms of the GNU General +# Public License ("GPL") version 2 as published by the Free Software +# Foundation. + + +try: + import wx +except ImportError: + raise ImportError, "You need to install the wxpython lib for this script" + + +class RootFrame(wx.Frame): + Y_OFFSET = 100 + RECT_HEIGHT = 100 + RECT_SPACE = 50 + EVENT_MARKING_WIDTH = 5 + + def __init__(self, sched_tracer, title, parent = None, id = -1): + wx.Frame.__init__(self, parent, id, title) + + (self.screen_width, self.screen_height) = wx.GetDisplaySize() + self.screen_width -= 10 + self.screen_height -= 10 + self.zoom = 0.5 + self.scroll_scale = 20 + self.sched_tracer = sched_tracer + self.sched_tracer.set_root_win(self) + (self.ts_start, self.ts_end) = sched_tracer.interval() + self.update_width_virtual() + self.nr_rects = sched_tracer.nr_rectangles() + 1 + self.height_virtual = RootFrame.Y_OFFSET + (self.nr_rects * (RootFrame.RECT_HEIGHT + RootFrame.RECT_SPACE)) + + # whole window panel + self.panel = wx.Panel(self, size=(self.screen_width, self.screen_height)) + + # scrollable container + self.scroll = wx.ScrolledWindow(self.panel) + self.scroll.SetScrollbars(self.scroll_scale, self.scroll_scale, self.width_virtual / self.scroll_scale, self.height_virtual / self.scroll_scale) + self.scroll.EnableScrolling(True, True) + self.scroll.SetFocus() + + # scrollable drawing area + self.scroll_panel = wx.Panel(self.scroll, size=(self.screen_width - 15, self.screen_height / 2)) + self.scroll_panel.Bind(wx.EVT_PAINT, self.on_paint) + self.scroll_panel.Bind(wx.EVT_KEY_DOWN, self.on_key_press) + self.scroll_panel.Bind(wx.EVT_LEFT_DOWN, self.on_mouse_down) + self.scroll.Bind(wx.EVT_PAINT, self.on_paint) + self.scroll.Bind(wx.EVT_KEY_DOWN, self.on_key_press) + self.scroll.Bind(wx.EVT_LEFT_DOWN, self.on_mouse_down) + + self.scroll.Fit() + self.Fit() + + self.scroll_panel.SetDimensions(-1, -1, self.width_virtual, self.height_virtual, wx.SIZE_USE_EXISTING) + + self.txt = None + + self.Show(True) + + def us_to_px(self, val): + return val / (10 ** 3) * self.zoom + + def px_to_us(self, val): + return (val / self.zoom) * (10 ** 3) + + def scroll_start(self): + (x, y) = self.scroll.GetViewStart() + return (x * self.scroll_scale, y * self.scroll_scale) + + def scroll_start_us(self): + (x, y) = self.scroll_start() + return self.px_to_us(x) + + def paint_rectangle_zone(self, nr, color, top_color, start, end): + offset_px = self.us_to_px(start - self.ts_start) + width_px = self.us_to_px(end - self.ts_start) + + offset_py = RootFrame.Y_OFFSET + (nr * (RootFrame.RECT_HEIGHT + RootFrame.RECT_SPACE)) + width_py = RootFrame.RECT_HEIGHT + + dc = self.dc + + if top_color is not None: + (r, g, b) = top_color + top_color = wx.Colour(r, g, b) + brush = wx.Brush(top_color, wx.SOLID) + dc.SetBrush(brush) + dc.DrawRectangle(offset_px, offset_py, width_px, RootFrame.EVENT_MARKING_WIDTH) + width_py -= RootFrame.EVENT_MARKING_WIDTH + offset_py += RootFrame.EVENT_MARKING_WIDTH + + (r ,g, b) = color + color = wx.Colour(r, g, b) + brush = wx.Brush(color, wx.SOLID) + dc.SetBrush(brush) + dc.DrawRectangle(offset_px, offset_py, width_px, width_py) + + def update_rectangles(self, dc, start, end): + start += self.ts_start + end += self.ts_start + self.sched_tracer.fill_zone(start, end) + + def on_paint(self, event): + dc = wx.PaintDC(self.scroll_panel) + self.dc = dc + + width = min(self.width_virtual, self.screen_width) + (x, y) = self.scroll_start() + start = self.px_to_us(x) + end = self.px_to_us(x + width) + self.update_rectangles(dc, start, end) + + def rect_from_ypixel(self, y): + y -= RootFrame.Y_OFFSET + rect = y / (RootFrame.RECT_HEIGHT + RootFrame.RECT_SPACE) + height = y % (RootFrame.RECT_HEIGHT + RootFrame.RECT_SPACE) + + if rect < 0 or rect > self.nr_rects - 1 or height > RootFrame.RECT_HEIGHT: + return -1 + + return rect + + def update_summary(self, txt): + if self.txt: + self.txt.Destroy() + self.txt = wx.StaticText(self.panel, -1, txt, (0, (self.screen_height / 2) + 50)) + + + def on_mouse_down(self, event): + (x, y) = event.GetPositionTuple() + rect = self.rect_from_ypixel(y) + if rect == -1: + return + + t = self.px_to_us(x) + self.ts_start + + self.sched_tracer.mouse_down(rect, t) + + + def update_width_virtual(self): + self.width_virtual = self.us_to_px(self.ts_end - self.ts_start) + + def __zoom(self, x): + self.update_width_virtual() + (xpos, ypos) = self.scroll.GetViewStart() + xpos = self.us_to_px(x) / self.scroll_scale + self.scroll.SetScrollbars(self.scroll_scale, self.scroll_scale, self.width_virtual / self.scroll_scale, self.height_virtual / self.scroll_scale, xpos, ypos) + self.Refresh() + + def zoom_in(self): + x = self.scroll_start_us() + self.zoom *= 2 + self.__zoom(x) + + def zoom_out(self): + x = self.scroll_start_us() + self.zoom /= 2 + self.__zoom(x) + + + def on_key_press(self, event): + key = event.GetRawKeyCode() + if key == ord("+"): + self.zoom_in() + return + if key == ord("-"): + self.zoom_out() + return + + key = event.GetKeyCode() + (x, y) = self.scroll.GetViewStart() + if key == wx.WXK_RIGHT: + self.scroll.Scroll(x + 1, y) + elif key == wx.WXK_LEFT: + self.scroll.Scroll(x - 1, y) + elif key == wx.WXK_DOWN: + self.scroll.Scroll(x, y + 1) + elif key == wx.WXK_UP: + self.scroll.Scroll(x, y - 1) diff --git a/tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Util.py b/tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Util.py new file mode 100644 index 00000000..15c84002 --- /dev/null +++ b/tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Util.py @@ -0,0 +1,86 @@ +# Util.py - Python extension for perf script, miscellaneous utility code +# +# Copyright (C) 2010 by Tom Zanussi <tzanussi@gmail.com> +# +# This software may be distributed under the terms of the GNU General +# Public License ("GPL") version 2 as published by the Free Software +# Foundation. + +import errno, os + +FUTEX_WAIT = 0 +FUTEX_WAKE = 1 +FUTEX_PRIVATE_FLAG = 128 +FUTEX_CLOCK_REALTIME = 256 +FUTEX_CMD_MASK = ~(FUTEX_PRIVATE_FLAG | FUTEX_CLOCK_REALTIME) + +NSECS_PER_SEC = 1000000000 + +def avg(total, n): + return total / n + +def nsecs(secs, nsecs): + return secs * NSECS_PER_SEC + nsecs + +def nsecs_secs(nsecs): + return nsecs / NSECS_PER_SEC + +def nsecs_nsecs(nsecs): + return nsecs % NSECS_PER_SEC + +def nsecs_str(nsecs): + str = "%5u.%09u" % (nsecs_secs(nsecs), nsecs_nsecs(nsecs)), + return str + +def add_stats(dict, key, value): + if not dict.has_key(key): + dict[key] = (value, value, value, 1) + else: + min, max, avg, count = dict[key] + if value < min: + min = value + if value > max: + max = value + avg = (avg + value) / 2 + dict[key] = (min, max, avg, count + 1) + +def clear_term(): + print("\x1b[H\x1b[2J") + +audit_package_warned = False + +try: + import audit + machine_to_id = { + 'x86_64': audit.MACH_86_64, + 'alpha' : audit.MACH_ALPHA, + 'ia64' : audit.MACH_IA64, + 'ppc' : audit.MACH_PPC, + 'ppc64' : audit.MACH_PPC64, + 's390' : audit.MACH_S390, + 's390x' : audit.MACH_S390X, + 'i386' : audit.MACH_X86, + 'i586' : audit.MACH_X86, + 'i686' : audit.MACH_X86, + } + try: + machine_to_id['armeb'] = audit.MACH_ARMEB + except: + pass + machine_id = machine_to_id[os.uname()[4]] +except: + if not audit_package_warned: + audit_package_warned = True + print "Install the audit-libs-python package to get syscall names" + +def syscall_name(id): + try: + return audit.audit_syscall_to_name(id, machine_id) + except: + return str(id) + +def strerror(nr): + try: + return errno.errorcode[abs(nr)] + except: + return "Unknown %d errno" % nr diff --git a/tools/perf/scripts/python/bin/failed-syscalls-by-pid-record b/tools/perf/scripts/python/bin/failed-syscalls-by-pid-record new file mode 100644 index 00000000..8104895a --- /dev/null +++ b/tools/perf/scripts/python/bin/failed-syscalls-by-pid-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -e raw_syscalls:sys_exit $@ diff --git a/tools/perf/scripts/python/bin/failed-syscalls-by-pid-report b/tools/perf/scripts/python/bin/failed-syscalls-by-pid-report new file mode 100644 index 00000000..fda5096d --- /dev/null +++ b/tools/perf/scripts/python/bin/failed-syscalls-by-pid-report @@ -0,0 +1,10 @@ +#!/bin/bash +# description: system-wide failed syscalls, by pid +# args: [comm] +if [ $# -gt 0 ] ; then + if ! expr match "$1" "-" > /dev/null ; then + comm=$1 + shift + fi +fi +perf script $@ -s "$PERF_EXEC_PATH"/scripts/python/failed-syscalls-by-pid.py $comm diff --git a/tools/perf/scripts/python/bin/futex-contention-record b/tools/perf/scripts/python/bin/futex-contention-record new file mode 100644 index 00000000..b1495c9a --- /dev/null +++ b/tools/perf/scripts/python/bin/futex-contention-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -e syscalls:sys_enter_futex -e syscalls:sys_exit_futex $@ diff --git a/tools/perf/scripts/python/bin/futex-contention-report b/tools/perf/scripts/python/bin/futex-contention-report new file mode 100644 index 00000000..6c442710 --- /dev/null +++ b/tools/perf/scripts/python/bin/futex-contention-report @@ -0,0 +1,4 @@ +#!/bin/bash +# description: futext contention measurement + +perf script $@ -s "$PERF_EXEC_PATH"/scripts/python/futex-contention.py diff --git a/tools/perf/scripts/python/bin/netdev-times-record b/tools/perf/scripts/python/bin/netdev-times-record new file mode 100644 index 00000000..558754b8 --- /dev/null +++ b/tools/perf/scripts/python/bin/netdev-times-record @@ -0,0 +1,8 @@ +#!/bin/bash +perf record -e net:net_dev_xmit -e net:net_dev_queue \ + -e net:netif_receive_skb -e net:netif_rx \ + -e skb:consume_skb -e skb:kfree_skb \ + -e skb:skb_copy_datagram_iovec -e napi:napi_poll \ + -e irq:irq_handler_entry -e irq:irq_handler_exit \ + -e irq:softirq_entry -e irq:softirq_exit \ + -e irq:softirq_raise $@ diff --git a/tools/perf/scripts/python/bin/netdev-times-report b/tools/perf/scripts/python/bin/netdev-times-report new file mode 100644 index 00000000..8f759291 --- /dev/null +++ b/tools/perf/scripts/python/bin/netdev-times-report @@ -0,0 +1,5 @@ +#!/bin/bash +# description: display a process of packet and processing time +# args: [tx] [rx] [dev=] [debug] + +perf script -s "$PERF_EXEC_PATH"/scripts/python/netdev-times.py $@ diff --git a/tools/perf/scripts/python/bin/sched-migration-record b/tools/perf/scripts/python/bin/sched-migration-record new file mode 100644 index 00000000..7493fddb --- /dev/null +++ b/tools/perf/scripts/python/bin/sched-migration-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -m 16384 -e sched:sched_wakeup -e sched:sched_wakeup_new -e sched:sched_switch -e sched:sched_migrate_task $@ diff --git a/tools/perf/scripts/python/bin/sched-migration-report b/tools/perf/scripts/python/bin/sched-migration-report new file mode 100644 index 00000000..68b037a1 --- /dev/null +++ b/tools/perf/scripts/python/bin/sched-migration-report @@ -0,0 +1,3 @@ +#!/bin/bash +# description: sched migration overview +perf script $@ -s "$PERF_EXEC_PATH"/scripts/python/sched-migration.py diff --git a/tools/perf/scripts/python/bin/sctop-record b/tools/perf/scripts/python/bin/sctop-record new file mode 100644 index 00000000..4efbfaa7 --- /dev/null +++ b/tools/perf/scripts/python/bin/sctop-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -e raw_syscalls:sys_enter $@ diff --git a/tools/perf/scripts/python/bin/sctop-report b/tools/perf/scripts/python/bin/sctop-report new file mode 100644 index 00000000..c32db294 --- /dev/null +++ b/tools/perf/scripts/python/bin/sctop-report @@ -0,0 +1,24 @@ +#!/bin/bash +# description: syscall top +# args: [comm] [interval] +n_args=0 +for i in "$@" +do + if expr match "$i" "-" > /dev/null ; then + break + fi + n_args=$(( $n_args + 1 )) +done +if [ "$n_args" -gt 2 ] ; then + echo "usage: sctop-report [comm] [interval]" + exit +fi +if [ "$n_args" -gt 1 ] ; then + comm=$1 + interval=$2 + shift 2 +elif [ "$n_args" -gt 0 ] ; then + interval=$1 + shift +fi +perf script $@ -s "$PERF_EXEC_PATH"/scripts/python/sctop.py $comm $interval diff --git a/tools/perf/scripts/python/bin/syscall-counts-by-pid-record b/tools/perf/scripts/python/bin/syscall-counts-by-pid-record new file mode 100644 index 00000000..4efbfaa7 --- /dev/null +++ b/tools/perf/scripts/python/bin/syscall-counts-by-pid-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -e raw_syscalls:sys_enter $@ diff --git a/tools/perf/scripts/python/bin/syscall-counts-by-pid-report b/tools/perf/scripts/python/bin/syscall-counts-by-pid-report new file mode 100644 index 00000000..16eb8d65 --- /dev/null +++ b/tools/perf/scripts/python/bin/syscall-counts-by-pid-report @@ -0,0 +1,10 @@ +#!/bin/bash +# description: system-wide syscall counts, by pid +# args: [comm] +if [ $# -gt 0 ] ; then + if ! expr match "$1" "-" > /dev/null ; then + comm=$1 + shift + fi +fi +perf script $@ -s "$PERF_EXEC_PATH"/scripts/python/syscall-counts-by-pid.py $comm diff --git a/tools/perf/scripts/python/bin/syscall-counts-record b/tools/perf/scripts/python/bin/syscall-counts-record new file mode 100644 index 00000000..4efbfaa7 --- /dev/null +++ b/tools/perf/scripts/python/bin/syscall-counts-record @@ -0,0 +1,2 @@ +#!/bin/bash +perf record -e raw_syscalls:sys_enter $@ diff --git a/tools/perf/scripts/python/bin/syscall-counts-report b/tools/perf/scripts/python/bin/syscall-counts-report new file mode 100644 index 00000000..0f0e9d45 --- /dev/null +++ b/tools/perf/scripts/python/bin/syscall-counts-report @@ -0,0 +1,10 @@ +#!/bin/bash +# description: system-wide syscall counts +# args: [comm] +if [ $# -gt 0 ] ; then + if ! expr match "$1" "-" > /dev/null ; then + comm=$1 + shift + fi +fi +perf script $@ -s "$PERF_EXEC_PATH"/scripts/python/syscall-counts.py $comm diff --git a/tools/perf/scripts/python/check-perf-trace.py b/tools/perf/scripts/python/check-perf-trace.py new file mode 100644 index 00000000..4647a769 --- /dev/null +++ b/tools/perf/scripts/python/check-perf-trace.py @@ -0,0 +1,82 @@ +# perf script event handlers, generated by perf script -g python +# (c) 2010, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 +# +# This script tests basic functionality such as flag and symbol +# strings, common_xxx() calls back into perf, begin, end, unhandled +# events, etc. Basically, if this script runs successfully and +# displays expected results, Python scripting support should be ok. + +import os +import sys + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/Perf-Trace-Util/lib/Perf/Trace') + +from Core import * +from perf_trace_context import * + +unhandled = autodict() + +def trace_begin(): + print "trace_begin" + pass + +def trace_end(): + print_unhandled() + +def irq__softirq_entry(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + vec): + print_header(event_name, common_cpu, common_secs, common_nsecs, + common_pid, common_comm) + + print_uncommon(context) + + print "vec=%s\n" % \ + (symbol_str("irq__softirq_entry", "vec", vec)), + +def kmem__kmalloc(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + call_site, ptr, bytes_req, bytes_alloc, + gfp_flags): + print_header(event_name, common_cpu, common_secs, common_nsecs, + common_pid, common_comm) + + print_uncommon(context) + + print "call_site=%u, ptr=%u, bytes_req=%u, " \ + "bytes_alloc=%u, gfp_flags=%s\n" % \ + (call_site, ptr, bytes_req, bytes_alloc, + + flag_str("kmem__kmalloc", "gfp_flags", gfp_flags)), + +def trace_unhandled(event_name, context, event_fields_dict): + try: + unhandled[event_name] += 1 + except TypeError: + unhandled[event_name] = 1 + +def print_header(event_name, cpu, secs, nsecs, pid, comm): + print "%-20s %5u %05u.%09u %8u %-20s " % \ + (event_name, cpu, secs, nsecs, pid, comm), + +# print trace fields not included in handler args +def print_uncommon(context): + print "common_preempt_count=%d, common_flags=%s, common_lock_depth=%d, " \ + % (common_pc(context), trace_flag_str(common_flags(context)), \ + common_lock_depth(context)) + +def print_unhandled(): + keys = unhandled.keys() + if not keys: + return + + print "\nunhandled events:\n\n", + + print "%-40s %10s\n" % ("event", "count"), + print "%-40s %10s\n" % ("----------------------------------------", \ + "-----------"), + + for event_name in keys: + print "%-40s %10d\n" % (event_name, unhandled[event_name]) diff --git a/tools/perf/scripts/python/failed-syscalls-by-pid.py b/tools/perf/scripts/python/failed-syscalls-by-pid.py new file mode 100644 index 00000000..85805fac --- /dev/null +++ b/tools/perf/scripts/python/failed-syscalls-by-pid.py @@ -0,0 +1,73 @@ +# failed system call counts, by pid +# (c) 2010, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 +# +# Displays system-wide failed system call totals, broken down by pid. +# If a [comm] arg is specified, only syscalls called by [comm] are displayed. + +import os +import sys + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/Perf-Trace-Util/lib/Perf/Trace') + +from perf_trace_context import * +from Core import * +from Util import * + +usage = "perf script -s syscall-counts-by-pid.py [comm|pid]\n"; + +for_comm = None +for_pid = None + +if len(sys.argv) > 2: + sys.exit(usage) + +if len(sys.argv) > 1: + try: + for_pid = int(sys.argv[1]) + except: + for_comm = sys.argv[1] + +syscalls = autodict() + +def trace_begin(): + print "Press control+C to stop and show the summary" + +def trace_end(): + print_error_totals() + +def raw_syscalls__sys_exit(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + id, ret): + if (for_comm and common_comm != for_comm) or \ + (for_pid and common_pid != for_pid ): + return + + if ret < 0: + try: + syscalls[common_comm][common_pid][id][ret] += 1 + except TypeError: + syscalls[common_comm][common_pid][id][ret] = 1 + +def print_error_totals(): + if for_comm is not None: + print "\nsyscall errors for %s:\n\n" % (for_comm), + else: + print "\nsyscall errors:\n\n", + + print "%-30s %10s\n" % ("comm [pid]", "count"), + print "%-30s %10s\n" % ("------------------------------", \ + "----------"), + + comm_keys = syscalls.keys() + for comm in comm_keys: + pid_keys = syscalls[comm].keys() + for pid in pid_keys: + print "\n%s [%d]\n" % (comm, pid), + id_keys = syscalls[comm][pid].keys() + for id in id_keys: + print " syscall: %-16s\n" % syscall_name(id), + ret_keys = syscalls[comm][pid][id].keys() + for ret, val in sorted(syscalls[comm][pid][id].iteritems(), key = lambda(k, v): (v, k), reverse = True): + print " err = %-20s %10d\n" % (strerror(ret), val), diff --git a/tools/perf/scripts/python/futex-contention.py b/tools/perf/scripts/python/futex-contention.py new file mode 100644 index 00000000..11e70a38 --- /dev/null +++ b/tools/perf/scripts/python/futex-contention.py @@ -0,0 +1,50 @@ +# futex contention +# (c) 2010, Arnaldo Carvalho de Melo <acme@redhat.com> +# Licensed under the terms of the GNU GPL License version 2 +# +# Translation of: +# +# http://sourceware.org/systemtap/wiki/WSFutexContention +# +# to perf python scripting. +# +# Measures futex contention + +import os, sys +sys.path.append(os.environ['PERF_EXEC_PATH'] + '/scripts/python/Perf-Trace-Util/lib/Perf/Trace') +from Util import * + +process_names = {} +thread_thislock = {} +thread_blocktime = {} + +lock_waits = {} # long-lived stats on (tid,lock) blockage elapsed time +process_names = {} # long-lived pid-to-execname mapping + +def syscalls__sys_enter_futex(event, ctxt, cpu, s, ns, tid, comm, + nr, uaddr, op, val, utime, uaddr2, val3): + cmd = op & FUTEX_CMD_MASK + if cmd != FUTEX_WAIT: + return # we don't care about originators of WAKE events + + process_names[tid] = comm + thread_thislock[tid] = uaddr + thread_blocktime[tid] = nsecs(s, ns) + +def syscalls__sys_exit_futex(event, ctxt, cpu, s, ns, tid, comm, + nr, ret): + if thread_blocktime.has_key(tid): + elapsed = nsecs(s, ns) - thread_blocktime[tid] + add_stats(lock_waits, (tid, thread_thislock[tid]), elapsed) + del thread_blocktime[tid] + del thread_thislock[tid] + +def trace_begin(): + print "Press control+C to stop and show the summary" + +def trace_end(): + for (tid, lock) in lock_waits: + min, max, avg, count = lock_waits[tid, lock] + print "%s[%d] lock %x contended %d times, %d avg ns" % \ + (process_names[tid], tid, lock, count, avg) + diff --git a/tools/perf/scripts/python/netdev-times.py b/tools/perf/scripts/python/netdev-times.py new file mode 100644 index 00000000..9aa0a329 --- /dev/null +++ b/tools/perf/scripts/python/netdev-times.py @@ -0,0 +1,464 @@ +# Display a process of packets and processed time. +# It helps us to investigate networking or network device. +# +# options +# tx: show only tx chart +# rx: show only rx chart +# dev=: show only thing related to specified device +# debug: work with debug mode. It shows buffer status. + +import os +import sys + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/Perf-Trace-Util/lib/Perf/Trace') + +from perf_trace_context import * +from Core import * +from Util import * + +all_event_list = []; # insert all tracepoint event related with this script +irq_dic = {}; # key is cpu and value is a list which stacks irqs + # which raise NET_RX softirq +net_rx_dic = {}; # key is cpu and value include time of NET_RX softirq-entry + # and a list which stacks receive +receive_hunk_list = []; # a list which include a sequence of receive events +rx_skb_list = []; # received packet list for matching + # skb_copy_datagram_iovec + +buffer_budget = 65536; # the budget of rx_skb_list, tx_queue_list and + # tx_xmit_list +of_count_rx_skb_list = 0; # overflow count + +tx_queue_list = []; # list of packets which pass through dev_queue_xmit +of_count_tx_queue_list = 0; # overflow count + +tx_xmit_list = []; # list of packets which pass through dev_hard_start_xmit +of_count_tx_xmit_list = 0; # overflow count + +tx_free_list = []; # list of packets which is freed + +# options +show_tx = 0; +show_rx = 0; +dev = 0; # store a name of device specified by option "dev=" +debug = 0; + +# indices of event_info tuple +EINFO_IDX_NAME= 0 +EINFO_IDX_CONTEXT=1 +EINFO_IDX_CPU= 2 +EINFO_IDX_TIME= 3 +EINFO_IDX_PID= 4 +EINFO_IDX_COMM= 5 + +# Calculate a time interval(msec) from src(nsec) to dst(nsec) +def diff_msec(src, dst): + return (dst - src) / 1000000.0 + +# Display a process of transmitting a packet +def print_transmit(hunk): + if dev != 0 and hunk['dev'].find(dev) < 0: + return + print "%7s %5d %6d.%06dsec %12.3fmsec %12.3fmsec" % \ + (hunk['dev'], hunk['len'], + nsecs_secs(hunk['queue_t']), + nsecs_nsecs(hunk['queue_t'])/1000, + diff_msec(hunk['queue_t'], hunk['xmit_t']), + diff_msec(hunk['xmit_t'], hunk['free_t'])) + +# Format for displaying rx packet processing +PF_IRQ_ENTRY= " irq_entry(+%.3fmsec irq=%d:%s)" +PF_SOFT_ENTRY=" softirq_entry(+%.3fmsec)" +PF_NAPI_POLL= " napi_poll_exit(+%.3fmsec %s)" +PF_JOINT= " |" +PF_WJOINT= " | |" +PF_NET_RECV= " |---netif_receive_skb(+%.3fmsec skb=%x len=%d)" +PF_NET_RX= " |---netif_rx(+%.3fmsec skb=%x)" +PF_CPY_DGRAM= " | skb_copy_datagram_iovec(+%.3fmsec %d:%s)" +PF_KFREE_SKB= " | kfree_skb(+%.3fmsec location=%x)" +PF_CONS_SKB= " | consume_skb(+%.3fmsec)" + +# Display a process of received packets and interrputs associated with +# a NET_RX softirq +def print_receive(hunk): + show_hunk = 0 + irq_list = hunk['irq_list'] + cpu = irq_list[0]['cpu'] + base_t = irq_list[0]['irq_ent_t'] + # check if this hunk should be showed + if dev != 0: + for i in range(len(irq_list)): + if irq_list[i]['name'].find(dev) >= 0: + show_hunk = 1 + break + else: + show_hunk = 1 + if show_hunk == 0: + return + + print "%d.%06dsec cpu=%d" % \ + (nsecs_secs(base_t), nsecs_nsecs(base_t)/1000, cpu) + for i in range(len(irq_list)): + print PF_IRQ_ENTRY % \ + (diff_msec(base_t, irq_list[i]['irq_ent_t']), + irq_list[i]['irq'], irq_list[i]['name']) + print PF_JOINT + irq_event_list = irq_list[i]['event_list'] + for j in range(len(irq_event_list)): + irq_event = irq_event_list[j] + if irq_event['event'] == 'netif_rx': + print PF_NET_RX % \ + (diff_msec(base_t, irq_event['time']), + irq_event['skbaddr']) + print PF_JOINT + print PF_SOFT_ENTRY % \ + diff_msec(base_t, hunk['sirq_ent_t']) + print PF_JOINT + event_list = hunk['event_list'] + for i in range(len(event_list)): + event = event_list[i] + if event['event_name'] == 'napi_poll': + print PF_NAPI_POLL % \ + (diff_msec(base_t, event['event_t']), event['dev']) + if i == len(event_list) - 1: + print "" + else: + print PF_JOINT + else: + print PF_NET_RECV % \ + (diff_msec(base_t, event['event_t']), event['skbaddr'], + event['len']) + if 'comm' in event.keys(): + print PF_WJOINT + print PF_CPY_DGRAM % \ + (diff_msec(base_t, event['comm_t']), + event['pid'], event['comm']) + elif 'handle' in event.keys(): + print PF_WJOINT + if event['handle'] == "kfree_skb": + print PF_KFREE_SKB % \ + (diff_msec(base_t, + event['comm_t']), + event['location']) + elif event['handle'] == "consume_skb": + print PF_CONS_SKB % \ + diff_msec(base_t, + event['comm_t']) + print PF_JOINT + +def trace_begin(): + global show_tx + global show_rx + global dev + global debug + + for i in range(len(sys.argv)): + if i == 0: + continue + arg = sys.argv[i] + if arg == 'tx': + show_tx = 1 + elif arg =='rx': + show_rx = 1 + elif arg.find('dev=',0, 4) >= 0: + dev = arg[4:] + elif arg == 'debug': + debug = 1 + if show_tx == 0 and show_rx == 0: + show_tx = 1 + show_rx = 1 + +def trace_end(): + # order all events in time + all_event_list.sort(lambda a,b :cmp(a[EINFO_IDX_TIME], + b[EINFO_IDX_TIME])) + # process all events + for i in range(len(all_event_list)): + event_info = all_event_list[i] + name = event_info[EINFO_IDX_NAME] + if name == 'irq__softirq_exit': + handle_irq_softirq_exit(event_info) + elif name == 'irq__softirq_entry': + handle_irq_softirq_entry(event_info) + elif name == 'irq__softirq_raise': + handle_irq_softirq_raise(event_info) + elif name == 'irq__irq_handler_entry': + handle_irq_handler_entry(event_info) + elif name == 'irq__irq_handler_exit': + handle_irq_handler_exit(event_info) + elif name == 'napi__napi_poll': + handle_napi_poll(event_info) + elif name == 'net__netif_receive_skb': + handle_netif_receive_skb(event_info) + elif name == 'net__netif_rx': + handle_netif_rx(event_info) + elif name == 'skb__skb_copy_datagram_iovec': + handle_skb_copy_datagram_iovec(event_info) + elif name == 'net__net_dev_queue': + handle_net_dev_queue(event_info) + elif name == 'net__net_dev_xmit': + handle_net_dev_xmit(event_info) + elif name == 'skb__kfree_skb': + handle_kfree_skb(event_info) + elif name == 'skb__consume_skb': + handle_consume_skb(event_info) + # display receive hunks + if show_rx: + for i in range(len(receive_hunk_list)): + print_receive(receive_hunk_list[i]) + # display transmit hunks + if show_tx: + print " dev len Qdisc " \ + " netdevice free" + for i in range(len(tx_free_list)): + print_transmit(tx_free_list[i]) + if debug: + print "debug buffer status" + print "----------------------------" + print "xmit Qdisc:remain:%d overflow:%d" % \ + (len(tx_queue_list), of_count_tx_queue_list) + print "xmit netdevice:remain:%d overflow:%d" % \ + (len(tx_xmit_list), of_count_tx_xmit_list) + print "receive:remain:%d overflow:%d" % \ + (len(rx_skb_list), of_count_rx_skb_list) + +# called from perf, when it finds a correspoinding event +def irq__softirq_entry(name, context, cpu, sec, nsec, pid, comm, vec): + if symbol_str("irq__softirq_entry", "vec", vec) != "NET_RX": + return + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, vec) + all_event_list.append(event_info) + +def irq__softirq_exit(name, context, cpu, sec, nsec, pid, comm, vec): + if symbol_str("irq__softirq_entry", "vec", vec) != "NET_RX": + return + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, vec) + all_event_list.append(event_info) + +def irq__softirq_raise(name, context, cpu, sec, nsec, pid, comm, vec): + if symbol_str("irq__softirq_entry", "vec", vec) != "NET_RX": + return + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, vec) + all_event_list.append(event_info) + +def irq__irq_handler_entry(name, context, cpu, sec, nsec, pid, comm, + irq, irq_name): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, + irq, irq_name) + all_event_list.append(event_info) + +def irq__irq_handler_exit(name, context, cpu, sec, nsec, pid, comm, irq, ret): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, irq, ret) + all_event_list.append(event_info) + +def napi__napi_poll(name, context, cpu, sec, nsec, pid, comm, napi, dev_name): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, + napi, dev_name) + all_event_list.append(event_info) + +def net__netif_receive_skb(name, context, cpu, sec, nsec, pid, comm, skbaddr, + skblen, dev_name): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, + skbaddr, skblen, dev_name) + all_event_list.append(event_info) + +def net__netif_rx(name, context, cpu, sec, nsec, pid, comm, skbaddr, + skblen, dev_name): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, + skbaddr, skblen, dev_name) + all_event_list.append(event_info) + +def net__net_dev_queue(name, context, cpu, sec, nsec, pid, comm, + skbaddr, skblen, dev_name): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, + skbaddr, skblen, dev_name) + all_event_list.append(event_info) + +def net__net_dev_xmit(name, context, cpu, sec, nsec, pid, comm, + skbaddr, skblen, rc, dev_name): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, + skbaddr, skblen, rc ,dev_name) + all_event_list.append(event_info) + +def skb__kfree_skb(name, context, cpu, sec, nsec, pid, comm, + skbaddr, protocol, location): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, + skbaddr, protocol, location) + all_event_list.append(event_info) + +def skb__consume_skb(name, context, cpu, sec, nsec, pid, comm, skbaddr): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, + skbaddr) + all_event_list.append(event_info) + +def skb__skb_copy_datagram_iovec(name, context, cpu, sec, nsec, pid, comm, + skbaddr, skblen): + event_info = (name, context, cpu, nsecs(sec, nsec), pid, comm, + skbaddr, skblen) + all_event_list.append(event_info) + +def handle_irq_handler_entry(event_info): + (name, context, cpu, time, pid, comm, irq, irq_name) = event_info + if cpu not in irq_dic.keys(): + irq_dic[cpu] = [] + irq_record = {'irq':irq, 'name':irq_name, 'cpu':cpu, 'irq_ent_t':time} + irq_dic[cpu].append(irq_record) + +def handle_irq_handler_exit(event_info): + (name, context, cpu, time, pid, comm, irq, ret) = event_info + if cpu not in irq_dic.keys(): + return + irq_record = irq_dic[cpu].pop() + if irq != irq_record['irq']: + return + irq_record.update({'irq_ext_t':time}) + # if an irq doesn't include NET_RX softirq, drop. + if 'event_list' in irq_record.keys(): + irq_dic[cpu].append(irq_record) + +def handle_irq_softirq_raise(event_info): + (name, context, cpu, time, pid, comm, vec) = event_info + if cpu not in irq_dic.keys() \ + or len(irq_dic[cpu]) == 0: + return + irq_record = irq_dic[cpu].pop() + if 'event_list' in irq_record.keys(): + irq_event_list = irq_record['event_list'] + else: + irq_event_list = [] + irq_event_list.append({'time':time, 'event':'sirq_raise'}) + irq_record.update({'event_list':irq_event_list}) + irq_dic[cpu].append(irq_record) + +def handle_irq_softirq_entry(event_info): + (name, context, cpu, time, pid, comm, vec) = event_info + net_rx_dic[cpu] = {'sirq_ent_t':time, 'event_list':[]} + +def handle_irq_softirq_exit(event_info): + (name, context, cpu, time, pid, comm, vec) = event_info + irq_list = [] + event_list = 0 + if cpu in irq_dic.keys(): + irq_list = irq_dic[cpu] + del irq_dic[cpu] + if cpu in net_rx_dic.keys(): + sirq_ent_t = net_rx_dic[cpu]['sirq_ent_t'] + event_list = net_rx_dic[cpu]['event_list'] + del net_rx_dic[cpu] + if irq_list == [] or event_list == 0: + return + rec_data = {'sirq_ent_t':sirq_ent_t, 'sirq_ext_t':time, + 'irq_list':irq_list, 'event_list':event_list} + # merge information realted to a NET_RX softirq + receive_hunk_list.append(rec_data) + +def handle_napi_poll(event_info): + (name, context, cpu, time, pid, comm, napi, dev_name) = event_info + if cpu in net_rx_dic.keys(): + event_list = net_rx_dic[cpu]['event_list'] + rec_data = {'event_name':'napi_poll', + 'dev':dev_name, 'event_t':time} + event_list.append(rec_data) + +def handle_netif_rx(event_info): + (name, context, cpu, time, pid, comm, + skbaddr, skblen, dev_name) = event_info + if cpu not in irq_dic.keys() \ + or len(irq_dic[cpu]) == 0: + return + irq_record = irq_dic[cpu].pop() + if 'event_list' in irq_record.keys(): + irq_event_list = irq_record['event_list'] + else: + irq_event_list = [] + irq_event_list.append({'time':time, 'event':'netif_rx', + 'skbaddr':skbaddr, 'skblen':skblen, 'dev_name':dev_name}) + irq_record.update({'event_list':irq_event_list}) + irq_dic[cpu].append(irq_record) + +def handle_netif_receive_skb(event_info): + global of_count_rx_skb_list + + (name, context, cpu, time, pid, comm, + skbaddr, skblen, dev_name) = event_info + if cpu in net_rx_dic.keys(): + rec_data = {'event_name':'netif_receive_skb', + 'event_t':time, 'skbaddr':skbaddr, 'len':skblen} + event_list = net_rx_dic[cpu]['event_list'] + event_list.append(rec_data) + rx_skb_list.insert(0, rec_data) + if len(rx_skb_list) > buffer_budget: + rx_skb_list.pop() + of_count_rx_skb_list += 1 + +def handle_net_dev_queue(event_info): + global of_count_tx_queue_list + + (name, context, cpu, time, pid, comm, + skbaddr, skblen, dev_name) = event_info + skb = {'dev':dev_name, 'skbaddr':skbaddr, 'len':skblen, 'queue_t':time} + tx_queue_list.insert(0, skb) + if len(tx_queue_list) > buffer_budget: + tx_queue_list.pop() + of_count_tx_queue_list += 1 + +def handle_net_dev_xmit(event_info): + global of_count_tx_xmit_list + + (name, context, cpu, time, pid, comm, + skbaddr, skblen, rc, dev_name) = event_info + if rc == 0: # NETDEV_TX_OK + for i in range(len(tx_queue_list)): + skb = tx_queue_list[i] + if skb['skbaddr'] == skbaddr: + skb['xmit_t'] = time + tx_xmit_list.insert(0, skb) + del tx_queue_list[i] + if len(tx_xmit_list) > buffer_budget: + tx_xmit_list.pop() + of_count_tx_xmit_list += 1 + return + +def handle_kfree_skb(event_info): + (name, context, cpu, time, pid, comm, + skbaddr, protocol, location) = event_info + for i in range(len(tx_queue_list)): + skb = tx_queue_list[i] + if skb['skbaddr'] == skbaddr: + del tx_queue_list[i] + return + for i in range(len(tx_xmit_list)): + skb = tx_xmit_list[i] + if skb['skbaddr'] == skbaddr: + skb['free_t'] = time + tx_free_list.append(skb) + del tx_xmit_list[i] + return + for i in range(len(rx_skb_list)): + rec_data = rx_skb_list[i] + if rec_data['skbaddr'] == skbaddr: + rec_data.update({'handle':"kfree_skb", + 'comm':comm, 'pid':pid, 'comm_t':time}) + del rx_skb_list[i] + return + +def handle_consume_skb(event_info): + (name, context, cpu, time, pid, comm, skbaddr) = event_info + for i in range(len(tx_xmit_list)): + skb = tx_xmit_list[i] + if skb['skbaddr'] == skbaddr: + skb['free_t'] = time + tx_free_list.append(skb) + del tx_xmit_list[i] + return + +def handle_skb_copy_datagram_iovec(event_info): + (name, context, cpu, time, pid, comm, skbaddr, skblen) = event_info + for i in range(len(rx_skb_list)): + rec_data = rx_skb_list[i] + if skbaddr == rec_data['skbaddr']: + rec_data.update({'handle':"skb_copy_datagram_iovec", + 'comm':comm, 'pid':pid, 'comm_t':time}) + del rx_skb_list[i] + return diff --git a/tools/perf/scripts/python/sched-migration.py b/tools/perf/scripts/python/sched-migration.py new file mode 100644 index 00000000..74d55ec0 --- /dev/null +++ b/tools/perf/scripts/python/sched-migration.py @@ -0,0 +1,461 @@ +#!/usr/bin/python +# +# Cpu task migration overview toy +# +# Copyright (C) 2010 Frederic Weisbecker <fweisbec@gmail.com> +# +# perf script event handlers have been generated by perf script -g python +# +# This software is distributed under the terms of the GNU General +# Public License ("GPL") version 2 as published by the Free Software +# Foundation. + + +import os +import sys + +from collections import defaultdict +from UserList import UserList + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/Perf-Trace-Util/lib/Perf/Trace') +sys.path.append('scripts/python/Perf-Trace-Util/lib/Perf/Trace') + +from perf_trace_context import * +from Core import * +from SchedGui import * + + +threads = { 0 : "idle"} + +def thread_name(pid): + return "%s:%d" % (threads[pid], pid) + +class RunqueueEventUnknown: + @staticmethod + def color(): + return None + + def __repr__(self): + return "unknown" + +class RunqueueEventSleep: + @staticmethod + def color(): + return (0, 0, 0xff) + + def __init__(self, sleeper): + self.sleeper = sleeper + + def __repr__(self): + return "%s gone to sleep" % thread_name(self.sleeper) + +class RunqueueEventWakeup: + @staticmethod + def color(): + return (0xff, 0xff, 0) + + def __init__(self, wakee): + self.wakee = wakee + + def __repr__(self): + return "%s woke up" % thread_name(self.wakee) + +class RunqueueEventFork: + @staticmethod + def color(): + return (0, 0xff, 0) + + def __init__(self, child): + self.child = child + + def __repr__(self): + return "new forked task %s" % thread_name(self.child) + +class RunqueueMigrateIn: + @staticmethod + def color(): + return (0, 0xf0, 0xff) + + def __init__(self, new): + self.new = new + + def __repr__(self): + return "task migrated in %s" % thread_name(self.new) + +class RunqueueMigrateOut: + @staticmethod + def color(): + return (0xff, 0, 0xff) + + def __init__(self, old): + self.old = old + + def __repr__(self): + return "task migrated out %s" % thread_name(self.old) + +class RunqueueSnapshot: + def __init__(self, tasks = [0], event = RunqueueEventUnknown()): + self.tasks = tuple(tasks) + self.event = event + + def sched_switch(self, prev, prev_state, next): + event = RunqueueEventUnknown() + + if taskState(prev_state) == "R" and next in self.tasks \ + and prev in self.tasks: + return self + + if taskState(prev_state) != "R": + event = RunqueueEventSleep(prev) + + next_tasks = list(self.tasks[:]) + if prev in self.tasks: + if taskState(prev_state) != "R": + next_tasks.remove(prev) + elif taskState(prev_state) == "R": + next_tasks.append(prev) + + if next not in next_tasks: + next_tasks.append(next) + + return RunqueueSnapshot(next_tasks, event) + + def migrate_out(self, old): + if old not in self.tasks: + return self + next_tasks = [task for task in self.tasks if task != old] + + return RunqueueSnapshot(next_tasks, RunqueueMigrateOut(old)) + + def __migrate_in(self, new, event): + if new in self.tasks: + self.event = event + return self + next_tasks = self.tasks[:] + tuple([new]) + + return RunqueueSnapshot(next_tasks, event) + + def migrate_in(self, new): + return self.__migrate_in(new, RunqueueMigrateIn(new)) + + def wake_up(self, new): + return self.__migrate_in(new, RunqueueEventWakeup(new)) + + def wake_up_new(self, new): + return self.__migrate_in(new, RunqueueEventFork(new)) + + def load(self): + """ Provide the number of tasks on the runqueue. + Don't count idle""" + return len(self.tasks) - 1 + + def __repr__(self): + ret = self.tasks.__repr__() + ret += self.origin_tostring() + + return ret + +class TimeSlice: + def __init__(self, start, prev): + self.start = start + self.prev = prev + self.end = start + # cpus that triggered the event + self.event_cpus = [] + if prev is not None: + self.total_load = prev.total_load + self.rqs = prev.rqs.copy() + else: + self.rqs = defaultdict(RunqueueSnapshot) + self.total_load = 0 + + def __update_total_load(self, old_rq, new_rq): + diff = new_rq.load() - old_rq.load() + self.total_load += diff + + def sched_switch(self, ts_list, prev, prev_state, next, cpu): + old_rq = self.prev.rqs[cpu] + new_rq = old_rq.sched_switch(prev, prev_state, next) + + if old_rq is new_rq: + return + + self.rqs[cpu] = new_rq + self.__update_total_load(old_rq, new_rq) + ts_list.append(self) + self.event_cpus = [cpu] + + def migrate(self, ts_list, new, old_cpu, new_cpu): + if old_cpu == new_cpu: + return + old_rq = self.prev.rqs[old_cpu] + out_rq = old_rq.migrate_out(new) + self.rqs[old_cpu] = out_rq + self.__update_total_load(old_rq, out_rq) + + new_rq = self.prev.rqs[new_cpu] + in_rq = new_rq.migrate_in(new) + self.rqs[new_cpu] = in_rq + self.__update_total_load(new_rq, in_rq) + + ts_list.append(self) + + if old_rq is not out_rq: + self.event_cpus.append(old_cpu) + self.event_cpus.append(new_cpu) + + def wake_up(self, ts_list, pid, cpu, fork): + old_rq = self.prev.rqs[cpu] + if fork: + new_rq = old_rq.wake_up_new(pid) + else: + new_rq = old_rq.wake_up(pid) + + if new_rq is old_rq: + return + self.rqs[cpu] = new_rq + self.__update_total_load(old_rq, new_rq) + ts_list.append(self) + self.event_cpus = [cpu] + + def next(self, t): + self.end = t + return TimeSlice(t, self) + +class TimeSliceList(UserList): + def __init__(self, arg = []): + self.data = arg + + def get_time_slice(self, ts): + if len(self.data) == 0: + slice = TimeSlice(ts, TimeSlice(-1, None)) + else: + slice = self.data[-1].next(ts) + return slice + + def find_time_slice(self, ts): + start = 0 + end = len(self.data) + found = -1 + searching = True + while searching: + if start == end or start == end - 1: + searching = False + + i = (end + start) / 2 + if self.data[i].start <= ts and self.data[i].end >= ts: + found = i + end = i + continue + + if self.data[i].end < ts: + start = i + + elif self.data[i].start > ts: + end = i + + return found + + def set_root_win(self, win): + self.root_win = win + + def mouse_down(self, cpu, t): + idx = self.find_time_slice(t) + if idx == -1: + return + + ts = self[idx] + rq = ts.rqs[cpu] + raw = "CPU: %d\n" % cpu + raw += "Last event : %s\n" % rq.event.__repr__() + raw += "Timestamp : %d.%06d\n" % (ts.start / (10 ** 9), (ts.start % (10 ** 9)) / 1000) + raw += "Duration : %6d us\n" % ((ts.end - ts.start) / (10 ** 6)) + raw += "Load = %d\n" % rq.load() + for t in rq.tasks: + raw += "%s \n" % thread_name(t) + + self.root_win.update_summary(raw) + + def update_rectangle_cpu(self, slice, cpu): + rq = slice.rqs[cpu] + + if slice.total_load != 0: + load_rate = rq.load() / float(slice.total_load) + else: + load_rate = 0 + + red_power = int(0xff - (0xff * load_rate)) + color = (0xff, red_power, red_power) + + top_color = None + + if cpu in slice.event_cpus: + top_color = rq.event.color() + + self.root_win.paint_rectangle_zone(cpu, color, top_color, slice.start, slice.end) + + def fill_zone(self, start, end): + i = self.find_time_slice(start) + if i == -1: + return + + for i in xrange(i, len(self.data)): + timeslice = self.data[i] + if timeslice.start > end: + return + + for cpu in timeslice.rqs: + self.update_rectangle_cpu(timeslice, cpu) + + def interval(self): + if len(self.data) == 0: + return (0, 0) + + return (self.data[0].start, self.data[-1].end) + + def nr_rectangles(self): + last_ts = self.data[-1] + max_cpu = 0 + for cpu in last_ts.rqs: + if cpu > max_cpu: + max_cpu = cpu + return max_cpu + + +class SchedEventProxy: + def __init__(self): + self.current_tsk = defaultdict(lambda : -1) + self.timeslices = TimeSliceList() + + def sched_switch(self, headers, prev_comm, prev_pid, prev_prio, prev_state, + next_comm, next_pid, next_prio): + """ Ensure the task we sched out this cpu is really the one + we logged. Otherwise we may have missed traces """ + + on_cpu_task = self.current_tsk[headers.cpu] + + if on_cpu_task != -1 and on_cpu_task != prev_pid: + print "Sched switch event rejected ts: %s cpu: %d prev: %s(%d) next: %s(%d)" % \ + (headers.ts_format(), headers.cpu, prev_comm, prev_pid, next_comm, next_pid) + + threads[prev_pid] = prev_comm + threads[next_pid] = next_comm + self.current_tsk[headers.cpu] = next_pid + + ts = self.timeslices.get_time_slice(headers.ts()) + ts.sched_switch(self.timeslices, prev_pid, prev_state, next_pid, headers.cpu) + + def migrate(self, headers, pid, prio, orig_cpu, dest_cpu): + ts = self.timeslices.get_time_slice(headers.ts()) + ts.migrate(self.timeslices, pid, orig_cpu, dest_cpu) + + def wake_up(self, headers, comm, pid, success, target_cpu, fork): + if success == 0: + return + ts = self.timeslices.get_time_slice(headers.ts()) + ts.wake_up(self.timeslices, pid, target_cpu, fork) + + +def trace_begin(): + global parser + parser = SchedEventProxy() + +def trace_end(): + app = wx.App(False) + timeslices = parser.timeslices + frame = RootFrame(timeslices, "Migration") + app.MainLoop() + +def sched__sched_stat_runtime(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, runtime, vruntime): + pass + +def sched__sched_stat_iowait(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, delay): + pass + +def sched__sched_stat_sleep(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, delay): + pass + +def sched__sched_stat_wait(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, delay): + pass + +def sched__sched_process_fork(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + parent_comm, parent_pid, child_comm, child_pid): + pass + +def sched__sched_process_wait(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, prio): + pass + +def sched__sched_process_exit(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, prio): + pass + +def sched__sched_process_free(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, prio): + pass + +def sched__sched_migrate_task(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, prio, orig_cpu, + dest_cpu): + headers = EventHeaders(common_cpu, common_secs, common_nsecs, + common_pid, common_comm) + parser.migrate(headers, pid, prio, orig_cpu, dest_cpu) + +def sched__sched_switch(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + prev_comm, prev_pid, prev_prio, prev_state, + next_comm, next_pid, next_prio): + + headers = EventHeaders(common_cpu, common_secs, common_nsecs, + common_pid, common_comm) + parser.sched_switch(headers, prev_comm, prev_pid, prev_prio, prev_state, + next_comm, next_pid, next_prio) + +def sched__sched_wakeup_new(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, prio, success, + target_cpu): + headers = EventHeaders(common_cpu, common_secs, common_nsecs, + common_pid, common_comm) + parser.wake_up(headers, comm, pid, success, target_cpu, 1) + +def sched__sched_wakeup(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, prio, success, + target_cpu): + headers = EventHeaders(common_cpu, common_secs, common_nsecs, + common_pid, common_comm) + parser.wake_up(headers, comm, pid, success, target_cpu, 0) + +def sched__sched_wait_task(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid, prio): + pass + +def sched__sched_kthread_stop_ret(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + ret): + pass + +def sched__sched_kthread_stop(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + comm, pid): + pass + +def trace_unhandled(event_name, context, common_cpu, common_secs, common_nsecs, + common_pid, common_comm): + pass diff --git a/tools/perf/scripts/python/sctop.py b/tools/perf/scripts/python/sctop.py new file mode 100644 index 00000000..42c267e2 --- /dev/null +++ b/tools/perf/scripts/python/sctop.py @@ -0,0 +1,75 @@ +# system call top +# (c) 2010, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 +# +# Periodically displays system-wide system call totals, broken down by +# syscall. If a [comm] arg is specified, only syscalls called by +# [comm] are displayed. If an [interval] arg is specified, the display +# will be refreshed every [interval] seconds. The default interval is +# 3 seconds. + +import os, sys, thread, time + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/Perf-Trace-Util/lib/Perf/Trace') + +from perf_trace_context import * +from Core import * +from Util import * + +usage = "perf script -s sctop.py [comm] [interval]\n"; + +for_comm = None +default_interval = 3 +interval = default_interval + +if len(sys.argv) > 3: + sys.exit(usage) + +if len(sys.argv) > 2: + for_comm = sys.argv[1] + interval = int(sys.argv[2]) +elif len(sys.argv) > 1: + try: + interval = int(sys.argv[1]) + except ValueError: + for_comm = sys.argv[1] + interval = default_interval + +syscalls = autodict() + +def trace_begin(): + thread.start_new_thread(print_syscall_totals, (interval,)) + pass + +def raw_syscalls__sys_enter(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + id, args): + if for_comm is not None: + if common_comm != for_comm: + return + try: + syscalls[id] += 1 + except TypeError: + syscalls[id] = 1 + +def print_syscall_totals(interval): + while 1: + clear_term() + if for_comm is not None: + print "\nsyscall events for %s:\n\n" % (for_comm), + else: + print "\nsyscall events:\n\n", + + print "%-40s %10s\n" % ("event", "count"), + print "%-40s %10s\n" % ("----------------------------------------", \ + "----------"), + + for id, val in sorted(syscalls.iteritems(), key = lambda(k, v): (v, k), \ + reverse = True): + try: + print "%-40s %10d\n" % (syscall_name(id), val), + except TypeError: + pass + syscalls.clear() + time.sleep(interval) diff --git a/tools/perf/scripts/python/syscall-counts-by-pid.py b/tools/perf/scripts/python/syscall-counts-by-pid.py new file mode 100644 index 00000000..c64d1c55 --- /dev/null +++ b/tools/perf/scripts/python/syscall-counts-by-pid.py @@ -0,0 +1,69 @@ +# system call counts, by pid +# (c) 2010, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 +# +# Displays system-wide system call totals, broken down by syscall. +# If a [comm] arg is specified, only syscalls called by [comm] are displayed. + +import os, sys + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/Perf-Trace-Util/lib/Perf/Trace') + +from perf_trace_context import * +from Core import * +from Util import syscall_name + +usage = "perf script -s syscall-counts-by-pid.py [comm]\n"; + +for_comm = None +for_pid = None + +if len(sys.argv) > 2: + sys.exit(usage) + +if len(sys.argv) > 1: + try: + for_pid = int(sys.argv[1]) + except: + for_comm = sys.argv[1] + +syscalls = autodict() + +def trace_begin(): + print "Press control+C to stop and show the summary" + +def trace_end(): + print_syscall_totals() + +def raw_syscalls__sys_enter(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + id, args): + + if (for_comm and common_comm != for_comm) or \ + (for_pid and common_pid != for_pid ): + return + try: + syscalls[common_comm][common_pid][id] += 1 + except TypeError: + syscalls[common_comm][common_pid][id] = 1 + +def print_syscall_totals(): + if for_comm is not None: + print "\nsyscall events for %s:\n\n" % (for_comm), + else: + print "\nsyscall events by comm/pid:\n\n", + + print "%-40s %10s\n" % ("comm [pid]/syscalls", "count"), + print "%-40s %10s\n" % ("----------------------------------------", \ + "----------"), + + comm_keys = syscalls.keys() + for comm in comm_keys: + pid_keys = syscalls[comm].keys() + for pid in pid_keys: + print "\n%s [%d]\n" % (comm, pid), + id_keys = syscalls[comm][pid].keys() + for id, val in sorted(syscalls[comm][pid].iteritems(), \ + key = lambda(k, v): (v, k), reverse = True): + print " %-38s %10d\n" % (syscall_name(id), val), diff --git a/tools/perf/scripts/python/syscall-counts.py b/tools/perf/scripts/python/syscall-counts.py new file mode 100644 index 00000000..b435d3f1 --- /dev/null +++ b/tools/perf/scripts/python/syscall-counts.py @@ -0,0 +1,59 @@ +# system call counts +# (c) 2010, Tom Zanussi <tzanussi@gmail.com> +# Licensed under the terms of the GNU GPL License version 2 +# +# Displays system-wide system call totals, broken down by syscall. +# If a [comm] arg is specified, only syscalls called by [comm] are displayed. + +import os +import sys + +sys.path.append(os.environ['PERF_EXEC_PATH'] + \ + '/scripts/python/Perf-Trace-Util/lib/Perf/Trace') + +from perf_trace_context import * +from Core import * +from Util import syscall_name + +usage = "perf script -s syscall-counts.py [comm]\n"; + +for_comm = None + +if len(sys.argv) > 2: + sys.exit(usage) + +if len(sys.argv) > 1: + for_comm = sys.argv[1] + +syscalls = autodict() + +def trace_begin(): + print "Press control+C to stop and show the summary" + +def trace_end(): + print_syscall_totals() + +def raw_syscalls__sys_enter(event_name, context, common_cpu, + common_secs, common_nsecs, common_pid, common_comm, + id, args): + if for_comm is not None: + if common_comm != for_comm: + return + try: + syscalls[id] += 1 + except TypeError: + syscalls[id] = 1 + +def print_syscall_totals(): + if for_comm is not None: + print "\nsyscall events for %s:\n\n" % (for_comm), + else: + print "\nsyscall events:\n\n", + + print "%-40s %10s\n" % ("event", "count"), + print "%-40s %10s\n" % ("----------------------------------------", \ + "-----------"), + + for id, val in sorted(syscalls.iteritems(), key = lambda(k, v): (v, k), \ + reverse = True): + print "%-40s %10d\n" % (syscall_name(id), val), diff --git a/tools/perf/util/PERF-VERSION-GEN b/tools/perf/util/PERF-VERSION-GEN new file mode 100755 index 00000000..ad73300f --- /dev/null +++ b/tools/perf/util/PERF-VERSION-GEN @@ -0,0 +1,42 @@ +#!/bin/sh + +if [ $# -eq 1 ] ; then + OUTPUT=$1 +fi + +GVF=${OUTPUT}PERF-VERSION-FILE + +LF=' +' + +# First check if there is a .git to get the version from git describe +# otherwise try to get the version from the kernel makefile +if test -d ../../.git -o -f ../../.git && + VN=$(git describe --abbrev=4 HEAD 2>/dev/null) && + case "$VN" in + *$LF*) (exit 1) ;; + v[0-9]*) + git update-index -q --refresh + test -z "$(git diff-index --name-only HEAD --)" || + VN="$VN-dirty" ;; + esac +then + VN=$(echo "$VN" | sed -e 's/-/./g'); +else + VN=$(MAKEFLAGS= make -sC ../.. kernelversion) +fi + +VN=$(expr "$VN" : v*'\(.*\)') + +if test -r $GVF +then + VC=$(sed -e 's/^PERF_VERSION = //' <$GVF) +else + VC=unset +fi +test "$VN" = "$VC" || { + echo >&2 "PERF_VERSION = $VN" + echo "PERF_VERSION = $VN" >$GVF +} + + diff --git a/tools/perf/util/abspath.c b/tools/perf/util/abspath.c new file mode 100644 index 00000000..0e76affe --- /dev/null +++ b/tools/perf/util/abspath.c @@ -0,0 +1,37 @@ +#include "cache.h" + +static const char *get_pwd_cwd(void) +{ + static char cwd[PATH_MAX + 1]; + char *pwd; + struct stat cwd_stat, pwd_stat; + if (getcwd(cwd, PATH_MAX) == NULL) + return NULL; + pwd = getenv("PWD"); + if (pwd && strcmp(pwd, cwd)) { + stat(cwd, &cwd_stat); + if (!stat(pwd, &pwd_stat) && + pwd_stat.st_dev == cwd_stat.st_dev && + pwd_stat.st_ino == cwd_stat.st_ino) { + strlcpy(cwd, pwd, PATH_MAX); + } + } + return cwd; +} + +const char *make_nonrelative_path(const char *path) +{ + static char buf[PATH_MAX + 1]; + + if (is_absolute_path(path)) { + if (strlcpy(buf, path, PATH_MAX) >= PATH_MAX) + die("Too long path: %.*s", 60, path); + } else { + const char *cwd = get_pwd_cwd(); + if (!cwd) + die("Cannot determine the current working directory"); + if (snprintf(buf, PATH_MAX, "%s/%s", cwd, path) >= PATH_MAX) + die("Too long path: %.*s", 60, path); + } + return buf; +} diff --git a/tools/perf/util/alias.c b/tools/perf/util/alias.c new file mode 100644 index 00000000..b8144e80 --- /dev/null +++ b/tools/perf/util/alias.c @@ -0,0 +1,77 @@ +#include "cache.h" + +static const char *alias_key; +static char *alias_val; + +static int alias_lookup_cb(const char *k, const char *v, void *cb __used) +{ + if (!prefixcmp(k, "alias.") && !strcmp(k+6, alias_key)) { + if (!v) + return config_error_nonbool(k); + alias_val = strdup(v); + return 0; + } + return 0; +} + +char *alias_lookup(const char *alias) +{ + alias_key = alias; + alias_val = NULL; + perf_config(alias_lookup_cb, NULL); + return alias_val; +} + +int split_cmdline(char *cmdline, const char ***argv) +{ + int src, dst, count = 0, size = 16; + char quoted = 0; + + *argv = malloc(sizeof(char*) * size); + + /* split alias_string */ + (*argv)[count++] = cmdline; + for (src = dst = 0; cmdline[src];) { + char c = cmdline[src]; + if (!quoted && isspace(c)) { + cmdline[dst++] = 0; + while (cmdline[++src] + && isspace(cmdline[src])) + ; /* skip */ + if (count >= size) { + size += 16; + *argv = realloc(*argv, sizeof(char*) * size); + } + (*argv)[count++] = cmdline + dst; + } else if (!quoted && (c == '\'' || c == '"')) { + quoted = c; + src++; + } else if (c == quoted) { + quoted = 0; + src++; + } else { + if (c == '\\' && quoted != '\'') { + src++; + c = cmdline[src]; + if (!c) { + free(*argv); + *argv = NULL; + return error("cmdline ends with \\"); + } + } + cmdline[dst++] = c; + src++; + } + } + + cmdline[dst] = 0; + + if (quoted) { + free(*argv); + *argv = NULL; + return error("unclosed quote"); + } + + return count; +} + diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c new file mode 100644 index 00000000..e01af2b1 --- /dev/null +++ b/tools/perf/util/annotate.c @@ -0,0 +1,605 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Parts came from builtin-annotate.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ + +#include "util.h" +#include "build-id.h" +#include "color.h" +#include "cache.h" +#include "symbol.h" +#include "debug.h" +#include "annotate.h" +#include <pthread.h> + +int symbol__annotate_init(struct map *map __used, struct symbol *sym) +{ + struct annotation *notes = symbol__annotation(sym); + pthread_mutex_init(¬es->lock, NULL); + return 0; +} + +int symbol__alloc_hist(struct symbol *sym, int nevents) +{ + struct annotation *notes = symbol__annotation(sym); + size_t sizeof_sym_hist = (sizeof(struct sym_hist) + + (sym->end - sym->start) * sizeof(u64)); + + notes->src = zalloc(sizeof(*notes->src) + nevents * sizeof_sym_hist); + if (notes->src == NULL) + return -1; + notes->src->sizeof_sym_hist = sizeof_sym_hist; + notes->src->nr_histograms = nevents; + INIT_LIST_HEAD(¬es->src->source); + return 0; +} + +void symbol__annotate_zero_histograms(struct symbol *sym) +{ + struct annotation *notes = symbol__annotation(sym); + + pthread_mutex_lock(¬es->lock); + if (notes->src != NULL) + memset(notes->src->histograms, 0, + notes->src->nr_histograms * notes->src->sizeof_sym_hist); + pthread_mutex_unlock(¬es->lock); +} + +int symbol__inc_addr_samples(struct symbol *sym, struct map *map, + int evidx, u64 addr) +{ + unsigned offset; + struct annotation *notes; + struct sym_hist *h; + + notes = symbol__annotation(sym); + if (notes->src == NULL) + return -ENOMEM; + + pr_debug3("%s: addr=%#" PRIx64 "\n", __func__, map->unmap_ip(map, addr)); + + if (addr >= sym->end) + return 0; + + offset = addr - sym->start; + h = annotation__histogram(notes, evidx); + h->sum++; + h->addr[offset]++; + + pr_debug3("%#" PRIx64 " %s: period++ [addr: %#" PRIx64 ", %#" PRIx64 + ", evidx=%d] => %" PRIu64 "\n", sym->start, sym->name, + addr, addr - sym->start, evidx, h->addr[offset]); + return 0; +} + +static struct objdump_line *objdump_line__new(s64 offset, char *line, size_t privsize) +{ + struct objdump_line *self = malloc(sizeof(*self) + privsize); + + if (self != NULL) { + self->offset = offset; + self->line = line; + } + + return self; +} + +void objdump_line__free(struct objdump_line *self) +{ + free(self->line); + free(self); +} + +static void objdump__add_line(struct list_head *head, struct objdump_line *line) +{ + list_add_tail(&line->node, head); +} + +struct objdump_line *objdump__get_next_ip_line(struct list_head *head, + struct objdump_line *pos) +{ + list_for_each_entry_continue(pos, head, node) + if (pos->offset >= 0) + return pos; + + return NULL; +} + +static int objdump_line__print(struct objdump_line *oline, struct symbol *sym, + int evidx, u64 len, int min_pcnt, + int printed, int max_lines, + struct objdump_line *queue) +{ + static const char *prev_line; + static const char *prev_color; + + if (oline->offset != -1) { + const char *path = NULL; + unsigned int hits = 0; + double percent = 0.0; + const char *color; + struct annotation *notes = symbol__annotation(sym); + struct source_line *src_line = notes->src->lines; + struct sym_hist *h = annotation__histogram(notes, evidx); + s64 offset = oline->offset; + struct objdump_line *next; + + next = objdump__get_next_ip_line(¬es->src->source, oline); + + while (offset < (s64)len && + (next == NULL || offset < next->offset)) { + if (src_line) { + if (path == NULL) + path = src_line[offset].path; + percent += src_line[offset].percent; + } else + hits += h->addr[offset]; + + ++offset; + } + + if (src_line == NULL && h->sum) + percent = 100.0 * hits / h->sum; + + if (percent < min_pcnt) + return -1; + + if (max_lines && printed >= max_lines) + return 1; + + if (queue != NULL) { + list_for_each_entry_from(queue, ¬es->src->source, node) { + if (queue == oline) + break; + objdump_line__print(queue, sym, evidx, len, + 0, 0, 1, NULL); + } + } + + color = get_percent_color(percent); + + /* + * Also color the filename and line if needed, with + * the same color than the percentage. Don't print it + * twice for close colored addr with the same filename:line + */ + if (path) { + if (!prev_line || strcmp(prev_line, path) + || color != prev_color) { + color_fprintf(stdout, color, " %s", path); + prev_line = path; + prev_color = color; + } + } + + color_fprintf(stdout, color, " %7.2f", percent); + printf(" : "); + color_fprintf(stdout, PERF_COLOR_BLUE, "%s\n", oline->line); + } else if (max_lines && printed >= max_lines) + return 1; + else { + if (queue) + return -1; + + if (!*oline->line) + printf(" :\n"); + else + printf(" : %s\n", oline->line); + } + + return 0; +} + +static int symbol__parse_objdump_line(struct symbol *sym, struct map *map, + FILE *file, size_t privsize) +{ + struct annotation *notes = symbol__annotation(sym); + struct objdump_line *objdump_line; + char *line = NULL, *tmp, *tmp2, *c; + size_t line_len; + s64 line_ip, offset = -1; + + if (getline(&line, &line_len, file) < 0) + return -1; + + if (!line) + return -1; + + while (line_len != 0 && isspace(line[line_len - 1])) + line[--line_len] = '\0'; + + c = strchr(line, '\n'); + if (c) + *c = 0; + + line_ip = -1; + + /* + * Strip leading spaces: + */ + tmp = line; + while (*tmp) { + if (*tmp != ' ') + break; + tmp++; + } + + if (*tmp) { + /* + * Parse hexa addresses followed by ':' + */ + line_ip = strtoull(tmp, &tmp2, 16); + if (*tmp2 != ':' || tmp == tmp2 || tmp2[1] == '\0') + line_ip = -1; + } + + if (line_ip != -1) { + u64 start = map__rip_2objdump(map, sym->start), + end = map__rip_2objdump(map, sym->end); + + offset = line_ip - start; + if (offset < 0 || (u64)line_ip > end) + offset = -1; + } + + objdump_line = objdump_line__new(offset, line, privsize); + if (objdump_line == NULL) { + free(line); + return -1; + } + objdump__add_line(¬es->src->source, objdump_line); + + return 0; +} + +int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize) +{ + struct dso *dso = map->dso; + char *filename = dso__build_id_filename(dso, NULL, 0); + bool free_filename = true; + char command[PATH_MAX * 2]; + FILE *file; + int err = 0; + char symfs_filename[PATH_MAX]; + + if (filename) { + snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", + symbol_conf.symfs, filename); + } + + if (filename == NULL) { + if (dso->has_build_id) { + pr_err("Can't annotate %s: not enough memory\n", + sym->name); + return -ENOMEM; + } + goto fallback; + } else if (readlink(symfs_filename, command, sizeof(command)) < 0 || + strstr(command, "[kernel.kallsyms]") || + access(symfs_filename, R_OK)) { + free(filename); +fallback: + /* + * If we don't have build-ids or the build-id file isn't in the + * cache, or is just a kallsyms file, well, lets hope that this + * DSO is the same as when 'perf record' ran. + */ + filename = dso->long_name; + snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", + symbol_conf.symfs, filename); + free_filename = false; + } + + if (dso->symtab_type == SYMTAB__KALLSYMS) { + char bf[BUILD_ID_SIZE * 2 + 16] = " with build id "; + char *build_id_msg = NULL; + + if (dso->annotate_warned) + goto out_free_filename; + + if (dso->has_build_id) { + build_id__sprintf(dso->build_id, + sizeof(dso->build_id), bf + 15); + build_id_msg = bf; + } + err = -ENOENT; + dso->annotate_warned = 1; + pr_err("Can't annotate %s: No vmlinux file%s was found in the " + "path.\nPlease use 'perf buildid-cache -av vmlinux' or " + "--vmlinux vmlinux.\n", + sym->name, build_id_msg ?: ""); + goto out_free_filename; + } + + pr_debug("%s: filename=%s, sym=%s, start=%#" PRIx64 ", end=%#" PRIx64 "\n", __func__, + filename, sym->name, map->unmap_ip(map, sym->start), + map->unmap_ip(map, sym->end)); + + pr_debug("annotating [%p] %30s : [%p] %30s\n", + dso, dso->long_name, sym, sym->name); + + snprintf(command, sizeof(command), + "objdump --start-address=0x%016" PRIx64 + " --stop-address=0x%016" PRIx64 " -dS -C %s|grep -v %s|expand", + map__rip_2objdump(map, sym->start), + map__rip_2objdump(map, sym->end), + symfs_filename, filename); + + pr_debug("Executing: %s\n", command); + + file = popen(command, "r"); + if (!file) + goto out_free_filename; + + while (!feof(file)) + if (symbol__parse_objdump_line(sym, map, file, privsize) < 0) + break; + + pclose(file); +out_free_filename: + if (free_filename) + free(filename); + return err; +} + +static void insert_source_line(struct rb_root *root, struct source_line *src_line) +{ + struct source_line *iter; + struct rb_node **p = &root->rb_node; + struct rb_node *parent = NULL; + + while (*p != NULL) { + parent = *p; + iter = rb_entry(parent, struct source_line, node); + + if (src_line->percent > iter->percent) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&src_line->node, parent, p); + rb_insert_color(&src_line->node, root); +} + +static void symbol__free_source_line(struct symbol *sym, int len) +{ + struct annotation *notes = symbol__annotation(sym); + struct source_line *src_line = notes->src->lines; + int i; + + for (i = 0; i < len; i++) + free(src_line[i].path); + + free(src_line); + notes->src->lines = NULL; +} + +/* Get the filename:line for the colored entries */ +static int symbol__get_source_line(struct symbol *sym, struct map *map, + int evidx, struct rb_root *root, int len, + const char *filename) +{ + u64 start; + int i; + char cmd[PATH_MAX * 2]; + struct source_line *src_line; + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + + if (!h->sum) + return 0; + + src_line = notes->src->lines = calloc(len, sizeof(struct source_line)); + if (!notes->src->lines) + return -1; + + start = map->unmap_ip(map, sym->start); + + for (i = 0; i < len; i++) { + char *path = NULL; + size_t line_len; + u64 offset; + FILE *fp; + + src_line[i].percent = 100.0 * h->addr[i] / h->sum; + if (src_line[i].percent <= 0.5) + continue; + + offset = start + i; + sprintf(cmd, "addr2line -e %s %016" PRIx64, filename, offset); + fp = popen(cmd, "r"); + if (!fp) + continue; + + if (getline(&path, &line_len, fp) < 0 || !line_len) + goto next; + + src_line[i].path = malloc(sizeof(char) * line_len + 1); + if (!src_line[i].path) + goto next; + + strcpy(src_line[i].path, path); + insert_source_line(root, &src_line[i]); + + next: + pclose(fp); + } + + return 0; +} + +static void print_summary(struct rb_root *root, const char *filename) +{ + struct source_line *src_line; + struct rb_node *node; + + printf("\nSorted summary for file %s\n", filename); + printf("----------------------------------------------\n\n"); + + if (RB_EMPTY_ROOT(root)) { + printf(" Nothing higher than %1.1f%%\n", MIN_GREEN); + return; + } + + node = rb_first(root); + while (node) { + double percent; + const char *color; + char *path; + + src_line = rb_entry(node, struct source_line, node); + percent = src_line->percent; + color = get_percent_color(percent); + path = src_line->path; + + color_fprintf(stdout, color, " %7.2f %s", percent, path); + node = rb_next(node); + } +} + +static void symbol__annotate_hits(struct symbol *sym, int evidx) +{ + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + u64 len = sym->end - sym->start, offset; + + for (offset = 0; offset < len; ++offset) + if (h->addr[offset] != 0) + printf("%*" PRIx64 ": %" PRIu64 "\n", BITS_PER_LONG / 2, + sym->start + offset, h->addr[offset]); + printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum); +} + +int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, + bool full_paths, int min_pcnt, int max_lines, + int context) +{ + struct dso *dso = map->dso; + const char *filename = dso->long_name, *d_filename; + struct annotation *notes = symbol__annotation(sym); + struct objdump_line *pos, *queue = NULL; + int printed = 2, queue_len = 0; + int more = 0; + u64 len; + + if (full_paths) + d_filename = filename; + else + d_filename = basename(filename); + + len = sym->end - sym->start; + + printf(" Percent | Source code & Disassembly of %s\n", d_filename); + printf("------------------------------------------------\n"); + + if (verbose) + symbol__annotate_hits(sym, evidx); + + list_for_each_entry(pos, ¬es->src->source, node) { + if (context && queue == NULL) { + queue = pos; + queue_len = 0; + } + + switch (objdump_line__print(pos, sym, evidx, len, min_pcnt, + printed, max_lines, queue)) { + case 0: + ++printed; + if (context) { + printed += queue_len; + queue = NULL; + queue_len = 0; + } + break; + case 1: + /* filtered by max_lines */ + ++more; + break; + case -1: + default: + /* + * Filtered by min_pcnt or non IP lines when + * context != 0 + */ + if (!context) + break; + if (queue_len == context) + queue = list_entry(queue->node.next, typeof(*queue), node); + else + ++queue_len; + break; + } + } + + return more; +} + +void symbol__annotate_zero_histogram(struct symbol *sym, int evidx) +{ + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + + memset(h, 0, notes->src->sizeof_sym_hist); +} + +void symbol__annotate_decay_histogram(struct symbol *sym, int evidx) +{ + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + struct objdump_line *pos; + int len = sym->end - sym->start; + + h->sum = 0; + + list_for_each_entry(pos, ¬es->src->source, node) { + if (pos->offset != -1 && pos->offset < len) { + h->addr[pos->offset] = h->addr[pos->offset] * 7 / 8; + h->sum += h->addr[pos->offset]; + } + } +} + +void objdump_line_list__purge(struct list_head *head) +{ + struct objdump_line *pos, *n; + + list_for_each_entry_safe(pos, n, head, node) { + list_del(&pos->node); + objdump_line__free(pos); + } +} + +int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, + bool print_lines, bool full_paths, int min_pcnt, + int max_lines) +{ + struct dso *dso = map->dso; + const char *filename = dso->long_name; + struct rb_root source_line = RB_ROOT; + u64 len; + + if (symbol__annotate(sym, map, 0) < 0) + return -1; + + len = sym->end - sym->start; + + if (print_lines) { + symbol__get_source_line(sym, map, evidx, &source_line, + len, filename); + print_summary(&source_line, filename); + } + + symbol__annotate_printf(sym, map, evidx, full_paths, + min_pcnt, max_lines, 0); + if (print_lines) + symbol__free_source_line(sym, len); + + objdump_line_list__purge(&symbol__annotation(sym)->src->source); + + return 0; +} diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h new file mode 100644 index 00000000..c2c28689 --- /dev/null +++ b/tools/perf/util/annotate.h @@ -0,0 +1,103 @@ +#ifndef __PERF_ANNOTATE_H +#define __PERF_ANNOTATE_H + +#include <stdbool.h> +#include "types.h" +#include "symbol.h" +#include <linux/list.h> +#include <linux/rbtree.h> + +struct objdump_line { + struct list_head node; + s64 offset; + char *line; +}; + +void objdump_line__free(struct objdump_line *self); +struct objdump_line *objdump__get_next_ip_line(struct list_head *head, + struct objdump_line *pos); + +struct sym_hist { + u64 sum; + u64 addr[0]; +}; + +struct source_line { + struct rb_node node; + double percent; + char *path; +}; + +/** struct annotated_source - symbols with hits have this attached as in sannotation + * + * @histogram: Array of addr hit histograms per event being monitored + * @lines: If 'print_lines' is specified, per source code line percentages + * @source: source parsed from objdump -dS + * + * lines is allocated, percentages calculated and all sorted by percentage + * when the annotation is about to be presented, so the percentages are for + * one of the entries in the histogram array, i.e. for the event/counter being + * presented. It is deallocated right after symbol__{tui,tty,etc}_annotate + * returns. + */ +struct annotated_source { + struct list_head source; + struct source_line *lines; + int nr_histograms; + int sizeof_sym_hist; + struct sym_hist histograms[0]; +}; + +struct annotation { + pthread_mutex_t lock; + struct annotated_source *src; +}; + +struct sannotation { + struct annotation annotation; + struct symbol symbol; +}; + +static inline struct sym_hist *annotation__histogram(struct annotation *notes, int idx) +{ + return (((void *)¬es->src->histograms) + + (notes->src->sizeof_sym_hist * idx)); +} + +static inline struct annotation *symbol__annotation(struct symbol *sym) +{ + struct sannotation *a = container_of(sym, struct sannotation, symbol); + return &a->annotation; +} + +int symbol__inc_addr_samples(struct symbol *sym, struct map *map, + int evidx, u64 addr); +int symbol__alloc_hist(struct symbol *sym, int nevents); +void symbol__annotate_zero_histograms(struct symbol *sym); + +int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize); +int symbol__annotate_init(struct map *map __used, struct symbol *sym); +int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, + bool full_paths, int min_pcnt, int max_lines, + int context); +void symbol__annotate_zero_histogram(struct symbol *sym, int evidx); +void symbol__annotate_decay_histogram(struct symbol *sym, int evidx); +void objdump_line_list__purge(struct list_head *head); + +int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, + bool print_lines, bool full_paths, int min_pcnt, + int max_lines); + +#ifdef NO_NEWT_SUPPORT +static inline int symbol__tui_annotate(struct symbol *sym __used, + struct map *map __used, + int evidx __used, int refresh __used) +{ + return 0; +} +#else +int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx, + int refresh); +#endif + +#endif /* __PERF_ANNOTATE_H */ diff --git a/tools/perf/util/bitmap.c b/tools/perf/util/bitmap.c new file mode 100644 index 00000000..5e230aca --- /dev/null +++ b/tools/perf/util/bitmap.c @@ -0,0 +1,21 @@ +/* + * From lib/bitmap.c + * Helper functions for bitmap.h. + * + * This source code is licensed under the GNU General Public License, + * Version 2. See the file COPYING for more details. + */ +#include <linux/bitmap.h> + +int __bitmap_weight(const unsigned long *bitmap, int bits) +{ + int k, w = 0, lim = bits/BITS_PER_LONG; + + for (k = 0; k < lim; k++) + w += hweight_long(bitmap[k]); + + if (bits % BITS_PER_LONG) + w += hweight_long(bitmap[k] & BITMAP_LAST_WORD_MASK(bits)); + + return w; +} diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c new file mode 100644 index 00000000..a91cd99f --- /dev/null +++ b/tools/perf/util/build-id.c @@ -0,0 +1,82 @@ +/* + * build-id.c + * + * build-id support + * + * Copyright (C) 2009, 2010 Red Hat Inc. + * Copyright (C) 2009, 2010 Arnaldo Carvalho de Melo <acme@redhat.com> + */ +#include "util.h" +#include <stdio.h> +#include "build-id.h" +#include "event.h" +#include "symbol.h" +#include <linux/kernel.h> +#include "debug.h" + +static int build_id__mark_dso_hit(union perf_event *event, + struct perf_sample *sample __used, + struct perf_evsel *evsel __used, + struct perf_session *session) +{ + struct addr_location al; + u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + struct thread *thread = perf_session__findnew(session, event->ip.pid); + + if (thread == NULL) { + pr_err("problem processing %d event, skipping it.\n", + event->header.type); + return -1; + } + + thread__find_addr_map(thread, session, cpumode, MAP__FUNCTION, + event->ip.pid, event->ip.ip, &al); + + if (al.map != NULL) + al.map->dso->hit = 1; + + return 0; +} + +static int perf_event__exit_del_thread(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) +{ + struct thread *thread = perf_session__findnew(session, event->fork.tid); + + dump_printf("(%d:%d):(%d:%d)\n", event->fork.pid, event->fork.tid, + event->fork.ppid, event->fork.ptid); + + if (thread) { + rb_erase(&thread->rb_node, &session->threads); + session->last_match = NULL; + thread__delete(thread); + } + + return 0; +} + +struct perf_event_ops build_id__mark_dso_hit_ops = { + .sample = build_id__mark_dso_hit, + .mmap = perf_event__process_mmap, + .fork = perf_event__process_task, + .exit = perf_event__exit_del_thread, +}; + +char *dso__build_id_filename(struct dso *self, char *bf, size_t size) +{ + char build_id_hex[BUILD_ID_SIZE * 2 + 1]; + + if (!self->has_build_id) + return NULL; + + build_id__sprintf(self->build_id, sizeof(self->build_id), build_id_hex); + if (bf == NULL) { + if (asprintf(&bf, "%s/.build-id/%.2s/%s", buildid_dir, + build_id_hex, build_id_hex + 2) < 0) + return NULL; + } else + snprintf(bf, size, "%s/.build-id/%.2s/%s", buildid_dir, + build_id_hex, build_id_hex + 2); + return bf; +} diff --git a/tools/perf/util/build-id.h b/tools/perf/util/build-id.h new file mode 100644 index 00000000..5dafb00e --- /dev/null +++ b/tools/perf/util/build-id.h @@ -0,0 +1,10 @@ +#ifndef PERF_BUILD_ID_H_ +#define PERF_BUILD_ID_H_ 1 + +#include "session.h" + +extern struct perf_event_ops build_id__mark_dso_hit_ops; + +char *dso__build_id_filename(struct dso *self, char *bf, size_t size); + +#endif diff --git a/tools/perf/util/cache.h b/tools/perf/util/cache.h new file mode 100644 index 00000000..fc5e5a09 --- /dev/null +++ b/tools/perf/util/cache.h @@ -0,0 +1,90 @@ +#ifndef __PERF_CACHE_H +#define __PERF_CACHE_H + +#include <stdbool.h> +#include "util.h" +#include "strbuf.h" +#include "../perf.h" + +#define CMD_EXEC_PATH "--exec-path" +#define CMD_PERF_DIR "--perf-dir=" +#define CMD_WORK_TREE "--work-tree=" +#define CMD_DEBUGFS_DIR "--debugfs-dir=" + +#define PERF_DIR_ENVIRONMENT "PERF_DIR" +#define PERF_WORK_TREE_ENVIRONMENT "PERF_WORK_TREE" +#define EXEC_PATH_ENVIRONMENT "PERF_EXEC_PATH" +#define DEFAULT_PERF_DIR_ENVIRONMENT ".perf" +#define PERF_DEBUGFS_ENVIRONMENT "PERF_DEBUGFS_DIR" + +typedef int (*config_fn_t)(const char *, const char *, void *); +extern int perf_default_config(const char *, const char *, void *); +extern int perf_config(config_fn_t fn, void *); +extern int perf_config_int(const char *, const char *); +extern int perf_config_bool(const char *, const char *); +extern int config_error_nonbool(const char *); +extern const char *perf_config_dirname(const char *, const char *); + +/* pager.c */ +extern void setup_pager(void); +extern const char *pager_program; +extern int pager_in_use(void); +extern int pager_use_color; + +extern int use_browser; + +#ifdef NO_NEWT_SUPPORT +static inline void setup_browser(bool fallback_to_pager) +{ + if (fallback_to_pager) + setup_pager(); +} +static inline void exit_browser(bool wait_for_ok __used) {} +#else +void setup_browser(bool fallback_to_pager); +void exit_browser(bool wait_for_ok); +#endif + +char *alias_lookup(const char *alias); +int split_cmdline(char *cmdline, const char ***argv); + +#define alloc_nr(x) (((x)+16)*3/2) + +/* + * Realloc the buffer pointed at by variable 'x' so that it can hold + * at least 'nr' entries; the number of entries currently allocated + * is 'alloc', using the standard growing factor alloc_nr() macro. + * + * DO NOT USE any expression with side-effect for 'x' or 'alloc'. + */ +#define ALLOC_GROW(x, nr, alloc) \ + do { \ + if ((nr) > alloc) { \ + if (alloc_nr(alloc) < (nr)) \ + alloc = (nr); \ + else \ + alloc = alloc_nr(alloc); \ + x = xrealloc((x), alloc * sizeof(*(x))); \ + } \ + } while(0) + + +static inline int is_absolute_path(const char *path) +{ + return path[0] == '/'; +} + +const char *make_nonrelative_path(const char *path); +char *strip_path_suffix(const char *path, const char *suffix); + +extern char *mkpath(const char *fmt, ...) __attribute__((format (printf, 1, 2))); +extern char *perf_path(const char *fmt, ...) __attribute__((format (printf, 1, 2))); + +extern char *perf_pathdup(const char *fmt, ...) + __attribute__((format (printf, 1, 2))); + +#ifdef NO_STRLCPY +extern size_t strlcpy(char *dest, const char *src, size_t size); +#endif + +#endif /* __PERF_CACHE_H */ diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c new file mode 100644 index 00000000..9f7106a8 --- /dev/null +++ b/tools/perf/util/callchain.c @@ -0,0 +1,461 @@ +/* + * Copyright (C) 2009-2011, Frederic Weisbecker <fweisbec@gmail.com> + * + * Handle the callchains from the stream in an ad-hoc radix tree and then + * sort them in an rbtree. + * + * Using a radix for code path provides a fast retrieval and factorizes + * memory use. Also that lets us use the paths in a hierarchical graph view. + * + */ + +#include <stdlib.h> +#include <stdio.h> +#include <stdbool.h> +#include <errno.h> +#include <math.h> + +#include "util.h" +#include "callchain.h" + +bool ip_callchain__valid(struct ip_callchain *chain, + const union perf_event *event) +{ + unsigned int chain_size = event->header.size; + chain_size -= (unsigned long)&event->ip.__more_data - (unsigned long)event; + return chain->nr * sizeof(u64) <= chain_size; +} + +#define chain_for_each_child(child, parent) \ + list_for_each_entry(child, &parent->children, siblings) + +#define chain_for_each_child_safe(child, next, parent) \ + list_for_each_entry_safe(child, next, &parent->children, siblings) + +static void +rb_insert_callchain(struct rb_root *root, struct callchain_node *chain, + enum chain_mode mode) +{ + struct rb_node **p = &root->rb_node; + struct rb_node *parent = NULL; + struct callchain_node *rnode; + u64 chain_cumul = callchain_cumul_hits(chain); + + while (*p) { + u64 rnode_cumul; + + parent = *p; + rnode = rb_entry(parent, struct callchain_node, rb_node); + rnode_cumul = callchain_cumul_hits(rnode); + + switch (mode) { + case CHAIN_FLAT: + if (rnode->hit < chain->hit) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + break; + case CHAIN_GRAPH_ABS: /* Falldown */ + case CHAIN_GRAPH_REL: + if (rnode_cumul < chain_cumul) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + break; + case CHAIN_NONE: + default: + break; + } + } + + rb_link_node(&chain->rb_node, parent, p); + rb_insert_color(&chain->rb_node, root); +} + +static void +__sort_chain_flat(struct rb_root *rb_root, struct callchain_node *node, + u64 min_hit) +{ + struct callchain_node *child; + + chain_for_each_child(child, node) + __sort_chain_flat(rb_root, child, min_hit); + + if (node->hit && node->hit >= min_hit) + rb_insert_callchain(rb_root, node, CHAIN_FLAT); +} + +/* + * Once we get every callchains from the stream, we can now + * sort them by hit + */ +static void +sort_chain_flat(struct rb_root *rb_root, struct callchain_root *root, + u64 min_hit, struct callchain_param *param __used) +{ + __sort_chain_flat(rb_root, &root->node, min_hit); +} + +static void __sort_chain_graph_abs(struct callchain_node *node, + u64 min_hit) +{ + struct callchain_node *child; + + node->rb_root = RB_ROOT; + + chain_for_each_child(child, node) { + __sort_chain_graph_abs(child, min_hit); + if (callchain_cumul_hits(child) >= min_hit) + rb_insert_callchain(&node->rb_root, child, + CHAIN_GRAPH_ABS); + } +} + +static void +sort_chain_graph_abs(struct rb_root *rb_root, struct callchain_root *chain_root, + u64 min_hit, struct callchain_param *param __used) +{ + __sort_chain_graph_abs(&chain_root->node, min_hit); + rb_root->rb_node = chain_root->node.rb_root.rb_node; +} + +static void __sort_chain_graph_rel(struct callchain_node *node, + double min_percent) +{ + struct callchain_node *child; + u64 min_hit; + + node->rb_root = RB_ROOT; + min_hit = ceil(node->children_hit * min_percent); + + chain_for_each_child(child, node) { + __sort_chain_graph_rel(child, min_percent); + if (callchain_cumul_hits(child) >= min_hit) + rb_insert_callchain(&node->rb_root, child, + CHAIN_GRAPH_REL); + } +} + +static void +sort_chain_graph_rel(struct rb_root *rb_root, struct callchain_root *chain_root, + u64 min_hit __used, struct callchain_param *param) +{ + __sort_chain_graph_rel(&chain_root->node, param->min_percent / 100.0); + rb_root->rb_node = chain_root->node.rb_root.rb_node; +} + +int callchain_register_param(struct callchain_param *param) +{ + switch (param->mode) { + case CHAIN_GRAPH_ABS: + param->sort = sort_chain_graph_abs; + break; + case CHAIN_GRAPH_REL: + param->sort = sort_chain_graph_rel; + break; + case CHAIN_FLAT: + param->sort = sort_chain_flat; + break; + case CHAIN_NONE: + default: + return -1; + } + return 0; +} + +/* + * Create a child for a parent. If inherit_children, then the new child + * will become the new parent of it's parent children + */ +static struct callchain_node * +create_child(struct callchain_node *parent, bool inherit_children) +{ + struct callchain_node *new; + + new = zalloc(sizeof(*new)); + if (!new) { + perror("not enough memory to create child for code path tree"); + return NULL; + } + new->parent = parent; + INIT_LIST_HEAD(&new->children); + INIT_LIST_HEAD(&new->val); + + if (inherit_children) { + struct callchain_node *next; + + list_splice(&parent->children, &new->children); + INIT_LIST_HEAD(&parent->children); + + chain_for_each_child(next, new) + next->parent = new; + } + list_add_tail(&new->siblings, &parent->children); + + return new; +} + + +/* + * Fill the node with callchain values + */ +static void +fill_node(struct callchain_node *node, struct callchain_cursor *cursor) +{ + struct callchain_cursor_node *cursor_node; + + node->val_nr = cursor->nr - cursor->pos; + if (!node->val_nr) + pr_warning("Warning: empty node in callchain tree\n"); + + cursor_node = callchain_cursor_current(cursor); + + while (cursor_node) { + struct callchain_list *call; + + call = zalloc(sizeof(*call)); + if (!call) { + perror("not enough memory for the code path tree"); + return; + } + call->ip = cursor_node->ip; + call->ms.sym = cursor_node->sym; + call->ms.map = cursor_node->map; + list_add_tail(&call->list, &node->val); + + callchain_cursor_advance(cursor); + cursor_node = callchain_cursor_current(cursor); + } +} + +static void +add_child(struct callchain_node *parent, + struct callchain_cursor *cursor, + u64 period) +{ + struct callchain_node *new; + + new = create_child(parent, false); + fill_node(new, cursor); + + new->children_hit = 0; + new->hit = period; +} + +/* + * Split the parent in two parts (a new child is created) and + * give a part of its callchain to the created child. + * Then create another child to host the given callchain of new branch + */ +static void +split_add_child(struct callchain_node *parent, + struct callchain_cursor *cursor, + struct callchain_list *to_split, + u64 idx_parents, u64 idx_local, u64 period) +{ + struct callchain_node *new; + struct list_head *old_tail; + unsigned int idx_total = idx_parents + idx_local; + + /* split */ + new = create_child(parent, true); + + /* split the callchain and move a part to the new child */ + old_tail = parent->val.prev; + list_del_range(&to_split->list, old_tail); + new->val.next = &to_split->list; + new->val.prev = old_tail; + to_split->list.prev = &new->val; + old_tail->next = &new->val; + + /* split the hits */ + new->hit = parent->hit; + new->children_hit = parent->children_hit; + parent->children_hit = callchain_cumul_hits(new); + new->val_nr = parent->val_nr - idx_local; + parent->val_nr = idx_local; + + /* create a new child for the new branch if any */ + if (idx_total < cursor->nr) { + parent->hit = 0; + add_child(parent, cursor, period); + parent->children_hit += period; + } else { + parent->hit = period; + } +} + +static int +append_chain(struct callchain_node *root, + struct callchain_cursor *cursor, + u64 period); + +static void +append_chain_children(struct callchain_node *root, + struct callchain_cursor *cursor, + u64 period) +{ + struct callchain_node *rnode; + + /* lookup in childrens */ + chain_for_each_child(rnode, root) { + unsigned int ret = append_chain(rnode, cursor, period); + + if (!ret) + goto inc_children_hit; + } + /* nothing in children, add to the current node */ + add_child(root, cursor, period); + +inc_children_hit: + root->children_hit += period; +} + +static int +append_chain(struct callchain_node *root, + struct callchain_cursor *cursor, + u64 period) +{ + struct callchain_cursor_node *curr_snap = cursor->curr; + struct callchain_list *cnode; + u64 start = cursor->pos; + bool found = false; + u64 matches; + + /* + * Lookup in the current node + * If we have a symbol, then compare the start to match + * anywhere inside a function. + */ + list_for_each_entry(cnode, &root->val, list) { + struct callchain_cursor_node *node; + struct symbol *sym; + + node = callchain_cursor_current(cursor); + if (!node) + break; + + sym = node->sym; + + if (cnode->ms.sym && sym) { + if (cnode->ms.sym->start != sym->start) + break; + } else if (cnode->ip != node->ip) + break; + + if (!found) + found = true; + + callchain_cursor_advance(cursor); + } + + /* matches not, relay on the parent */ + if (!found) { + cursor->curr = curr_snap; + cursor->pos = start; + return -1; + } + + matches = cursor->pos - start; + + /* we match only a part of the node. Split it and add the new chain */ + if (matches < root->val_nr) { + split_add_child(root, cursor, cnode, start, matches, period); + return 0; + } + + /* we match 100% of the path, increment the hit */ + if (matches == root->val_nr && cursor->pos == cursor->nr) { + root->hit += period; + return 0; + } + + /* We match the node and still have a part remaining */ + append_chain_children(root, cursor, period); + + return 0; +} + +int callchain_append(struct callchain_root *root, + struct callchain_cursor *cursor, + u64 period) +{ + if (!cursor->nr) + return 0; + + callchain_cursor_commit(cursor); + + append_chain_children(&root->node, cursor, period); + + if (cursor->nr > root->max_depth) + root->max_depth = cursor->nr; + + return 0; +} + +static int +merge_chain_branch(struct callchain_cursor *cursor, + struct callchain_node *dst, struct callchain_node *src) +{ + struct callchain_cursor_node **old_last = cursor->last; + struct callchain_node *child, *next_child; + struct callchain_list *list, *next_list; + int old_pos = cursor->nr; + int err = 0; + + list_for_each_entry_safe(list, next_list, &src->val, list) { + callchain_cursor_append(cursor, list->ip, + list->ms.map, list->ms.sym); + list_del(&list->list); + free(list); + } + + if (src->hit) { + callchain_cursor_commit(cursor); + append_chain_children(dst, cursor, src->hit); + } + + chain_for_each_child_safe(child, next_child, src) { + err = merge_chain_branch(cursor, dst, child); + if (err) + break; + + list_del(&child->siblings); + free(child); + } + + cursor->nr = old_pos; + cursor->last = old_last; + + return err; +} + +int callchain_merge(struct callchain_cursor *cursor, + struct callchain_root *dst, struct callchain_root *src) +{ + return merge_chain_branch(cursor, &dst->node, &src->node); +} + +int callchain_cursor_append(struct callchain_cursor *cursor, + u64 ip, struct map *map, struct symbol *sym) +{ + struct callchain_cursor_node *node = *cursor->last; + + if (!node) { + node = calloc(sizeof(*node), 1); + if (!node) + return -ENOMEM; + + *cursor->last = node; + } + + node->ip = ip; + node->map = map; + node->sym = sym; + + cursor->nr++; + + cursor->last = &node->next; + + return 0; +} diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h new file mode 100644 index 00000000..1a79df9f --- /dev/null +++ b/tools/perf/util/callchain.h @@ -0,0 +1,135 @@ +#ifndef __PERF_CALLCHAIN_H +#define __PERF_CALLCHAIN_H + +#include "../perf.h" +#include <linux/list.h> +#include <linux/rbtree.h> +#include "event.h" +#include "symbol.h" + +enum chain_mode { + CHAIN_NONE, + CHAIN_FLAT, + CHAIN_GRAPH_ABS, + CHAIN_GRAPH_REL +}; + +struct callchain_node { + struct callchain_node *parent; + struct list_head siblings; + struct list_head children; + struct list_head val; + struct rb_node rb_node; /* to sort nodes in an rbtree */ + struct rb_root rb_root; /* sorted tree of children */ + unsigned int val_nr; + u64 hit; + u64 children_hit; +}; + +struct callchain_root { + u64 max_depth; + struct callchain_node node; +}; + +struct callchain_param; + +typedef void (*sort_chain_func_t)(struct rb_root *, struct callchain_root *, + u64, struct callchain_param *); + +struct callchain_param { + enum chain_mode mode; + u32 print_limit; + double min_percent; + sort_chain_func_t sort; +}; + +struct callchain_list { + u64 ip; + struct map_symbol ms; + struct list_head list; +}; + +/* + * A callchain cursor is a single linked list that + * let one feed a callchain progressively. + * It keeps persitent allocated entries to minimize + * allocations. + */ +struct callchain_cursor_node { + u64 ip; + struct map *map; + struct symbol *sym; + struct callchain_cursor_node *next; +}; + +struct callchain_cursor { + u64 nr; + struct callchain_cursor_node *first; + struct callchain_cursor_node **last; + u64 pos; + struct callchain_cursor_node *curr; +}; + +static inline void callchain_init(struct callchain_root *root) +{ + INIT_LIST_HEAD(&root->node.siblings); + INIT_LIST_HEAD(&root->node.children); + INIT_LIST_HEAD(&root->node.val); + + root->node.parent = NULL; + root->node.hit = 0; + root->node.children_hit = 0; + root->max_depth = 0; +} + +static inline u64 callchain_cumul_hits(struct callchain_node *node) +{ + return node->hit + node->children_hit; +} + +int callchain_register_param(struct callchain_param *param); +int callchain_append(struct callchain_root *root, + struct callchain_cursor *cursor, + u64 period); + +int callchain_merge(struct callchain_cursor *cursor, + struct callchain_root *dst, struct callchain_root *src); + +bool ip_callchain__valid(struct ip_callchain *chain, + const union perf_event *event); +/* + * Initialize a cursor before adding entries inside, but keep + * the previously allocated entries as a cache. + */ +static inline void callchain_cursor_reset(struct callchain_cursor *cursor) +{ + cursor->nr = 0; + cursor->last = &cursor->first; +} + +int callchain_cursor_append(struct callchain_cursor *cursor, u64 ip, + struct map *map, struct symbol *sym); + +/* Close a cursor writing session. Initialize for the reader */ +static inline void callchain_cursor_commit(struct callchain_cursor *cursor) +{ + cursor->curr = cursor->first; + cursor->pos = 0; +} + +/* Cursor reading iteration helpers */ +static inline struct callchain_cursor_node * +callchain_cursor_current(struct callchain_cursor *cursor) +{ + if (cursor->pos == cursor->nr) + return NULL; + + return cursor->curr; +} + +static inline void callchain_cursor_advance(struct callchain_cursor *cursor) +{ + cursor->curr = cursor->curr->next; + cursor->pos++; +} +#endif /* __PERF_CALLCHAIN_H */ diff --git a/tools/perf/util/cgroup.c b/tools/perf/util/cgroup.c new file mode 100644 index 00000000..96bee5c4 --- /dev/null +++ b/tools/perf/util/cgroup.c @@ -0,0 +1,178 @@ +#include "util.h" +#include "../perf.h" +#include "parse-options.h" +#include "evsel.h" +#include "cgroup.h" +#include "debugfs.h" /* MAX_PATH, STR() */ +#include "evlist.h" + +int nr_cgroups; + +static int +cgroupfs_find_mountpoint(char *buf, size_t maxlen) +{ + FILE *fp; + char mountpoint[MAX_PATH+1], tokens[MAX_PATH+1], type[MAX_PATH+1]; + char *token, *saved_ptr = NULL; + int found = 0; + + fp = fopen("/proc/mounts", "r"); + if (!fp) + return -1; + + /* + * in order to handle split hierarchy, we need to scan /proc/mounts + * and inspect every cgroupfs mount point to find one that has + * perf_event subsystem + */ + while (fscanf(fp, "%*s %"STR(MAX_PATH)"s %"STR(MAX_PATH)"s %" + STR(MAX_PATH)"s %*d %*d\n", + mountpoint, type, tokens) == 3) { + + if (!strcmp(type, "cgroup")) { + + token = strtok_r(tokens, ",", &saved_ptr); + + while (token != NULL) { + if (!strcmp(token, "perf_event")) { + found = 1; + break; + } + token = strtok_r(NULL, ",", &saved_ptr); + } + } + if (found) + break; + } + fclose(fp); + if (!found) + return -1; + + if (strlen(mountpoint) < maxlen) { + strcpy(buf, mountpoint); + return 0; + } + return -1; +} + +static int open_cgroup(char *name) +{ + char path[MAX_PATH+1]; + char mnt[MAX_PATH+1]; + int fd; + + + if (cgroupfs_find_mountpoint(mnt, MAX_PATH+1)) + return -1; + + snprintf(path, MAX_PATH, "%s/%s", mnt, name); + + fd = open(path, O_RDONLY); + if (fd == -1) + fprintf(stderr, "no access to cgroup %s\n", path); + + return fd; +} + +static int add_cgroup(struct perf_evlist *evlist, char *str) +{ + struct perf_evsel *counter; + struct cgroup_sel *cgrp = NULL; + int n; + /* + * check if cgrp is already defined, if so we reuse it + */ + list_for_each_entry(counter, &evlist->entries, node) { + cgrp = counter->cgrp; + if (!cgrp) + continue; + if (!strcmp(cgrp->name, str)) + break; + + cgrp = NULL; + } + + if (!cgrp) { + cgrp = zalloc(sizeof(*cgrp)); + if (!cgrp) + return -1; + + cgrp->name = str; + + cgrp->fd = open_cgroup(str); + if (cgrp->fd == -1) { + free(cgrp); + return -1; + } + } + + /* + * find corresponding event + * if add cgroup N, then need to find event N + */ + n = 0; + list_for_each_entry(counter, &evlist->entries, node) { + if (n == nr_cgroups) + goto found; + n++; + } + if (cgrp->refcnt == 0) + free(cgrp); + + return -1; +found: + cgrp->refcnt++; + counter->cgrp = cgrp; + return 0; +} + +void close_cgroup(struct cgroup_sel *cgrp) +{ + if (!cgrp) + return; + + /* XXX: not reentrant */ + if (--cgrp->refcnt == 0) { + close(cgrp->fd); + free(cgrp->name); + free(cgrp); + } +} + +int parse_cgroups(const struct option *opt __used, const char *str, + int unset __used) +{ + struct perf_evlist *evlist = *(struct perf_evlist **)opt->value; + const char *p, *e, *eos = str + strlen(str); + char *s; + int ret; + + if (list_empty(&evlist->entries)) { + fprintf(stderr, "must define events before cgroups\n"); + return -1; + } + + for (;;) { + p = strchr(str, ','); + e = p ? p : eos; + + /* allow empty cgroups, i.e., skip */ + if (e - str) { + /* termination added */ + s = strndup(str, e - str); + if (!s) + return -1; + ret = add_cgroup(evlist, s); + if (ret) { + free(s); + return -1; + } + } + /* nr_cgroups is increased een for empty cgroups */ + nr_cgroups++; + if (!p) + break; + str = p+1; + } + return 0; +} diff --git a/tools/perf/util/cgroup.h b/tools/perf/util/cgroup.h new file mode 100644 index 00000000..89acd6de --- /dev/null +++ b/tools/perf/util/cgroup.h @@ -0,0 +1,17 @@ +#ifndef __CGROUP_H__ +#define __CGROUP_H__ + +struct option; + +struct cgroup_sel { + char *name; + int fd; + int refcnt; +}; + + +extern int nr_cgroups; /* number of explicit cgroups defined */ +extern void close_cgroup(struct cgroup_sel *cgrp); +extern int parse_cgroups(const struct option *opt, const char *str, int unset); + +#endif /* __CGROUP_H__ */ diff --git a/tools/perf/util/color.c b/tools/perf/util/color.c new file mode 100644 index 00000000..e191eb9a --- /dev/null +++ b/tools/perf/util/color.c @@ -0,0 +1,324 @@ +#include "cache.h" +#include "color.h" + +int perf_use_color_default = -1; + +static int parse_color(const char *name, int len) +{ + static const char * const color_names[] = { + "normal", "black", "red", "green", "yellow", + "blue", "magenta", "cyan", "white" + }; + char *end; + int i; + + for (i = 0; i < (int)ARRAY_SIZE(color_names); i++) { + const char *str = color_names[i]; + if (!strncasecmp(name, str, len) && !str[len]) + return i - 1; + } + i = strtol(name, &end, 10); + if (end - name == len && i >= -1 && i <= 255) + return i; + return -2; +} + +static int parse_attr(const char *name, int len) +{ + static const int attr_values[] = { 1, 2, 4, 5, 7 }; + static const char * const attr_names[] = { + "bold", "dim", "ul", "blink", "reverse" + }; + unsigned int i; + + for (i = 0; i < ARRAY_SIZE(attr_names); i++) { + const char *str = attr_names[i]; + if (!strncasecmp(name, str, len) && !str[len]) + return attr_values[i]; + } + return -1; +} + +void color_parse(const char *value, const char *var, char *dst) +{ + color_parse_mem(value, strlen(value), var, dst); +} + +void color_parse_mem(const char *value, int value_len, const char *var, + char *dst) +{ + const char *ptr = value; + int len = value_len; + int attr = -1; + int fg = -2; + int bg = -2; + + if (!strncasecmp(value, "reset", len)) { + strcpy(dst, PERF_COLOR_RESET); + return; + } + + /* [fg [bg]] [attr] */ + while (len > 0) { + const char *word = ptr; + int val, wordlen = 0; + + while (len > 0 && !isspace(word[wordlen])) { + wordlen++; + len--; + } + + ptr = word + wordlen; + while (len > 0 && isspace(*ptr)) { + ptr++; + len--; + } + + val = parse_color(word, wordlen); + if (val >= -1) { + if (fg == -2) { + fg = val; + continue; + } + if (bg == -2) { + bg = val; + continue; + } + goto bad; + } + val = parse_attr(word, wordlen); + if (val < 0 || attr != -1) + goto bad; + attr = val; + } + + if (attr >= 0 || fg >= 0 || bg >= 0) { + int sep = 0; + + *dst++ = '\033'; + *dst++ = '['; + if (attr >= 0) { + *dst++ = '0' + attr; + sep++; + } + if (fg >= 0) { + if (sep++) + *dst++ = ';'; + if (fg < 8) { + *dst++ = '3'; + *dst++ = '0' + fg; + } else { + dst += sprintf(dst, "38;5;%d", fg); + } + } + if (bg >= 0) { + if (sep++) + *dst++ = ';'; + if (bg < 8) { + *dst++ = '4'; + *dst++ = '0' + bg; + } else { + dst += sprintf(dst, "48;5;%d", bg); + } + } + *dst++ = 'm'; + } + *dst = 0; + return; +bad: + die("bad color value '%.*s' for variable '%s'", value_len, value, var); +} + +int perf_config_colorbool(const char *var, const char *value, int stdout_is_tty) +{ + if (value) { + if (!strcasecmp(value, "never")) + return 0; + if (!strcasecmp(value, "always")) + return 1; + if (!strcasecmp(value, "auto")) + goto auto_color; + } + + /* Missing or explicit false to turn off colorization */ + if (!perf_config_bool(var, value)) + return 0; + + /* any normal truth value defaults to 'auto' */ + auto_color: + if (stdout_is_tty < 0) + stdout_is_tty = isatty(1); + if (stdout_is_tty || (pager_in_use() && pager_use_color)) { + char *term = getenv("TERM"); + if (term && strcmp(term, "dumb")) + return 1; + } + return 0; +} + +int perf_color_default_config(const char *var, const char *value, void *cb) +{ + if (!strcmp(var, "color.ui")) { + perf_use_color_default = perf_config_colorbool(var, value, -1); + return 0; + } + + return perf_default_config(var, value, cb); +} + +static int __color_vsnprintf(char *bf, size_t size, const char *color, + const char *fmt, va_list args, const char *trail) +{ + int r = 0; + + /* + * Auto-detect: + */ + if (perf_use_color_default < 0) { + if (isatty(1) || pager_in_use()) + perf_use_color_default = 1; + else + perf_use_color_default = 0; + } + + if (perf_use_color_default && *color) + r += snprintf(bf, size, "%s", color); + r += vsnprintf(bf + r, size - r, fmt, args); + if (perf_use_color_default && *color) + r += snprintf(bf + r, size - r, "%s", PERF_COLOR_RESET); + if (trail) + r += snprintf(bf + r, size - r, "%s", trail); + return r; +} + +static int __color_vfprintf(FILE *fp, const char *color, const char *fmt, + va_list args, const char *trail) +{ + int r = 0; + + /* + * Auto-detect: + */ + if (perf_use_color_default < 0) { + if (isatty(1) || pager_in_use()) + perf_use_color_default = 1; + else + perf_use_color_default = 0; + } + + if (perf_use_color_default && *color) + r += fprintf(fp, "%s", color); + r += vfprintf(fp, fmt, args); + if (perf_use_color_default && *color) + r += fprintf(fp, "%s", PERF_COLOR_RESET); + if (trail) + r += fprintf(fp, "%s", trail); + return r; +} + +int color_vsnprintf(char *bf, size_t size, const char *color, + const char *fmt, va_list args) +{ + return __color_vsnprintf(bf, size, color, fmt, args, NULL); +} + +int color_vfprintf(FILE *fp, const char *color, const char *fmt, va_list args) +{ + return __color_vfprintf(fp, color, fmt, args, NULL); +} + +int color_snprintf(char *bf, size_t size, const char *color, + const char *fmt, ...) +{ + va_list args; + int r; + + va_start(args, fmt); + r = color_vsnprintf(bf, size, color, fmt, args); + va_end(args); + return r; +} + +int color_fprintf(FILE *fp, const char *color, const char *fmt, ...) +{ + va_list args; + int r; + + va_start(args, fmt); + r = color_vfprintf(fp, color, fmt, args); + va_end(args); + return r; +} + +int color_fprintf_ln(FILE *fp, const char *color, const char *fmt, ...) +{ + va_list args; + int r; + va_start(args, fmt); + r = __color_vfprintf(fp, color, fmt, args, "\n"); + va_end(args); + return r; +} + +/* + * This function splits the buffer by newlines and colors the lines individually. + * + * Returns 0 on success. + */ +int color_fwrite_lines(FILE *fp, const char *color, + size_t count, const char *buf) +{ + if (!*color) + return fwrite(buf, count, 1, fp) != 1; + + while (count) { + char *p = memchr(buf, '\n', count); + + if (p != buf && (fputs(color, fp) < 0 || + fwrite(buf, p ? (size_t)(p - buf) : count, 1, fp) != 1 || + fputs(PERF_COLOR_RESET, fp) < 0)) + return -1; + if (!p) + return 0; + if (fputc('\n', fp) < 0) + return -1; + count -= p + 1 - buf; + buf = p + 1; + } + return 0; +} + +const char *get_percent_color(double percent) +{ + const char *color = PERF_COLOR_NORMAL; + + /* + * We color high-overhead entries in red, mid-overhead + * entries in green - and keep the low overhead places + * normal: + */ + if (percent >= MIN_RED) + color = PERF_COLOR_RED; + else { + if (percent > MIN_GREEN) + color = PERF_COLOR_GREEN; + } + return color; +} + +int percent_color_fprintf(FILE *fp, const char *fmt, double percent) +{ + int r; + const char *color; + + color = get_percent_color(percent); + r = color_fprintf(fp, color, fmt, percent); + + return r; +} + +int percent_color_snprintf(char *bf, size_t size, const char *fmt, double percent) +{ + const char *color = get_percent_color(percent); + return color_snprintf(bf, size, color, fmt, percent); +} diff --git a/tools/perf/util/color.h b/tools/perf/util/color.h new file mode 100644 index 00000000..dea082b7 --- /dev/null +++ b/tools/perf/util/color.h @@ -0,0 +1,46 @@ +#ifndef __PERF_COLOR_H +#define __PERF_COLOR_H + +/* "\033[1;38;5;2xx;48;5;2xxm\0" is 23 bytes */ +#define COLOR_MAXLEN 24 + +#define PERF_COLOR_NORMAL "" +#define PERF_COLOR_RESET "\033[m" +#define PERF_COLOR_BOLD "\033[1m" +#define PERF_COLOR_RED "\033[31m" +#define PERF_COLOR_GREEN "\033[32m" +#define PERF_COLOR_YELLOW "\033[33m" +#define PERF_COLOR_BLUE "\033[34m" +#define PERF_COLOR_MAGENTA "\033[35m" +#define PERF_COLOR_CYAN "\033[36m" +#define PERF_COLOR_BG_RED "\033[41m" + +#define MIN_GREEN 0.5 +#define MIN_RED 5.0 + +/* + * This variable stores the value of color.ui + */ +extern int perf_use_color_default; + + +/* + * Use this instead of perf_default_config if you need the value of color.ui. + */ +int perf_color_default_config(const char *var, const char *value, void *cb); + +int perf_config_colorbool(const char *var, const char *value, int stdout_is_tty); +void color_parse(const char *value, const char *var, char *dst); +void color_parse_mem(const char *value, int len, const char *var, char *dst); +int color_vsnprintf(char *bf, size_t size, const char *color, + const char *fmt, va_list args); +int color_vfprintf(FILE *fp, const char *color, const char *fmt, va_list args); +int color_fprintf(FILE *fp, const char *color, const char *fmt, ...); +int color_snprintf(char *bf, size_t size, const char *color, const char *fmt, ...); +int color_fprintf_ln(FILE *fp, const char *color, const char *fmt, ...); +int color_fwrite_lines(FILE *fp, const char *color, size_t count, const char *buf); +int percent_color_snprintf(char *bf, size_t size, const char *fmt, double percent); +int percent_color_fprintf(FILE *fp, const char *fmt, double percent); +const char *get_percent_color(double percent); + +#endif /* __PERF_COLOR_H */ diff --git a/tools/perf/util/config.c b/tools/perf/util/config.c new file mode 100644 index 00000000..6c86eca8 --- /dev/null +++ b/tools/perf/util/config.c @@ -0,0 +1,485 @@ +/* + * GIT - The information manager from hell + * + * Copyright (C) Linus Torvalds, 2005 + * Copyright (C) Johannes Schindelin, 2005 + * + */ +#include "util.h" +#include "cache.h" +#include "exec_cmd.h" + +#define MAXNAME (256) + +#define DEBUG_CACHE_DIR ".debug" + + +char buildid_dir[MAXPATHLEN]; /* root dir for buildid, binary cache */ + +static FILE *config_file; +static const char *config_file_name; +static int config_linenr; +static int config_file_eof; + +static const char *config_exclusive_filename; + +static int get_next_char(void) +{ + int c; + FILE *f; + + c = '\n'; + if ((f = config_file) != NULL) { + c = fgetc(f); + if (c == '\r') { + /* DOS like systems */ + c = fgetc(f); + if (c != '\n') { + ungetc(c, f); + c = '\r'; + } + } + if (c == '\n') + config_linenr++; + if (c == EOF) { + config_file_eof = 1; + c = '\n'; + } + } + return c; +} + +static char *parse_value(void) +{ + static char value[1024]; + int quote = 0, comment = 0, space = 0; + size_t len = 0; + + for (;;) { + int c = get_next_char(); + + if (len >= sizeof(value) - 1) + return NULL; + if (c == '\n') { + if (quote) + return NULL; + value[len] = 0; + return value; + } + if (comment) + continue; + if (isspace(c) && !quote) { + space = 1; + continue; + } + if (!quote) { + if (c == ';' || c == '#') { + comment = 1; + continue; + } + } + if (space) { + if (len) + value[len++] = ' '; + space = 0; + } + if (c == '\\') { + c = get_next_char(); + switch (c) { + case '\n': + continue; + case 't': + c = '\t'; + break; + case 'b': + c = '\b'; + break; + case 'n': + c = '\n'; + break; + /* Some characters escape as themselves */ + case '\\': case '"': + break; + /* Reject unknown escape sequences */ + default: + return NULL; + } + value[len++] = c; + continue; + } + if (c == '"') { + quote = 1-quote; + continue; + } + value[len++] = c; + } +} + +static inline int iskeychar(int c) +{ + return isalnum(c) || c == '-'; +} + +static int get_value(config_fn_t fn, void *data, char *name, unsigned int len) +{ + int c; + char *value; + + /* Get the full name */ + for (;;) { + c = get_next_char(); + if (config_file_eof) + break; + if (!iskeychar(c)) + break; + name[len++] = c; + if (len >= MAXNAME) + return -1; + } + name[len] = 0; + while (c == ' ' || c == '\t') + c = get_next_char(); + + value = NULL; + if (c != '\n') { + if (c != '=') + return -1; + value = parse_value(); + if (!value) + return -1; + } + return fn(name, value, data); +} + +static int get_extended_base_var(char *name, int baselen, int c) +{ + do { + if (c == '\n') + return -1; + c = get_next_char(); + } while (isspace(c)); + + /* We require the format to be '[base "extension"]' */ + if (c != '"') + return -1; + name[baselen++] = '.'; + + for (;;) { + int ch = get_next_char(); + + if (ch == '\n') + return -1; + if (ch == '"') + break; + if (ch == '\\') { + ch = get_next_char(); + if (ch == '\n') + return -1; + } + name[baselen++] = ch; + if (baselen > MAXNAME / 2) + return -1; + } + + /* Final ']' */ + if (get_next_char() != ']') + return -1; + return baselen; +} + +static int get_base_var(char *name) +{ + int baselen = 0; + + for (;;) { + int c = get_next_char(); + if (config_file_eof) + return -1; + if (c == ']') + return baselen; + if (isspace(c)) + return get_extended_base_var(name, baselen, c); + if (!iskeychar(c) && c != '.') + return -1; + if (baselen > MAXNAME / 2) + return -1; + name[baselen++] = tolower(c); + } +} + +static int perf_parse_file(config_fn_t fn, void *data) +{ + int comment = 0; + int baselen = 0; + static char var[MAXNAME]; + + /* U+FEFF Byte Order Mark in UTF8 */ + static const unsigned char *utf8_bom = (unsigned char *) "\xef\xbb\xbf"; + const unsigned char *bomptr = utf8_bom; + + for (;;) { + int c = get_next_char(); + if (bomptr && *bomptr) { + /* We are at the file beginning; skip UTF8-encoded BOM + * if present. Sane editors won't put this in on their + * own, but e.g. Windows Notepad will do it happily. */ + if ((unsigned char) c == *bomptr) { + bomptr++; + continue; + } else { + /* Do not tolerate partial BOM. */ + if (bomptr != utf8_bom) + break; + /* No BOM at file beginning. Cool. */ + bomptr = NULL; + } + } + if (c == '\n') { + if (config_file_eof) + return 0; + comment = 0; + continue; + } + if (comment || isspace(c)) + continue; + if (c == '#' || c == ';') { + comment = 1; + continue; + } + if (c == '[') { + baselen = get_base_var(var); + if (baselen <= 0) + break; + var[baselen++] = '.'; + var[baselen] = 0; + continue; + } + if (!isalpha(c)) + break; + var[baselen] = tolower(c); + if (get_value(fn, data, var, baselen+1) < 0) + break; + } + die("bad config file line %d in %s", config_linenr, config_file_name); +} + +static int parse_unit_factor(const char *end, unsigned long *val) +{ + if (!*end) + return 1; + else if (!strcasecmp(end, "k")) { + *val *= 1024; + return 1; + } + else if (!strcasecmp(end, "m")) { + *val *= 1024 * 1024; + return 1; + } + else if (!strcasecmp(end, "g")) { + *val *= 1024 * 1024 * 1024; + return 1; + } + return 0; +} + +static int perf_parse_long(const char *value, long *ret) +{ + if (value && *value) { + char *end; + long val = strtol(value, &end, 0); + unsigned long factor = 1; + if (!parse_unit_factor(end, &factor)) + return 0; + *ret = val * factor; + return 1; + } + return 0; +} + +static void die_bad_config(const char *name) +{ + if (config_file_name) + die("bad config value for '%s' in %s", name, config_file_name); + die("bad config value for '%s'", name); +} + +int perf_config_int(const char *name, const char *value) +{ + long ret = 0; + if (!perf_parse_long(value, &ret)) + die_bad_config(name); + return ret; +} + +static int perf_config_bool_or_int(const char *name, const char *value, int *is_bool) +{ + *is_bool = 1; + if (!value) + return 1; + if (!*value) + return 0; + if (!strcasecmp(value, "true") || !strcasecmp(value, "yes") || !strcasecmp(value, "on")) + return 1; + if (!strcasecmp(value, "false") || !strcasecmp(value, "no") || !strcasecmp(value, "off")) + return 0; + *is_bool = 0; + return perf_config_int(name, value); +} + +int perf_config_bool(const char *name, const char *value) +{ + int discard; + return !!perf_config_bool_or_int(name, value, &discard); +} + +const char *perf_config_dirname(const char *name, const char *value) +{ + if (!name) + return NULL; + return value; +} + +static int perf_default_core_config(const char *var __used, const char *value __used) +{ + /* Add other config variables here and to Documentation/config.txt. */ + return 0; +} + +int perf_default_config(const char *var, const char *value, void *dummy __used) +{ + if (!prefixcmp(var, "core.")) + return perf_default_core_config(var, value); + + /* Add other config variables here and to Documentation/config.txt. */ + return 0; +} + +static int perf_config_from_file(config_fn_t fn, const char *filename, void *data) +{ + int ret; + FILE *f = fopen(filename, "r"); + + ret = -1; + if (f) { + config_file = f; + config_file_name = filename; + config_linenr = 1; + config_file_eof = 0; + ret = perf_parse_file(fn, data); + fclose(f); + config_file_name = NULL; + } + return ret; +} + +static const char *perf_etc_perfconfig(void) +{ + static const char *system_wide; + if (!system_wide) + system_wide = system_path(ETC_PERFCONFIG); + return system_wide; +} + +static int perf_env_bool(const char *k, int def) +{ + const char *v = getenv(k); + return v ? perf_config_bool(k, v) : def; +} + +static int perf_config_system(void) +{ + return !perf_env_bool("PERF_CONFIG_NOSYSTEM", 0); +} + +static int perf_config_global(void) +{ + return !perf_env_bool("PERF_CONFIG_NOGLOBAL", 0); +} + +int perf_config(config_fn_t fn, void *data) +{ + int ret = 0, found = 0; + const char *home = NULL; + + /* Setting $PERF_CONFIG makes perf read _only_ the given config file. */ + if (config_exclusive_filename) + return perf_config_from_file(fn, config_exclusive_filename, data); + if (perf_config_system() && !access(perf_etc_perfconfig(), R_OK)) { + ret += perf_config_from_file(fn, perf_etc_perfconfig(), + data); + found += 1; + } + + home = getenv("HOME"); + if (perf_config_global() && home) { + char *user_config = strdup(mkpath("%s/.perfconfig", home)); + if (!access(user_config, R_OK)) { + ret += perf_config_from_file(fn, user_config, data); + found += 1; + } + free(user_config); + } + + if (found == 0) + return -1; + return ret; +} + +/* + * Call this to report error for your variable that should not + * get a boolean value (i.e. "[my] var" means "true"). + */ +int config_error_nonbool(const char *var) +{ + return error("Missing value for '%s'", var); +} + +struct buildid_dir_config { + char *dir; +}; + +static int buildid_dir_command_config(const char *var, const char *value, + void *data) +{ + struct buildid_dir_config *c = data; + const char *v; + + /* same dir for all commands */ + if (!prefixcmp(var, "buildid.") && !strcmp(var + 8, "dir")) { + v = perf_config_dirname(var, value); + if (!v) + return -1; + strncpy(c->dir, v, MAXPATHLEN-1); + c->dir[MAXPATHLEN-1] = '\0'; + } + return 0; +} + +static void check_buildid_dir_config(void) +{ + struct buildid_dir_config c; + c.dir = buildid_dir; + perf_config(buildid_dir_command_config, &c); +} + +void set_buildid_dir(void) +{ + buildid_dir[0] = '\0'; + + /* try config file */ + check_buildid_dir_config(); + + /* default to $HOME/.debug */ + if (buildid_dir[0] == '\0') { + char *v = getenv("HOME"); + if (v) { + snprintf(buildid_dir, MAXPATHLEN-1, "%s/%s", + v, DEBUG_CACHE_DIR); + } else { + strncpy(buildid_dir, DEBUG_CACHE_DIR, MAXPATHLEN-1); + } + buildid_dir[MAXPATHLEN-1] = '\0'; + } + /* for communicating with external commands */ + setenv("PERF_BUILDID_DIR", buildid_dir, 1); +} diff --git a/tools/perf/util/cpumap.c b/tools/perf/util/cpumap.c new file mode 100644 index 00000000..6893eec6 --- /dev/null +++ b/tools/perf/util/cpumap.c @@ -0,0 +1,184 @@ +#include "util.h" +#include "../perf.h" +#include "cpumap.h" +#include <assert.h> +#include <stdio.h> + +static struct cpu_map *cpu_map__default_new(void) +{ + struct cpu_map *cpus; + int nr_cpus; + + nr_cpus = sysconf(_SC_NPROCESSORS_ONLN); + if (nr_cpus < 0) + return NULL; + + cpus = malloc(sizeof(*cpus) + nr_cpus * sizeof(int)); + if (cpus != NULL) { + int i; + for (i = 0; i < nr_cpus; ++i) + cpus->map[i] = i; + + cpus->nr = nr_cpus; + } + + return cpus; +} + +static struct cpu_map *cpu_map__trim_new(int nr_cpus, int *tmp_cpus) +{ + size_t payload_size = nr_cpus * sizeof(int); + struct cpu_map *cpus = malloc(sizeof(*cpus) + payload_size); + + if (cpus != NULL) { + cpus->nr = nr_cpus; + memcpy(cpus->map, tmp_cpus, payload_size); + } + + return cpus; +} + +static struct cpu_map *cpu_map__read_all_cpu_map(void) +{ + struct cpu_map *cpus = NULL; + FILE *onlnf; + int nr_cpus = 0; + int *tmp_cpus = NULL, *tmp; + int max_entries = 0; + int n, cpu, prev; + char sep; + + onlnf = fopen("/sys/devices/system/cpu/online", "r"); + if (!onlnf) + return cpu_map__default_new(); + + sep = 0; + prev = -1; + for (;;) { + n = fscanf(onlnf, "%u%c", &cpu, &sep); + if (n <= 0) + break; + if (prev >= 0) { + int new_max = nr_cpus + cpu - prev - 1; + + if (new_max >= max_entries) { + max_entries = new_max + MAX_NR_CPUS / 2; + tmp = realloc(tmp_cpus, max_entries * sizeof(int)); + if (tmp == NULL) + goto out_free_tmp; + tmp_cpus = tmp; + } + + while (++prev < cpu) + tmp_cpus[nr_cpus++] = prev; + } + if (nr_cpus == max_entries) { + max_entries += MAX_NR_CPUS; + tmp = realloc(tmp_cpus, max_entries * sizeof(int)); + if (tmp == NULL) + goto out_free_tmp; + tmp_cpus = tmp; + } + + tmp_cpus[nr_cpus++] = cpu; + if (n == 2 && sep == '-') + prev = cpu; + else + prev = -1; + if (n == 1 || sep == '\n') + break; + } + + if (nr_cpus > 0) + cpus = cpu_map__trim_new(nr_cpus, tmp_cpus); + else + cpus = cpu_map__default_new(); +out_free_tmp: + free(tmp_cpus); + fclose(onlnf); + return cpus; +} + +struct cpu_map *cpu_map__new(const char *cpu_list) +{ + struct cpu_map *cpus = NULL; + unsigned long start_cpu, end_cpu = 0; + char *p = NULL; + int i, nr_cpus = 0; + int *tmp_cpus = NULL, *tmp; + int max_entries = 0; + + if (!cpu_list) + return cpu_map__read_all_cpu_map(); + + if (!isdigit(*cpu_list)) + goto out; + + while (isdigit(*cpu_list)) { + p = NULL; + start_cpu = strtoul(cpu_list, &p, 0); + if (start_cpu >= INT_MAX + || (*p != '\0' && *p != ',' && *p != '-')) + goto invalid; + + if (*p == '-') { + cpu_list = ++p; + p = NULL; + end_cpu = strtoul(cpu_list, &p, 0); + + if (end_cpu >= INT_MAX || (*p != '\0' && *p != ',')) + goto invalid; + + if (end_cpu < start_cpu) + goto invalid; + } else { + end_cpu = start_cpu; + } + + for (; start_cpu <= end_cpu; start_cpu++) { + /* check for duplicates */ + for (i = 0; i < nr_cpus; i++) + if (tmp_cpus[i] == (int)start_cpu) + goto invalid; + + if (nr_cpus == max_entries) { + max_entries += MAX_NR_CPUS; + tmp = realloc(tmp_cpus, max_entries * sizeof(int)); + if (tmp == NULL) + goto invalid; + tmp_cpus = tmp; + } + tmp_cpus[nr_cpus++] = (int)start_cpu; + } + if (*p) + ++p; + + cpu_list = p; + } + + if (nr_cpus > 0) + cpus = cpu_map__trim_new(nr_cpus, tmp_cpus); + else + cpus = cpu_map__default_new(); +invalid: + free(tmp_cpus); +out: + return cpus; +} + +struct cpu_map *cpu_map__dummy_new(void) +{ + struct cpu_map *cpus = malloc(sizeof(*cpus) + sizeof(int)); + + if (cpus != NULL) { + cpus->nr = 1; + cpus->map[0] = -1; + } + + return cpus; +} + +void cpu_map__delete(struct cpu_map *map) +{ + free(map); +} diff --git a/tools/perf/util/cpumap.h b/tools/perf/util/cpumap.h new file mode 100644 index 00000000..072c0a37 --- /dev/null +++ b/tools/perf/util/cpumap.h @@ -0,0 +1,13 @@ +#ifndef __PERF_CPUMAP_H +#define __PERF_CPUMAP_H + +struct cpu_map { + int nr; + int map[]; +}; + +struct cpu_map *cpu_map__new(const char *cpu_list); +struct cpu_map *cpu_map__dummy_new(void); +void cpu_map__delete(struct cpu_map *map); + +#endif /* __PERF_CPUMAP_H */ diff --git a/tools/perf/util/ctype.c b/tools/perf/util/ctype.c new file mode 100644 index 00000000..35073621 --- /dev/null +++ b/tools/perf/util/ctype.c @@ -0,0 +1,39 @@ +/* + * Sane locale-independent, ASCII ctype. + * + * No surprises, and works with signed and unsigned chars. + */ +#include "cache.h" + +enum { + S = GIT_SPACE, + A = GIT_ALPHA, + D = GIT_DIGIT, + G = GIT_GLOB_SPECIAL, /* *, ?, [, \\ */ + R = GIT_REGEX_SPECIAL, /* $, (, ), +, ., ^, {, | * */ + P = GIT_PRINT_EXTRA, /* printable - alpha - digit - glob - regex */ + + PS = GIT_SPACE | GIT_PRINT_EXTRA, +}; + +unsigned char sane_ctype[256] = { +/* 0 1 2 3 4 5 6 7 8 9 A B C D E F */ + + 0, 0, 0, 0, 0, 0, 0, 0, 0, S, S, 0, 0, S, 0, 0, /* 0.. 15 */ + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 16.. 31 */ + PS,P, P, P, R, P, P, P, R, R, G, R, P, P, R, P, /* 32.. 47 */ + D, D, D, D, D, D, D, D, D, D, P, P, P, P, P, G, /* 48.. 63 */ + P, A, A, A, A, A, A, A, A, A, A, A, A, A, A, A, /* 64.. 79 */ + A, A, A, A, A, A, A, A, A, A, A, G, G, P, R, P, /* 80.. 95 */ + P, A, A, A, A, A, A, A, A, A, A, A, A, A, A, A, /* 96..111 */ + A, A, A, A, A, A, A, A, A, A, A, R, R, P, P, 0, /* 112..127 */ + /* Nothing in the 128.. range */ +}; + +const char *graph_line = + "_____________________________________________________________________" + "_____________________________________________________________________"; +const char *graph_dotted_line = + "---------------------------------------------------------------------" + "---------------------------------------------------------------------" + "---------------------------------------------------------------------"; diff --git a/tools/perf/util/debug.c b/tools/perf/util/debug.c new file mode 100644 index 00000000..155749d7 --- /dev/null +++ b/tools/perf/util/debug.c @@ -0,0 +1,104 @@ +/* For general debugging purposes */ + +#include "../perf.h" + +#include <string.h> +#include <stdarg.h> +#include <stdio.h> + +#include "cache.h" +#include "color.h" +#include "event.h" +#include "debug.h" +#include "util.h" + +int verbose; +bool dump_trace = false, quiet = false; + +int eprintf(int level, const char *fmt, ...) +{ + va_list args; + int ret = 0; + + if (verbose >= level) { + va_start(args, fmt); + if (use_browser > 0) + ret = ui_helpline__show_help(fmt, args); + else + ret = vfprintf(stderr, fmt, args); + va_end(args); + } + + return ret; +} + +int dump_printf(const char *fmt, ...) +{ + va_list args; + int ret = 0; + + if (dump_trace) { + va_start(args, fmt); + ret = vprintf(fmt, args); + va_end(args); + } + + return ret; +} + +#ifdef NO_NEWT_SUPPORT +void ui__warning(const char *format, ...) +{ + va_list args; + + va_start(args, format); + vfprintf(stderr, format, args); + va_end(args); +} +#endif + +void ui__warning_paranoid(void) +{ + ui__warning("Permission error - are you root?\n" + "Consider tweaking /proc/sys/kernel/perf_event_paranoid:\n" + " -1 - Not paranoid at all\n" + " 0 - Disallow raw tracepoint access for unpriv\n" + " 1 - Disallow cpu events for unpriv\n" + " 2 - Disallow kernel profiling for unpriv\n"); +} + +void trace_event(union perf_event *event) +{ + unsigned char *raw_event = (void *)event; + const char *color = PERF_COLOR_BLUE; + int i, j; + + if (!dump_trace) + return; + + printf("."); + color_fprintf(stdout, color, "\n. ... raw event: size %d bytes\n", + event->header.size); + + for (i = 0; i < event->header.size; i++) { + if ((i & 15) == 0) { + printf("."); + color_fprintf(stdout, color, " %04x: ", i); + } + + color_fprintf(stdout, color, " %02x", raw_event[i]); + + if (((i & 15) == 15) || i == event->header.size-1) { + color_fprintf(stdout, color, " "); + for (j = 0; j < 15-(i & 15); j++) + color_fprintf(stdout, color, " "); + for (j = i & ~15; j <= i; j++) { + color_fprintf(stdout, color, "%c", + isprint(raw_event[j]) ? + raw_event[j] : '.'); + } + color_fprintf(stdout, color, "\n"); + } + } + printf(".\n"); +} diff --git a/tools/perf/util/debug.h b/tools/perf/util/debug.h new file mode 100644 index 00000000..fd53db47 --- /dev/null +++ b/tools/perf/util/debug.h @@ -0,0 +1,41 @@ +/* For debugging general purposes */ +#ifndef __PERF_DEBUG_H +#define __PERF_DEBUG_H + +#include <stdbool.h> +#include "event.h" + +extern int verbose; +extern bool quiet, dump_trace; + +int dump_printf(const char *fmt, ...) __attribute__((format(printf, 1, 2))); +void trace_event(union perf_event *event); + +struct ui_progress; + +#ifdef NO_NEWT_SUPPORT +static inline int ui_helpline__show_help(const char *format __used, va_list ap __used) +{ + return 0; +} + +static inline struct ui_progress *ui_progress__new(const char *title __used, + u64 total __used) +{ + return (struct ui_progress *)1; +} + +static inline void ui_progress__update(struct ui_progress *self __used, + u64 curr __used) {} + +static inline void ui_progress__delete(struct ui_progress *self __used) {} +#else +extern char ui_helpline__last_msg[]; +int ui_helpline__show_help(const char *format, va_list ap); +#include "ui/progress.h" +#endif + +void ui__warning(const char *format, ...) __attribute__((format(printf, 1, 2))); +void ui__warning_paranoid(void); + +#endif /* __PERF_DEBUG_H */ diff --git a/tools/perf/util/debugfs.c b/tools/perf/util/debugfs.c new file mode 100644 index 00000000..a88fefc0 --- /dev/null +++ b/tools/perf/util/debugfs.c @@ -0,0 +1,240 @@ +#include "util.h" +#include "debugfs.h" +#include "cache.h" + +static int debugfs_premounted; +static char debugfs_mountpoint[MAX_PATH+1]; + +static const char *debugfs_known_mountpoints[] = { + "/sys/kernel/debug/", + "/debug/", + 0, +}; + +/* use this to force a umount */ +void debugfs_force_cleanup(void) +{ + debugfs_find_mountpoint(); + debugfs_premounted = 0; + debugfs_umount(); +} + +/* construct a full path to a debugfs element */ +int debugfs_make_path(const char *element, char *buffer, int size) +{ + int len; + + if (strlen(debugfs_mountpoint) == 0) { + buffer[0] = '\0'; + return -1; + } + + len = strlen(debugfs_mountpoint) + strlen(element) + 1; + if (len >= size) + return len+1; + + snprintf(buffer, size-1, "%s/%s", debugfs_mountpoint, element); + return 0; +} + +static int debugfs_found; + +/* find the path to the mounted debugfs */ +const char *debugfs_find_mountpoint(void) +{ + const char **ptr; + char type[100]; + FILE *fp; + + if (debugfs_found) + return (const char *) debugfs_mountpoint; + + ptr = debugfs_known_mountpoints; + while (*ptr) { + if (debugfs_valid_mountpoint(*ptr) == 0) { + debugfs_found = 1; + strcpy(debugfs_mountpoint, *ptr); + return debugfs_mountpoint; + } + ptr++; + } + + /* give up and parse /proc/mounts */ + fp = fopen("/proc/mounts", "r"); + if (fp == NULL) + die("Can't open /proc/mounts for read"); + + while (fscanf(fp, "%*s %" + STR(MAX_PATH) + "s %99s %*s %*d %*d\n", + debugfs_mountpoint, type) == 2) { + if (strcmp(type, "debugfs") == 0) + break; + } + fclose(fp); + + if (strcmp(type, "debugfs") != 0) + return NULL; + + debugfs_found = 1; + + return debugfs_mountpoint; +} + +/* verify that a mountpoint is actually a debugfs instance */ + +int debugfs_valid_mountpoint(const char *debugfs) +{ + struct statfs st_fs; + + if (statfs(debugfs, &st_fs) < 0) + return -ENOENT; + else if (st_fs.f_type != (long) DEBUGFS_MAGIC) + return -ENOENT; + + return 0; +} + + +int debugfs_valid_entry(const char *path) +{ + struct stat st; + + if (stat(path, &st)) + return -errno; + + return 0; +} + +/* mount the debugfs somewhere if it's not mounted */ + +char *debugfs_mount(const char *mountpoint) +{ + /* see if it's already mounted */ + if (debugfs_find_mountpoint()) { + debugfs_premounted = 1; + return debugfs_mountpoint; + } + + /* if not mounted and no argument */ + if (mountpoint == NULL) { + /* see if environment variable set */ + mountpoint = getenv(PERF_DEBUGFS_ENVIRONMENT); + /* if no environment variable, use default */ + if (mountpoint == NULL) + mountpoint = "/sys/kernel/debug"; + } + + if (mount(NULL, mountpoint, "debugfs", 0, NULL) < 0) + return NULL; + + /* save the mountpoint */ + strncpy(debugfs_mountpoint, mountpoint, sizeof(debugfs_mountpoint)); + debugfs_found = 1; + + return debugfs_mountpoint; +} + +/* umount the debugfs */ + +int debugfs_umount(void) +{ + char umountcmd[128]; + int ret; + + /* if it was already mounted, leave it */ + if (debugfs_premounted) + return 0; + + /* make sure it's a valid mount point */ + ret = debugfs_valid_mountpoint(debugfs_mountpoint); + if (ret) + return ret; + + snprintf(umountcmd, sizeof(umountcmd), + "/bin/umount %s", debugfs_mountpoint); + return system(umountcmd); +} + +int debugfs_write(const char *entry, const char *value) +{ + char path[MAX_PATH+1]; + int ret, count; + int fd; + + /* construct the path */ + snprintf(path, sizeof(path), "%s/%s", debugfs_mountpoint, entry); + + /* verify that it exists */ + ret = debugfs_valid_entry(path); + if (ret) + return ret; + + /* get how many chars we're going to write */ + count = strlen(value); + + /* open the debugfs entry */ + fd = open(path, O_RDWR); + if (fd < 0) + return -errno; + + while (count > 0) { + /* write it */ + ret = write(fd, value, count); + if (ret <= 0) { + if (ret == EAGAIN) + continue; + close(fd); + return -errno; + } + count -= ret; + } + + /* close it */ + close(fd); + + /* return success */ + return 0; +} + +/* + * read a debugfs entry + * returns the number of chars read or a negative errno + */ +int debugfs_read(const char *entry, char *buffer, size_t size) +{ + char path[MAX_PATH+1]; + int ret; + int fd; + + /* construct the path */ + snprintf(path, sizeof(path), "%s/%s", debugfs_mountpoint, entry); + + /* verify that it exists */ + ret = debugfs_valid_entry(path); + if (ret) + return ret; + + /* open the debugfs entry */ + fd = open(path, O_RDONLY); + if (fd < 0) + return -errno; + + do { + /* read it */ + ret = read(fd, buffer, size); + if (ret == 0) { + close(fd); + return EOF; + } + } while (ret < 0 && errno == EAGAIN); + + /* close it */ + close(fd); + + /* make *sure* there's a null character at the end */ + buffer[ret] = '\0'; + + /* return the number of chars read */ + return ret; +} diff --git a/tools/perf/util/debugfs.h b/tools/perf/util/debugfs.h new file mode 100644 index 00000000..83a02879 --- /dev/null +++ b/tools/perf/util/debugfs.h @@ -0,0 +1,25 @@ +#ifndef __DEBUGFS_H__ +#define __DEBUGFS_H__ + +#include <sys/mount.h> + +#ifndef MAX_PATH +# define MAX_PATH 256 +#endif + +#ifndef STR +# define _STR(x) #x +# define STR(x) _STR(x) +#endif + +extern const char *debugfs_find_mountpoint(void); +extern int debugfs_valid_mountpoint(const char *debugfs); +extern int debugfs_valid_entry(const char *path); +extern char *debugfs_mount(const char *mountpoint); +extern int debugfs_umount(void); +extern int debugfs_write(const char *entry, const char *value); +extern int debugfs_read(const char *entry, char *buffer, size_t size); +extern void debugfs_force_cleanup(void); +extern int debugfs_make_path(const char *element, char *buffer, int size); + +#endif /* __DEBUGFS_H__ */ diff --git a/tools/perf/util/environment.c b/tools/perf/util/environment.c new file mode 100644 index 00000000..275b0ee3 --- /dev/null +++ b/tools/perf/util/environment.c @@ -0,0 +1,9 @@ +/* + * We put all the perf config variables in this same object + * file, so that programs can link against the config parser + * without having to link against all the rest of perf. + */ +#include "cache.h" + +const char *pager_program; +int pager_use_color = 1; diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c new file mode 100644 index 00000000..3c1b8a63 --- /dev/null +++ b/tools/perf/util/event.c @@ -0,0 +1,802 @@ +#include <linux/types.h> +#include "event.h" +#include "debug.h" +#include "session.h" +#include "sort.h" +#include "string.h" +#include "strlist.h" +#include "thread.h" +#include "thread_map.h" + +static const char *perf_event__names[] = { + [0] = "TOTAL", + [PERF_RECORD_MMAP] = "MMAP", + [PERF_RECORD_LOST] = "LOST", + [PERF_RECORD_COMM] = "COMM", + [PERF_RECORD_EXIT] = "EXIT", + [PERF_RECORD_THROTTLE] = "THROTTLE", + [PERF_RECORD_UNTHROTTLE] = "UNTHROTTLE", + [PERF_RECORD_FORK] = "FORK", + [PERF_RECORD_READ] = "READ", + [PERF_RECORD_SAMPLE] = "SAMPLE", + [PERF_RECORD_HEADER_ATTR] = "ATTR", + [PERF_RECORD_HEADER_EVENT_TYPE] = "EVENT_TYPE", + [PERF_RECORD_HEADER_TRACING_DATA] = "TRACING_DATA", + [PERF_RECORD_HEADER_BUILD_ID] = "BUILD_ID", + [PERF_RECORD_FINISHED_ROUND] = "FINISHED_ROUND", +}; + +const char *perf_event__name(unsigned int id) +{ + if (id >= ARRAY_SIZE(perf_event__names)) + return "INVALID"; + if (!perf_event__names[id]) + return "UNKNOWN"; + return perf_event__names[id]; +} + +static struct perf_sample synth_sample = { + .pid = -1, + .tid = -1, + .time = -1, + .stream_id = -1, + .cpu = -1, + .period = 1, +}; + +static pid_t perf_event__synthesize_comm(union perf_event *event, pid_t pid, + int full, perf_event__handler_t process, + struct perf_session *session) +{ + char filename[PATH_MAX]; + char bf[BUFSIZ]; + FILE *fp; + size_t size = 0; + DIR *tasks; + struct dirent dirent, *next; + pid_t tgid = 0; + + snprintf(filename, sizeof(filename), "/proc/%d/status", pid); + + fp = fopen(filename, "r"); + if (fp == NULL) { +out_race: + /* + * We raced with a task exiting - just return: + */ + pr_debug("couldn't open %s\n", filename); + return 0; + } + + memset(&event->comm, 0, sizeof(event->comm)); + + while (!event->comm.comm[0] || !event->comm.pid) { + if (fgets(bf, sizeof(bf), fp) == NULL) { + pr_warning("couldn't get COMM and pgid, malformed %s\n", filename); + goto out; + } + + if (memcmp(bf, "Name:", 5) == 0) { + char *name = bf + 5; + while (*name && isspace(*name)) + ++name; + size = strlen(name) - 1; + memcpy(event->comm.comm, name, size++); + } else if (memcmp(bf, "Tgid:", 5) == 0) { + char *tgids = bf + 5; + while (*tgids && isspace(*tgids)) + ++tgids; + tgid = event->comm.pid = atoi(tgids); + } + } + + event->comm.header.type = PERF_RECORD_COMM; + size = ALIGN(size, sizeof(u64)); + memset(event->comm.comm + size, 0, session->id_hdr_size); + event->comm.header.size = (sizeof(event->comm) - + (sizeof(event->comm.comm) - size) + + session->id_hdr_size); + if (!full) { + event->comm.tid = pid; + + process(event, &synth_sample, session); + goto out; + } + + snprintf(filename, sizeof(filename), "/proc/%d/task", pid); + + tasks = opendir(filename); + if (tasks == NULL) + goto out_race; + + while (!readdir_r(tasks, &dirent, &next) && next) { + char *end; + pid = strtol(dirent.d_name, &end, 10); + if (*end) + continue; + + event->comm.tid = pid; + + process(event, &synth_sample, session); + } + + closedir(tasks); +out: + fclose(fp); + + return tgid; +} + +static int perf_event__synthesize_mmap_events(union perf_event *event, + pid_t pid, pid_t tgid, + perf_event__handler_t process, + struct perf_session *session) +{ + char filename[PATH_MAX]; + FILE *fp; + + snprintf(filename, sizeof(filename), "/proc/%d/maps", pid); + + fp = fopen(filename, "r"); + if (fp == NULL) { + /* + * We raced with a task exiting - just return: + */ + pr_debug("couldn't open %s\n", filename); + return -1; + } + + event->header.type = PERF_RECORD_MMAP; + /* + * Just like the kernel, see __perf_event_mmap in kernel/perf_event.c + */ + event->header.misc = PERF_RECORD_MISC_USER; + + while (1) { + char bf[BUFSIZ], *pbf = bf; + int n; + size_t size; + if (fgets(bf, sizeof(bf), fp) == NULL) + break; + + /* 00400000-0040c000 r-xp 00000000 fd:01 41038 /bin/cat */ + n = hex2u64(pbf, &event->mmap.start); + if (n < 0) + continue; + pbf += n + 1; + n = hex2u64(pbf, &event->mmap.len); + if (n < 0) + continue; + pbf += n + 3; + if (*pbf == 'x') { /* vm_exec */ + char *execname = strchr(bf, '/'); + + /* Catch VDSO */ + if (execname == NULL) + execname = strstr(bf, "[vdso]"); + + if (execname == NULL) + continue; + + pbf += 3; + n = hex2u64(pbf, &event->mmap.pgoff); + + size = strlen(execname); + execname[size - 1] = '\0'; /* Remove \n */ + memcpy(event->mmap.filename, execname, size); + size = ALIGN(size, sizeof(u64)); + event->mmap.len -= event->mmap.start; + event->mmap.header.size = (sizeof(event->mmap) - + (sizeof(event->mmap.filename) - size)); + memset(event->mmap.filename + size, 0, session->id_hdr_size); + event->mmap.header.size += session->id_hdr_size; + event->mmap.pid = tgid; + event->mmap.tid = pid; + + process(event, &synth_sample, session); + } + } + + fclose(fp); + return 0; +} + +int perf_event__synthesize_modules(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine) +{ + struct rb_node *nd; + struct map_groups *kmaps = &machine->kmaps; + union perf_event *event = zalloc((sizeof(event->mmap) + + session->id_hdr_size)); + if (event == NULL) { + pr_debug("Not enough memory synthesizing mmap event " + "for kernel modules\n"); + return -1; + } + + event->header.type = PERF_RECORD_MMAP; + + /* + * kernel uses 0 for user space maps, see kernel/perf_event.c + * __perf_event_mmap + */ + if (machine__is_host(machine)) + event->header.misc = PERF_RECORD_MISC_KERNEL; + else + event->header.misc = PERF_RECORD_MISC_GUEST_KERNEL; + + for (nd = rb_first(&kmaps->maps[MAP__FUNCTION]); + nd; nd = rb_next(nd)) { + size_t size; + struct map *pos = rb_entry(nd, struct map, rb_node); + + if (pos->dso->kernel) + continue; + + size = ALIGN(pos->dso->long_name_len + 1, sizeof(u64)); + event->mmap.header.type = PERF_RECORD_MMAP; + event->mmap.header.size = (sizeof(event->mmap) - + (sizeof(event->mmap.filename) - size)); + memset(event->mmap.filename + size, 0, session->id_hdr_size); + event->mmap.header.size += session->id_hdr_size; + event->mmap.start = pos->start; + event->mmap.len = pos->end - pos->start; + event->mmap.pid = machine->pid; + + memcpy(event->mmap.filename, pos->dso->long_name, + pos->dso->long_name_len + 1); + process(event, &synth_sample, session); + } + + free(event); + return 0; +} + +static int __event__synthesize_thread(union perf_event *comm_event, + union perf_event *mmap_event, + pid_t pid, perf_event__handler_t process, + struct perf_session *session) +{ + pid_t tgid = perf_event__synthesize_comm(comm_event, pid, 1, process, + session); + if (tgid == -1) + return -1; + return perf_event__synthesize_mmap_events(mmap_event, pid, tgid, + process, session); +} + +int perf_event__synthesize_thread_map(struct thread_map *threads, + perf_event__handler_t process, + struct perf_session *session) +{ + union perf_event *comm_event, *mmap_event; + int err = -1, thread; + + comm_event = malloc(sizeof(comm_event->comm) + session->id_hdr_size); + if (comm_event == NULL) + goto out; + + mmap_event = malloc(sizeof(mmap_event->mmap) + session->id_hdr_size); + if (mmap_event == NULL) + goto out_free_comm; + + err = 0; + for (thread = 0; thread < threads->nr; ++thread) { + if (__event__synthesize_thread(comm_event, mmap_event, + threads->map[thread], + process, session)) { + err = -1; + break; + } + } + free(mmap_event); +out_free_comm: + free(comm_event); +out: + return err; +} + +int perf_event__synthesize_threads(perf_event__handler_t process, + struct perf_session *session) +{ + DIR *proc; + struct dirent dirent, *next; + union perf_event *comm_event, *mmap_event; + int err = -1; + + comm_event = malloc(sizeof(comm_event->comm) + session->id_hdr_size); + if (comm_event == NULL) + goto out; + + mmap_event = malloc(sizeof(mmap_event->mmap) + session->id_hdr_size); + if (mmap_event == NULL) + goto out_free_comm; + + proc = opendir("/proc"); + if (proc == NULL) + goto out_free_mmap; + + while (!readdir_r(proc, &dirent, &next) && next) { + char *end; + pid_t pid = strtol(dirent.d_name, &end, 10); + + if (*end) /* only interested in proper numerical dirents */ + continue; + + __event__synthesize_thread(comm_event, mmap_event, pid, + process, session); + } + + closedir(proc); + err = 0; +out_free_mmap: + free(mmap_event); +out_free_comm: + free(comm_event); +out: + return err; +} + +struct process_symbol_args { + const char *name; + u64 start; +}; + +static int find_symbol_cb(void *arg, const char *name, char type, + u64 start, u64 end __used) +{ + struct process_symbol_args *args = arg; + + /* + * Must be a function or at least an alias, as in PARISC64, where "_text" is + * an 'A' to the same address as "_stext". + */ + if (!(symbol_type__is_a(type, MAP__FUNCTION) || + type == 'A') || strcmp(name, args->name)) + return 0; + + args->start = start; + return 1; +} + +int perf_event__synthesize_kernel_mmap(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine, + const char *symbol_name) +{ + size_t size; + const char *filename, *mmap_name; + char path[PATH_MAX]; + char name_buff[PATH_MAX]; + struct map *map; + int err; + /* + * We should get this from /sys/kernel/sections/.text, but till that is + * available use this, and after it is use this as a fallback for older + * kernels. + */ + struct process_symbol_args args = { .name = symbol_name, }; + union perf_event *event = zalloc((sizeof(event->mmap) + + session->id_hdr_size)); + if (event == NULL) { + pr_debug("Not enough memory synthesizing mmap event " + "for kernel modules\n"); + return -1; + } + + mmap_name = machine__mmap_name(machine, name_buff, sizeof(name_buff)); + if (machine__is_host(machine)) { + /* + * kernel uses PERF_RECORD_MISC_USER for user space maps, + * see kernel/perf_event.c __perf_event_mmap + */ + event->header.misc = PERF_RECORD_MISC_KERNEL; + filename = "/proc/kallsyms"; + } else { + event->header.misc = PERF_RECORD_MISC_GUEST_KERNEL; + if (machine__is_default_guest(machine)) + filename = (char *) symbol_conf.default_guest_kallsyms; + else { + sprintf(path, "%s/proc/kallsyms", machine->root_dir); + filename = path; + } + } + + if (kallsyms__parse(filename, &args, find_symbol_cb) <= 0) + return -ENOENT; + + map = machine->vmlinux_maps[MAP__FUNCTION]; + size = snprintf(event->mmap.filename, sizeof(event->mmap.filename), + "%s%s", mmap_name, symbol_name) + 1; + size = ALIGN(size, sizeof(u64)); + event->mmap.header.type = PERF_RECORD_MMAP; + event->mmap.header.size = (sizeof(event->mmap) - + (sizeof(event->mmap.filename) - size) + session->id_hdr_size); + event->mmap.pgoff = args.start; + event->mmap.start = map->start; + event->mmap.len = map->end - event->mmap.start; + event->mmap.pid = machine->pid; + + err = process(event, &synth_sample, session); + free(event); + + return err; +} + +int perf_event__process_comm(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) +{ + struct thread *thread = perf_session__findnew(session, event->comm.tid); + + dump_printf(": %s:%d\n", event->comm.comm, event->comm.tid); + + if (thread == NULL || thread__set_comm(thread, event->comm.comm)) { + dump_printf("problem processing PERF_RECORD_COMM, skipping event.\n"); + return -1; + } + + return 0; +} + +int perf_event__process_lost(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) +{ + dump_printf(": id:%" PRIu64 ": lost:%" PRIu64 "\n", + event->lost.id, event->lost.lost); + session->hists.stats.total_lost += event->lost.lost; + return 0; +} + +static void perf_event__set_kernel_mmap_len(union perf_event *event, + struct map **maps) +{ + maps[MAP__FUNCTION]->start = event->mmap.start; + maps[MAP__FUNCTION]->end = event->mmap.start + event->mmap.len; + /* + * Be a bit paranoid here, some perf.data file came with + * a zero sized synthesized MMAP event for the kernel. + */ + if (maps[MAP__FUNCTION]->end == 0) + maps[MAP__FUNCTION]->end = ~0ULL; +} + +static int perf_event__process_kernel_mmap(union perf_event *event, + struct perf_session *session) +{ + struct map *map; + char kmmap_prefix[PATH_MAX]; + struct machine *machine; + enum dso_kernel_type kernel_type; + bool is_kernel_mmap; + + machine = perf_session__findnew_machine(session, event->mmap.pid); + if (!machine) { + pr_err("Can't find id %d's machine\n", event->mmap.pid); + goto out_problem; + } + + machine__mmap_name(machine, kmmap_prefix, sizeof(kmmap_prefix)); + if (machine__is_host(machine)) + kernel_type = DSO_TYPE_KERNEL; + else + kernel_type = DSO_TYPE_GUEST_KERNEL; + + is_kernel_mmap = memcmp(event->mmap.filename, + kmmap_prefix, + strlen(kmmap_prefix)) == 0; + if (event->mmap.filename[0] == '/' || + (!is_kernel_mmap && event->mmap.filename[0] == '[')) { + + char short_module_name[1024]; + char *name, *dot; + + if (event->mmap.filename[0] == '/') { + name = strrchr(event->mmap.filename, '/'); + if (name == NULL) + goto out_problem; + + ++name; /* skip / */ + dot = strrchr(name, '.'); + if (dot == NULL) + goto out_problem; + snprintf(short_module_name, sizeof(short_module_name), + "[%.*s]", (int)(dot - name), name); + strxfrchar(short_module_name, '-', '_'); + } else + strcpy(short_module_name, event->mmap.filename); + + map = machine__new_module(machine, event->mmap.start, + event->mmap.filename); + if (map == NULL) + goto out_problem; + + name = strdup(short_module_name); + if (name == NULL) + goto out_problem; + + map->dso->short_name = name; + map->dso->sname_alloc = 1; + map->end = map->start + event->mmap.len; + } else if (is_kernel_mmap) { + const char *symbol_name = (event->mmap.filename + + strlen(kmmap_prefix)); + /* + * Should be there already, from the build-id table in + * the header. + */ + struct dso *kernel = __dsos__findnew(&machine->kernel_dsos, + kmmap_prefix); + if (kernel == NULL) + goto out_problem; + + kernel->kernel = kernel_type; + if (__machine__create_kernel_maps(machine, kernel) < 0) + goto out_problem; + + perf_event__set_kernel_mmap_len(event, machine->vmlinux_maps); + + /* + * Avoid using a zero address (kptr_restrict) for the ref reloc + * symbol. Effectively having zero here means that at record + * time /proc/sys/kernel/kptr_restrict was non zero. + */ + if (event->mmap.pgoff != 0) { + perf_session__set_kallsyms_ref_reloc_sym(machine->vmlinux_maps, + symbol_name, + event->mmap.pgoff); + } + + if (machine__is_default_guest(machine)) { + /* + * preload dso of guest kernel and modules + */ + dso__load(kernel, machine->vmlinux_maps[MAP__FUNCTION], + NULL); + } + } + return 0; +out_problem: + return -1; +} + +int perf_event__process_mmap(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) +{ + struct machine *machine; + struct thread *thread; + struct map *map; + u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + int ret = 0; + + dump_printf(" %d/%d: [%#" PRIx64 "(%#" PRIx64 ") @ %#" PRIx64 "]: %s\n", + event->mmap.pid, event->mmap.tid, event->mmap.start, + event->mmap.len, event->mmap.pgoff, event->mmap.filename); + + if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL || + cpumode == PERF_RECORD_MISC_KERNEL) { + ret = perf_event__process_kernel_mmap(event, session); + if (ret < 0) + goto out_problem; + return 0; + } + + machine = perf_session__find_host_machine(session); + if (machine == NULL) + goto out_problem; + thread = perf_session__findnew(session, event->mmap.pid); + if (thread == NULL) + goto out_problem; + map = map__new(&machine->user_dsos, event->mmap.start, + event->mmap.len, event->mmap.pgoff, + event->mmap.pid, event->mmap.filename, + MAP__FUNCTION); + if (map == NULL) + goto out_problem; + + thread__insert_map(thread, map); + return 0; + +out_problem: + dump_printf("problem processing PERF_RECORD_MMAP, skipping event.\n"); + return 0; +} + +int perf_event__process_task(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) +{ + struct thread *thread = perf_session__findnew(session, event->fork.tid); + struct thread *parent = perf_session__findnew(session, event->fork.ptid); + + dump_printf("(%d:%d):(%d:%d)\n", event->fork.pid, event->fork.tid, + event->fork.ppid, event->fork.ptid); + + if (event->header.type == PERF_RECORD_EXIT) { + perf_session__remove_thread(session, thread); + return 0; + } + + if (thread == NULL || parent == NULL || + thread__fork(thread, parent) < 0) { + dump_printf("problem processing PERF_RECORD_FORK, skipping event.\n"); + return -1; + } + + return 0; +} + +int perf_event__process(union perf_event *event, struct perf_sample *sample, + struct perf_session *session) +{ + switch (event->header.type) { + case PERF_RECORD_COMM: + perf_event__process_comm(event, sample, session); + break; + case PERF_RECORD_MMAP: + perf_event__process_mmap(event, sample, session); + break; + case PERF_RECORD_FORK: + case PERF_RECORD_EXIT: + perf_event__process_task(event, sample, session); + break; + case PERF_RECORD_LOST: + perf_event__process_lost(event, sample, session); + default: + break; + } + + return 0; +} + +void thread__find_addr_map(struct thread *self, + struct perf_session *session, u8 cpumode, + enum map_type type, pid_t pid, u64 addr, + struct addr_location *al) +{ + struct map_groups *mg = &self->mg; + struct machine *machine = NULL; + + al->thread = self; + al->addr = addr; + al->cpumode = cpumode; + al->filtered = false; + + if (cpumode == PERF_RECORD_MISC_KERNEL && perf_host) { + al->level = 'k'; + machine = perf_session__find_host_machine(session); + if (machine == NULL) { + al->map = NULL; + return; + } + mg = &machine->kmaps; + } else if (cpumode == PERF_RECORD_MISC_USER && perf_host) { + al->level = '.'; + machine = perf_session__find_host_machine(session); + } else if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL && perf_guest) { + al->level = 'g'; + machine = perf_session__find_machine(session, pid); + if (machine == NULL) { + al->map = NULL; + return; + } + mg = &machine->kmaps; + } else { + /* + * 'u' means guest os user space. + * TODO: We don't support guest user space. Might support late. + */ + if (cpumode == PERF_RECORD_MISC_GUEST_USER && perf_guest) + al->level = 'u'; + else + al->level = 'H'; + al->map = NULL; + + if ((cpumode == PERF_RECORD_MISC_GUEST_USER || + cpumode == PERF_RECORD_MISC_GUEST_KERNEL) && + !perf_guest) + al->filtered = true; + if ((cpumode == PERF_RECORD_MISC_USER || + cpumode == PERF_RECORD_MISC_KERNEL) && + !perf_host) + al->filtered = true; + + return; + } +try_again: + al->map = map_groups__find(mg, type, al->addr); + if (al->map == NULL) { + /* + * If this is outside of all known maps, and is a negative + * address, try to look it up in the kernel dso, as it might be + * a vsyscall or vdso (which executes in user-mode). + * + * XXX This is nasty, we should have a symbol list in the + * "[vdso]" dso, but for now lets use the old trick of looking + * in the whole kernel symbol list. + */ + if ((long long)al->addr < 0 && + cpumode == PERF_RECORD_MISC_USER && + machine && mg != &machine->kmaps) { + mg = &machine->kmaps; + goto try_again; + } + } else + al->addr = al->map->map_ip(al->map, al->addr); +} + +void thread__find_addr_location(struct thread *self, + struct perf_session *session, u8 cpumode, + enum map_type type, pid_t pid, u64 addr, + struct addr_location *al, + symbol_filter_t filter) +{ + thread__find_addr_map(self, session, cpumode, type, pid, addr, al); + if (al->map != NULL) + al->sym = map__find_symbol(al->map, al->addr, filter); + else + al->sym = NULL; +} + +int perf_event__preprocess_sample(const union perf_event *event, + struct perf_session *session, + struct addr_location *al, + struct perf_sample *sample, + symbol_filter_t filter) +{ + u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + struct thread *thread = perf_session__findnew(session, event->ip.pid); + + if (thread == NULL) + return -1; + + if (symbol_conf.comm_list && + !strlist__has_entry(symbol_conf.comm_list, thread->comm)) + goto out_filtered; + + dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid); + /* + * Have we already created the kernel maps for the host machine? + * + * This should have happened earlier, when we processed the kernel MMAP + * events, but for older perf.data files there was no such thing, so do + * it now. + */ + if (cpumode == PERF_RECORD_MISC_KERNEL && + session->host_machine.vmlinux_maps[MAP__FUNCTION] == NULL) + machine__create_kernel_maps(&session->host_machine); + + thread__find_addr_map(thread, session, cpumode, MAP__FUNCTION, + event->ip.pid, event->ip.ip, al); + dump_printf(" ...... dso: %s\n", + al->map ? al->map->dso->long_name : + al->level == 'H' ? "[hypervisor]" : "<not found>"); + al->sym = NULL; + al->cpu = sample->cpu; + + if (al->map) { + if (symbol_conf.dso_list && + (!al->map || !al->map->dso || + !(strlist__has_entry(symbol_conf.dso_list, + al->map->dso->short_name) || + (al->map->dso->short_name != al->map->dso->long_name && + strlist__has_entry(symbol_conf.dso_list, + al->map->dso->long_name))))) + goto out_filtered; + + al->sym = map__find_symbol(al->map, al->addr, filter); + } + + if (symbol_conf.sym_list && al->sym && + !strlist__has_entry(symbol_conf.sym_list, al->sym->name)) + goto out_filtered; + + return 0; + +out_filtered: + al->filtered = true; + return 0; +} diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h new file mode 100644 index 00000000..1d7f6648 --- /dev/null +++ b/tools/perf/util/event.h @@ -0,0 +1,191 @@ +#ifndef __PERF_RECORD_H +#define __PERF_RECORD_H + +#include <limits.h> + +#include "../perf.h" +#include "map.h" + +/* + * PERF_SAMPLE_IP | PERF_SAMPLE_TID | * + */ +struct ip_event { + struct perf_event_header header; + u64 ip; + u32 pid, tid; + unsigned char __more_data[]; +}; + +struct mmap_event { + struct perf_event_header header; + u32 pid, tid; + u64 start; + u64 len; + u64 pgoff; + char filename[PATH_MAX]; +}; + +struct comm_event { + struct perf_event_header header; + u32 pid, tid; + char comm[16]; +}; + +struct fork_event { + struct perf_event_header header; + u32 pid, ppid; + u32 tid, ptid; + u64 time; +}; + +struct lost_event { + struct perf_event_header header; + u64 id; + u64 lost; +}; + +/* + * PERF_FORMAT_ENABLED | PERF_FORMAT_RUNNING | PERF_FORMAT_ID + */ +struct read_event { + struct perf_event_header header; + u32 pid, tid; + u64 value; + u64 time_enabled; + u64 time_running; + u64 id; +}; + + +#define PERF_SAMPLE_MASK \ + (PERF_SAMPLE_IP | PERF_SAMPLE_TID | \ + PERF_SAMPLE_TIME | PERF_SAMPLE_ADDR | \ + PERF_SAMPLE_ID | PERF_SAMPLE_STREAM_ID | \ + PERF_SAMPLE_CPU | PERF_SAMPLE_PERIOD) + +struct sample_event { + struct perf_event_header header; + u64 array[]; +}; + +struct perf_sample { + u64 ip; + u32 pid, tid; + u64 time; + u64 addr; + u64 id; + u64 stream_id; + u64 period; + u32 cpu; + u32 raw_size; + void *raw_data; + struct ip_callchain *callchain; +}; + +#define BUILD_ID_SIZE 20 + +struct build_id_event { + struct perf_event_header header; + pid_t pid; + u8 build_id[ALIGN(BUILD_ID_SIZE, sizeof(u64))]; + char filename[]; +}; + +enum perf_user_event_type { /* above any possible kernel type */ + PERF_RECORD_USER_TYPE_START = 64, + PERF_RECORD_HEADER_ATTR = 64, + PERF_RECORD_HEADER_EVENT_TYPE = 65, + PERF_RECORD_HEADER_TRACING_DATA = 66, + PERF_RECORD_HEADER_BUILD_ID = 67, + PERF_RECORD_FINISHED_ROUND = 68, + PERF_RECORD_HEADER_MAX +}; + +struct attr_event { + struct perf_event_header header; + struct perf_event_attr attr; + u64 id[]; +}; + +#define MAX_EVENT_NAME 64 + +struct perf_trace_event_type { + u64 event_id; + char name[MAX_EVENT_NAME]; +}; + +struct event_type_event { + struct perf_event_header header; + struct perf_trace_event_type event_type; +}; + +struct tracing_data_event { + struct perf_event_header header; + u32 size; +}; + +union perf_event { + struct perf_event_header header; + struct ip_event ip; + struct mmap_event mmap; + struct comm_event comm; + struct fork_event fork; + struct lost_event lost; + struct read_event read; + struct sample_event sample; + struct attr_event attr; + struct event_type_event event_type; + struct tracing_data_event tracing_data; + struct build_id_event build_id; +}; + +void perf_event__print_totals(void); + +struct perf_session; +struct thread_map; + +typedef int (*perf_event__handler_synth_t)(union perf_event *event, + struct perf_session *session); +typedef int (*perf_event__handler_t)(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session); + +int perf_event__synthesize_thread_map(struct thread_map *threads, + perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_threads(perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_kernel_mmap(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine, + const char *symbol_name); + +int perf_event__synthesize_modules(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine); + +int perf_event__process_comm(union perf_event *event, struct perf_sample *sample, + struct perf_session *session); +int perf_event__process_lost(union perf_event *event, struct perf_sample *sample, + struct perf_session *session); +int perf_event__process_mmap(union perf_event *event, struct perf_sample *sample, + struct perf_session *session); +int perf_event__process_task(union perf_event *event, struct perf_sample *sample, + struct perf_session *session); +int perf_event__process(union perf_event *event, struct perf_sample *sample, + struct perf_session *session); + +struct addr_location; +int perf_event__preprocess_sample(const union perf_event *self, + struct perf_session *session, + struct addr_location *al, + struct perf_sample *sample, + symbol_filter_t filter); + +const char *perf_event__name(unsigned int id); + +int perf_event__parse_sample(const union perf_event *event, u64 type, + int sample_size, bool sample_id_all, + struct perf_sample *sample); + +#endif /* __PERF_RECORD_H */ diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c new file mode 100644 index 00000000..b021ea92 --- /dev/null +++ b/tools/perf/util/evlist.c @@ -0,0 +1,500 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Parts came from builtin-{top,stat,record}.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ +#include <poll.h> +#include "cpumap.h" +#include "thread_map.h" +#include "evlist.h" +#include "evsel.h" +#include "util.h" + +#include <sys/mman.h> + +#include <linux/bitops.h> +#include <linux/hash.h> + +#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) +#define SID(e, x, y) xyarray__entry(e->sample_id, x, y) + +void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads) +{ + int i; + + for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i) + INIT_HLIST_HEAD(&evlist->heads[i]); + INIT_LIST_HEAD(&evlist->entries); + perf_evlist__set_maps(evlist, cpus, threads); +} + +struct perf_evlist *perf_evlist__new(struct cpu_map *cpus, + struct thread_map *threads) +{ + struct perf_evlist *evlist = zalloc(sizeof(*evlist)); + + if (evlist != NULL) + perf_evlist__init(evlist, cpus, threads); + + return evlist; +} + +static void perf_evlist__purge(struct perf_evlist *evlist) +{ + struct perf_evsel *pos, *n; + + list_for_each_entry_safe(pos, n, &evlist->entries, node) { + list_del_init(&pos->node); + perf_evsel__delete(pos); + } + + evlist->nr_entries = 0; +} + +void perf_evlist__exit(struct perf_evlist *evlist) +{ + free(evlist->mmap); + free(evlist->pollfd); + evlist->mmap = NULL; + evlist->pollfd = NULL; +} + +void perf_evlist__delete(struct perf_evlist *evlist) +{ + perf_evlist__purge(evlist); + perf_evlist__exit(evlist); + free(evlist); +} + +void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry) +{ + list_add_tail(&entry->node, &evlist->entries); + ++evlist->nr_entries; +} + +int perf_evlist__add_default(struct perf_evlist *evlist) +{ + struct perf_event_attr attr = { + .type = PERF_TYPE_HARDWARE, + .config = PERF_COUNT_HW_CPU_CYCLES, + }; + struct perf_evsel *evsel = perf_evsel__new(&attr, 0); + + if (evsel == NULL) + return -ENOMEM; + + perf_evlist__add(evlist, evsel); + return 0; +} + +int perf_evlist__alloc_pollfd(struct perf_evlist *evlist) +{ + int nfds = evlist->cpus->nr * evlist->threads->nr * evlist->nr_entries; + evlist->pollfd = malloc(sizeof(struct pollfd) * nfds); + return evlist->pollfd != NULL ? 0 : -ENOMEM; +} + +void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd) +{ + fcntl(fd, F_SETFL, O_NONBLOCK); + evlist->pollfd[evlist->nr_fds].fd = fd; + evlist->pollfd[evlist->nr_fds].events = POLLIN; + evlist->nr_fds++; +} + +static void perf_evlist__id_hash(struct perf_evlist *evlist, + struct perf_evsel *evsel, + int cpu, int thread, u64 id) +{ + int hash; + struct perf_sample_id *sid = SID(evsel, cpu, thread); + + sid->id = id; + sid->evsel = evsel; + hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS); + hlist_add_head(&sid->node, &evlist->heads[hash]); +} + +void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel, + int cpu, int thread, u64 id) +{ + perf_evlist__id_hash(evlist, evsel, cpu, thread, id); + evsel->id[evsel->ids++] = id; +} + +static int perf_evlist__id_add_fd(struct perf_evlist *evlist, + struct perf_evsel *evsel, + int cpu, int thread, int fd) +{ + u64 read_data[4] = { 0, }; + int id_idx = 1; /* The first entry is the counter value */ + + if (!(evsel->attr.read_format & PERF_FORMAT_ID) || + read(fd, &read_data, sizeof(read_data)) == -1) + return -1; + + if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) + ++id_idx; + if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) + ++id_idx; + + perf_evlist__id_add(evlist, evsel, cpu, thread, read_data[id_idx]); + return 0; +} + +struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id) +{ + struct hlist_head *head; + struct hlist_node *pos; + struct perf_sample_id *sid; + int hash; + + if (evlist->nr_entries == 1) + return list_entry(evlist->entries.next, struct perf_evsel, node); + + hash = hash_64(id, PERF_EVLIST__HLIST_BITS); + head = &evlist->heads[hash]; + + hlist_for_each_entry(sid, pos, head, node) + if (sid->id == id) + return sid->evsel; + return NULL; +} + +union perf_event *perf_evlist__mmap_read(struct perf_evlist *evlist, int idx) +{ + /* XXX Move this to perf.c, making it generally available */ + unsigned int page_size = sysconf(_SC_PAGE_SIZE); + struct perf_mmap *md = &evlist->mmap[idx]; + unsigned int head = perf_mmap__read_head(md); + unsigned int old = md->prev; + unsigned char *data = md->base + page_size; + union perf_event *event = NULL; + + if (evlist->overwrite) { + /* + * If we're further behind than half the buffer, there's a chance + * the writer will bite our tail and mess up the samples under us. + * + * If we somehow ended up ahead of the head, we got messed up. + * + * In either case, truncate and restart at head. + */ + int diff = head - old; + if (diff > md->mask / 2 || diff < 0) { + fprintf(stderr, "WARNING: failed to keep up with mmap data.\n"); + + /* + * head points to a known good entry, start there. + */ + old = head; + } + } + + if (old != head) { + size_t size; + + event = (union perf_event *)&data[old & md->mask]; + size = event->header.size; + + /* + * Event straddles the mmap boundary -- header should always + * be inside due to u64 alignment of output. + */ + if ((old & md->mask) + size != ((old + size) & md->mask)) { + unsigned int offset = old; + unsigned int len = min(sizeof(*event), size), cpy; + void *dst = &evlist->event_copy; + + do { + cpy = min(md->mask + 1 - (offset & md->mask), len); + memcpy(dst, &data[offset & md->mask], cpy); + offset += cpy; + dst += cpy; + len -= cpy; + } while (len); + + event = &evlist->event_copy; + } + + old += size; + } + + md->prev = old; + + if (!evlist->overwrite) + perf_mmap__write_tail(md, old); + + return event; +} + +void perf_evlist__munmap(struct perf_evlist *evlist) +{ + int i; + + for (i = 0; i < evlist->nr_mmaps; i++) { + if (evlist->mmap[i].base != NULL) { + munmap(evlist->mmap[i].base, evlist->mmap_len); + evlist->mmap[i].base = NULL; + } + } + + free(evlist->mmap); + evlist->mmap = NULL; +} + +int perf_evlist__alloc_mmap(struct perf_evlist *evlist) +{ + evlist->nr_mmaps = evlist->cpus->nr; + if (evlist->cpus->map[0] == -1) + evlist->nr_mmaps = evlist->threads->nr; + evlist->mmap = zalloc(evlist->nr_mmaps * sizeof(struct perf_mmap)); + return evlist->mmap != NULL ? 0 : -ENOMEM; +} + +static int __perf_evlist__mmap(struct perf_evlist *evlist, + int idx, int prot, int mask, int fd) +{ + evlist->mmap[idx].prev = 0; + evlist->mmap[idx].mask = mask; + evlist->mmap[idx].base = mmap(NULL, evlist->mmap_len, prot, + MAP_SHARED, fd, 0); + if (evlist->mmap[idx].base == MAP_FAILED) + return -1; + + perf_evlist__add_pollfd(evlist, fd); + return 0; +} + +static int perf_evlist__mmap_per_cpu(struct perf_evlist *evlist, int prot, int mask) +{ + struct perf_evsel *evsel; + int cpu, thread; + + for (cpu = 0; cpu < evlist->cpus->nr; cpu++) { + int output = -1; + + for (thread = 0; thread < evlist->threads->nr; thread++) { + list_for_each_entry(evsel, &evlist->entries, node) { + int fd = FD(evsel, cpu, thread); + + if (output == -1) { + output = fd; + if (__perf_evlist__mmap(evlist, cpu, + prot, mask, output) < 0) + goto out_unmap; + } else { + if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, output) != 0) + goto out_unmap; + } + + if ((evsel->attr.read_format & PERF_FORMAT_ID) && + perf_evlist__id_add_fd(evlist, evsel, cpu, thread, fd) < 0) + goto out_unmap; + } + } + } + + return 0; + +out_unmap: + for (cpu = 0; cpu < evlist->cpus->nr; cpu++) { + if (evlist->mmap[cpu].base != NULL) { + munmap(evlist->mmap[cpu].base, evlist->mmap_len); + evlist->mmap[cpu].base = NULL; + } + } + return -1; +} + +static int perf_evlist__mmap_per_thread(struct perf_evlist *evlist, int prot, int mask) +{ + struct perf_evsel *evsel; + int thread; + + for (thread = 0; thread < evlist->threads->nr; thread++) { + int output = -1; + + list_for_each_entry(evsel, &evlist->entries, node) { + int fd = FD(evsel, 0, thread); + + if (output == -1) { + output = fd; + if (__perf_evlist__mmap(evlist, thread, + prot, mask, output) < 0) + goto out_unmap; + } else { + if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, output) != 0) + goto out_unmap; + } + + if ((evsel->attr.read_format & PERF_FORMAT_ID) && + perf_evlist__id_add_fd(evlist, evsel, 0, thread, fd) < 0) + goto out_unmap; + } + } + + return 0; + +out_unmap: + for (thread = 0; thread < evlist->threads->nr; thread++) { + if (evlist->mmap[thread].base != NULL) { + munmap(evlist->mmap[thread].base, evlist->mmap_len); + evlist->mmap[thread].base = NULL; + } + } + return -1; +} + +/** perf_evlist__mmap - Create per cpu maps to receive events + * + * @evlist - list of events + * @pages - map length in pages + * @overwrite - overwrite older events? + * + * If overwrite is false the user needs to signal event consuption using: + * + * struct perf_mmap *m = &evlist->mmap[cpu]; + * unsigned int head = perf_mmap__read_head(m); + * + * perf_mmap__write_tail(m, head) + * + * Using perf_evlist__read_on_cpu does this automatically. + */ +int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite) +{ + unsigned int page_size = sysconf(_SC_PAGE_SIZE); + int mask = pages * page_size - 1; + struct perf_evsel *evsel; + const struct cpu_map *cpus = evlist->cpus; + const struct thread_map *threads = evlist->threads; + int prot = PROT_READ | (overwrite ? 0 : PROT_WRITE); + + if (evlist->mmap == NULL && perf_evlist__alloc_mmap(evlist) < 0) + return -ENOMEM; + + if (evlist->pollfd == NULL && perf_evlist__alloc_pollfd(evlist) < 0) + return -ENOMEM; + + evlist->overwrite = overwrite; + evlist->mmap_len = (pages + 1) * page_size; + + list_for_each_entry(evsel, &evlist->entries, node) { + if ((evsel->attr.read_format & PERF_FORMAT_ID) && + evsel->sample_id == NULL && + perf_evsel__alloc_id(evsel, cpus->nr, threads->nr) < 0) + return -ENOMEM; + } + + if (evlist->cpus->map[0] == -1) + return perf_evlist__mmap_per_thread(evlist, prot, mask); + + return perf_evlist__mmap_per_cpu(evlist, prot, mask); +} + +int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid, + pid_t target_tid, const char *cpu_list) +{ + evlist->threads = thread_map__new(target_pid, target_tid); + + if (evlist->threads == NULL) + return -1; + + if (cpu_list == NULL && target_tid != -1) + evlist->cpus = cpu_map__dummy_new(); + else + evlist->cpus = cpu_map__new(cpu_list); + + if (evlist->cpus == NULL) + goto out_delete_threads; + + return 0; + +out_delete_threads: + thread_map__delete(evlist->threads); + return -1; +} + +void perf_evlist__delete_maps(struct perf_evlist *evlist) +{ + cpu_map__delete(evlist->cpus); + thread_map__delete(evlist->threads); + evlist->cpus = NULL; + evlist->threads = NULL; +} + +int perf_evlist__set_filters(struct perf_evlist *evlist) +{ + const struct thread_map *threads = evlist->threads; + const struct cpu_map *cpus = evlist->cpus; + struct perf_evsel *evsel; + char *filter; + int thread; + int cpu; + int err; + int fd; + + list_for_each_entry(evsel, &evlist->entries, node) { + filter = evsel->filter; + if (!filter) + continue; + for (cpu = 0; cpu < cpus->nr; cpu++) { + for (thread = 0; thread < threads->nr; thread++) { + fd = FD(evsel, cpu, thread); + err = ioctl(fd, PERF_EVENT_IOC_SET_FILTER, filter); + if (err) + return err; + } + } + } + + return 0; +} + +bool perf_evlist__valid_sample_type(const struct perf_evlist *evlist) +{ + struct perf_evsel *pos, *first; + + pos = first = list_entry(evlist->entries.next, struct perf_evsel, node); + + list_for_each_entry_continue(pos, &evlist->entries, node) { + if (first->attr.sample_type != pos->attr.sample_type) + return false; + } + + return true; +} + +u64 perf_evlist__sample_type(const struct perf_evlist *evlist) +{ + struct perf_evsel *first; + + first = list_entry(evlist->entries.next, struct perf_evsel, node); + return first->attr.sample_type; +} + +bool perf_evlist__valid_sample_id_all(const struct perf_evlist *evlist) +{ + struct perf_evsel *pos, *first; + + pos = first = list_entry(evlist->entries.next, struct perf_evsel, node); + + list_for_each_entry_continue(pos, &evlist->entries, node) { + if (first->attr.sample_id_all != pos->attr.sample_id_all) + return false; + } + + return true; +} + +bool perf_evlist__sample_id_all(const struct perf_evlist *evlist) +{ + struct perf_evsel *first; + + first = list_entry(evlist->entries.next, struct perf_evsel, node); + return first->attr.sample_id_all; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h new file mode 100644 index 00000000..b2b86237 --- /dev/null +++ b/tools/perf/util/evlist.h @@ -0,0 +1,74 @@ +#ifndef __PERF_EVLIST_H +#define __PERF_EVLIST_H 1 + +#include <linux/list.h> +#include "../perf.h" +#include "event.h" + +struct pollfd; +struct thread_map; +struct cpu_map; + +#define PERF_EVLIST__HLIST_BITS 8 +#define PERF_EVLIST__HLIST_SIZE (1 << PERF_EVLIST__HLIST_BITS) + +struct perf_evlist { + struct list_head entries; + struct hlist_head heads[PERF_EVLIST__HLIST_SIZE]; + int nr_entries; + int nr_fds; + int nr_mmaps; + int mmap_len; + bool overwrite; + union perf_event event_copy; + struct perf_mmap *mmap; + struct pollfd *pollfd; + struct thread_map *threads; + struct cpu_map *cpus; +}; + +struct perf_evsel; + +struct perf_evlist *perf_evlist__new(struct cpu_map *cpus, + struct thread_map *threads); +void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads); +void perf_evlist__exit(struct perf_evlist *evlist); +void perf_evlist__delete(struct perf_evlist *evlist); + +void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); +int perf_evlist__add_default(struct perf_evlist *evlist); + +void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel, + int cpu, int thread, u64 id); + +int perf_evlist__alloc_pollfd(struct perf_evlist *evlist); +void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd); + +struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id); + +union perf_event *perf_evlist__mmap_read(struct perf_evlist *self, int idx); + +int perf_evlist__alloc_mmap(struct perf_evlist *evlist); +int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite); +void perf_evlist__munmap(struct perf_evlist *evlist); + +static inline void perf_evlist__set_maps(struct perf_evlist *evlist, + struct cpu_map *cpus, + struct thread_map *threads) +{ + evlist->cpus = cpus; + evlist->threads = threads; +} + +int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid, + pid_t target_tid, const char *cpu_list); +void perf_evlist__delete_maps(struct perf_evlist *evlist); +int perf_evlist__set_filters(struct perf_evlist *evlist); + +u64 perf_evlist__sample_type(const struct perf_evlist *evlist); +bool perf_evlist__sample_id_all(const const struct perf_evlist *evlist); + +bool perf_evlist__valid_sample_type(const struct perf_evlist *evlist); +bool perf_evlist__valid_sample_id_all(const struct perf_evlist *evlist); +#endif /* __PERF_EVLIST_H */ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c new file mode 100644 index 00000000..ad2183c9 --- /dev/null +++ b/tools/perf/util/evsel.c @@ -0,0 +1,441 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Parts came from builtin-{top,stat,record}.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ + +#include "evsel.h" +#include "evlist.h" +#include "util.h" +#include "cpumap.h" +#include "thread_map.h" + +#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) + +int __perf_evsel__sample_size(u64 sample_type) +{ + u64 mask = sample_type & PERF_SAMPLE_MASK; + int size = 0; + int i; + + for (i = 0; i < 64; i++) { + if (mask & (1ULL << i)) + size++; + } + + size *= sizeof(u64); + + return size; +} + +void perf_evsel__init(struct perf_evsel *evsel, + struct perf_event_attr *attr, int idx) +{ + evsel->idx = idx; + evsel->attr = *attr; + INIT_LIST_HEAD(&evsel->node); +} + +struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx) +{ + struct perf_evsel *evsel = zalloc(sizeof(*evsel)); + + if (evsel != NULL) + perf_evsel__init(evsel, attr, idx); + + return evsel; +} + +int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads) +{ + int cpu, thread; + evsel->fd = xyarray__new(ncpus, nthreads, sizeof(int)); + + if (evsel->fd) { + for (cpu = 0; cpu < ncpus; cpu++) { + for (thread = 0; thread < nthreads; thread++) { + FD(evsel, cpu, thread) = -1; + } + } + } + + return evsel->fd != NULL ? 0 : -ENOMEM; +} + +int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads) +{ + evsel->sample_id = xyarray__new(ncpus, nthreads, sizeof(struct perf_sample_id)); + if (evsel->sample_id == NULL) + return -ENOMEM; + + evsel->id = zalloc(ncpus * nthreads * sizeof(u64)); + if (evsel->id == NULL) { + xyarray__delete(evsel->sample_id); + evsel->sample_id = NULL; + return -ENOMEM; + } + + return 0; +} + +int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus) +{ + evsel->counts = zalloc((sizeof(*evsel->counts) + + (ncpus * sizeof(struct perf_counts_values)))); + return evsel->counts != NULL ? 0 : -ENOMEM; +} + +void perf_evsel__free_fd(struct perf_evsel *evsel) +{ + xyarray__delete(evsel->fd); + evsel->fd = NULL; +} + +void perf_evsel__free_id(struct perf_evsel *evsel) +{ + xyarray__delete(evsel->sample_id); + evsel->sample_id = NULL; + free(evsel->id); + evsel->id = NULL; +} + +void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads) +{ + int cpu, thread; + + for (cpu = 0; cpu < ncpus; cpu++) + for (thread = 0; thread < nthreads; ++thread) { + close(FD(evsel, cpu, thread)); + FD(evsel, cpu, thread) = -1; + } +} + +void perf_evsel__exit(struct perf_evsel *evsel) +{ + assert(list_empty(&evsel->node)); + xyarray__delete(evsel->fd); + xyarray__delete(evsel->sample_id); + free(evsel->id); +} + +void perf_evsel__delete(struct perf_evsel *evsel) +{ + perf_evsel__exit(evsel); + close_cgroup(evsel->cgrp); + free(evsel->name); + free(evsel); +} + +int __perf_evsel__read_on_cpu(struct perf_evsel *evsel, + int cpu, int thread, bool scale) +{ + struct perf_counts_values count; + size_t nv = scale ? 3 : 1; + + if (FD(evsel, cpu, thread) < 0) + return -EINVAL; + + if (evsel->counts == NULL && perf_evsel__alloc_counts(evsel, cpu + 1) < 0) + return -ENOMEM; + + if (readn(FD(evsel, cpu, thread), &count, nv * sizeof(u64)) < 0) + return -errno; + + if (scale) { + if (count.run == 0) + count.val = 0; + else if (count.run < count.ena) + count.val = (u64)((double)count.val * count.ena / count.run + 0.5); + } else + count.ena = count.run = 0; + + evsel->counts->cpu[cpu] = count; + return 0; +} + +int __perf_evsel__read(struct perf_evsel *evsel, + int ncpus, int nthreads, bool scale) +{ + size_t nv = scale ? 3 : 1; + int cpu, thread; + struct perf_counts_values *aggr = &evsel->counts->aggr, count; + + aggr->val = aggr->ena = aggr->run = 0; + + for (cpu = 0; cpu < ncpus; cpu++) { + for (thread = 0; thread < nthreads; thread++) { + if (FD(evsel, cpu, thread) < 0) + continue; + + if (readn(FD(evsel, cpu, thread), + &count, nv * sizeof(u64)) < 0) + return -errno; + + aggr->val += count.val; + if (scale) { + aggr->ena += count.ena; + aggr->run += count.run; + } + } + } + + evsel->counts->scaled = 0; + if (scale) { + if (aggr->run == 0) { + evsel->counts->scaled = -1; + aggr->val = 0; + return 0; + } + + if (aggr->run < aggr->ena) { + evsel->counts->scaled = 1; + aggr->val = (u64)((double)aggr->val * aggr->ena / aggr->run + 0.5); + } + } else + aggr->ena = aggr->run = 0; + + return 0; +} + +static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads, bool group) +{ + int cpu, thread; + unsigned long flags = 0; + int pid = -1; + + if (evsel->fd == NULL && + perf_evsel__alloc_fd(evsel, cpus->nr, threads->nr) < 0) + return -1; + + if (evsel->cgrp) { + flags = PERF_FLAG_PID_CGROUP; + pid = evsel->cgrp->fd; + } + + for (cpu = 0; cpu < cpus->nr; cpu++) { + int group_fd = -1; + + for (thread = 0; thread < threads->nr; thread++) { + + if (!evsel->cgrp) + pid = threads->map[thread]; + + FD(evsel, cpu, thread) = sys_perf_event_open(&evsel->attr, + pid, + cpus->map[cpu], + group_fd, flags); + if (FD(evsel, cpu, thread) < 0) + goto out_close; + + if (group && group_fd == -1) + group_fd = FD(evsel, cpu, thread); + } + } + + return 0; + +out_close: + do { + while (--thread >= 0) { + close(FD(evsel, cpu, thread)); + FD(evsel, cpu, thread) = -1; + } + thread = threads->nr; + } while (--cpu >= 0); + return -1; +} + +static struct { + struct cpu_map map; + int cpus[1]; +} empty_cpu_map = { + .map.nr = 1, + .cpus = { -1, }, +}; + +static struct { + struct thread_map map; + int threads[1]; +} empty_thread_map = { + .map.nr = 1, + .threads = { -1, }, +}; + +int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads, bool group) +{ + if (cpus == NULL) { + /* Work around old compiler warnings about strict aliasing */ + cpus = &empty_cpu_map.map; + } + + if (threads == NULL) + threads = &empty_thread_map.map; + + return __perf_evsel__open(evsel, cpus, threads, group); +} + +int perf_evsel__open_per_cpu(struct perf_evsel *evsel, + struct cpu_map *cpus, bool group) +{ + return __perf_evsel__open(evsel, cpus, &empty_thread_map.map, group); +} + +int perf_evsel__open_per_thread(struct perf_evsel *evsel, + struct thread_map *threads, bool group) +{ + return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group); +} + +static int perf_event__parse_id_sample(const union perf_event *event, u64 type, + struct perf_sample *sample) +{ + const u64 *array = event->sample.array; + + array += ((event->header.size - + sizeof(event->header)) / sizeof(u64)) - 1; + + if (type & PERF_SAMPLE_CPU) { + u32 *p = (u32 *)array; + sample->cpu = *p; + array--; + } + + if (type & PERF_SAMPLE_STREAM_ID) { + sample->stream_id = *array; + array--; + } + + if (type & PERF_SAMPLE_ID) { + sample->id = *array; + array--; + } + + if (type & PERF_SAMPLE_TIME) { + sample->time = *array; + array--; + } + + if (type & PERF_SAMPLE_TID) { + u32 *p = (u32 *)array; + sample->pid = p[0]; + sample->tid = p[1]; + } + + return 0; +} + +static bool sample_overlap(const union perf_event *event, + const void *offset, u64 size) +{ + const void *base = event; + + if (offset + size > base + event->header.size) + return true; + + return false; +} + +int perf_event__parse_sample(const union perf_event *event, u64 type, + int sample_size, bool sample_id_all, + struct perf_sample *data) +{ + const u64 *array; + + data->cpu = data->pid = data->tid = -1; + data->stream_id = data->id = data->time = -1ULL; + data->period = 1; + + if (event->header.type != PERF_RECORD_SAMPLE) { + if (!sample_id_all) + return 0; + return perf_event__parse_id_sample(event, type, data); + } + + array = event->sample.array; + + if (sample_size + sizeof(event->header) > event->header.size) + return -EFAULT; + + if (type & PERF_SAMPLE_IP) { + data->ip = event->ip.ip; + array++; + } + + if (type & PERF_SAMPLE_TID) { + u32 *p = (u32 *)array; + data->pid = p[0]; + data->tid = p[1]; + array++; + } + + if (type & PERF_SAMPLE_TIME) { + data->time = *array; + array++; + } + + if (type & PERF_SAMPLE_ADDR) { + data->addr = *array; + array++; + } + + data->id = -1ULL; + if (type & PERF_SAMPLE_ID) { + data->id = *array; + array++; + } + + if (type & PERF_SAMPLE_STREAM_ID) { + data->stream_id = *array; + array++; + } + + if (type & PERF_SAMPLE_CPU) { + u32 *p = (u32 *)array; + data->cpu = *p; + array++; + } + + if (type & PERF_SAMPLE_PERIOD) { + data->period = *array; + array++; + } + + if (type & PERF_SAMPLE_READ) { + fprintf(stderr, "PERF_SAMPLE_READ is unsuported for now\n"); + return -1; + } + + if (type & PERF_SAMPLE_CALLCHAIN) { + if (sample_overlap(event, array, sizeof(data->callchain->nr))) + return -EFAULT; + + data->callchain = (struct ip_callchain *)array; + + if (sample_overlap(event, array, data->callchain->nr)) + return -EFAULT; + + array += 1 + data->callchain->nr; + } + + if (type & PERF_SAMPLE_RAW) { + u32 *p = (u32 *)array; + + if (sample_overlap(event, array, sizeof(u32))) + return -EFAULT; + + data->raw_size = *p; + p++; + + if (sample_overlap(event, p, data->raw_size)) + return -EFAULT; + + data->raw_data = p; + } + + return 0; +} diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h new file mode 100644 index 00000000..7e9366e4 --- /dev/null +++ b/tools/perf/util/evsel.h @@ -0,0 +1,159 @@ +#ifndef __PERF_EVSEL_H +#define __PERF_EVSEL_H 1 + +#include <linux/list.h> +#include <stdbool.h> +#include "../../../include/linux/perf_event.h" +#include "types.h" +#include "xyarray.h" +#include "cgroup.h" +#include "hist.h" + +struct perf_counts_values { + union { + struct { + u64 val; + u64 ena; + u64 run; + }; + u64 values[3]; + }; +}; + +struct perf_counts { + s8 scaled; + struct perf_counts_values aggr; + struct perf_counts_values cpu[]; +}; + +struct perf_evsel; + +/* + * Per fd, to map back from PERF_SAMPLE_ID to evsel, only used when there are + * more than one entry in the evlist. + */ +struct perf_sample_id { + struct hlist_node node; + u64 id; + struct perf_evsel *evsel; +}; + +/** struct perf_evsel - event selector + * + * @name - Can be set to retain the original event name passed by the user, + * so that when showing results in tools such as 'perf stat', we + * show the name used, not some alias. + */ +struct perf_evsel { + struct list_head node; + struct perf_event_attr attr; + char *filter; + struct xyarray *fd; + struct xyarray *sample_id; + u64 *id; + struct perf_counts *counts; + int idx; + int ids; + struct hists hists; + char *name; + union { + void *priv; + off_t id_offset; + }; + struct cgroup_sel *cgrp; +}; + +struct cpu_map; +struct thread_map; +struct perf_evlist; + +struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx); +void perf_evsel__init(struct perf_evsel *evsel, + struct perf_event_attr *attr, int idx); +void perf_evsel__exit(struct perf_evsel *evsel); +void perf_evsel__delete(struct perf_evsel *evsel); + +int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads); +int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads); +int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus); +void perf_evsel__free_fd(struct perf_evsel *evsel); +void perf_evsel__free_id(struct perf_evsel *evsel); +void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads); + +int perf_evsel__open_per_cpu(struct perf_evsel *evsel, + struct cpu_map *cpus, bool group); +int perf_evsel__open_per_thread(struct perf_evsel *evsel, + struct thread_map *threads, bool group); +int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads, bool group); + +#define perf_evsel__match(evsel, t, c) \ + (evsel->attr.type == PERF_TYPE_##t && \ + evsel->attr.config == PERF_COUNT_##c) + +int __perf_evsel__read_on_cpu(struct perf_evsel *evsel, + int cpu, int thread, bool scale); + +/** + * perf_evsel__read_on_cpu - Read out the results on a CPU and thread + * + * @evsel - event selector to read value + * @cpu - CPU of interest + * @thread - thread of interest + */ +static inline int perf_evsel__read_on_cpu(struct perf_evsel *evsel, + int cpu, int thread) +{ + return __perf_evsel__read_on_cpu(evsel, cpu, thread, false); +} + +/** + * perf_evsel__read_on_cpu_scaled - Read out the results on a CPU and thread, scaled + * + * @evsel - event selector to read value + * @cpu - CPU of interest + * @thread - thread of interest + */ +static inline int perf_evsel__read_on_cpu_scaled(struct perf_evsel *evsel, + int cpu, int thread) +{ + return __perf_evsel__read_on_cpu(evsel, cpu, thread, true); +} + +int __perf_evsel__read(struct perf_evsel *evsel, int ncpus, int nthreads, + bool scale); + +/** + * perf_evsel__read - Read the aggregate results on all CPUs + * + * @evsel - event selector to read value + * @ncpus - Number of cpus affected, from zero + * @nthreads - Number of threads affected, from zero + */ +static inline int perf_evsel__read(struct perf_evsel *evsel, + int ncpus, int nthreads) +{ + return __perf_evsel__read(evsel, ncpus, nthreads, false); +} + +/** + * perf_evsel__read_scaled - Read the aggregate results on all CPUs, scaled + * + * @evsel - event selector to read value + * @ncpus - Number of cpus affected, from zero + * @nthreads - Number of threads affected, from zero + */ +static inline int perf_evsel__read_scaled(struct perf_evsel *evsel, + int ncpus, int nthreads) +{ + return __perf_evsel__read(evsel, ncpus, nthreads, true); +} + +int __perf_evsel__sample_size(u64 sample_type); + +static inline int perf_evsel__sample_size(struct perf_evsel *evsel) +{ + return __perf_evsel__sample_size(evsel->attr.sample_type); +} + +#endif /* __PERF_EVSEL_H */ diff --git a/tools/perf/util/exec_cmd.c b/tools/perf/util/exec_cmd.c new file mode 100644 index 00000000..7adf4ad1 --- /dev/null +++ b/tools/perf/util/exec_cmd.c @@ -0,0 +1,148 @@ +#include "cache.h" +#include "exec_cmd.h" +#include "quote.h" + +#include <string.h> + +#define MAX_ARGS 32 + +static const char *argv_exec_path; +static const char *argv0_path; + +const char *system_path(const char *path) +{ + static const char *prefix = PREFIX; + struct strbuf d = STRBUF_INIT; + + if (is_absolute_path(path)) + return path; + + strbuf_addf(&d, "%s/%s", prefix, path); + path = strbuf_detach(&d, NULL); + return path; +} + +const char *perf_extract_argv0_path(const char *argv0) +{ + const char *slash; + + if (!argv0 || !*argv0) + return NULL; + slash = argv0 + strlen(argv0); + + while (argv0 <= slash && !is_dir_sep(*slash)) + slash--; + + if (slash >= argv0) { + argv0_path = strndup(argv0, slash - argv0); + return argv0_path ? slash + 1 : NULL; + } + + return argv0; +} + +void perf_set_argv_exec_path(const char *exec_path) +{ + argv_exec_path = exec_path; + /* + * Propagate this setting to external programs. + */ + setenv(EXEC_PATH_ENVIRONMENT, exec_path, 1); +} + + +/* Returns the highest-priority, location to look for perf programs. */ +const char *perf_exec_path(void) +{ + const char *env; + + if (argv_exec_path) + return argv_exec_path; + + env = getenv(EXEC_PATH_ENVIRONMENT); + if (env && *env) { + return env; + } + + return system_path(PERF_EXEC_PATH); +} + +static void add_path(struct strbuf *out, const char *path) +{ + if (path && *path) { + if (is_absolute_path(path)) + strbuf_addstr(out, path); + else + strbuf_addstr(out, make_nonrelative_path(path)); + + strbuf_addch(out, PATH_SEP); + } +} + +void setup_path(void) +{ + const char *old_path = getenv("PATH"); + struct strbuf new_path = STRBUF_INIT; + + add_path(&new_path, perf_exec_path()); + add_path(&new_path, argv0_path); + + if (old_path) + strbuf_addstr(&new_path, old_path); + else + strbuf_addstr(&new_path, "/usr/local/bin:/usr/bin:/bin"); + + setenv("PATH", new_path.buf, 1); + + strbuf_release(&new_path); +} + +static const char **prepare_perf_cmd(const char **argv) +{ + int argc; + const char **nargv; + + for (argc = 0; argv[argc]; argc++) + ; /* just counting */ + nargv = malloc(sizeof(*nargv) * (argc + 2)); + + nargv[0] = "perf"; + for (argc = 0; argv[argc]; argc++) + nargv[argc + 1] = argv[argc]; + nargv[argc + 1] = NULL; + return nargv; +} + +int execv_perf_cmd(const char **argv) { + const char **nargv = prepare_perf_cmd(argv); + + /* execvp() can only ever return if it fails */ + execvp("perf", (char **)nargv); + + free(nargv); + return -1; +} + + +int execl_perf_cmd(const char *cmd,...) +{ + int argc; + const char *argv[MAX_ARGS + 1]; + const char *arg; + va_list param; + + va_start(param, cmd); + argv[0] = cmd; + argc = 1; + while (argc < MAX_ARGS) { + arg = argv[argc++] = va_arg(param, char *); + if (!arg) + break; + } + va_end(param); + if (MAX_ARGS <= argc) + return error("too many args to run %s", cmd); + + argv[argc] = NULL; + return execv_perf_cmd(argv); +} diff --git a/tools/perf/util/exec_cmd.h b/tools/perf/util/exec_cmd.h new file mode 100644 index 00000000..bc4b9159 --- /dev/null +++ b/tools/perf/util/exec_cmd.h @@ -0,0 +1,12 @@ +#ifndef __PERF_EXEC_CMD_H +#define __PERF_EXEC_CMD_H + +extern void perf_set_argv_exec_path(const char *exec_path); +extern const char *perf_extract_argv0_path(const char *path); +extern const char *perf_exec_path(void); +extern void setup_path(void); +extern int execv_perf_cmd(const char **argv); /* NULL terminated */ +extern int execl_perf_cmd(const char *cmd, ...); +extern const char *system_path(const char *path); + +#endif /* __PERF_EXEC_CMD_H */ diff --git a/tools/perf/util/generate-cmdlist.sh b/tools/perf/util/generate-cmdlist.sh new file mode 100755 index 00000000..f06f6fd1 --- /dev/null +++ b/tools/perf/util/generate-cmdlist.sh @@ -0,0 +1,24 @@ +#!/bin/sh + +echo "/* Automatically generated by $0 */ +struct cmdname_help +{ + char name[16]; + char help[80]; +}; + +static struct cmdname_help common_cmds[] = {" + +sed -n -e 's/^perf-\([^ ]*\)[ ].* common.*/\1/p' command-list.txt | +sort | +while read cmd +do + sed -n ' + /^NAME/,/perf-'"$cmd"'/H + ${ + x + s/.*perf-'"$cmd"' - \(.*\)/ {"'"$cmd"'", "\1"},/ + p + }' "Documentation/perf-$cmd.txt" +done +echo "};" diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c new file mode 100644 index 00000000..cb2959a3 --- /dev/null +++ b/tools/perf/util/header.c @@ -0,0 +1,1185 @@ +#define _FILE_OFFSET_BITS 64 + +#include <sys/types.h> +#include <byteswap.h> +#include <unistd.h> +#include <stdio.h> +#include <stdlib.h> +#include <linux/list.h> +#include <linux/kernel.h> + +#include "evlist.h" +#include "evsel.h" +#include "util.h" +#include "header.h" +#include "../perf.h" +#include "trace-event.h" +#include "session.h" +#include "symbol.h" +#include "debug.h" + +static bool no_buildid_cache = false; + +static int event_count; +static struct perf_trace_event_type *events; + +int perf_header__push_event(u64 id, const char *name) +{ + if (strlen(name) > MAX_EVENT_NAME) + pr_warning("Event %s will be truncated\n", name); + + if (!events) { + events = malloc(sizeof(struct perf_trace_event_type)); + if (events == NULL) + return -ENOMEM; + } else { + struct perf_trace_event_type *nevents; + + nevents = realloc(events, (event_count + 1) * sizeof(*events)); + if (nevents == NULL) + return -ENOMEM; + events = nevents; + } + memset(&events[event_count], 0, sizeof(struct perf_trace_event_type)); + events[event_count].event_id = id; + strncpy(events[event_count].name, name, MAX_EVENT_NAME - 1); + event_count++; + return 0; +} + +char *perf_header__find_event(u64 id) +{ + int i; + for (i = 0 ; i < event_count; i++) { + if (events[i].event_id == id) + return events[i].name; + } + return NULL; +} + +static const char *__perf_magic = "PERFFILE"; + +#define PERF_MAGIC (*(u64 *)__perf_magic) + +struct perf_file_attr { + struct perf_event_attr attr; + struct perf_file_section ids; +}; + +void perf_header__set_feat(struct perf_header *header, int feat) +{ + set_bit(feat, header->adds_features); +} + +void perf_header__clear_feat(struct perf_header *header, int feat) +{ + clear_bit(feat, header->adds_features); +} + +bool perf_header__has_feat(const struct perf_header *header, int feat) +{ + return test_bit(feat, header->adds_features); +} + +static int do_write(int fd, const void *buf, size_t size) +{ + while (size) { + int ret = write(fd, buf, size); + + if (ret < 0) + return -errno; + + size -= ret; + buf += ret; + } + + return 0; +} + +#define NAME_ALIGN 64 + +static int write_padded(int fd, const void *bf, size_t count, + size_t count_aligned) +{ + static const char zero_buf[NAME_ALIGN]; + int err = do_write(fd, bf, count); + + if (!err) + err = do_write(fd, zero_buf, count_aligned - count); + + return err; +} + +#define dsos__for_each_with_build_id(pos, head) \ + list_for_each_entry(pos, head, node) \ + if (!pos->has_build_id) \ + continue; \ + else + +static int __dsos__write_buildid_table(struct list_head *head, pid_t pid, + u16 misc, int fd) +{ + struct dso *pos; + + dsos__for_each_with_build_id(pos, head) { + int err; + struct build_id_event b; + size_t len; + + if (!pos->hit) + continue; + len = pos->long_name_len + 1; + len = ALIGN(len, NAME_ALIGN); + memset(&b, 0, sizeof(b)); + memcpy(&b.build_id, pos->build_id, sizeof(pos->build_id)); + b.pid = pid; + b.header.misc = misc; + b.header.size = sizeof(b) + len; + err = do_write(fd, &b, sizeof(b)); + if (err < 0) + return err; + err = write_padded(fd, pos->long_name, + pos->long_name_len + 1, len); + if (err < 0) + return err; + } + + return 0; +} + +static int machine__write_buildid_table(struct machine *machine, int fd) +{ + int err; + u16 kmisc = PERF_RECORD_MISC_KERNEL, + umisc = PERF_RECORD_MISC_USER; + + if (!machine__is_host(machine)) { + kmisc = PERF_RECORD_MISC_GUEST_KERNEL; + umisc = PERF_RECORD_MISC_GUEST_USER; + } + + err = __dsos__write_buildid_table(&machine->kernel_dsos, machine->pid, + kmisc, fd); + if (err == 0) + err = __dsos__write_buildid_table(&machine->user_dsos, + machine->pid, umisc, fd); + return err; +} + +static int dsos__write_buildid_table(struct perf_header *header, int fd) +{ + struct perf_session *session = container_of(header, + struct perf_session, header); + struct rb_node *nd; + int err = machine__write_buildid_table(&session->host_machine, fd); + + if (err) + return err; + + for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { + struct machine *pos = rb_entry(nd, struct machine, rb_node); + err = machine__write_buildid_table(pos, fd); + if (err) + break; + } + return err; +} + +int build_id_cache__add_s(const char *sbuild_id, const char *debugdir, + const char *name, bool is_kallsyms) +{ + const size_t size = PATH_MAX; + char *realname, *filename = malloc(size), + *linkname = malloc(size), *targetname; + int len, err = -1; + + if (is_kallsyms) { + if (symbol_conf.kptr_restrict) { + pr_debug("Not caching a kptr_restrict'ed /proc/kallsyms\n"); + return 0; + } + realname = (char *)name; + } else + realname = realpath(name, NULL); + + if (realname == NULL || filename == NULL || linkname == NULL) + goto out_free; + + len = snprintf(filename, size, "%s%s%s", + debugdir, is_kallsyms ? "/" : "", realname); + if (mkdir_p(filename, 0755)) + goto out_free; + + snprintf(filename + len, sizeof(filename) - len, "/%s", sbuild_id); + + if (access(filename, F_OK)) { + if (is_kallsyms) { + if (copyfile("/proc/kallsyms", filename)) + goto out_free; + } else if (link(realname, filename) && copyfile(name, filename)) + goto out_free; + } + + len = snprintf(linkname, size, "%s/.build-id/%.2s", + debugdir, sbuild_id); + + if (access(linkname, X_OK) && mkdir_p(linkname, 0755)) + goto out_free; + + snprintf(linkname + len, size - len, "/%s", sbuild_id + 2); + targetname = filename + strlen(debugdir) - 5; + memcpy(targetname, "../..", 5); + + if (symlink(targetname, linkname) == 0) + err = 0; +out_free: + if (!is_kallsyms) + free(realname); + free(filename); + free(linkname); + return err; +} + +static int build_id_cache__add_b(const u8 *build_id, size_t build_id_size, + const char *name, const char *debugdir, + bool is_kallsyms) +{ + char sbuild_id[BUILD_ID_SIZE * 2 + 1]; + + build_id__sprintf(build_id, build_id_size, sbuild_id); + + return build_id_cache__add_s(sbuild_id, debugdir, name, is_kallsyms); +} + +int build_id_cache__remove_s(const char *sbuild_id, const char *debugdir) +{ + const size_t size = PATH_MAX; + char *filename = malloc(size), + *linkname = malloc(size); + int err = -1; + + if (filename == NULL || linkname == NULL) + goto out_free; + + snprintf(linkname, size, "%s/.build-id/%.2s/%s", + debugdir, sbuild_id, sbuild_id + 2); + + if (access(linkname, F_OK)) + goto out_free; + + if (readlink(linkname, filename, size) < 0) + goto out_free; + + if (unlink(linkname)) + goto out_free; + + /* + * Since the link is relative, we must make it absolute: + */ + snprintf(linkname, size, "%s/.build-id/%.2s/%s", + debugdir, sbuild_id, filename); + + if (unlink(linkname)) + goto out_free; + + err = 0; +out_free: + free(filename); + free(linkname); + return err; +} + +static int dso__cache_build_id(struct dso *dso, const char *debugdir) +{ + bool is_kallsyms = dso->kernel && dso->long_name[0] != '/'; + + return build_id_cache__add_b(dso->build_id, sizeof(dso->build_id), + dso->long_name, debugdir, is_kallsyms); +} + +static int __dsos__cache_build_ids(struct list_head *head, const char *debugdir) +{ + struct dso *pos; + int err = 0; + + dsos__for_each_with_build_id(pos, head) + if (dso__cache_build_id(pos, debugdir)) + err = -1; + + return err; +} + +static int machine__cache_build_ids(struct machine *machine, const char *debugdir) +{ + int ret = __dsos__cache_build_ids(&machine->kernel_dsos, debugdir); + ret |= __dsos__cache_build_ids(&machine->user_dsos, debugdir); + return ret; +} + +static int perf_session__cache_build_ids(struct perf_session *session) +{ + struct rb_node *nd; + int ret; + char debugdir[PATH_MAX]; + + snprintf(debugdir, sizeof(debugdir), "%s", buildid_dir); + + if (mkdir(debugdir, 0755) != 0 && errno != EEXIST) + return -1; + + ret = machine__cache_build_ids(&session->host_machine, debugdir); + + for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { + struct machine *pos = rb_entry(nd, struct machine, rb_node); + ret |= machine__cache_build_ids(pos, debugdir); + } + return ret ? -1 : 0; +} + +static bool machine__read_build_ids(struct machine *machine, bool with_hits) +{ + bool ret = __dsos__read_build_ids(&machine->kernel_dsos, with_hits); + ret |= __dsos__read_build_ids(&machine->user_dsos, with_hits); + return ret; +} + +static bool perf_session__read_build_ids(struct perf_session *session, bool with_hits) +{ + struct rb_node *nd; + bool ret = machine__read_build_ids(&session->host_machine, with_hits); + + for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { + struct machine *pos = rb_entry(nd, struct machine, rb_node); + ret |= machine__read_build_ids(pos, with_hits); + } + + return ret; +} + +static int perf_header__adds_write(struct perf_header *header, + struct perf_evlist *evlist, int fd) +{ + int nr_sections; + struct perf_session *session; + struct perf_file_section *feat_sec; + int sec_size; + u64 sec_start; + int idx = 0, err; + + session = container_of(header, struct perf_session, header); + + if (perf_header__has_feat(header, HEADER_BUILD_ID && + !perf_session__read_build_ids(session, true))) + perf_header__clear_feat(header, HEADER_BUILD_ID); + + nr_sections = bitmap_weight(header->adds_features, HEADER_FEAT_BITS); + if (!nr_sections) + return 0; + + feat_sec = calloc(sizeof(*feat_sec), nr_sections); + if (feat_sec == NULL) + return -ENOMEM; + + sec_size = sizeof(*feat_sec) * nr_sections; + + sec_start = header->data_offset + header->data_size; + lseek(fd, sec_start + sec_size, SEEK_SET); + + if (perf_header__has_feat(header, HEADER_TRACE_INFO)) { + struct perf_file_section *trace_sec; + + trace_sec = &feat_sec[idx++]; + + /* Write trace info */ + trace_sec->offset = lseek(fd, 0, SEEK_CUR); + read_tracing_data(fd, &evlist->entries); + trace_sec->size = lseek(fd, 0, SEEK_CUR) - trace_sec->offset; + } + + if (perf_header__has_feat(header, HEADER_BUILD_ID)) { + struct perf_file_section *buildid_sec; + + buildid_sec = &feat_sec[idx++]; + + /* Write build-ids */ + buildid_sec->offset = lseek(fd, 0, SEEK_CUR); + err = dsos__write_buildid_table(header, fd); + if (err < 0) { + pr_debug("failed to write buildid table\n"); + goto out_free; + } + buildid_sec->size = lseek(fd, 0, SEEK_CUR) - + buildid_sec->offset; + if (!no_buildid_cache) + perf_session__cache_build_ids(session); + } + + lseek(fd, sec_start, SEEK_SET); + err = do_write(fd, feat_sec, sec_size); + if (err < 0) + pr_debug("failed to write feature section\n"); +out_free: + free(feat_sec); + return err; +} + +int perf_header__write_pipe(int fd) +{ + struct perf_pipe_file_header f_header; + int err; + + f_header = (struct perf_pipe_file_header){ + .magic = PERF_MAGIC, + .size = sizeof(f_header), + }; + + err = do_write(fd, &f_header, sizeof(f_header)); + if (err < 0) { + pr_debug("failed to write perf pipe header\n"); + return err; + } + + return 0; +} + +int perf_session__write_header(struct perf_session *session, + struct perf_evlist *evlist, + int fd, bool at_exit) +{ + struct perf_file_header f_header; + struct perf_file_attr f_attr; + struct perf_header *header = &session->header; + struct perf_evsel *attr, *pair = NULL; + int err; + + lseek(fd, sizeof(f_header), SEEK_SET); + + if (session->evlist != evlist) + pair = list_entry(session->evlist->entries.next, struct perf_evsel, node); + + list_for_each_entry(attr, &evlist->entries, node) { + attr->id_offset = lseek(fd, 0, SEEK_CUR); + err = do_write(fd, attr->id, attr->ids * sizeof(u64)); + if (err < 0) { +out_err_write: + pr_debug("failed to write perf header\n"); + return err; + } + if (session->evlist != evlist) { + err = do_write(fd, pair->id, pair->ids * sizeof(u64)); + if (err < 0) + goto out_err_write; + attr->ids += pair->ids; + pair = list_entry(pair->node.next, struct perf_evsel, node); + } + } + + header->attr_offset = lseek(fd, 0, SEEK_CUR); + + list_for_each_entry(attr, &evlist->entries, node) { + f_attr = (struct perf_file_attr){ + .attr = attr->attr, + .ids = { + .offset = attr->id_offset, + .size = attr->ids * sizeof(u64), + } + }; + err = do_write(fd, &f_attr, sizeof(f_attr)); + if (err < 0) { + pr_debug("failed to write perf header attribute\n"); + return err; + } + } + + header->event_offset = lseek(fd, 0, SEEK_CUR); + header->event_size = event_count * sizeof(struct perf_trace_event_type); + if (events) { + err = do_write(fd, events, header->event_size); + if (err < 0) { + pr_debug("failed to write perf header events\n"); + return err; + } + } + + header->data_offset = lseek(fd, 0, SEEK_CUR); + + if (at_exit) { + err = perf_header__adds_write(header, evlist, fd); + if (err < 0) + return err; + } + + f_header = (struct perf_file_header){ + .magic = PERF_MAGIC, + .size = sizeof(f_header), + .attr_size = sizeof(f_attr), + .attrs = { + .offset = header->attr_offset, + .size = evlist->nr_entries * sizeof(f_attr), + }, + .data = { + .offset = header->data_offset, + .size = header->data_size, + }, + .event_types = { + .offset = header->event_offset, + .size = header->event_size, + }, + }; + + memcpy(&f_header.adds_features, &header->adds_features, sizeof(header->adds_features)); + + lseek(fd, 0, SEEK_SET); + err = do_write(fd, &f_header, sizeof(f_header)); + if (err < 0) { + pr_debug("failed to write perf header\n"); + return err; + } + lseek(fd, header->data_offset + header->data_size, SEEK_SET); + + header->frozen = 1; + return 0; +} + +static int perf_header__getbuffer64(struct perf_header *header, + int fd, void *buf, size_t size) +{ + if (readn(fd, buf, size) <= 0) + return -1; + + if (header->needs_swap) + mem_bswap_64(buf, size); + + return 0; +} + +int perf_header__process_sections(struct perf_header *header, int fd, + int (*process)(struct perf_file_section *section, + struct perf_header *ph, + int feat, int fd)) +{ + struct perf_file_section *feat_sec; + int nr_sections; + int sec_size; + int idx = 0; + int err = -1, feat = 1; + + nr_sections = bitmap_weight(header->adds_features, HEADER_FEAT_BITS); + if (!nr_sections) + return 0; + + feat_sec = calloc(sizeof(*feat_sec), nr_sections); + if (!feat_sec) + return -1; + + sec_size = sizeof(*feat_sec) * nr_sections; + + lseek(fd, header->data_offset + header->data_size, SEEK_SET); + + if (perf_header__getbuffer64(header, fd, feat_sec, sec_size)) + goto out_free; + + err = 0; + while (idx < nr_sections && feat < HEADER_LAST_FEATURE) { + if (perf_header__has_feat(header, feat)) { + struct perf_file_section *sec = &feat_sec[idx++]; + + err = process(sec, header, feat, fd); + if (err < 0) + break; + } + ++feat; + } +out_free: + free(feat_sec); + return err; +} + +int perf_file_header__read(struct perf_file_header *header, + struct perf_header *ph, int fd) +{ + lseek(fd, 0, SEEK_SET); + + if (readn(fd, header, sizeof(*header)) <= 0 || + memcmp(&header->magic, __perf_magic, sizeof(header->magic))) + return -1; + + if (header->attr_size != sizeof(struct perf_file_attr)) { + u64 attr_size = bswap_64(header->attr_size); + + if (attr_size != sizeof(struct perf_file_attr)) + return -1; + + mem_bswap_64(header, offsetof(struct perf_file_header, + adds_features)); + ph->needs_swap = true; + } + + if (header->size != sizeof(*header)) { + /* Support the previous format */ + if (header->size == offsetof(typeof(*header), adds_features)) + bitmap_zero(header->adds_features, HEADER_FEAT_BITS); + else + return -1; + } + + memcpy(&ph->adds_features, &header->adds_features, + sizeof(ph->adds_features)); + /* + * FIXME: hack that assumes that if we need swap the perf.data file + * may be coming from an arch with a different word-size, ergo different + * DEFINE_BITMAP format, investigate more later, but for now its mostly + * safe to assume that we have a build-id section. Trace files probably + * have several other issues in this realm anyway... + */ + if (ph->needs_swap) { + memset(&ph->adds_features, 0, sizeof(ph->adds_features)); + perf_header__set_feat(ph, HEADER_BUILD_ID); + } + + ph->event_offset = header->event_types.offset; + ph->event_size = header->event_types.size; + ph->data_offset = header->data.offset; + ph->data_size = header->data.size; + return 0; +} + +static int __event_process_build_id(struct build_id_event *bev, + char *filename, + struct perf_session *session) +{ + int err = -1; + struct list_head *head; + struct machine *machine; + u16 misc; + struct dso *dso; + enum dso_kernel_type dso_type; + + machine = perf_session__findnew_machine(session, bev->pid); + if (!machine) + goto out; + + misc = bev->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + + switch (misc) { + case PERF_RECORD_MISC_KERNEL: + dso_type = DSO_TYPE_KERNEL; + head = &machine->kernel_dsos; + break; + case PERF_RECORD_MISC_GUEST_KERNEL: + dso_type = DSO_TYPE_GUEST_KERNEL; + head = &machine->kernel_dsos; + break; + case PERF_RECORD_MISC_USER: + case PERF_RECORD_MISC_GUEST_USER: + dso_type = DSO_TYPE_USER; + head = &machine->user_dsos; + break; + default: + goto out; + } + + dso = __dsos__findnew(head, filename); + if (dso != NULL) { + char sbuild_id[BUILD_ID_SIZE * 2 + 1]; + + dso__set_build_id(dso, &bev->build_id); + + if (filename[0] == '[') + dso->kernel = dso_type; + + build_id__sprintf(dso->build_id, sizeof(dso->build_id), + sbuild_id); + pr_debug("build id event received for %s: %s\n", + dso->long_name, sbuild_id); + } + + err = 0; +out: + return err; +} + +static int perf_header__read_build_ids_abi_quirk(struct perf_header *header, + int input, u64 offset, u64 size) +{ + struct perf_session *session = container_of(header, struct perf_session, header); + struct { + struct perf_event_header header; + u8 build_id[ALIGN(BUILD_ID_SIZE, sizeof(u64))]; + char filename[0]; + } old_bev; + struct build_id_event bev; + char filename[PATH_MAX]; + u64 limit = offset + size; + + while (offset < limit) { + ssize_t len; + + if (read(input, &old_bev, sizeof(old_bev)) != sizeof(old_bev)) + return -1; + + if (header->needs_swap) + perf_event_header__bswap(&old_bev.header); + + len = old_bev.header.size - sizeof(old_bev); + if (read(input, filename, len) != len) + return -1; + + bev.header = old_bev.header; + bev.pid = 0; + memcpy(bev.build_id, old_bev.build_id, sizeof(bev.build_id)); + __event_process_build_id(&bev, filename, session); + + offset += bev.header.size; + } + + return 0; +} + +static int perf_header__read_build_ids(struct perf_header *header, + int input, u64 offset, u64 size) +{ + struct perf_session *session = container_of(header, struct perf_session, header); + struct build_id_event bev; + char filename[PATH_MAX]; + u64 limit = offset + size, orig_offset = offset; + int err = -1; + + while (offset < limit) { + ssize_t len; + + if (read(input, &bev, sizeof(bev)) != sizeof(bev)) + goto out; + + if (header->needs_swap) + perf_event_header__bswap(&bev.header); + + len = bev.header.size - sizeof(bev); + if (read(input, filename, len) != len) + goto out; + /* + * The a1645ce1 changeset: + * + * "perf: 'perf kvm' tool for monitoring guest performance from host" + * + * Added a field to struct build_id_event that broke the file + * format. + * + * Since the kernel build-id is the first entry, process the + * table using the old format if the well known + * '[kernel.kallsyms]' string for the kernel build-id has the + * first 4 characters chopped off (where the pid_t sits). + */ + if (memcmp(filename, "nel.kallsyms]", 13) == 0) { + if (lseek(input, orig_offset, SEEK_SET) == (off_t)-1) + return -1; + return perf_header__read_build_ids_abi_quirk(header, input, offset, size); + } + + __event_process_build_id(&bev, filename, session); + + offset += bev.header.size; + } + err = 0; +out: + return err; +} + +static int perf_file_section__process(struct perf_file_section *section, + struct perf_header *ph, + int feat, int fd) +{ + if (lseek(fd, section->offset, SEEK_SET) == (off_t)-1) { + pr_debug("Failed to lseek to %" PRIu64 " offset for feature " + "%d, continuing...\n", section->offset, feat); + return 0; + } + + switch (feat) { + case HEADER_TRACE_INFO: + trace_report(fd, false); + break; + + case HEADER_BUILD_ID: + if (perf_header__read_build_ids(ph, fd, section->offset, section->size)) + pr_debug("Failed to read buildids, continuing...\n"); + break; + default: + pr_debug("unknown feature %d, continuing...\n", feat); + } + + return 0; +} + +static int perf_file_header__read_pipe(struct perf_pipe_file_header *header, + struct perf_header *ph, int fd, + bool repipe) +{ + if (readn(fd, header, sizeof(*header)) <= 0 || + memcmp(&header->magic, __perf_magic, sizeof(header->magic))) + return -1; + + if (repipe && do_write(STDOUT_FILENO, header, sizeof(*header)) < 0) + return -1; + + if (header->size != sizeof(*header)) { + u64 size = bswap_64(header->size); + + if (size != sizeof(*header)) + return -1; + + ph->needs_swap = true; + } + + return 0; +} + +static int perf_header__read_pipe(struct perf_session *session, int fd) +{ + struct perf_header *header = &session->header; + struct perf_pipe_file_header f_header; + + if (perf_file_header__read_pipe(&f_header, header, fd, + session->repipe) < 0) { + pr_debug("incompatible file format\n"); + return -EINVAL; + } + + session->fd = fd; + + return 0; +} + +int perf_session__read_header(struct perf_session *session, int fd) +{ + struct perf_header *header = &session->header; + struct perf_file_header f_header; + struct perf_file_attr f_attr; + u64 f_id; + int nr_attrs, nr_ids, i, j; + + session->evlist = perf_evlist__new(NULL, NULL); + if (session->evlist == NULL) + return -ENOMEM; + + if (session->fd_pipe) + return perf_header__read_pipe(session, fd); + + if (perf_file_header__read(&f_header, header, fd) < 0) { + pr_debug("incompatible file format\n"); + return -EINVAL; + } + + nr_attrs = f_header.attrs.size / sizeof(f_attr); + lseek(fd, f_header.attrs.offset, SEEK_SET); + + for (i = 0; i < nr_attrs; i++) { + struct perf_evsel *evsel; + off_t tmp; + + if (readn(fd, &f_attr, sizeof(f_attr)) <= 0) + goto out_errno; + + if (header->needs_swap) + perf_event__attr_swap(&f_attr.attr); + + tmp = lseek(fd, 0, SEEK_CUR); + evsel = perf_evsel__new(&f_attr.attr, i); + + if (evsel == NULL) + goto out_delete_evlist; + /* + * Do it before so that if perf_evsel__alloc_id fails, this + * entry gets purged too at perf_evlist__delete(). + */ + perf_evlist__add(session->evlist, evsel); + + nr_ids = f_attr.ids.size / sizeof(u64); + /* + * We don't have the cpu and thread maps on the header, so + * for allocating the perf_sample_id table we fake 1 cpu and + * hattr->ids threads. + */ + if (perf_evsel__alloc_id(evsel, 1, nr_ids)) + goto out_delete_evlist; + + lseek(fd, f_attr.ids.offset, SEEK_SET); + + for (j = 0; j < nr_ids; j++) { + if (perf_header__getbuffer64(header, fd, &f_id, sizeof(f_id))) + goto out_errno; + + perf_evlist__id_add(session->evlist, evsel, 0, j, f_id); + } + + lseek(fd, tmp, SEEK_SET); + } + + if (f_header.event_types.size) { + lseek(fd, f_header.event_types.offset, SEEK_SET); + events = malloc(f_header.event_types.size); + if (events == NULL) + return -ENOMEM; + if (perf_header__getbuffer64(header, fd, events, + f_header.event_types.size)) + goto out_errno; + event_count = f_header.event_types.size / sizeof(struct perf_trace_event_type); + } + + perf_header__process_sections(header, fd, perf_file_section__process); + + lseek(fd, header->data_offset, SEEK_SET); + + header->frozen = 1; + return 0; +out_errno: + return -errno; + +out_delete_evlist: + perf_evlist__delete(session->evlist); + session->evlist = NULL; + return -ENOMEM; +} + +int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, + perf_event__handler_t process, + struct perf_session *session) +{ + union perf_event *ev; + size_t size; + int err; + + size = sizeof(struct perf_event_attr); + size = ALIGN(size, sizeof(u64)); + size += sizeof(struct perf_event_header); + size += ids * sizeof(u64); + + ev = malloc(size); + + if (ev == NULL) + return -ENOMEM; + + ev->attr.attr = *attr; + memcpy(ev->attr.id, id, ids * sizeof(u64)); + + ev->attr.header.type = PERF_RECORD_HEADER_ATTR; + ev->attr.header.size = size; + + err = process(ev, NULL, session); + + free(ev); + + return err; +} + +int perf_session__synthesize_attrs(struct perf_session *session, + perf_event__handler_t process) +{ + struct perf_evsel *attr; + int err = 0; + + list_for_each_entry(attr, &session->evlist->entries, node) { + err = perf_event__synthesize_attr(&attr->attr, attr->ids, + attr->id, process, session); + if (err) { + pr_debug("failed to create perf header attribute\n"); + return err; + } + } + + return err; +} + +int perf_event__process_attr(union perf_event *event, + struct perf_session *session) +{ + unsigned int i, ids, n_ids; + struct perf_evsel *evsel; + + if (session->evlist == NULL) { + session->evlist = perf_evlist__new(NULL, NULL); + if (session->evlist == NULL) + return -ENOMEM; + } + + evsel = perf_evsel__new(&event->attr.attr, + session->evlist->nr_entries); + if (evsel == NULL) + return -ENOMEM; + + perf_evlist__add(session->evlist, evsel); + + ids = event->header.size; + ids -= (void *)&event->attr.id - (void *)event; + n_ids = ids / sizeof(u64); + /* + * We don't have the cpu and thread maps on the header, so + * for allocating the perf_sample_id table we fake 1 cpu and + * hattr->ids threads. + */ + if (perf_evsel__alloc_id(evsel, 1, n_ids)) + return -ENOMEM; + + for (i = 0; i < n_ids; i++) { + perf_evlist__id_add(session->evlist, evsel, 0, i, + event->attr.id[i]); + } + + perf_session__update_sample_type(session); + + return 0; +} + +int perf_event__synthesize_event_type(u64 event_id, char *name, + perf_event__handler_t process, + struct perf_session *session) +{ + union perf_event ev; + size_t size = 0; + int err = 0; + + memset(&ev, 0, sizeof(ev)); + + ev.event_type.event_type.event_id = event_id; + memset(ev.event_type.event_type.name, 0, MAX_EVENT_NAME); + strncpy(ev.event_type.event_type.name, name, MAX_EVENT_NAME - 1); + + ev.event_type.header.type = PERF_RECORD_HEADER_EVENT_TYPE; + size = strlen(name); + size = ALIGN(size, sizeof(u64)); + ev.event_type.header.size = sizeof(ev.event_type) - + (sizeof(ev.event_type.event_type.name) - size); + + err = process(&ev, NULL, session); + + return err; +} + +int perf_event__synthesize_event_types(perf_event__handler_t process, + struct perf_session *session) +{ + struct perf_trace_event_type *type; + int i, err = 0; + + for (i = 0; i < event_count; i++) { + type = &events[i]; + + err = perf_event__synthesize_event_type(type->event_id, + type->name, process, + session); + if (err) { + pr_debug("failed to create perf header event type\n"); + return err; + } + } + + return err; +} + +int perf_event__process_event_type(union perf_event *event, + struct perf_session *session __unused) +{ + if (perf_header__push_event(event->event_type.event_type.event_id, + event->event_type.event_type.name) < 0) + return -ENOMEM; + + return 0; +} + +int perf_event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, + perf_event__handler_t process, + struct perf_session *session __unused) +{ + union perf_event ev; + ssize_t size = 0, aligned_size = 0, padding; + int err __used = 0; + + memset(&ev, 0, sizeof(ev)); + + ev.tracing_data.header.type = PERF_RECORD_HEADER_TRACING_DATA; + size = read_tracing_data_size(fd, &evlist->entries); + if (size <= 0) + return size; + aligned_size = ALIGN(size, sizeof(u64)); + padding = aligned_size - size; + ev.tracing_data.header.size = sizeof(ev.tracing_data); + ev.tracing_data.size = aligned_size; + + process(&ev, NULL, session); + + err = read_tracing_data(fd, &evlist->entries); + write_padded(fd, NULL, 0, padding); + + return aligned_size; +} + +int perf_event__process_tracing_data(union perf_event *event, + struct perf_session *session) +{ + ssize_t size_read, padding, size = event->tracing_data.size; + off_t offset = lseek(session->fd, 0, SEEK_CUR); + char buf[BUFSIZ]; + + /* setup for reading amidst mmap */ + lseek(session->fd, offset + sizeof(struct tracing_data_event), + SEEK_SET); + + size_read = trace_report(session->fd, session->repipe); + + padding = ALIGN(size_read, sizeof(u64)) - size_read; + + if (read(session->fd, buf, padding) < 0) + die("reading input file"); + if (session->repipe) { + int retw = write(STDOUT_FILENO, buf, padding); + if (retw <= 0 || retw != padding) + die("repiping tracing data padding"); + } + + if (size_read + padding != size) + die("tracing data size mismatch"); + + return size_read + padding; +} + +int perf_event__synthesize_build_id(struct dso *pos, u16 misc, + perf_event__handler_t process, + struct machine *machine, + struct perf_session *session) +{ + union perf_event ev; + size_t len; + int err = 0; + + if (!pos->hit) + return err; + + memset(&ev, 0, sizeof(ev)); + + len = pos->long_name_len + 1; + len = ALIGN(len, NAME_ALIGN); + memcpy(&ev.build_id.build_id, pos->build_id, sizeof(pos->build_id)); + ev.build_id.header.type = PERF_RECORD_HEADER_BUILD_ID; + ev.build_id.header.misc = misc; + ev.build_id.pid = machine->pid; + ev.build_id.header.size = sizeof(ev.build_id) + len; + memcpy(&ev.build_id.filename, pos->long_name, pos->long_name_len); + + err = process(&ev, NULL, session); + + return err; +} + +int perf_event__process_build_id(union perf_event *event, + struct perf_session *session) +{ + __event_process_build_id(&event->build_id, + event->build_id.filename, + session); + return 0; +} + +void disable_buildid_cache(void) +{ + no_buildid_cache = true; +} diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h new file mode 100644 index 00000000..18862567 --- /dev/null +++ b/tools/perf/util/header.h @@ -0,0 +1,107 @@ +#ifndef __PERF_HEADER_H +#define __PERF_HEADER_H + +#include "../../../include/linux/perf_event.h" +#include <sys/types.h> +#include <stdbool.h> +#include "types.h" +#include "event.h" + +#include <linux/bitmap.h> + +enum { + HEADER_TRACE_INFO = 1, + HEADER_BUILD_ID, + HEADER_LAST_FEATURE, +}; + +#define HEADER_FEAT_BITS 256 + +struct perf_file_section { + u64 offset; + u64 size; +}; + +struct perf_file_header { + u64 magic; + u64 size; + u64 attr_size; + struct perf_file_section attrs; + struct perf_file_section data; + struct perf_file_section event_types; + DECLARE_BITMAP(adds_features, HEADER_FEAT_BITS); +}; + +struct perf_pipe_file_header { + u64 magic; + u64 size; +}; + +struct perf_header; + +int perf_file_header__read(struct perf_file_header *header, + struct perf_header *ph, int fd); + +struct perf_header { + int frozen; + bool needs_swap; + s64 attr_offset; + u64 data_offset; + u64 data_size; + u64 event_offset; + u64 event_size; + DECLARE_BITMAP(adds_features, HEADER_FEAT_BITS); +}; + +struct perf_evlist; + +int perf_session__read_header(struct perf_session *session, int fd); +int perf_session__write_header(struct perf_session *session, + struct perf_evlist *evlist, + int fd, bool at_exit); +int perf_header__write_pipe(int fd); + +int perf_header__push_event(u64 id, const char *name); +char *perf_header__find_event(u64 id); + +void perf_header__set_feat(struct perf_header *header, int feat); +void perf_header__clear_feat(struct perf_header *header, int feat); +bool perf_header__has_feat(const struct perf_header *header, int feat); + +int perf_header__process_sections(struct perf_header *header, int fd, + int (*process)(struct perf_file_section *section, + struct perf_header *ph, + int feat, int fd)); + +int build_id_cache__add_s(const char *sbuild_id, const char *debugdir, + const char *name, bool is_kallsyms); +int build_id_cache__remove_s(const char *sbuild_id, const char *debugdir); + +int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, + perf_event__handler_t process, + struct perf_session *session); +int perf_session__synthesize_attrs(struct perf_session *session, + perf_event__handler_t process); +int perf_event__process_attr(union perf_event *event, struct perf_session *session); + +int perf_event__synthesize_event_type(u64 event_id, char *name, + perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_event_types(perf_event__handler_t process, + struct perf_session *session); +int perf_event__process_event_type(union perf_event *event, + struct perf_session *session); + +int perf_event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, + perf_event__handler_t process, + struct perf_session *session); +int perf_event__process_tracing_data(union perf_event *event, + struct perf_session *session); + +int perf_event__synthesize_build_id(struct dso *pos, u16 misc, + perf_event__handler_t process, + struct machine *machine, + struct perf_session *session); +int perf_event__process_build_id(union perf_event *event, + struct perf_session *session); +#endif /* __PERF_HEADER_H */ diff --git a/tools/perf/util/help.c b/tools/perf/util/help.c new file mode 100644 index 00000000..6f2975a0 --- /dev/null +++ b/tools/perf/util/help.c @@ -0,0 +1,338 @@ +#include "cache.h" +#include "../builtin.h" +#include "exec_cmd.h" +#include "levenshtein.h" +#include "help.h" + +void add_cmdname(struct cmdnames *cmds, const char *name, size_t len) +{ + struct cmdname *ent = malloc(sizeof(*ent) + len + 1); + + ent->len = len; + memcpy(ent->name, name, len); + ent->name[len] = 0; + + ALLOC_GROW(cmds->names, cmds->cnt + 1, cmds->alloc); + cmds->names[cmds->cnt++] = ent; +} + +static void clean_cmdnames(struct cmdnames *cmds) +{ + unsigned int i; + + for (i = 0; i < cmds->cnt; ++i) + free(cmds->names[i]); + free(cmds->names); + cmds->cnt = 0; + cmds->alloc = 0; +} + +static int cmdname_compare(const void *a_, const void *b_) +{ + struct cmdname *a = *(struct cmdname **)a_; + struct cmdname *b = *(struct cmdname **)b_; + return strcmp(a->name, b->name); +} + +static void uniq(struct cmdnames *cmds) +{ + unsigned int i, j; + + if (!cmds->cnt) + return; + + for (i = j = 1; i < cmds->cnt; i++) + if (strcmp(cmds->names[i]->name, cmds->names[i-1]->name)) + cmds->names[j++] = cmds->names[i]; + + cmds->cnt = j; +} + +void exclude_cmds(struct cmdnames *cmds, struct cmdnames *excludes) +{ + size_t ci, cj, ei; + int cmp; + + ci = cj = ei = 0; + while (ci < cmds->cnt && ei < excludes->cnt) { + cmp = strcmp(cmds->names[ci]->name, excludes->names[ei]->name); + if (cmp < 0) + cmds->names[cj++] = cmds->names[ci++]; + else if (cmp == 0) + ci++, ei++; + else if (cmp > 0) + ei++; + } + + while (ci < cmds->cnt) + cmds->names[cj++] = cmds->names[ci++]; + + cmds->cnt = cj; +} + +static void pretty_print_string_list(struct cmdnames *cmds, int longest) +{ + int cols = 1, rows; + int space = longest + 1; /* min 1 SP between words */ + struct winsize win; + int max_cols; + int i, j; + + get_term_dimensions(&win); + max_cols = win.ws_col - 1; /* don't print *on* the edge */ + + if (space < max_cols) + cols = max_cols / space; + rows = (cmds->cnt + cols - 1) / cols; + + for (i = 0; i < rows; i++) { + printf(" "); + + for (j = 0; j < cols; j++) { + unsigned int n = j * rows + i; + unsigned int size = space; + + if (n >= cmds->cnt) + break; + if (j == cols-1 || n + rows >= cmds->cnt) + size = 1; + printf("%-*s", size, cmds->names[n]->name); + } + putchar('\n'); + } +} + +static int is_executable(const char *name) +{ + struct stat st; + + if (stat(name, &st) || /* stat, not lstat */ + !S_ISREG(st.st_mode)) + return 0; + + return st.st_mode & S_IXUSR; +} + +static void list_commands_in_dir(struct cmdnames *cmds, + const char *path, + const char *prefix) +{ + int prefix_len; + DIR *dir = opendir(path); + struct dirent *de; + struct strbuf buf = STRBUF_INIT; + int len; + + if (!dir) + return; + if (!prefix) + prefix = "perf-"; + prefix_len = strlen(prefix); + + strbuf_addf(&buf, "%s/", path); + len = buf.len; + + while ((de = readdir(dir)) != NULL) { + int entlen; + + if (prefixcmp(de->d_name, prefix)) + continue; + + strbuf_setlen(&buf, len); + strbuf_addstr(&buf, de->d_name); + if (!is_executable(buf.buf)) + continue; + + entlen = strlen(de->d_name) - prefix_len; + if (has_extension(de->d_name, ".exe")) + entlen -= 4; + + add_cmdname(cmds, de->d_name + prefix_len, entlen); + } + closedir(dir); + strbuf_release(&buf); +} + +void load_command_list(const char *prefix, + struct cmdnames *main_cmds, + struct cmdnames *other_cmds) +{ + const char *env_path = getenv("PATH"); + const char *exec_path = perf_exec_path(); + + if (exec_path) { + list_commands_in_dir(main_cmds, exec_path, prefix); + qsort(main_cmds->names, main_cmds->cnt, + sizeof(*main_cmds->names), cmdname_compare); + uniq(main_cmds); + } + + if (env_path) { + char *paths, *path, *colon; + path = paths = strdup(env_path); + while (1) { + if ((colon = strchr(path, PATH_SEP))) + *colon = 0; + if (!exec_path || strcmp(path, exec_path)) + list_commands_in_dir(other_cmds, path, prefix); + + if (!colon) + break; + path = colon + 1; + } + free(paths); + + qsort(other_cmds->names, other_cmds->cnt, + sizeof(*other_cmds->names), cmdname_compare); + uniq(other_cmds); + } + exclude_cmds(other_cmds, main_cmds); +} + +void list_commands(const char *title, struct cmdnames *main_cmds, + struct cmdnames *other_cmds) +{ + unsigned int i, longest = 0; + + for (i = 0; i < main_cmds->cnt; i++) + if (longest < main_cmds->names[i]->len) + longest = main_cmds->names[i]->len; + for (i = 0; i < other_cmds->cnt; i++) + if (longest < other_cmds->names[i]->len) + longest = other_cmds->names[i]->len; + + if (main_cmds->cnt) { + const char *exec_path = perf_exec_path(); + printf("available %s in '%s'\n", title, exec_path); + printf("----------------"); + mput_char('-', strlen(title) + strlen(exec_path)); + putchar('\n'); + pretty_print_string_list(main_cmds, longest); + putchar('\n'); + } + + if (other_cmds->cnt) { + printf("%s available from elsewhere on your $PATH\n", title); + printf("---------------------------------------"); + mput_char('-', strlen(title)); + putchar('\n'); + pretty_print_string_list(other_cmds, longest); + putchar('\n'); + } +} + +int is_in_cmdlist(struct cmdnames *c, const char *s) +{ + unsigned int i; + + for (i = 0; i < c->cnt; i++) + if (!strcmp(s, c->names[i]->name)) + return 1; + return 0; +} + +static int autocorrect; +static struct cmdnames aliases; + +static int perf_unknown_cmd_config(const char *var, const char *value, void *cb) +{ + if (!strcmp(var, "help.autocorrect")) + autocorrect = perf_config_int(var,value); + /* Also use aliases for command lookup */ + if (!prefixcmp(var, "alias.")) + add_cmdname(&aliases, var + 6, strlen(var + 6)); + + return perf_default_config(var, value, cb); +} + +static int levenshtein_compare(const void *p1, const void *p2) +{ + const struct cmdname *const *c1 = p1, *const *c2 = p2; + const char *s1 = (*c1)->name, *s2 = (*c2)->name; + int l1 = (*c1)->len; + int l2 = (*c2)->len; + return l1 != l2 ? l1 - l2 : strcmp(s1, s2); +} + +static void add_cmd_list(struct cmdnames *cmds, struct cmdnames *old) +{ + unsigned int i; + + ALLOC_GROW(cmds->names, cmds->cnt + old->cnt, cmds->alloc); + + for (i = 0; i < old->cnt; i++) + cmds->names[cmds->cnt++] = old->names[i]; + free(old->names); + old->cnt = 0; + old->names = NULL; +} + +const char *help_unknown_cmd(const char *cmd) +{ + unsigned int i, n = 0, best_similarity = 0; + struct cmdnames main_cmds, other_cmds; + + memset(&main_cmds, 0, sizeof(main_cmds)); + memset(&other_cmds, 0, sizeof(main_cmds)); + memset(&aliases, 0, sizeof(aliases)); + + perf_config(perf_unknown_cmd_config, NULL); + + load_command_list("perf-", &main_cmds, &other_cmds); + + add_cmd_list(&main_cmds, &aliases); + add_cmd_list(&main_cmds, &other_cmds); + qsort(main_cmds.names, main_cmds.cnt, + sizeof(main_cmds.names), cmdname_compare); + uniq(&main_cmds); + + if (main_cmds.cnt) { + /* This reuses cmdname->len for similarity index */ + for (i = 0; i < main_cmds.cnt; ++i) + main_cmds.names[i]->len = + levenshtein(cmd, main_cmds.names[i]->name, 0, 2, 1, 4); + + qsort(main_cmds.names, main_cmds.cnt, + sizeof(*main_cmds.names), levenshtein_compare); + + best_similarity = main_cmds.names[0]->len; + n = 1; + while (n < main_cmds.cnt && best_similarity == main_cmds.names[n]->len) + ++n; + } + + if (autocorrect && n == 1) { + const char *assumed = main_cmds.names[0]->name; + + main_cmds.names[0] = NULL; + clean_cmdnames(&main_cmds); + fprintf(stderr, "WARNING: You called a perf program named '%s', " + "which does not exist.\n" + "Continuing under the assumption that you meant '%s'\n", + cmd, assumed); + if (autocorrect > 0) { + fprintf(stderr, "in %0.1f seconds automatically...\n", + (float)autocorrect/10.0); + poll(NULL, 0, autocorrect * 100); + } + return assumed; + } + + fprintf(stderr, "perf: '%s' is not a perf-command. See 'perf --help'.\n", cmd); + + if (main_cmds.cnt && best_similarity < 6) { + fprintf(stderr, "\nDid you mean %s?\n", + n < 2 ? "this": "one of these"); + + for (i = 0; i < n; i++) + fprintf(stderr, "\t%s\n", main_cmds.names[i]->name); + } + + exit(1); +} + +int cmd_version(int argc __used, const char **argv __used, const char *prefix __used) +{ + printf("perf version %s\n", perf_version_string); + return 0; +} diff --git a/tools/perf/util/help.h b/tools/perf/util/help.h new file mode 100644 index 00000000..7f5c6ded --- /dev/null +++ b/tools/perf/util/help.h @@ -0,0 +1,29 @@ +#ifndef __PERF_HELP_H +#define __PERF_HELP_H + +struct cmdnames { + size_t alloc; + size_t cnt; + struct cmdname { + size_t len; /* also used for similarity index in help.c */ + char name[FLEX_ARRAY]; + } **names; +}; + +static inline void mput_char(char c, unsigned int num) +{ + while(num--) + putchar(c); +} + +void load_command_list(const char *prefix, + struct cmdnames *main_cmds, + struct cmdnames *other_cmds); +void add_cmdname(struct cmdnames *cmds, const char *name, size_t len); +/* Here we require that excludes is a sorted list. */ +void exclude_cmds(struct cmdnames *cmds, struct cmdnames *excludes); +int is_in_cmdlist(struct cmdnames *c, const char *s); +void list_commands(const char *title, struct cmdnames *main_cmds, + struct cmdnames *other_cmds); + +#endif /* __PERF_HELP_H */ diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c new file mode 100644 index 00000000..fb695991 --- /dev/null +++ b/tools/perf/util/hist.c @@ -0,0 +1,1013 @@ +#include "annotate.h" +#include "util.h" +#include "build-id.h" +#include "hist.h" +#include "session.h" +#include "sort.h" +#include <math.h> + +enum hist_filter { + HIST_FILTER__DSO, + HIST_FILTER__THREAD, + HIST_FILTER__PARENT, +}; + +struct callchain_param callchain_param = { + .mode = CHAIN_GRAPH_REL, + .min_percent = 0.5 +}; + +u16 hists__col_len(struct hists *self, enum hist_column col) +{ + return self->col_len[col]; +} + +void hists__set_col_len(struct hists *self, enum hist_column col, u16 len) +{ + self->col_len[col] = len; +} + +bool hists__new_col_len(struct hists *self, enum hist_column col, u16 len) +{ + if (len > hists__col_len(self, col)) { + hists__set_col_len(self, col, len); + return true; + } + return false; +} + +static void hists__reset_col_len(struct hists *self) +{ + enum hist_column col; + + for (col = 0; col < HISTC_NR_COLS; ++col) + hists__set_col_len(self, col, 0); +} + +static void hists__calc_col_len(struct hists *self, struct hist_entry *h) +{ + u16 len; + + if (h->ms.sym) + hists__new_col_len(self, HISTC_SYMBOL, h->ms.sym->namelen); + else { + const unsigned int unresolved_col_width = BITS_PER_LONG / 4; + + if (hists__col_len(self, HISTC_DSO) < unresolved_col_width && + !symbol_conf.col_width_list_str && !symbol_conf.field_sep && + !symbol_conf.dso_list) + hists__set_col_len(self, HISTC_DSO, + unresolved_col_width); + } + + len = thread__comm_len(h->thread); + if (hists__new_col_len(self, HISTC_COMM, len)) + hists__set_col_len(self, HISTC_THREAD, len + 6); + + if (h->ms.map) { + len = dso__name_len(h->ms.map->dso); + hists__new_col_len(self, HISTC_DSO, len); + } +} + +static void hist_entry__add_cpumode_period(struct hist_entry *self, + unsigned int cpumode, u64 period) +{ + switch (cpumode) { + case PERF_RECORD_MISC_KERNEL: + self->period_sys += period; + break; + case PERF_RECORD_MISC_USER: + self->period_us += period; + break; + case PERF_RECORD_MISC_GUEST_KERNEL: + self->period_guest_sys += period; + break; + case PERF_RECORD_MISC_GUEST_USER: + self->period_guest_us += period; + break; + default: + break; + } +} + +/* + * histogram, sorted on item, collects periods + */ + +static struct hist_entry *hist_entry__new(struct hist_entry *template) +{ + size_t callchain_size = symbol_conf.use_callchain ? sizeof(struct callchain_root) : 0; + struct hist_entry *self = malloc(sizeof(*self) + callchain_size); + + if (self != NULL) { + *self = *template; + self->nr_events = 1; + if (self->ms.map) + self->ms.map->referenced = true; + if (symbol_conf.use_callchain) + callchain_init(self->callchain); + } + + return self; +} + +static void hists__inc_nr_entries(struct hists *self, struct hist_entry *h) +{ + if (!h->filtered) { + hists__calc_col_len(self, h); + ++self->nr_entries; + } +} + +static u8 symbol__parent_filter(const struct symbol *parent) +{ + if (symbol_conf.exclude_other && parent == NULL) + return 1 << HIST_FILTER__PARENT; + return 0; +} + +struct hist_entry *__hists__add_entry(struct hists *self, + struct addr_location *al, + struct symbol *sym_parent, u64 period) +{ + struct rb_node **p = &self->entries.rb_node; + struct rb_node *parent = NULL; + struct hist_entry *he; + struct hist_entry entry = { + .thread = al->thread, + .ms = { + .map = al->map, + .sym = al->sym, + }, + .cpu = al->cpu, + .ip = al->addr, + .level = al->level, + .period = period, + .parent = sym_parent, + .filtered = symbol__parent_filter(sym_parent), + }; + int cmp; + + while (*p != NULL) { + parent = *p; + he = rb_entry(parent, struct hist_entry, rb_node); + + cmp = hist_entry__cmp(&entry, he); + + if (!cmp) { + he->period += period; + ++he->nr_events; + + /* If the map of an existing hist_entry has + * become out-of-date due to an exec() or + * similar, update it. Otherwise we will + * mis-adjust symbol addresses when computing + * the history counter to increment. + */ + if (he->ms.map != entry.ms.map) { + he->ms.map = entry.ms.map; + if (he->ms.map) + he->ms.map->referenced = true; + } + goto out; + } + + if (cmp < 0) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + he = hist_entry__new(&entry); + if (!he) + return NULL; + rb_link_node(&he->rb_node, parent, p); + rb_insert_color(&he->rb_node, &self->entries); + hists__inc_nr_entries(self, he); +out: + hist_entry__add_cpumode_period(he, al->cpumode, period); + return he; +} + +int64_t +hist_entry__cmp(struct hist_entry *left, struct hist_entry *right) +{ + struct sort_entry *se; + int64_t cmp = 0; + + list_for_each_entry(se, &hist_entry__sort_list, list) { + cmp = se->se_cmp(left, right); + if (cmp) + break; + } + + return cmp; +} + +int64_t +hist_entry__collapse(struct hist_entry *left, struct hist_entry *right) +{ + struct sort_entry *se; + int64_t cmp = 0; + + list_for_each_entry(se, &hist_entry__sort_list, list) { + int64_t (*f)(struct hist_entry *, struct hist_entry *); + + f = se->se_collapse ?: se->se_cmp; + + cmp = f(left, right); + if (cmp) + break; + } + + return cmp; +} + +void hist_entry__free(struct hist_entry *he) +{ + free(he); +} + +/* + * collapse the histogram + */ + +static bool hists__collapse_insert_entry(struct hists *self, + struct rb_root *root, + struct hist_entry *he) +{ + struct rb_node **p = &root->rb_node; + struct rb_node *parent = NULL; + struct hist_entry *iter; + int64_t cmp; + + while (*p != NULL) { + parent = *p; + iter = rb_entry(parent, struct hist_entry, rb_node); + + cmp = hist_entry__collapse(iter, he); + + if (!cmp) { + iter->period += he->period; + if (symbol_conf.use_callchain) { + callchain_cursor_reset(&self->callchain_cursor); + callchain_merge(&self->callchain_cursor, iter->callchain, + he->callchain); + } + hist_entry__free(he); + return false; + } + + if (cmp < 0) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&he->rb_node, parent, p); + rb_insert_color(&he->rb_node, root); + return true; +} + +void hists__collapse_resort(struct hists *self) +{ + struct rb_root tmp; + struct rb_node *next; + struct hist_entry *n; + + if (!sort__need_collapse) + return; + + tmp = RB_ROOT; + next = rb_first(&self->entries); + self->nr_entries = 0; + hists__reset_col_len(self); + + while (next) { + n = rb_entry(next, struct hist_entry, rb_node); + next = rb_next(&n->rb_node); + + rb_erase(&n->rb_node, &self->entries); + if (hists__collapse_insert_entry(self, &tmp, n)) + hists__inc_nr_entries(self, n); + } + + self->entries = tmp; +} + +/* + * reverse the map, sort on period. + */ + +static void __hists__insert_output_entry(struct rb_root *entries, + struct hist_entry *he, + u64 min_callchain_hits) +{ + struct rb_node **p = &entries->rb_node; + struct rb_node *parent = NULL; + struct hist_entry *iter; + + if (symbol_conf.use_callchain) + callchain_param.sort(&he->sorted_chain, he->callchain, + min_callchain_hits, &callchain_param); + + while (*p != NULL) { + parent = *p; + iter = rb_entry(parent, struct hist_entry, rb_node); + + if (he->period > iter->period) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&he->rb_node, parent, p); + rb_insert_color(&he->rb_node, entries); +} + +void hists__output_resort(struct hists *self) +{ + struct rb_root tmp; + struct rb_node *next; + struct hist_entry *n; + u64 min_callchain_hits; + + min_callchain_hits = self->stats.total_period * (callchain_param.min_percent / 100); + + tmp = RB_ROOT; + next = rb_first(&self->entries); + + self->nr_entries = 0; + hists__reset_col_len(self); + + while (next) { + n = rb_entry(next, struct hist_entry, rb_node); + next = rb_next(&n->rb_node); + + rb_erase(&n->rb_node, &self->entries); + __hists__insert_output_entry(&tmp, n, min_callchain_hits); + hists__inc_nr_entries(self, n); + } + + self->entries = tmp; +} + +static size_t callchain__fprintf_left_margin(FILE *fp, int left_margin) +{ + int i; + int ret = fprintf(fp, " "); + + for (i = 0; i < left_margin; i++) + ret += fprintf(fp, " "); + + return ret; +} + +static size_t ipchain__fprintf_graph_line(FILE *fp, int depth, int depth_mask, + int left_margin) +{ + int i; + size_t ret = callchain__fprintf_left_margin(fp, left_margin); + + for (i = 0; i < depth; i++) + if (depth_mask & (1 << i)) + ret += fprintf(fp, "| "); + else + ret += fprintf(fp, " "); + + ret += fprintf(fp, "\n"); + + return ret; +} + +static size_t ipchain__fprintf_graph(FILE *fp, struct callchain_list *chain, + int depth, int depth_mask, int period, + u64 total_samples, u64 hits, + int left_margin) +{ + int i; + size_t ret = 0; + + ret += callchain__fprintf_left_margin(fp, left_margin); + for (i = 0; i < depth; i++) { + if (depth_mask & (1 << i)) + ret += fprintf(fp, "|"); + else + ret += fprintf(fp, " "); + if (!period && i == depth - 1) { + double percent; + + percent = hits * 100.0 / total_samples; + ret += percent_color_fprintf(fp, "--%2.2f%%-- ", percent); + } else + ret += fprintf(fp, "%s", " "); + } + if (chain->ms.sym) + ret += fprintf(fp, "%s\n", chain->ms.sym->name); + else + ret += fprintf(fp, "%p\n", (void *)(long)chain->ip); + + return ret; +} + +static struct symbol *rem_sq_bracket; +static struct callchain_list rem_hits; + +static void init_rem_hits(void) +{ + rem_sq_bracket = malloc(sizeof(*rem_sq_bracket) + 6); + if (!rem_sq_bracket) { + fprintf(stderr, "Not enough memory to display remaining hits\n"); + return; + } + + strcpy(rem_sq_bracket->name, "[...]"); + rem_hits.ms.sym = rem_sq_bracket; +} + +static size_t __callchain__fprintf_graph(FILE *fp, struct callchain_node *self, + u64 total_samples, int depth, + int depth_mask, int left_margin) +{ + struct rb_node *node, *next; + struct callchain_node *child; + struct callchain_list *chain; + int new_depth_mask = depth_mask; + u64 new_total; + u64 remaining; + size_t ret = 0; + int i; + uint entries_printed = 0; + + if (callchain_param.mode == CHAIN_GRAPH_REL) + new_total = self->children_hit; + else + new_total = total_samples; + + remaining = new_total; + + node = rb_first(&self->rb_root); + while (node) { + u64 cumul; + + child = rb_entry(node, struct callchain_node, rb_node); + cumul = callchain_cumul_hits(child); + remaining -= cumul; + + /* + * The depth mask manages the output of pipes that show + * the depth. We don't want to keep the pipes of the current + * level for the last child of this depth. + * Except if we have remaining filtered hits. They will + * supersede the last child + */ + next = rb_next(node); + if (!next && (callchain_param.mode != CHAIN_GRAPH_REL || !remaining)) + new_depth_mask &= ~(1 << (depth - 1)); + + /* + * But we keep the older depth mask for the line separator + * to keep the level link until we reach the last child + */ + ret += ipchain__fprintf_graph_line(fp, depth, depth_mask, + left_margin); + i = 0; + list_for_each_entry(chain, &child->val, list) { + ret += ipchain__fprintf_graph(fp, chain, depth, + new_depth_mask, i++, + new_total, + cumul, + left_margin); + } + ret += __callchain__fprintf_graph(fp, child, new_total, + depth + 1, + new_depth_mask | (1 << depth), + left_margin); + node = next; + if (++entries_printed == callchain_param.print_limit) + break; + } + + if (callchain_param.mode == CHAIN_GRAPH_REL && + remaining && remaining != new_total) { + + if (!rem_sq_bracket) + return ret; + + new_depth_mask &= ~(1 << (depth - 1)); + + ret += ipchain__fprintf_graph(fp, &rem_hits, depth, + new_depth_mask, 0, new_total, + remaining, left_margin); + } + + return ret; +} + +static size_t callchain__fprintf_graph(FILE *fp, struct callchain_node *self, + u64 total_samples, int left_margin) +{ + struct callchain_list *chain; + bool printed = false; + int i = 0; + int ret = 0; + u32 entries_printed = 0; + + list_for_each_entry(chain, &self->val, list) { + if (!i++ && sort__first_dimension == SORT_SYM) + continue; + + if (!printed) { + ret += callchain__fprintf_left_margin(fp, left_margin); + ret += fprintf(fp, "|\n"); + ret += callchain__fprintf_left_margin(fp, left_margin); + ret += fprintf(fp, "---"); + + left_margin += 3; + printed = true; + } else + ret += callchain__fprintf_left_margin(fp, left_margin); + + if (chain->ms.sym) + ret += fprintf(fp, " %s\n", chain->ms.sym->name); + else + ret += fprintf(fp, " %p\n", (void *)(long)chain->ip); + + if (++entries_printed == callchain_param.print_limit) + break; + } + + ret += __callchain__fprintf_graph(fp, self, total_samples, 1, 1, left_margin); + + return ret; +} + +static size_t callchain__fprintf_flat(FILE *fp, struct callchain_node *self, + u64 total_samples) +{ + struct callchain_list *chain; + size_t ret = 0; + + if (!self) + return 0; + + ret += callchain__fprintf_flat(fp, self->parent, total_samples); + + + list_for_each_entry(chain, &self->val, list) { + if (chain->ip >= PERF_CONTEXT_MAX) + continue; + if (chain->ms.sym) + ret += fprintf(fp, " %s\n", chain->ms.sym->name); + else + ret += fprintf(fp, " %p\n", + (void *)(long)chain->ip); + } + + return ret; +} + +static size_t hist_entry_callchain__fprintf(FILE *fp, struct hist_entry *self, + u64 total_samples, int left_margin) +{ + struct rb_node *rb_node; + struct callchain_node *chain; + size_t ret = 0; + u32 entries_printed = 0; + + rb_node = rb_first(&self->sorted_chain); + while (rb_node) { + double percent; + + chain = rb_entry(rb_node, struct callchain_node, rb_node); + percent = chain->hit * 100.0 / total_samples; + switch (callchain_param.mode) { + case CHAIN_FLAT: + ret += percent_color_fprintf(fp, " %6.2f%%\n", + percent); + ret += callchain__fprintf_flat(fp, chain, total_samples); + break; + case CHAIN_GRAPH_ABS: /* Falldown */ + case CHAIN_GRAPH_REL: + ret += callchain__fprintf_graph(fp, chain, total_samples, + left_margin); + case CHAIN_NONE: + default: + break; + } + ret += fprintf(fp, "\n"); + if (++entries_printed == callchain_param.print_limit) + break; + rb_node = rb_next(rb_node); + } + + return ret; +} + +int hist_entry__snprintf(struct hist_entry *self, char *s, size_t size, + struct hists *hists, struct hists *pair_hists, + bool show_displacement, long displacement, + bool color, u64 session_total) +{ + struct sort_entry *se; + u64 period, total, period_sys, period_us, period_guest_sys, period_guest_us; + u64 nr_events; + const char *sep = symbol_conf.field_sep; + int ret; + + if (symbol_conf.exclude_other && !self->parent) + return 0; + + if (pair_hists) { + period = self->pair ? self->pair->period : 0; + nr_events = self->pair ? self->pair->nr_events : 0; + total = pair_hists->stats.total_period; + period_sys = self->pair ? self->pair->period_sys : 0; + period_us = self->pair ? self->pair->period_us : 0; + period_guest_sys = self->pair ? self->pair->period_guest_sys : 0; + period_guest_us = self->pair ? self->pair->period_guest_us : 0; + } else { + period = self->period; + nr_events = self->nr_events; + total = session_total; + period_sys = self->period_sys; + period_us = self->period_us; + period_guest_sys = self->period_guest_sys; + period_guest_us = self->period_guest_us; + } + + if (total) { + if (color) + ret = percent_color_snprintf(s, size, + sep ? "%.2f" : " %6.2f%%", + (period * 100.0) / total); + else + ret = snprintf(s, size, sep ? "%.2f" : " %6.2f%%", + (period * 100.0) / total); + if (symbol_conf.show_cpu_utilization) { + ret += percent_color_snprintf(s + ret, size - ret, + sep ? "%.2f" : " %6.2f%%", + (period_sys * 100.0) / total); + ret += percent_color_snprintf(s + ret, size - ret, + sep ? "%.2f" : " %6.2f%%", + (period_us * 100.0) / total); + if (perf_guest) { + ret += percent_color_snprintf(s + ret, + size - ret, + sep ? "%.2f" : " %6.2f%%", + (period_guest_sys * 100.0) / + total); + ret += percent_color_snprintf(s + ret, + size - ret, + sep ? "%.2f" : " %6.2f%%", + (period_guest_us * 100.0) / + total); + } + } + } else + ret = snprintf(s, size, sep ? "%" PRIu64 : "%12" PRIu64 " ", period); + + if (symbol_conf.show_nr_samples) { + if (sep) + ret += snprintf(s + ret, size - ret, "%c%" PRIu64, *sep, nr_events); + else + ret += snprintf(s + ret, size - ret, "%11" PRIu64, nr_events); + } + + if (pair_hists) { + char bf[32]; + double old_percent = 0, new_percent = 0, diff; + + if (total > 0) + old_percent = (period * 100.0) / total; + if (session_total > 0) + new_percent = (self->period * 100.0) / session_total; + + diff = new_percent - old_percent; + + if (fabs(diff) >= 0.01) + snprintf(bf, sizeof(bf), "%+4.2F%%", diff); + else + snprintf(bf, sizeof(bf), " "); + + if (sep) + ret += snprintf(s + ret, size - ret, "%c%s", *sep, bf); + else + ret += snprintf(s + ret, size - ret, "%11.11s", bf); + + if (show_displacement) { + if (displacement) + snprintf(bf, sizeof(bf), "%+4ld", displacement); + else + snprintf(bf, sizeof(bf), " "); + + if (sep) + ret += snprintf(s + ret, size - ret, "%c%s", *sep, bf); + else + ret += snprintf(s + ret, size - ret, "%6.6s", bf); + } + } + + list_for_each_entry(se, &hist_entry__sort_list, list) { + if (se->elide) + continue; + + ret += snprintf(s + ret, size - ret, "%s", sep ?: " "); + ret += se->se_snprintf(self, s + ret, size - ret, + hists__col_len(hists, se->se_width_idx)); + } + + return ret; +} + +int hist_entry__fprintf(struct hist_entry *self, struct hists *hists, + struct hists *pair_hists, bool show_displacement, + long displacement, FILE *fp, u64 session_total) +{ + char bf[512]; + hist_entry__snprintf(self, bf, sizeof(bf), hists, pair_hists, + show_displacement, displacement, + true, session_total); + return fprintf(fp, "%s\n", bf); +} + +static size_t hist_entry__fprintf_callchain(struct hist_entry *self, + struct hists *hists, FILE *fp, + u64 session_total) +{ + int left_margin = 0; + + if (sort__first_dimension == SORT_COMM) { + struct sort_entry *se = list_first_entry(&hist_entry__sort_list, + typeof(*se), list); + left_margin = hists__col_len(hists, se->se_width_idx); + left_margin -= thread__comm_len(self->thread); + } + + return hist_entry_callchain__fprintf(fp, self, session_total, + left_margin); +} + +size_t hists__fprintf(struct hists *self, struct hists *pair, + bool show_displacement, FILE *fp) +{ + struct sort_entry *se; + struct rb_node *nd; + size_t ret = 0; + unsigned long position = 1; + long displacement = 0; + unsigned int width; + const char *sep = symbol_conf.field_sep; + const char *col_width = symbol_conf.col_width_list_str; + + init_rem_hits(); + + fprintf(fp, "# %s", pair ? "Baseline" : "Overhead"); + + if (symbol_conf.show_nr_samples) { + if (sep) + fprintf(fp, "%cSamples", *sep); + else + fputs(" Samples ", fp); + } + + if (symbol_conf.show_cpu_utilization) { + if (sep) { + ret += fprintf(fp, "%csys", *sep); + ret += fprintf(fp, "%cus", *sep); + if (perf_guest) { + ret += fprintf(fp, "%cguest sys", *sep); + ret += fprintf(fp, "%cguest us", *sep); + } + } else { + ret += fprintf(fp, " sys "); + ret += fprintf(fp, " us "); + if (perf_guest) { + ret += fprintf(fp, " guest sys "); + ret += fprintf(fp, " guest us "); + } + } + } + + if (pair) { + if (sep) + ret += fprintf(fp, "%cDelta", *sep); + else + ret += fprintf(fp, " Delta "); + + if (show_displacement) { + if (sep) + ret += fprintf(fp, "%cDisplacement", *sep); + else + ret += fprintf(fp, " Displ"); + } + } + + list_for_each_entry(se, &hist_entry__sort_list, list) { + if (se->elide) + continue; + if (sep) { + fprintf(fp, "%c%s", *sep, se->se_header); + continue; + } + width = strlen(se->se_header); + if (symbol_conf.col_width_list_str) { + if (col_width) { + hists__set_col_len(self, se->se_width_idx, + atoi(col_width)); + col_width = strchr(col_width, ','); + if (col_width) + ++col_width; + } + } + if (!hists__new_col_len(self, se->se_width_idx, width)) + width = hists__col_len(self, se->se_width_idx); + fprintf(fp, " %*s", width, se->se_header); + } + fprintf(fp, "\n"); + + if (sep) + goto print_entries; + + fprintf(fp, "# ........"); + if (symbol_conf.show_nr_samples) + fprintf(fp, " .........."); + if (pair) { + fprintf(fp, " .........."); + if (show_displacement) + fprintf(fp, " ....."); + } + list_for_each_entry(se, &hist_entry__sort_list, list) { + unsigned int i; + + if (se->elide) + continue; + + fprintf(fp, " "); + width = hists__col_len(self, se->se_width_idx); + if (width == 0) + width = strlen(se->se_header); + for (i = 0; i < width; i++) + fprintf(fp, "."); + } + + fprintf(fp, "\n#\n"); + +print_entries: + for (nd = rb_first(&self->entries); nd; nd = rb_next(nd)) { + struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node); + + if (show_displacement) { + if (h->pair != NULL) + displacement = ((long)h->pair->position - + (long)position); + else + displacement = 0; + ++position; + } + ret += hist_entry__fprintf(h, self, pair, show_displacement, + displacement, fp, self->stats.total_period); + + if (symbol_conf.use_callchain) + ret += hist_entry__fprintf_callchain(h, self, fp, + self->stats.total_period); + if (h->ms.map == NULL && verbose > 1) { + __map_groups__fprintf_maps(&h->thread->mg, + MAP__FUNCTION, verbose, fp); + fprintf(fp, "%.10s end\n", graph_dotted_line); + } + } + + free(rem_sq_bracket); + + return ret; +} + +/* + * See hists__fprintf to match the column widths + */ +unsigned int hists__sort_list_width(struct hists *self) +{ + struct sort_entry *se; + int ret = 9; /* total % */ + + if (symbol_conf.show_cpu_utilization) { + ret += 7; /* count_sys % */ + ret += 6; /* count_us % */ + if (perf_guest) { + ret += 13; /* count_guest_sys % */ + ret += 12; /* count_guest_us % */ + } + } + + if (symbol_conf.show_nr_samples) + ret += 11; + + list_for_each_entry(se, &hist_entry__sort_list, list) + if (!se->elide) + ret += 2 + hists__col_len(self, se->se_width_idx); + + if (verbose) /* Addr + origin */ + ret += 3 + BITS_PER_LONG / 4; + + return ret; +} + +static void hists__remove_entry_filter(struct hists *self, struct hist_entry *h, + enum hist_filter filter) +{ + h->filtered &= ~(1 << filter); + if (h->filtered) + return; + + ++self->nr_entries; + if (h->ms.unfolded) + self->nr_entries += h->nr_rows; + h->row_offset = 0; + self->stats.total_period += h->period; + self->stats.nr_events[PERF_RECORD_SAMPLE] += h->nr_events; + + hists__calc_col_len(self, h); +} + +void hists__filter_by_dso(struct hists *self, const struct dso *dso) +{ + struct rb_node *nd; + + self->nr_entries = self->stats.total_period = 0; + self->stats.nr_events[PERF_RECORD_SAMPLE] = 0; + hists__reset_col_len(self); + + for (nd = rb_first(&self->entries); nd; nd = rb_next(nd)) { + struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node); + + if (symbol_conf.exclude_other && !h->parent) + continue; + + if (dso != NULL && (h->ms.map == NULL || h->ms.map->dso != dso)) { + h->filtered |= (1 << HIST_FILTER__DSO); + continue; + } + + hists__remove_entry_filter(self, h, HIST_FILTER__DSO); + } +} + +void hists__filter_by_thread(struct hists *self, const struct thread *thread) +{ + struct rb_node *nd; + + self->nr_entries = self->stats.total_period = 0; + self->stats.nr_events[PERF_RECORD_SAMPLE] = 0; + hists__reset_col_len(self); + + for (nd = rb_first(&self->entries); nd; nd = rb_next(nd)) { + struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node); + + if (thread != NULL && h->thread != thread) { + h->filtered |= (1 << HIST_FILTER__THREAD); + continue; + } + + hists__remove_entry_filter(self, h, HIST_FILTER__THREAD); + } +} + +int hist_entry__inc_addr_samples(struct hist_entry *he, int evidx, u64 ip) +{ + return symbol__inc_addr_samples(he->ms.sym, he->ms.map, evidx, ip); +} + +int hist_entry__annotate(struct hist_entry *he, size_t privsize) +{ + return symbol__annotate(he->ms.sym, he->ms.map, privsize); +} + +void hists__inc_nr_events(struct hists *self, u32 type) +{ + ++self->stats.nr_events[0]; + ++self->stats.nr_events[type]; +} + +size_t hists__fprintf_nr_events(struct hists *self, FILE *fp) +{ + int i; + size_t ret = 0; + + for (i = 0; i < PERF_RECORD_HEADER_MAX; ++i) { + const char *name; + + if (self->stats.nr_events[i] == 0) + continue; + + name = perf_event__name(i); + if (!strcmp(name, "UNKNOWN")) + continue; + + ret += fprintf(fp, "%16s events: %10d\n", name, + self->stats.nr_events[i]); + } + + return ret; +} diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h new file mode 100644 index 00000000..3beb97c4 --- /dev/null +++ b/tools/perf/util/hist.h @@ -0,0 +1,117 @@ +#ifndef __PERF_HIST_H +#define __PERF_HIST_H + +#include <linux/types.h> +#include "callchain.h" + +extern struct callchain_param callchain_param; + +struct hist_entry; +struct addr_location; +struct symbol; + +/* + * The kernel collects the number of events it couldn't send in a stretch and + * when possible sends this number in a PERF_RECORD_LOST event. The number of + * such "chunks" of lost events is stored in .nr_events[PERF_EVENT_LOST] while + * total_lost tells exactly how many events the kernel in fact lost, i.e. it is + * the sum of all struct lost_event.lost fields reported. + * + * The total_period is needed because by default auto-freq is used, so + * multipling nr_events[PERF_EVENT_SAMPLE] by a frequency isn't possible to get + * the total number of low level events, it is necessary to to sum all struct + * sample_event.period and stash the result in total_period. + */ +struct events_stats { + u64 total_period; + u64 total_lost; + u64 total_invalid_chains; + u32 nr_events[PERF_RECORD_HEADER_MAX]; + u32 nr_unknown_events; + u32 nr_invalid_chains; + u32 nr_unknown_id; +}; + +enum hist_column { + HISTC_SYMBOL, + HISTC_DSO, + HISTC_THREAD, + HISTC_COMM, + HISTC_PARENT, + HISTC_CPU, + HISTC_NR_COLS, /* Last entry */ +}; + +struct hists { + struct rb_root entries; + u64 nr_entries; + struct events_stats stats; + u64 event_stream; + u16 col_len[HISTC_NR_COLS]; + /* Best would be to reuse the session callchain cursor */ + struct callchain_cursor callchain_cursor; +}; + +struct hist_entry *__hists__add_entry(struct hists *self, + struct addr_location *al, + struct symbol *parent, u64 period); +extern int64_t hist_entry__cmp(struct hist_entry *, struct hist_entry *); +extern int64_t hist_entry__collapse(struct hist_entry *, struct hist_entry *); +int hist_entry__fprintf(struct hist_entry *self, struct hists *hists, + struct hists *pair_hists, bool show_displacement, + long displacement, FILE *fp, u64 total); +int hist_entry__snprintf(struct hist_entry *self, char *bf, size_t size, + struct hists *hists, struct hists *pair_hists, + bool show_displacement, long displacement, + bool color, u64 total); +void hist_entry__free(struct hist_entry *); + +void hists__output_resort(struct hists *self); +void hists__collapse_resort(struct hists *self); + +void hists__inc_nr_events(struct hists *self, u32 type); +size_t hists__fprintf_nr_events(struct hists *self, FILE *fp); + +size_t hists__fprintf(struct hists *self, struct hists *pair, + bool show_displacement, FILE *fp); + +int hist_entry__inc_addr_samples(struct hist_entry *self, int evidx, u64 addr); +int hist_entry__annotate(struct hist_entry *self, size_t privsize); + +void hists__filter_by_dso(struct hists *self, const struct dso *dso); +void hists__filter_by_thread(struct hists *self, const struct thread *thread); + +u16 hists__col_len(struct hists *self, enum hist_column col); +void hists__set_col_len(struct hists *self, enum hist_column col, u16 len); +bool hists__new_col_len(struct hists *self, enum hist_column col, u16 len); + +struct perf_evlist; + +#ifdef NO_NEWT_SUPPORT +static inline +int perf_evlist__tui_browse_hists(struct perf_evlist *evlist __used, + const char *help __used) +{ + return 0; +} + +static inline int hist_entry__tui_annotate(struct hist_entry *self __used, + int evidx __used) +{ + return 0; +} +#define KEY_LEFT -1 +#define KEY_RIGHT -2 +#else +#include <newt.h> +int hist_entry__tui_annotate(struct hist_entry *self, int evidx); + +#define KEY_LEFT NEWT_KEY_LEFT +#define KEY_RIGHT NEWT_KEY_RIGHT + +int perf_evlist__tui_browse_hists(struct perf_evlist *evlist, const char *help); +#endif + +unsigned int hists__sort_list_width(struct hists *self); + +#endif /* __PERF_HIST_H */ diff --git a/tools/perf/util/hweight.c b/tools/perf/util/hweight.c new file mode 100644 index 00000000..5c1d0d09 --- /dev/null +++ b/tools/perf/util/hweight.c @@ -0,0 +1,31 @@ +#include <linux/bitops.h> + +/** + * hweightN - returns the hamming weight of a N-bit word + * @x: the word to weigh + * + * The Hamming Weight of a number is the total number of bits set in it. + */ + +unsigned int hweight32(unsigned int w) +{ + unsigned int res = w - ((w >> 1) & 0x55555555); + res = (res & 0x33333333) + ((res >> 2) & 0x33333333); + res = (res + (res >> 4)) & 0x0F0F0F0F; + res = res + (res >> 8); + return (res + (res >> 16)) & 0x000000FF; +} + +unsigned long hweight64(__u64 w) +{ +#if BITS_PER_LONG == 32 + return hweight32((unsigned int)(w >> 32)) + hweight32((unsigned int)w); +#elif BITS_PER_LONG == 64 + __u64 res = w - ((w >> 1) & 0x5555555555555555ul); + res = (res & 0x3333333333333333ul) + ((res >> 2) & 0x3333333333333333ul); + res = (res + (res >> 4)) & 0x0F0F0F0F0F0F0F0Ful; + res = res + (res >> 8); + res = res + (res >> 16); + return (res + (res >> 32)) & 0x00000000000000FFul; +#endif +} diff --git a/tools/perf/util/include/asm/alternative-asm.h b/tools/perf/util/include/asm/alternative-asm.h new file mode 100644 index 00000000..6789d788 --- /dev/null +++ b/tools/perf/util/include/asm/alternative-asm.h @@ -0,0 +1,8 @@ +#ifndef _PERF_ASM_ALTERNATIVE_ASM_H +#define _PERF_ASM_ALTERNATIVE_ASM_H + +/* Just disable it so we can build arch/x86/lib/memcpy_64.S for perf bench: */ + +#define altinstruction_entry # + +#endif diff --git a/tools/perf/util/include/asm/asm-offsets.h b/tools/perf/util/include/asm/asm-offsets.h new file mode 100644 index 00000000..ed538942 --- /dev/null +++ b/tools/perf/util/include/asm/asm-offsets.h @@ -0,0 +1 @@ +/* stub */ diff --git a/tools/perf/util/include/asm/bug.h b/tools/perf/util/include/asm/bug.h new file mode 100644 index 00000000..7fcc6810 --- /dev/null +++ b/tools/perf/util/include/asm/bug.h @@ -0,0 +1,22 @@ +#ifndef _PERF_ASM_GENERIC_BUG_H +#define _PERF_ASM_GENERIC_BUG_H + +#define __WARN_printf(arg...) do { fprintf(stderr, arg); } while (0) + +#define WARN(condition, format...) ({ \ + int __ret_warn_on = !!(condition); \ + if (unlikely(__ret_warn_on)) \ + __WARN_printf(format); \ + unlikely(__ret_warn_on); \ +}) + +#define WARN_ONCE(condition, format...) ({ \ + static int __warned; \ + int __ret_warn_once = !!(condition); \ + \ + if (unlikely(__ret_warn_once)) \ + if (WARN(!__warned, format)) \ + __warned = 1; \ + unlikely(__ret_warn_once); \ +}) +#endif diff --git a/tools/perf/util/include/asm/byteorder.h b/tools/perf/util/include/asm/byteorder.h new file mode 100644 index 00000000..b722abe3 --- /dev/null +++ b/tools/perf/util/include/asm/byteorder.h @@ -0,0 +1,2 @@ +#include <asm/types.h> +#include "../../../../include/linux/swab.h" diff --git a/tools/perf/util/include/asm/cpufeature.h b/tools/perf/util/include/asm/cpufeature.h new file mode 100644 index 00000000..acffd5e4 --- /dev/null +++ b/tools/perf/util/include/asm/cpufeature.h @@ -0,0 +1,9 @@ + +#ifndef PERF_CPUFEATURE_H +#define PERF_CPUFEATURE_H + +/* cpufeature.h ... dummy header file for including arch/x86/lib/memcpy_64.S */ + +#define X86_FEATURE_REP_GOOD 0 + +#endif /* PERF_CPUFEATURE_H */ diff --git a/tools/perf/util/include/asm/dwarf2.h b/tools/perf/util/include/asm/dwarf2.h new file mode 100644 index 00000000..bb4198e7 --- /dev/null +++ b/tools/perf/util/include/asm/dwarf2.h @@ -0,0 +1,11 @@ + +#ifndef PERF_DWARF2_H +#define PERF_DWARF2_H + +/* dwarf2.h ... dummy header file for including arch/x86/lib/memcpy_64.S */ + +#define CFI_STARTPROC +#define CFI_ENDPROC + +#endif /* PERF_DWARF2_H */ + diff --git a/tools/perf/util/include/asm/hweight.h b/tools/perf/util/include/asm/hweight.h new file mode 100644 index 00000000..36cf26d4 --- /dev/null +++ b/tools/perf/util/include/asm/hweight.h @@ -0,0 +1,8 @@ +#ifndef PERF_HWEIGHT_H +#define PERF_HWEIGHT_H + +#include <linux/types.h> +unsigned int hweight32(unsigned int w); +unsigned long hweight64(__u64 w); + +#endif /* PERF_HWEIGHT_H */ diff --git a/tools/perf/util/include/asm/swab.h b/tools/perf/util/include/asm/swab.h new file mode 100644 index 00000000..ed538942 --- /dev/null +++ b/tools/perf/util/include/asm/swab.h @@ -0,0 +1 @@ +/* stub */ diff --git a/tools/perf/util/include/asm/system.h b/tools/perf/util/include/asm/system.h new file mode 100644 index 00000000..710cecca --- /dev/null +++ b/tools/perf/util/include/asm/system.h @@ -0,0 +1 @@ +/* Empty */ diff --git a/tools/perf/util/include/asm/uaccess.h b/tools/perf/util/include/asm/uaccess.h new file mode 100644 index 00000000..d0f72b8f --- /dev/null +++ b/tools/perf/util/include/asm/uaccess.h @@ -0,0 +1,14 @@ +#ifndef _PERF_ASM_UACCESS_H_ +#define _PERF_ASM_UACCESS_H_ + +#define __get_user(src, dest) \ +({ \ + (src) = *dest; \ + 0; \ +}) + +#define get_user __get_user + +#define access_ok(type, addr, size) 1 + +#endif diff --git a/tools/perf/util/include/dwarf-regs.h b/tools/perf/util/include/dwarf-regs.h new file mode 100644 index 00000000..cf6727e9 --- /dev/null +++ b/tools/perf/util/include/dwarf-regs.h @@ -0,0 +1,8 @@ +#ifndef _PERF_DWARF_REGS_H_ +#define _PERF_DWARF_REGS_H_ + +#ifdef DWARF_SUPPORT +const char *get_arch_regstr(unsigned int n); +#endif + +#endif diff --git a/tools/perf/util/include/linux/bitmap.h b/tools/perf/util/include/linux/bitmap.h new file mode 100644 index 00000000..eda4416e --- /dev/null +++ b/tools/perf/util/include/linux/bitmap.h @@ -0,0 +1,35 @@ +#ifndef _PERF_BITOPS_H +#define _PERF_BITOPS_H + +#include <string.h> +#include <linux/bitops.h> + +int __bitmap_weight(const unsigned long *bitmap, int bits); + +#define BITMAP_LAST_WORD_MASK(nbits) \ +( \ + ((nbits) % BITS_PER_LONG) ? \ + (1UL<<((nbits) % BITS_PER_LONG))-1 : ~0UL \ +) + +#define small_const_nbits(nbits) \ + (__builtin_constant_p(nbits) && (nbits) <= BITS_PER_LONG) + +static inline void bitmap_zero(unsigned long *dst, int nbits) +{ + if (small_const_nbits(nbits)) + *dst = 0UL; + else { + int len = BITS_TO_LONGS(nbits) * sizeof(unsigned long); + memset(dst, 0, len); + } +} + +static inline int bitmap_weight(const unsigned long *src, int nbits) +{ + if (small_const_nbits(nbits)) + return hweight_long(*src & BITMAP_LAST_WORD_MASK(nbits)); + return __bitmap_weight(src, nbits); +} + +#endif /* _PERF_BITOPS_H */ diff --git a/tools/perf/util/include/linux/bitops.h b/tools/perf/util/include/linux/bitops.h new file mode 100644 index 00000000..305c8484 --- /dev/null +++ b/tools/perf/util/include/linux/bitops.h @@ -0,0 +1,33 @@ +#ifndef _PERF_LINUX_BITOPS_H_ +#define _PERF_LINUX_BITOPS_H_ + +#include <linux/kernel.h> +#include <linux/compiler.h> +#include <asm/hweight.h> + +#define BITS_PER_LONG __WORDSIZE +#define BITS_PER_BYTE 8 +#define BITS_TO_LONGS(nr) DIV_ROUND_UP(nr, BITS_PER_BYTE * sizeof(long)) + +static inline void set_bit(int nr, unsigned long *addr) +{ + addr[nr / BITS_PER_LONG] |= 1UL << (nr % BITS_PER_LONG); +} + +static inline void clear_bit(int nr, unsigned long *addr) +{ + addr[nr / BITS_PER_LONG] &= ~(1UL << (nr % BITS_PER_LONG)); +} + +static __always_inline int test_bit(unsigned int nr, const unsigned long *addr) +{ + return ((1UL << (nr % BITS_PER_LONG)) & + (((unsigned long *)addr)[nr / BITS_PER_LONG])) != 0; +} + +static inline unsigned long hweight_long(unsigned long w) +{ + return sizeof(w) == 4 ? hweight32(w) : hweight64(w); +} + +#endif diff --git a/tools/perf/util/include/linux/compiler.h b/tools/perf/util/include/linux/compiler.h new file mode 100644 index 00000000..547628e9 --- /dev/null +++ b/tools/perf/util/include/linux/compiler.h @@ -0,0 +1,14 @@ +#ifndef _PERF_LINUX_COMPILER_H_ +#define _PERF_LINUX_COMPILER_H_ + +#ifndef __always_inline +#define __always_inline inline +#endif +#define __user +#ifndef __attribute_const__ +#define __attribute_const__ +#endif + +#define __used __attribute__((__unused__)) + +#endif diff --git a/tools/perf/util/include/linux/const.h b/tools/perf/util/include/linux/const.h new file mode 100644 index 00000000..1b476c9a --- /dev/null +++ b/tools/perf/util/include/linux/const.h @@ -0,0 +1 @@ +#include "../../../../include/linux/const.h" diff --git a/tools/perf/util/include/linux/ctype.h b/tools/perf/util/include/linux/ctype.h new file mode 100644 index 00000000..a53d4ee1 --- /dev/null +++ b/tools/perf/util/include/linux/ctype.h @@ -0,0 +1 @@ +#include "../util.h" diff --git a/tools/perf/util/include/linux/hash.h b/tools/perf/util/include/linux/hash.h new file mode 100644 index 00000000..201f5739 --- /dev/null +++ b/tools/perf/util/include/linux/hash.h @@ -0,0 +1,5 @@ +#include "../../../../include/linux/hash.h" + +#ifndef PERF_HASH_H +#define PERF_HASH_H +#endif diff --git a/tools/perf/util/include/linux/kernel.h b/tools/perf/util/include/linux/kernel.h new file mode 100644 index 00000000..1eb804fd --- /dev/null +++ b/tools/perf/util/include/linux/kernel.h @@ -0,0 +1,111 @@ +#ifndef PERF_LINUX_KERNEL_H_ +#define PERF_LINUX_KERNEL_H_ + +#include <stdarg.h> +#include <stdio.h> +#include <stdlib.h> +#include <assert.h> + +#define DIV_ROUND_UP(n,d) (((n) + (d) - 1) / (d)) + +#define ALIGN(x,a) __ALIGN_MASK(x,(typeof(x))(a)-1) +#define __ALIGN_MASK(x,mask) (((x)+(mask))&~(mask)) + +#ifndef offsetof +#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER) +#endif + +#ifndef container_of +/** + * container_of - cast a member of a structure out to the containing structure + * @ptr: the pointer to the member. + * @type: the type of the container struct this is embedded in. + * @member: the name of the member within the struct. + * + */ +#define container_of(ptr, type, member) ({ \ + const typeof(((type *)0)->member) * __mptr = (ptr); \ + (type *)((char *)__mptr - offsetof(type, member)); }) +#endif + +#define BUILD_BUG_ON_ZERO(e) (sizeof(struct { int:-!!(e); })) + +#ifndef max +#define max(x, y) ({ \ + typeof(x) _max1 = (x); \ + typeof(y) _max2 = (y); \ + (void) (&_max1 == &_max2); \ + _max1 > _max2 ? _max1 : _max2; }) +#endif + +#ifndef min +#define min(x, y) ({ \ + typeof(x) _min1 = (x); \ + typeof(y) _min2 = (y); \ + (void) (&_min1 == &_min2); \ + _min1 < _min2 ? _min1 : _min2; }) +#endif + +#ifndef BUG_ON +#define BUG_ON(cond) assert(!(cond)) +#endif + +/* + * Both need more care to handle endianness + * (Don't use bitmap_copy_le() for now) + */ +#define cpu_to_le64(x) (x) +#define cpu_to_le32(x) (x) + +static inline int +vscnprintf(char *buf, size_t size, const char *fmt, va_list args) +{ + int i; + ssize_t ssize = size; + + i = vsnprintf(buf, size, fmt, args); + + return (i >= ssize) ? (ssize - 1) : i; +} + +static inline int scnprintf(char * buf, size_t size, const char * fmt, ...) +{ + va_list args; + ssize_t ssize = size; + int i; + + va_start(args, fmt); + i = vsnprintf(buf, size, fmt, args); + va_end(args); + + return (i >= ssize) ? (ssize - 1) : i; +} + +static inline unsigned long +simple_strtoul(const char *nptr, char **endptr, int base) +{ + return strtoul(nptr, endptr, base); +} + +int eprintf(int level, + const char *fmt, ...) __attribute__((format(printf, 2, 3))); + +#ifndef pr_fmt +#define pr_fmt(fmt) fmt +#endif + +#define pr_err(fmt, ...) \ + eprintf(0, pr_fmt(fmt), ##__VA_ARGS__) +#define pr_warning(fmt, ...) \ + eprintf(0, pr_fmt(fmt), ##__VA_ARGS__) +#define pr_info(fmt, ...) \ + eprintf(0, pr_fmt(fmt), ##__VA_ARGS__) +#define pr_debug(fmt, ...) \ + eprintf(1, pr_fmt(fmt), ##__VA_ARGS__) +#define pr_debugN(n, fmt, ...) \ + eprintf(n, pr_fmt(fmt), ##__VA_ARGS__) +#define pr_debug2(fmt, ...) pr_debugN(2, pr_fmt(fmt), ##__VA_ARGS__) +#define pr_debug3(fmt, ...) pr_debugN(3, pr_fmt(fmt), ##__VA_ARGS__) +#define pr_debug4(fmt, ...) pr_debugN(4, pr_fmt(fmt), ##__VA_ARGS__) + +#endif diff --git a/tools/perf/util/include/linux/linkage.h b/tools/perf/util/include/linux/linkage.h new file mode 100644 index 00000000..06387cff --- /dev/null +++ b/tools/perf/util/include/linux/linkage.h @@ -0,0 +1,13 @@ + +#ifndef PERF_LINUX_LINKAGE_H_ +#define PERF_LINUX_LINKAGE_H_ + +/* linkage.h ... for including arch/x86/lib/memcpy_64.S */ + +#define ENTRY(name) \ + .globl name; \ + name: + +#define ENDPROC(name) + +#endif /* PERF_LINUX_LINKAGE_H_ */ diff --git a/tools/perf/util/include/linux/list.h b/tools/perf/util/include/linux/list.h new file mode 100644 index 00000000..1d928a0c --- /dev/null +++ b/tools/perf/util/include/linux/list.h @@ -0,0 +1,29 @@ +#include <linux/kernel.h> +#include <linux/prefetch.h> + +#include "../../../../include/linux/list.h" + +#ifndef PERF_LIST_H +#define PERF_LIST_H +/** + * list_del_range - deletes range of entries from list. + * @begin: first element in the range to delete from the list. + * @end: last element in the range to delete from the list. + * Note: list_empty on the range of entries does not return true after this, + * the entries is in an undefined state. + */ +static inline void list_del_range(struct list_head *begin, + struct list_head *end) +{ + begin->prev->next = end->next; + end->next->prev = begin->prev; +} + +/** + * list_for_each_from - iterate over a list from one of its nodes + * @pos: the &struct list_head to use as a loop cursor, from where to start + * @head: the head for your list. + */ +#define list_for_each_from(pos, head) \ + for (; pos != (head); pos = pos->next) +#endif diff --git a/tools/perf/util/include/linux/module.h b/tools/perf/util/include/linux/module.h new file mode 100644 index 00000000..b43e2dc2 --- /dev/null +++ b/tools/perf/util/include/linux/module.h @@ -0,0 +1,6 @@ +#ifndef PERF_LINUX_MODULE_H +#define PERF_LINUX_MODULE_H + +#define EXPORT_SYMBOL(name) + +#endif diff --git a/tools/perf/util/include/linux/poison.h b/tools/perf/util/include/linux/poison.h new file mode 100644 index 00000000..fef6dbc9 --- /dev/null +++ b/tools/perf/util/include/linux/poison.h @@ -0,0 +1 @@ +#include "../../../../include/linux/poison.h" diff --git a/tools/perf/util/include/linux/prefetch.h b/tools/perf/util/include/linux/prefetch.h new file mode 100644 index 00000000..7841e485 --- /dev/null +++ b/tools/perf/util/include/linux/prefetch.h @@ -0,0 +1,6 @@ +#ifndef PERF_LINUX_PREFETCH_H +#define PERF_LINUX_PREFETCH_H + +static inline void prefetch(void *a __attribute__((unused))) { } + +#endif diff --git a/tools/perf/util/include/linux/rbtree.h b/tools/perf/util/include/linux/rbtree.h new file mode 100644 index 00000000..7a243a14 --- /dev/null +++ b/tools/perf/util/include/linux/rbtree.h @@ -0,0 +1 @@ +#include "../../../../include/linux/rbtree.h" diff --git a/tools/perf/util/include/linux/string.h b/tools/perf/util/include/linux/string.h new file mode 100644 index 00000000..3b2f5900 --- /dev/null +++ b/tools/perf/util/include/linux/string.h @@ -0,0 +1 @@ +#include <string.h> diff --git a/tools/perf/util/include/linux/types.h b/tools/perf/util/include/linux/types.h new file mode 100644 index 00000000..12de3b81 --- /dev/null +++ b/tools/perf/util/include/linux/types.h @@ -0,0 +1,21 @@ +#ifndef _PERF_LINUX_TYPES_H_ +#define _PERF_LINUX_TYPES_H_ + +#include <asm/types.h> + +#define DECLARE_BITMAP(name,bits) \ + unsigned long name[BITS_TO_LONGS(bits)] + +struct list_head { + struct list_head *next, *prev; +}; + +struct hlist_head { + struct hlist_node *first; +}; + +struct hlist_node { + struct hlist_node *next, **pprev; +}; + +#endif diff --git a/tools/perf/util/levenshtein.c b/tools/perf/util/levenshtein.c new file mode 100644 index 00000000..e521d151 --- /dev/null +++ b/tools/perf/util/levenshtein.c @@ -0,0 +1,84 @@ +#include "cache.h" +#include "levenshtein.h" + +/* + * This function implements the Damerau-Levenshtein algorithm to + * calculate a distance between strings. + * + * Basically, it says how many letters need to be swapped, substituted, + * deleted from, or added to string1, at least, to get string2. + * + * The idea is to build a distance matrix for the substrings of both + * strings. To avoid a large space complexity, only the last three rows + * are kept in memory (if swaps had the same or higher cost as one deletion + * plus one insertion, only two rows would be needed). + * + * At any stage, "i + 1" denotes the length of the current substring of + * string1 that the distance is calculated for. + * + * row2 holds the current row, row1 the previous row (i.e. for the substring + * of string1 of length "i"), and row0 the row before that. + * + * In other words, at the start of the big loop, row2[j + 1] contains the + * Damerau-Levenshtein distance between the substring of string1 of length + * "i" and the substring of string2 of length "j + 1". + * + * All the big loop does is determine the partial minimum-cost paths. + * + * It does so by calculating the costs of the path ending in characters + * i (in string1) and j (in string2), respectively, given that the last + * operation is a substition, a swap, a deletion, or an insertion. + * + * This implementation allows the costs to be weighted: + * + * - w (as in "sWap") + * - s (as in "Substitution") + * - a (for insertion, AKA "Add") + * - d (as in "Deletion") + * + * Note that this algorithm calculates a distance _iff_ d == a. + */ +int levenshtein(const char *string1, const char *string2, + int w, int s, int a, int d) +{ + int len1 = strlen(string1), len2 = strlen(string2); + int *row0 = malloc(sizeof(int) * (len2 + 1)); + int *row1 = malloc(sizeof(int) * (len2 + 1)); + int *row2 = malloc(sizeof(int) * (len2 + 1)); + int i, j; + + for (j = 0; j <= len2; j++) + row1[j] = j * a; + for (i = 0; i < len1; i++) { + int *dummy; + + row2[0] = (i + 1) * d; + for (j = 0; j < len2; j++) { + /* substitution */ + row2[j + 1] = row1[j] + s * (string1[i] != string2[j]); + /* swap */ + if (i > 0 && j > 0 && string1[i - 1] == string2[j] && + string1[i] == string2[j - 1] && + row2[j + 1] > row0[j - 1] + w) + row2[j + 1] = row0[j - 1] + w; + /* deletion */ + if (row2[j + 1] > row1[j + 1] + d) + row2[j + 1] = row1[j + 1] + d; + /* insertion */ + if (row2[j + 1] > row2[j] + a) + row2[j + 1] = row2[j] + a; + } + + dummy = row0; + row0 = row1; + row1 = row2; + row2 = dummy; + } + + i = row1[len2]; + free(row0); + free(row1); + free(row2); + + return i; +} diff --git a/tools/perf/util/levenshtein.h b/tools/perf/util/levenshtein.h new file mode 100644 index 00000000..b0fcb6d8 --- /dev/null +++ b/tools/perf/util/levenshtein.h @@ -0,0 +1,8 @@ +#ifndef __PERF_LEVENSHTEIN_H +#define __PERF_LEVENSHTEIN_H + +int levenshtein(const char *string1, const char *string2, + int swap_penalty, int substition_penalty, + int insertion_penalty, int deletion_penalty); + +#endif /* __PERF_LEVENSHTEIN_H */ diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c new file mode 100644 index 00000000..a16ecab5 --- /dev/null +++ b/tools/perf/util/map.c @@ -0,0 +1,683 @@ +#include "symbol.h" +#include <errno.h> +#include <inttypes.h> +#include <limits.h> +#include <stdlib.h> +#include <string.h> +#include <stdio.h> +#include <unistd.h> +#include "map.h" + +const char *map_type__name[MAP__NR_TYPES] = { + [MAP__FUNCTION] = "Functions", + [MAP__VARIABLE] = "Variables", +}; + +static inline int is_anon_memory(const char *filename) +{ + return strcmp(filename, "//anon") == 0; +} + +void map__init(struct map *self, enum map_type type, + u64 start, u64 end, u64 pgoff, struct dso *dso) +{ + self->type = type; + self->start = start; + self->end = end; + self->pgoff = pgoff; + self->dso = dso; + self->map_ip = map__map_ip; + self->unmap_ip = map__unmap_ip; + RB_CLEAR_NODE(&self->rb_node); + self->groups = NULL; + self->referenced = false; +} + +struct map *map__new(struct list_head *dsos__list, u64 start, u64 len, + u64 pgoff, u32 pid, char *filename, + enum map_type type) +{ + struct map *self = malloc(sizeof(*self)); + + if (self != NULL) { + char newfilename[PATH_MAX]; + struct dso *dso; + int anon; + + anon = is_anon_memory(filename); + + if (anon) { + snprintf(newfilename, sizeof(newfilename), "/tmp/perf-%d.map", pid); + filename = newfilename; + } + + dso = __dsos__findnew(dsos__list, filename); + if (dso == NULL) + goto out_delete; + + map__init(self, type, start, start + len, pgoff, dso); + + if (anon) { +set_identity: + self->map_ip = self->unmap_ip = identity__map_ip; + } else if (strcmp(filename, "[vdso]") == 0) { + dso__set_loaded(dso, self->type); + goto set_identity; + } + } + return self; +out_delete: + free(self); + return NULL; +} + +void map__delete(struct map *self) +{ + free(self); +} + +void map__fixup_start(struct map *self) +{ + struct rb_root *symbols = &self->dso->symbols[self->type]; + struct rb_node *nd = rb_first(symbols); + if (nd != NULL) { + struct symbol *sym = rb_entry(nd, struct symbol, rb_node); + self->start = sym->start; + } +} + +void map__fixup_end(struct map *self) +{ + struct rb_root *symbols = &self->dso->symbols[self->type]; + struct rb_node *nd = rb_last(symbols); + if (nd != NULL) { + struct symbol *sym = rb_entry(nd, struct symbol, rb_node); + self->end = sym->end; + } +} + +#define DSO__DELETED "(deleted)" + +int map__load(struct map *self, symbol_filter_t filter) +{ + const char *name = self->dso->long_name; + int nr; + + if (dso__loaded(self->dso, self->type)) + return 0; + + nr = dso__load(self->dso, self, filter); + if (nr < 0) { + if (self->dso->has_build_id) { + char sbuild_id[BUILD_ID_SIZE * 2 + 1]; + + build_id__sprintf(self->dso->build_id, + sizeof(self->dso->build_id), + sbuild_id); + pr_warning("%s with build id %s not found", + name, sbuild_id); + } else + pr_warning("Failed to open %s", name); + + pr_warning(", continuing without symbols\n"); + return -1; + } else if (nr == 0) { + const size_t len = strlen(name); + const size_t real_len = len - sizeof(DSO__DELETED); + + if (len > sizeof(DSO__DELETED) && + strcmp(name + real_len + 1, DSO__DELETED) == 0) { + pr_warning("%.*s was updated, restart the long " + "running apps that use it!\n", + (int)real_len, name); + } else { + pr_warning("no symbols found in %s, maybe install " + "a debug package?\n", name); + } + + return -1; + } + /* + * Only applies to the kernel, as its symtabs aren't relative like the + * module ones. + */ + if (self->dso->kernel) + map__reloc_vmlinux(self); + + return 0; +} + +struct symbol *map__find_symbol(struct map *self, u64 addr, + symbol_filter_t filter) +{ + if (map__load(self, filter) < 0) + return NULL; + + return dso__find_symbol(self->dso, self->type, addr); +} + +struct symbol *map__find_symbol_by_name(struct map *self, const char *name, + symbol_filter_t filter) +{ + if (map__load(self, filter) < 0) + return NULL; + + if (!dso__sorted_by_name(self->dso, self->type)) + dso__sort_by_name(self->dso, self->type); + + return dso__find_symbol_by_name(self->dso, self->type, name); +} + +struct map *map__clone(struct map *self) +{ + struct map *map = malloc(sizeof(*self)); + + if (!map) + return NULL; + + memcpy(map, self, sizeof(*self)); + + return map; +} + +int map__overlap(struct map *l, struct map *r) +{ + if (l->start > r->start) { + struct map *t = l; + l = r; + r = t; + } + + if (l->end > r->start) + return 1; + + return 0; +} + +size_t map__fprintf(struct map *self, FILE *fp) +{ + return fprintf(fp, " %" PRIx64 "-%" PRIx64 " %" PRIx64 " %s\n", + self->start, self->end, self->pgoff, self->dso->name); +} + +/* + * objdump wants/reports absolute IPs for ET_EXEC, and RIPs for ET_DYN. + * map->dso->adjust_symbols==1 for ET_EXEC-like cases. + */ +u64 map__rip_2objdump(struct map *map, u64 rip) +{ + u64 addr = map->dso->adjust_symbols ? + map->unmap_ip(map, rip) : /* RIP -> IP */ + rip; + return addr; +} + +u64 map__objdump_2ip(struct map *map, u64 addr) +{ + u64 ip = map->dso->adjust_symbols ? + addr : + map->unmap_ip(map, addr); /* RIP -> IP */ + return ip; +} + +void map_groups__init(struct map_groups *self) +{ + int i; + for (i = 0; i < MAP__NR_TYPES; ++i) { + self->maps[i] = RB_ROOT; + INIT_LIST_HEAD(&self->removed_maps[i]); + } + self->machine = NULL; +} + +static void maps__delete(struct rb_root *self) +{ + struct rb_node *next = rb_first(self); + + while (next) { + struct map *pos = rb_entry(next, struct map, rb_node); + + next = rb_next(&pos->rb_node); + rb_erase(&pos->rb_node, self); + map__delete(pos); + } +} + +static void maps__delete_removed(struct list_head *self) +{ + struct map *pos, *n; + + list_for_each_entry_safe(pos, n, self, node) { + list_del(&pos->node); + map__delete(pos); + } +} + +void map_groups__exit(struct map_groups *self) +{ + int i; + + for (i = 0; i < MAP__NR_TYPES; ++i) { + maps__delete(&self->maps[i]); + maps__delete_removed(&self->removed_maps[i]); + } +} + +void map_groups__flush(struct map_groups *self) +{ + int type; + + for (type = 0; type < MAP__NR_TYPES; type++) { + struct rb_root *root = &self->maps[type]; + struct rb_node *next = rb_first(root); + + while (next) { + struct map *pos = rb_entry(next, struct map, rb_node); + next = rb_next(&pos->rb_node); + rb_erase(&pos->rb_node, root); + /* + * We may have references to this map, for + * instance in some hist_entry instances, so + * just move them to a separate list. + */ + list_add_tail(&pos->node, &self->removed_maps[pos->type]); + } + } +} + +struct symbol *map_groups__find_symbol(struct map_groups *self, + enum map_type type, u64 addr, + struct map **mapp, + symbol_filter_t filter) +{ + struct map *map = map_groups__find(self, type, addr); + + if (map != NULL) { + if (mapp != NULL) + *mapp = map; + return map__find_symbol(map, map->map_ip(map, addr), filter); + } + + return NULL; +} + +struct symbol *map_groups__find_symbol_by_name(struct map_groups *self, + enum map_type type, + const char *name, + struct map **mapp, + symbol_filter_t filter) +{ + struct rb_node *nd; + + for (nd = rb_first(&self->maps[type]); nd; nd = rb_next(nd)) { + struct map *pos = rb_entry(nd, struct map, rb_node); + struct symbol *sym = map__find_symbol_by_name(pos, name, filter); + + if (sym == NULL) + continue; + if (mapp != NULL) + *mapp = pos; + return sym; + } + + return NULL; +} + +size_t __map_groups__fprintf_maps(struct map_groups *self, + enum map_type type, int verbose, FILE *fp) +{ + size_t printed = fprintf(fp, "%s:\n", map_type__name[type]); + struct rb_node *nd; + + for (nd = rb_first(&self->maps[type]); nd; nd = rb_next(nd)) { + struct map *pos = rb_entry(nd, struct map, rb_node); + printed += fprintf(fp, "Map:"); + printed += map__fprintf(pos, fp); + if (verbose > 2) { + printed += dso__fprintf(pos->dso, type, fp); + printed += fprintf(fp, "--\n"); + } + } + + return printed; +} + +size_t map_groups__fprintf_maps(struct map_groups *self, int verbose, FILE *fp) +{ + size_t printed = 0, i; + for (i = 0; i < MAP__NR_TYPES; ++i) + printed += __map_groups__fprintf_maps(self, i, verbose, fp); + return printed; +} + +static size_t __map_groups__fprintf_removed_maps(struct map_groups *self, + enum map_type type, + int verbose, FILE *fp) +{ + struct map *pos; + size_t printed = 0; + + list_for_each_entry(pos, &self->removed_maps[type], node) { + printed += fprintf(fp, "Map:"); + printed += map__fprintf(pos, fp); + if (verbose > 1) { + printed += dso__fprintf(pos->dso, type, fp); + printed += fprintf(fp, "--\n"); + } + } + return printed; +} + +static size_t map_groups__fprintf_removed_maps(struct map_groups *self, + int verbose, FILE *fp) +{ + size_t printed = 0, i; + for (i = 0; i < MAP__NR_TYPES; ++i) + printed += __map_groups__fprintf_removed_maps(self, i, verbose, fp); + return printed; +} + +size_t map_groups__fprintf(struct map_groups *self, int verbose, FILE *fp) +{ + size_t printed = map_groups__fprintf_maps(self, verbose, fp); + printed += fprintf(fp, "Removed maps:\n"); + return printed + map_groups__fprintf_removed_maps(self, verbose, fp); +} + +int map_groups__fixup_overlappings(struct map_groups *self, struct map *map, + int verbose, FILE *fp) +{ + struct rb_root *root = &self->maps[map->type]; + struct rb_node *next = rb_first(root); + int err = 0; + + while (next) { + struct map *pos = rb_entry(next, struct map, rb_node); + next = rb_next(&pos->rb_node); + + if (!map__overlap(pos, map)) + continue; + + if (verbose >= 2) { + fputs("overlapping maps:\n", fp); + map__fprintf(map, fp); + map__fprintf(pos, fp); + } + + rb_erase(&pos->rb_node, root); + /* + * Now check if we need to create new maps for areas not + * overlapped by the new map: + */ + if (map->start > pos->start) { + struct map *before = map__clone(pos); + + if (before == NULL) { + err = -ENOMEM; + goto move_map; + } + + before->end = map->start - 1; + map_groups__insert(self, before); + if (verbose >= 2) + map__fprintf(before, fp); + } + + if (map->end < pos->end) { + struct map *after = map__clone(pos); + + if (after == NULL) { + err = -ENOMEM; + goto move_map; + } + + after->start = map->end + 1; + map_groups__insert(self, after); + if (verbose >= 2) + map__fprintf(after, fp); + } +move_map: + /* + * If we have references, just move them to a separate list. + */ + if (pos->referenced) + list_add_tail(&pos->node, &self->removed_maps[map->type]); + else + map__delete(pos); + + if (err) + return err; + } + + return 0; +} + +/* + * XXX This should not really _copy_ te maps, but refcount them. + */ +int map_groups__clone(struct map_groups *self, + struct map_groups *parent, enum map_type type) +{ + struct rb_node *nd; + for (nd = rb_first(&parent->maps[type]); nd; nd = rb_next(nd)) { + struct map *map = rb_entry(nd, struct map, rb_node); + struct map *new = map__clone(map); + if (new == NULL) + return -ENOMEM; + map_groups__insert(self, new); + } + return 0; +} + +static u64 map__reloc_map_ip(struct map *map, u64 ip) +{ + return ip + (s64)map->pgoff; +} + +static u64 map__reloc_unmap_ip(struct map *map, u64 ip) +{ + return ip - (s64)map->pgoff; +} + +void map__reloc_vmlinux(struct map *self) +{ + struct kmap *kmap = map__kmap(self); + s64 reloc; + + if (!kmap->ref_reloc_sym || !kmap->ref_reloc_sym->unrelocated_addr) + return; + + reloc = (kmap->ref_reloc_sym->unrelocated_addr - + kmap->ref_reloc_sym->addr); + + if (!reloc) + return; + + self->map_ip = map__reloc_map_ip; + self->unmap_ip = map__reloc_unmap_ip; + self->pgoff = reloc; +} + +void maps__insert(struct rb_root *maps, struct map *map) +{ + struct rb_node **p = &maps->rb_node; + struct rb_node *parent = NULL; + const u64 ip = map->start; + struct map *m; + + while (*p != NULL) { + parent = *p; + m = rb_entry(parent, struct map, rb_node); + if (ip < m->start) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&map->rb_node, parent, p); + rb_insert_color(&map->rb_node, maps); +} + +void maps__remove(struct rb_root *self, struct map *map) +{ + rb_erase(&map->rb_node, self); +} + +struct map *maps__find(struct rb_root *maps, u64 ip) +{ + struct rb_node **p = &maps->rb_node; + struct rb_node *parent = NULL; + struct map *m; + + while (*p != NULL) { + parent = *p; + m = rb_entry(parent, struct map, rb_node); + if (ip < m->start) + p = &(*p)->rb_left; + else if (ip > m->end) + p = &(*p)->rb_right; + else + return m; + } + + return NULL; +} + +int machine__init(struct machine *self, const char *root_dir, pid_t pid) +{ + map_groups__init(&self->kmaps); + RB_CLEAR_NODE(&self->rb_node); + INIT_LIST_HEAD(&self->user_dsos); + INIT_LIST_HEAD(&self->kernel_dsos); + + self->kmaps.machine = self; + self->pid = pid; + self->root_dir = strdup(root_dir); + return self->root_dir == NULL ? -ENOMEM : 0; +} + +static void dsos__delete(struct list_head *self) +{ + struct dso *pos, *n; + + list_for_each_entry_safe(pos, n, self, node) { + list_del(&pos->node); + dso__delete(pos); + } +} + +void machine__exit(struct machine *self) +{ + map_groups__exit(&self->kmaps); + dsos__delete(&self->user_dsos); + dsos__delete(&self->kernel_dsos); + free(self->root_dir); + self->root_dir = NULL; +} + +void machine__delete(struct machine *self) +{ + machine__exit(self); + free(self); +} + +struct machine *machines__add(struct rb_root *self, pid_t pid, + const char *root_dir) +{ + struct rb_node **p = &self->rb_node; + struct rb_node *parent = NULL; + struct machine *pos, *machine = malloc(sizeof(*machine)); + + if (!machine) + return NULL; + + if (machine__init(machine, root_dir, pid) != 0) { + free(machine); + return NULL; + } + + while (*p != NULL) { + parent = *p; + pos = rb_entry(parent, struct machine, rb_node); + if (pid < pos->pid) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&machine->rb_node, parent, p); + rb_insert_color(&machine->rb_node, self); + + return machine; +} + +struct machine *machines__find(struct rb_root *self, pid_t pid) +{ + struct rb_node **p = &self->rb_node; + struct rb_node *parent = NULL; + struct machine *machine; + struct machine *default_machine = NULL; + + while (*p != NULL) { + parent = *p; + machine = rb_entry(parent, struct machine, rb_node); + if (pid < machine->pid) + p = &(*p)->rb_left; + else if (pid > machine->pid) + p = &(*p)->rb_right; + else + return machine; + if (!machine->pid) + default_machine = machine; + } + + return default_machine; +} + +struct machine *machines__findnew(struct rb_root *self, pid_t pid) +{ + char path[PATH_MAX]; + const char *root_dir; + struct machine *machine = machines__find(self, pid); + + if (!machine || machine->pid != pid) { + if (pid == HOST_KERNEL_ID || pid == DEFAULT_GUEST_KERNEL_ID) + root_dir = ""; + else { + if (!symbol_conf.guestmount) + goto out; + sprintf(path, "%s/%d", symbol_conf.guestmount, pid); + if (access(path, R_OK)) { + pr_err("Can't access file %s\n", path); + goto out; + } + root_dir = path; + } + machine = machines__add(self, pid, root_dir); + } + +out: + return machine; +} + +void machines__process(struct rb_root *self, machine__process_t process, void *data) +{ + struct rb_node *nd; + + for (nd = rb_first(self); nd; nd = rb_next(nd)) { + struct machine *pos = rb_entry(nd, struct machine, rb_node); + process(pos, data); + } +} + +char *machine__mmap_name(struct machine *self, char *bf, size_t size) +{ + if (machine__is_host(self)) + snprintf(bf, size, "[%s]", "kernel.kallsyms"); + else if (machine__is_default_guest(self)) + snprintf(bf, size, "[%s]", "guest.kernel.kallsyms"); + else + snprintf(bf, size, "[%s.%d]", "guest.kernel.kallsyms", self->pid); + + return bf; +} diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h new file mode 100644 index 00000000..b397c038 --- /dev/null +++ b/tools/perf/util/map.h @@ -0,0 +1,237 @@ +#ifndef __PERF_MAP_H +#define __PERF_MAP_H + +#include <linux/compiler.h> +#include <linux/list.h> +#include <linux/rbtree.h> +#include <stdio.h> +#include <stdbool.h> +#include "types.h" + +enum map_type { + MAP__FUNCTION = 0, + MAP__VARIABLE, +}; + +#define MAP__NR_TYPES (MAP__VARIABLE + 1) + +extern const char *map_type__name[MAP__NR_TYPES]; + +struct dso; +struct ref_reloc_sym; +struct map_groups; +struct machine; + +struct map { + union { + struct rb_node rb_node; + struct list_head node; + }; + u64 start; + u64 end; + u8 /* enum map_type */ type; + bool referenced; + u32 priv; + u64 pgoff; + + /* ip -> dso rip */ + u64 (*map_ip)(struct map *, u64); + /* dso rip -> ip */ + u64 (*unmap_ip)(struct map *, u64); + + struct dso *dso; + struct map_groups *groups; +}; + +struct kmap { + struct ref_reloc_sym *ref_reloc_sym; + struct map_groups *kmaps; +}; + +struct map_groups { + struct rb_root maps[MAP__NR_TYPES]; + struct list_head removed_maps[MAP__NR_TYPES]; + struct machine *machine; +}; + +/* Native host kernel uses -1 as pid index in machine */ +#define HOST_KERNEL_ID (-1) +#define DEFAULT_GUEST_KERNEL_ID (0) + +struct machine { + struct rb_node rb_node; + pid_t pid; + char *root_dir; + struct list_head user_dsos; + struct list_head kernel_dsos; + struct map_groups kmaps; + struct map *vmlinux_maps[MAP__NR_TYPES]; +}; + +static inline +struct map *machine__kernel_map(struct machine *self, enum map_type type) +{ + return self->vmlinux_maps[type]; +} + +static inline struct kmap *map__kmap(struct map *self) +{ + return (struct kmap *)(self + 1); +} + +static inline u64 map__map_ip(struct map *map, u64 ip) +{ + return ip - map->start + map->pgoff; +} + +static inline u64 map__unmap_ip(struct map *map, u64 ip) +{ + return ip + map->start - map->pgoff; +} + +static inline u64 identity__map_ip(struct map *map __used, u64 ip) +{ + return ip; +} + + +/* rip/ip <-> addr suitable for passing to `objdump --start-address=` */ +u64 map__rip_2objdump(struct map *map, u64 rip); +u64 map__objdump_2ip(struct map *map, u64 addr); + +struct symbol; + +typedef int (*symbol_filter_t)(struct map *map, struct symbol *sym); + +void map__init(struct map *self, enum map_type type, + u64 start, u64 end, u64 pgoff, struct dso *dso); +struct map *map__new(struct list_head *dsos__list, u64 start, u64 len, + u64 pgoff, u32 pid, char *filename, + enum map_type type); +void map__delete(struct map *self); +struct map *map__clone(struct map *self); +int map__overlap(struct map *l, struct map *r); +size_t map__fprintf(struct map *self, FILE *fp); + +int map__load(struct map *self, symbol_filter_t filter); +struct symbol *map__find_symbol(struct map *self, + u64 addr, symbol_filter_t filter); +struct symbol *map__find_symbol_by_name(struct map *self, const char *name, + symbol_filter_t filter); +void map__fixup_start(struct map *self); +void map__fixup_end(struct map *self); + +void map__reloc_vmlinux(struct map *self); + +size_t __map_groups__fprintf_maps(struct map_groups *self, + enum map_type type, int verbose, FILE *fp); +void maps__insert(struct rb_root *maps, struct map *map); +void maps__remove(struct rb_root *self, struct map *map); +struct map *maps__find(struct rb_root *maps, u64 addr); +void map_groups__init(struct map_groups *self); +void map_groups__exit(struct map_groups *self); +int map_groups__clone(struct map_groups *self, + struct map_groups *parent, enum map_type type); +size_t map_groups__fprintf(struct map_groups *self, int verbose, FILE *fp); +size_t map_groups__fprintf_maps(struct map_groups *self, int verbose, FILE *fp); + +typedef void (*machine__process_t)(struct machine *self, void *data); + +void machines__process(struct rb_root *self, machine__process_t process, void *data); +struct machine *machines__add(struct rb_root *self, pid_t pid, + const char *root_dir); +struct machine *machines__find_host(struct rb_root *self); +struct machine *machines__find(struct rb_root *self, pid_t pid); +struct machine *machines__findnew(struct rb_root *self, pid_t pid); +char *machine__mmap_name(struct machine *self, char *bf, size_t size); +int machine__init(struct machine *self, const char *root_dir, pid_t pid); +void machine__exit(struct machine *self); +void machine__delete(struct machine *self); + +/* + * Default guest kernel is defined by parameter --guestkallsyms + * and --guestmodules + */ +static inline bool machine__is_default_guest(struct machine *self) +{ + return self ? self->pid == DEFAULT_GUEST_KERNEL_ID : false; +} + +static inline bool machine__is_host(struct machine *self) +{ + return self ? self->pid == HOST_KERNEL_ID : false; +} + +static inline void map_groups__insert(struct map_groups *self, struct map *map) +{ + maps__insert(&self->maps[map->type], map); + map->groups = self; +} + +static inline void map_groups__remove(struct map_groups *self, struct map *map) +{ + maps__remove(&self->maps[map->type], map); +} + +static inline struct map *map_groups__find(struct map_groups *self, + enum map_type type, u64 addr) +{ + return maps__find(&self->maps[type], addr); +} + +struct symbol *map_groups__find_symbol(struct map_groups *self, + enum map_type type, u64 addr, + struct map **mapp, + symbol_filter_t filter); + +struct symbol *map_groups__find_symbol_by_name(struct map_groups *self, + enum map_type type, + const char *name, + struct map **mapp, + symbol_filter_t filter); + +static inline +struct symbol *machine__find_kernel_symbol(struct machine *self, + enum map_type type, u64 addr, + struct map **mapp, + symbol_filter_t filter) +{ + return map_groups__find_symbol(&self->kmaps, type, addr, mapp, filter); +} + +static inline +struct symbol *machine__find_kernel_function(struct machine *self, u64 addr, + struct map **mapp, + symbol_filter_t filter) +{ + return machine__find_kernel_symbol(self, MAP__FUNCTION, addr, mapp, filter); +} + +static inline +struct symbol *map_groups__find_function_by_name(struct map_groups *self, + const char *name, struct map **mapp, + symbol_filter_t filter) +{ + return map_groups__find_symbol_by_name(self, MAP__FUNCTION, name, mapp, filter); +} + +static inline +struct symbol *machine__find_kernel_function_by_name(struct machine *self, + const char *name, + struct map **mapp, + symbol_filter_t filter) +{ + return map_groups__find_function_by_name(&self->kmaps, name, mapp, + filter); +} + +int map_groups__fixup_overlappings(struct map_groups *self, struct map *map, + int verbose, FILE *fp); + +struct map *map_groups__find_by_name(struct map_groups *self, + enum map_type type, const char *name); +struct map *machine__new_module(struct machine *self, u64 start, const char *filename); + +void map_groups__flush(struct map_groups *self); + +#endif /* __PERF_MAP_H */ diff --git a/tools/perf/util/pager.c b/tools/perf/util/pager.c new file mode 100644 index 00000000..1915de20 --- /dev/null +++ b/tools/perf/util/pager.c @@ -0,0 +1,96 @@ +#include "cache.h" +#include "run-command.h" +#include "sigchain.h" + +/* + * This is split up from the rest of git so that we can do + * something different on Windows. + */ + +static int spawned_pager; + +static void pager_preexec(void) +{ + /* + * Work around bug in "less" by not starting it until we + * have real input + */ + fd_set in; + + FD_ZERO(&in); + FD_SET(0, &in); + select(1, &in, NULL, &in, NULL); + + setenv("LESS", "FRSX", 0); +} + +static const char *pager_argv[] = { "sh", "-c", NULL, NULL }; +static struct child_process pager_process; + +static void wait_for_pager(void) +{ + fflush(stdout); + fflush(stderr); + /* signal EOF to pager */ + close(1); + close(2); + finish_command(&pager_process); +} + +static void wait_for_pager_signal(int signo) +{ + wait_for_pager(); + sigchain_pop(signo); + raise(signo); +} + +void setup_pager(void) +{ + const char *pager = getenv("PERF_PAGER"); + + if (!isatty(1)) + return; + if (!pager) { + if (!pager_program) + perf_config(perf_default_config, NULL); + pager = pager_program; + } + if (!pager) + pager = getenv("PAGER"); + if (!pager) + pager = "less"; + else if (!*pager || !strcmp(pager, "cat")) + return; + + spawned_pager = 1; /* means we are emitting to terminal */ + + /* spawn the pager */ + pager_argv[2] = pager; + pager_process.argv = pager_argv; + pager_process.in = -1; + pager_process.preexec_cb = pager_preexec; + + if (start_command(&pager_process)) + return; + + /* original process continues, but writes to the pipe */ + dup2(pager_process.in, 1); + if (isatty(2)) + dup2(pager_process.in, 2); + close(pager_process.in); + + /* this makes sure that the parent terminates after the pager */ + sigchain_push_common(wait_for_pager_signal); + atexit(wait_for_pager); +} + +int pager_in_use(void) +{ + const char *env; + + if (spawned_pager) + return 1; + + env = getenv("PERF_PAGER_IN_USE"); + return env ? perf_config_bool("PERF_PAGER_IN_USE", env) : 0; +} diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c new file mode 100644 index 00000000..41982c37 --- /dev/null +++ b/tools/perf/util/parse-events.c @@ -0,0 +1,1093 @@ +#include "../../../include/linux/hw_breakpoint.h" +#include "util.h" +#include "../perf.h" +#include "evlist.h" +#include "evsel.h" +#include "parse-options.h" +#include "parse-events.h" +#include "exec_cmd.h" +#include "string.h" +#include "symbol.h" +#include "cache.h" +#include "header.h" +#include "debugfs.h" + +struct event_symbol { + u8 type; + u64 config; + const char *symbol; + const char *alias; +}; + +enum event_result { + EVT_FAILED, + EVT_HANDLED, + EVT_HANDLED_ALL +}; + +char debugfs_path[MAXPATHLEN]; + +#define CHW(x) .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_##x +#define CSW(x) .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_##x + +static struct event_symbol event_symbols[] = { + { CHW(CPU_CYCLES), "cpu-cycles", "cycles" }, + { CHW(STALLED_CYCLES_FRONTEND), "stalled-cycles-frontend", "idle-cycles-frontend" }, + { CHW(STALLED_CYCLES_BACKEND), "stalled-cycles-backend", "idle-cycles-backend" }, + { CHW(INSTRUCTIONS), "instructions", "" }, + { CHW(CACHE_REFERENCES), "cache-references", "" }, + { CHW(CACHE_MISSES), "cache-misses", "" }, + { CHW(BRANCH_INSTRUCTIONS), "branch-instructions", "branches" }, + { CHW(BRANCH_MISSES), "branch-misses", "" }, + { CHW(BUS_CYCLES), "bus-cycles", "" }, + + { CSW(CPU_CLOCK), "cpu-clock", "" }, + { CSW(TASK_CLOCK), "task-clock", "" }, + { CSW(PAGE_FAULTS), "page-faults", "faults" }, + { CSW(PAGE_FAULTS_MIN), "minor-faults", "" }, + { CSW(PAGE_FAULTS_MAJ), "major-faults", "" }, + { CSW(CONTEXT_SWITCHES), "context-switches", "cs" }, + { CSW(CPU_MIGRATIONS), "cpu-migrations", "migrations" }, + { CSW(ALIGNMENT_FAULTS), "alignment-faults", "" }, + { CSW(EMULATION_FAULTS), "emulation-faults", "" }, +}; + +#define __PERF_EVENT_FIELD(config, name) \ + ((config & PERF_EVENT_##name##_MASK) >> PERF_EVENT_##name##_SHIFT) + +#define PERF_EVENT_RAW(config) __PERF_EVENT_FIELD(config, RAW) +#define PERF_EVENT_CONFIG(config) __PERF_EVENT_FIELD(config, CONFIG) +#define PERF_EVENT_TYPE(config) __PERF_EVENT_FIELD(config, TYPE) +#define PERF_EVENT_ID(config) __PERF_EVENT_FIELD(config, EVENT) + +static const char *hw_event_names[PERF_COUNT_HW_MAX] = { + "cycles", + "instructions", + "cache-references", + "cache-misses", + "branches", + "branch-misses", + "bus-cycles", + "stalled-cycles-frontend", + "stalled-cycles-backend", +}; + +static const char *sw_event_names[PERF_COUNT_SW_MAX] = { + "cpu-clock", + "task-clock", + "page-faults", + "context-switches", + "CPU-migrations", + "minor-faults", + "major-faults", + "alignment-faults", + "emulation-faults", +}; + +#define MAX_ALIASES 8 + +static const char *hw_cache[][MAX_ALIASES] = { + { "L1-dcache", "l1-d", "l1d", "L1-data", }, + { "L1-icache", "l1-i", "l1i", "L1-instruction", }, + { "LLC", "L2" }, + { "dTLB", "d-tlb", "Data-TLB", }, + { "iTLB", "i-tlb", "Instruction-TLB", }, + { "branch", "branches", "bpu", "btb", "bpc", }, +}; + +static const char *hw_cache_op[][MAX_ALIASES] = { + { "load", "loads", "read", }, + { "store", "stores", "write", }, + { "prefetch", "prefetches", "speculative-read", "speculative-load", }, +}; + +static const char *hw_cache_result[][MAX_ALIASES] = { + { "refs", "Reference", "ops", "access", }, + { "misses", "miss", }, +}; + +#define C(x) PERF_COUNT_HW_CACHE_##x +#define CACHE_READ (1 << C(OP_READ)) +#define CACHE_WRITE (1 << C(OP_WRITE)) +#define CACHE_PREFETCH (1 << C(OP_PREFETCH)) +#define COP(x) (1 << x) + +/* + * cache operartion stat + * L1I : Read and prefetch only + * ITLB and BPU : Read-only + */ +static unsigned long hw_cache_stat[C(MAX)] = { + [C(L1D)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH), + [C(L1I)] = (CACHE_READ | CACHE_PREFETCH), + [C(LL)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH), + [C(DTLB)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH), + [C(ITLB)] = (CACHE_READ), + [C(BPU)] = (CACHE_READ), +}; + +#define for_each_subsystem(sys_dir, sys_dirent, sys_next) \ + while (!readdir_r(sys_dir, &sys_dirent, &sys_next) && sys_next) \ + if (sys_dirent.d_type == DT_DIR && \ + (strcmp(sys_dirent.d_name, ".")) && \ + (strcmp(sys_dirent.d_name, ".."))) + +static int tp_event_has_id(struct dirent *sys_dir, struct dirent *evt_dir) +{ + char evt_path[MAXPATHLEN]; + int fd; + + snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", debugfs_path, + sys_dir->d_name, evt_dir->d_name); + fd = open(evt_path, O_RDONLY); + if (fd < 0) + return -EINVAL; + close(fd); + + return 0; +} + +#define for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next) \ + while (!readdir_r(evt_dir, &evt_dirent, &evt_next) && evt_next) \ + if (evt_dirent.d_type == DT_DIR && \ + (strcmp(evt_dirent.d_name, ".")) && \ + (strcmp(evt_dirent.d_name, "..")) && \ + (!tp_event_has_id(&sys_dirent, &evt_dirent))) + +#define MAX_EVENT_LENGTH 512 + + +struct tracepoint_path *tracepoint_id_to_path(u64 config) +{ + struct tracepoint_path *path = NULL; + DIR *sys_dir, *evt_dir; + struct dirent *sys_next, *evt_next, sys_dirent, evt_dirent; + char id_buf[4]; + int fd; + u64 id; + char evt_path[MAXPATHLEN]; + char dir_path[MAXPATHLEN]; + + if (debugfs_valid_mountpoint(debugfs_path)) + return NULL; + + sys_dir = opendir(debugfs_path); + if (!sys_dir) + return NULL; + + for_each_subsystem(sys_dir, sys_dirent, sys_next) { + + snprintf(dir_path, MAXPATHLEN, "%s/%s", debugfs_path, + sys_dirent.d_name); + evt_dir = opendir(dir_path); + if (!evt_dir) + continue; + + for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next) { + + snprintf(evt_path, MAXPATHLEN, "%s/%s/id", dir_path, + evt_dirent.d_name); + fd = open(evt_path, O_RDONLY); + if (fd < 0) + continue; + if (read(fd, id_buf, sizeof(id_buf)) < 0) { + close(fd); + continue; + } + close(fd); + id = atoll(id_buf); + if (id == config) { + closedir(evt_dir); + closedir(sys_dir); + path = zalloc(sizeof(*path)); + path->system = malloc(MAX_EVENT_LENGTH); + if (!path->system) { + free(path); + return NULL; + } + path->name = malloc(MAX_EVENT_LENGTH); + if (!path->name) { + free(path->system); + free(path); + return NULL; + } + strncpy(path->system, sys_dirent.d_name, + MAX_EVENT_LENGTH); + strncpy(path->name, evt_dirent.d_name, + MAX_EVENT_LENGTH); + return path; + } + } + closedir(evt_dir); + } + + closedir(sys_dir); + return NULL; +} + +#define TP_PATH_LEN (MAX_EVENT_LENGTH * 2 + 1) +static const char *tracepoint_id_to_name(u64 config) +{ + static char buf[TP_PATH_LEN]; + struct tracepoint_path *path; + + path = tracepoint_id_to_path(config); + if (path) { + snprintf(buf, TP_PATH_LEN, "%s:%s", path->system, path->name); + free(path->name); + free(path->system); + free(path); + } else + snprintf(buf, TP_PATH_LEN, "%s:%s", "unknown", "unknown"); + + return buf; +} + +static int is_cache_op_valid(u8 cache_type, u8 cache_op) +{ + if (hw_cache_stat[cache_type] & COP(cache_op)) + return 1; /* valid */ + else + return 0; /* invalid */ +} + +static char *event_cache_name(u8 cache_type, u8 cache_op, u8 cache_result) +{ + static char name[50]; + + if (cache_result) { + sprintf(name, "%s-%s-%s", hw_cache[cache_type][0], + hw_cache_op[cache_op][0], + hw_cache_result[cache_result][0]); + } else { + sprintf(name, "%s-%s", hw_cache[cache_type][0], + hw_cache_op[cache_op][1]); + } + + return name; +} + +const char *event_type(int type) +{ + switch (type) { + case PERF_TYPE_HARDWARE: + return "hardware"; + + case PERF_TYPE_SOFTWARE: + return "software"; + + case PERF_TYPE_TRACEPOINT: + return "tracepoint"; + + case PERF_TYPE_HW_CACHE: + return "hardware-cache"; + + default: + break; + } + + return "unknown"; +} + +const char *event_name(struct perf_evsel *evsel) +{ + u64 config = evsel->attr.config; + int type = evsel->attr.type; + + if (evsel->name) + return evsel->name; + + return __event_name(type, config); +} + +const char *__event_name(int type, u64 config) +{ + static char buf[32]; + + if (type == PERF_TYPE_RAW) { + sprintf(buf, "raw 0x%" PRIx64, config); + return buf; + } + + switch (type) { + case PERF_TYPE_HARDWARE: + if (config < PERF_COUNT_HW_MAX && hw_event_names[config]) + return hw_event_names[config]; + return "unknown-hardware"; + + case PERF_TYPE_HW_CACHE: { + u8 cache_type, cache_op, cache_result; + + cache_type = (config >> 0) & 0xff; + if (cache_type > PERF_COUNT_HW_CACHE_MAX) + return "unknown-ext-hardware-cache-type"; + + cache_op = (config >> 8) & 0xff; + if (cache_op > PERF_COUNT_HW_CACHE_OP_MAX) + return "unknown-ext-hardware-cache-op"; + + cache_result = (config >> 16) & 0xff; + if (cache_result > PERF_COUNT_HW_CACHE_RESULT_MAX) + return "unknown-ext-hardware-cache-result"; + + if (!is_cache_op_valid(cache_type, cache_op)) + return "invalid-cache"; + + return event_cache_name(cache_type, cache_op, cache_result); + } + + case PERF_TYPE_SOFTWARE: + if (config < PERF_COUNT_SW_MAX && sw_event_names[config]) + return sw_event_names[config]; + return "unknown-software"; + + case PERF_TYPE_TRACEPOINT: + return tracepoint_id_to_name(config); + + default: + break; + } + + return "unknown"; +} + +static int parse_aliases(const char **str, const char *names[][MAX_ALIASES], int size) +{ + int i, j; + int n, longest = -1; + + for (i = 0; i < size; i++) { + for (j = 0; j < MAX_ALIASES && names[i][j]; j++) { + n = strlen(names[i][j]); + if (n > longest && !strncasecmp(*str, names[i][j], n)) + longest = n; + } + if (longest > 0) { + *str += longest; + return i; + } + } + + return -1; +} + +static enum event_result +parse_generic_hw_event(const char **str, struct perf_event_attr *attr) +{ + const char *s = *str; + int cache_type = -1, cache_op = -1, cache_result = -1; + + cache_type = parse_aliases(&s, hw_cache, PERF_COUNT_HW_CACHE_MAX); + /* + * No fallback - if we cannot get a clear cache type + * then bail out: + */ + if (cache_type == -1) + return EVT_FAILED; + + while ((cache_op == -1 || cache_result == -1) && *s == '-') { + ++s; + + if (cache_op == -1) { + cache_op = parse_aliases(&s, hw_cache_op, + PERF_COUNT_HW_CACHE_OP_MAX); + if (cache_op >= 0) { + if (!is_cache_op_valid(cache_type, cache_op)) + return 0; + continue; + } + } + + if (cache_result == -1) { + cache_result = parse_aliases(&s, hw_cache_result, + PERF_COUNT_HW_CACHE_RESULT_MAX); + if (cache_result >= 0) + continue; + } + + /* + * Can't parse this as a cache op or result, so back up + * to the '-'. + */ + --s; + break; + } + + /* + * Fall back to reads: + */ + if (cache_op == -1) + cache_op = PERF_COUNT_HW_CACHE_OP_READ; + + /* + * Fall back to accesses: + */ + if (cache_result == -1) + cache_result = PERF_COUNT_HW_CACHE_RESULT_ACCESS; + + attr->config = cache_type | (cache_op << 8) | (cache_result << 16); + attr->type = PERF_TYPE_HW_CACHE; + + *str = s; + return EVT_HANDLED; +} + +static enum event_result +parse_single_tracepoint_event(char *sys_name, + const char *evt_name, + unsigned int evt_length, + struct perf_event_attr *attr, + const char **strp) +{ + char evt_path[MAXPATHLEN]; + char id_buf[4]; + u64 id; + int fd; + + snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", debugfs_path, + sys_name, evt_name); + + fd = open(evt_path, O_RDONLY); + if (fd < 0) + return EVT_FAILED; + + if (read(fd, id_buf, sizeof(id_buf)) < 0) { + close(fd); + return EVT_FAILED; + } + + close(fd); + id = atoll(id_buf); + attr->config = id; + attr->type = PERF_TYPE_TRACEPOINT; + *strp += strlen(sys_name) + evt_length + 1; /* + 1 for the ':' */ + + attr->sample_type |= PERF_SAMPLE_RAW; + attr->sample_type |= PERF_SAMPLE_TIME; + attr->sample_type |= PERF_SAMPLE_CPU; + + attr->sample_period = 1; + + + return EVT_HANDLED; +} + +/* sys + ':' + event + ':' + flags*/ +#define MAX_EVOPT_LEN (MAX_EVENT_LENGTH * 2 + 2 + 128) +static enum event_result +parse_multiple_tracepoint_event(const struct option *opt, char *sys_name, + const char *evt_exp, char *flags) +{ + char evt_path[MAXPATHLEN]; + struct dirent *evt_ent; + DIR *evt_dir; + + snprintf(evt_path, MAXPATHLEN, "%s/%s", debugfs_path, sys_name); + evt_dir = opendir(evt_path); + + if (!evt_dir) { + perror("Can't open event dir"); + return EVT_FAILED; + } + + while ((evt_ent = readdir(evt_dir))) { + char event_opt[MAX_EVOPT_LEN + 1]; + int len; + + if (!strcmp(evt_ent->d_name, ".") + || !strcmp(evt_ent->d_name, "..") + || !strcmp(evt_ent->d_name, "enable") + || !strcmp(evt_ent->d_name, "filter")) + continue; + + if (!strglobmatch(evt_ent->d_name, evt_exp)) + continue; + + len = snprintf(event_opt, MAX_EVOPT_LEN, "%s:%s%s%s", sys_name, + evt_ent->d_name, flags ? ":" : "", + flags ?: ""); + if (len < 0) + return EVT_FAILED; + + if (parse_events(opt, event_opt, 0)) + return EVT_FAILED; + } + + return EVT_HANDLED_ALL; +} + +static enum event_result +parse_tracepoint_event(const struct option *opt, const char **strp, + struct perf_event_attr *attr) +{ + const char *evt_name; + char *flags = NULL, *comma_loc; + char sys_name[MAX_EVENT_LENGTH]; + unsigned int sys_length, evt_length; + + if (debugfs_valid_mountpoint(debugfs_path)) + return 0; + + evt_name = strchr(*strp, ':'); + if (!evt_name) + return EVT_FAILED; + + sys_length = evt_name - *strp; + if (sys_length >= MAX_EVENT_LENGTH) + return 0; + + strncpy(sys_name, *strp, sys_length); + sys_name[sys_length] = '\0'; + evt_name = evt_name + 1; + + comma_loc = strchr(evt_name, ','); + if (comma_loc) { + /* take the event name up to the comma */ + evt_name = strndup(evt_name, comma_loc - evt_name); + } + flags = strchr(evt_name, ':'); + if (flags) { + /* split it out: */ + evt_name = strndup(evt_name, flags - evt_name); + flags++; + } + + evt_length = strlen(evt_name); + if (evt_length >= MAX_EVENT_LENGTH) + return EVT_FAILED; + if (strpbrk(evt_name, "*?")) { + *strp += strlen(sys_name) + evt_length + 1; /* 1 == the ':' */ + return parse_multiple_tracepoint_event(opt, sys_name, evt_name, + flags); + } else { + return parse_single_tracepoint_event(sys_name, evt_name, + evt_length, attr, strp); + } +} + +static enum event_result +parse_breakpoint_type(const char *type, const char **strp, + struct perf_event_attr *attr) +{ + int i; + + for (i = 0; i < 3; i++) { + if (!type[i]) + break; + + switch (type[i]) { + case 'r': + attr->bp_type |= HW_BREAKPOINT_R; + break; + case 'w': + attr->bp_type |= HW_BREAKPOINT_W; + break; + case 'x': + attr->bp_type |= HW_BREAKPOINT_X; + break; + default: + return EVT_FAILED; + } + } + if (!attr->bp_type) /* Default */ + attr->bp_type = HW_BREAKPOINT_R | HW_BREAKPOINT_W; + + *strp = type + i; + + return EVT_HANDLED; +} + +static enum event_result +parse_breakpoint_event(const char **strp, struct perf_event_attr *attr) +{ + const char *target; + const char *type; + char *endaddr; + u64 addr; + enum event_result err; + + target = strchr(*strp, ':'); + if (!target) + return EVT_FAILED; + + if (strncmp(*strp, "mem", target - *strp) != 0) + return EVT_FAILED; + + target++; + + addr = strtoull(target, &endaddr, 0); + if (target == endaddr) + return EVT_FAILED; + + attr->bp_addr = addr; + *strp = endaddr; + + type = strchr(target, ':'); + + /* If no type is defined, just rw as default */ + if (!type) { + attr->bp_type = HW_BREAKPOINT_R | HW_BREAKPOINT_W; + } else { + err = parse_breakpoint_type(++type, strp, attr); + if (err == EVT_FAILED) + return EVT_FAILED; + } + + /* + * We should find a nice way to override the access length + * Provide some defaults for now + */ + if (attr->bp_type == HW_BREAKPOINT_X) + attr->bp_len = sizeof(long); + else + attr->bp_len = HW_BREAKPOINT_LEN_4; + + attr->type = PERF_TYPE_BREAKPOINT; + + return EVT_HANDLED; +} + +static int check_events(const char *str, unsigned int i) +{ + int n; + + n = strlen(event_symbols[i].symbol); + if (!strncasecmp(str, event_symbols[i].symbol, n)) + return n; + + n = strlen(event_symbols[i].alias); + if (n) { + if (!strncasecmp(str, event_symbols[i].alias, n)) + return n; + } + + return 0; +} + +static enum event_result +parse_symbolic_event(const char **strp, struct perf_event_attr *attr) +{ + const char *str = *strp; + unsigned int i; + int n; + + for (i = 0; i < ARRAY_SIZE(event_symbols); i++) { + n = check_events(str, i); + if (n > 0) { + attr->type = event_symbols[i].type; + attr->config = event_symbols[i].config; + *strp = str + n; + return EVT_HANDLED; + } + } + return EVT_FAILED; +} + +static enum event_result +parse_raw_event(const char **strp, struct perf_event_attr *attr) +{ + const char *str = *strp; + u64 config; + int n; + + if (*str != 'r') + return EVT_FAILED; + n = hex2u64(str + 1, &config); + if (n > 0) { + *strp = str + n + 1; + attr->type = PERF_TYPE_RAW; + attr->config = config; + return EVT_HANDLED; + } + return EVT_FAILED; +} + +static enum event_result +parse_numeric_event(const char **strp, struct perf_event_attr *attr) +{ + const char *str = *strp; + char *endp; + unsigned long type; + u64 config; + + type = strtoul(str, &endp, 0); + if (endp > str && type < PERF_TYPE_MAX && *endp == ':') { + str = endp + 1; + config = strtoul(str, &endp, 0); + if (endp > str) { + attr->type = type; + attr->config = config; + *strp = endp; + return EVT_HANDLED; + } + } + return EVT_FAILED; +} + +static int +parse_event_modifier(const char **strp, struct perf_event_attr *attr) +{ + const char *str = *strp; + int exclude = 0; + int eu = 0, ek = 0, eh = 0, precise = 0; + + if (!*str) + return 0; + + if (*str == ',') + return 0; + + if (*str++ != ':') + return -1; + + while (*str) { + if (*str == 'u') { + if (!exclude) + exclude = eu = ek = eh = 1; + eu = 0; + } else if (*str == 'k') { + if (!exclude) + exclude = eu = ek = eh = 1; + ek = 0; + } else if (*str == 'h') { + if (!exclude) + exclude = eu = ek = eh = 1; + eh = 0; + } else if (*str == 'p') { + precise++; + } else + break; + + ++str; + } + if (str < *strp + 2) + return -1; + + *strp = str; + + attr->exclude_user = eu; + attr->exclude_kernel = ek; + attr->exclude_hv = eh; + attr->precise_ip = precise; + + return 0; +} + +/* + * Each event can have multiple symbolic names. + * Symbolic names are (almost) exactly matched. + */ +static enum event_result +parse_event_symbols(const struct option *opt, const char **str, + struct perf_event_attr *attr) +{ + enum event_result ret; + + ret = parse_tracepoint_event(opt, str, attr); + if (ret != EVT_FAILED) + goto modifier; + + ret = parse_raw_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; + + ret = parse_numeric_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; + + ret = parse_symbolic_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; + + ret = parse_generic_hw_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; + + ret = parse_breakpoint_event(str, attr); + if (ret != EVT_FAILED) + goto modifier; + + fprintf(stderr, "invalid or unsupported event: '%s'\n", *str); + fprintf(stderr, "Run 'perf list' for a list of valid events\n"); + return EVT_FAILED; + +modifier: + if (parse_event_modifier(str, attr) < 0) { + fprintf(stderr, "invalid event modifier: '%s'\n", *str); + fprintf(stderr, "Run 'perf list' for a list of valid events and modifiers\n"); + + return EVT_FAILED; + } + + return ret; +} + +int parse_events(const struct option *opt, const char *str, int unset __used) +{ + struct perf_evlist *evlist = *(struct perf_evlist **)opt->value; + struct perf_event_attr attr; + enum event_result ret; + const char *ostr; + + for (;;) { + ostr = str; + memset(&attr, 0, sizeof(attr)); + ret = parse_event_symbols(opt, &str, &attr); + if (ret == EVT_FAILED) + return -1; + + if (!(*str == 0 || *str == ',' || isspace(*str))) + return -1; + + if (ret != EVT_HANDLED_ALL) { + struct perf_evsel *evsel; + evsel = perf_evsel__new(&attr, evlist->nr_entries); + if (evsel == NULL) + return -1; + perf_evlist__add(evlist, evsel); + + evsel->name = calloc(str - ostr + 1, 1); + if (!evsel->name) + return -1; + strncpy(evsel->name, ostr, str - ostr); + } + + if (*str == 0) + break; + if (*str == ',') + ++str; + while (isspace(*str)) + ++str; + } + + return 0; +} + +int parse_filter(const struct option *opt, const char *str, + int unset __used) +{ + struct perf_evlist *evlist = *(struct perf_evlist **)opt->value; + struct perf_evsel *last = NULL; + + if (evlist->nr_entries > 0) + last = list_entry(evlist->entries.prev, struct perf_evsel, node); + + if (last == NULL || last->attr.type != PERF_TYPE_TRACEPOINT) { + fprintf(stderr, + "-F option should follow a -e tracepoint option\n"); + return -1; + } + + last->filter = strdup(str); + if (last->filter == NULL) { + fprintf(stderr, "not enough memory to hold filter string\n"); + return -1; + } + + return 0; +} + +static const char * const event_type_descriptors[] = { + "Hardware event", + "Software event", + "Tracepoint event", + "Hardware cache event", + "Raw hardware event descriptor", + "Hardware breakpoint", +}; + +/* + * Print the events from <debugfs_mount_point>/tracing/events + */ + +void print_tracepoint_events(const char *subsys_glob, const char *event_glob) +{ + DIR *sys_dir, *evt_dir; + struct dirent *sys_next, *evt_next, sys_dirent, evt_dirent; + char evt_path[MAXPATHLEN]; + char dir_path[MAXPATHLEN]; + + if (debugfs_valid_mountpoint(debugfs_path)) + return; + + sys_dir = opendir(debugfs_path); + if (!sys_dir) + return; + + for_each_subsystem(sys_dir, sys_dirent, sys_next) { + if (subsys_glob != NULL && + !strglobmatch(sys_dirent.d_name, subsys_glob)) + continue; + + snprintf(dir_path, MAXPATHLEN, "%s/%s", debugfs_path, + sys_dirent.d_name); + evt_dir = opendir(dir_path); + if (!evt_dir) + continue; + + for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next) { + if (event_glob != NULL && + !strglobmatch(evt_dirent.d_name, event_glob)) + continue; + + snprintf(evt_path, MAXPATHLEN, "%s:%s", + sys_dirent.d_name, evt_dirent.d_name); + printf(" %-50s [%s]\n", evt_path, + event_type_descriptors[PERF_TYPE_TRACEPOINT]); + } + closedir(evt_dir); + } + closedir(sys_dir); +} + +/* + * Check whether event is in <debugfs_mount_point>/tracing/events + */ + +int is_valid_tracepoint(const char *event_string) +{ + DIR *sys_dir, *evt_dir; + struct dirent *sys_next, *evt_next, sys_dirent, evt_dirent; + char evt_path[MAXPATHLEN]; + char dir_path[MAXPATHLEN]; + + if (debugfs_valid_mountpoint(debugfs_path)) + return 0; + + sys_dir = opendir(debugfs_path); + if (!sys_dir) + return 0; + + for_each_subsystem(sys_dir, sys_dirent, sys_next) { + + snprintf(dir_path, MAXPATHLEN, "%s/%s", debugfs_path, + sys_dirent.d_name); + evt_dir = opendir(dir_path); + if (!evt_dir) + continue; + + for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next) { + snprintf(evt_path, MAXPATHLEN, "%s:%s", + sys_dirent.d_name, evt_dirent.d_name); + if (!strcmp(evt_path, event_string)) { + closedir(evt_dir); + closedir(sys_dir); + return 1; + } + } + closedir(evt_dir); + } + closedir(sys_dir); + return 0; +} + +void print_events_type(u8 type) +{ + struct event_symbol *syms = event_symbols; + unsigned int i; + char name[64]; + + for (i = 0; i < ARRAY_SIZE(event_symbols); i++, syms++) { + if (type != syms->type) + continue; + + if (strlen(syms->alias)) + snprintf(name, sizeof(name), "%s OR %s", + syms->symbol, syms->alias); + else + snprintf(name, sizeof(name), "%s", syms->symbol); + + printf(" %-50s [%s]\n", name, + event_type_descriptors[type]); + } +} + +int print_hwcache_events(const char *event_glob) +{ + unsigned int type, op, i, printed = 0; + + for (type = 0; type < PERF_COUNT_HW_CACHE_MAX; type++) { + for (op = 0; op < PERF_COUNT_HW_CACHE_OP_MAX; op++) { + /* skip invalid cache type */ + if (!is_cache_op_valid(type, op)) + continue; + + for (i = 0; i < PERF_COUNT_HW_CACHE_RESULT_MAX; i++) { + char *name = event_cache_name(type, op, i); + + if (event_glob != NULL && !strglobmatch(name, event_glob)) + continue; + + printf(" %-50s [%s]\n", name, + event_type_descriptors[PERF_TYPE_HW_CACHE]); + ++printed; + } + } + } + + return printed; +} + +#define MAX_NAME_LEN 100 + +/* + * Print the help text for the event symbols: + */ +void print_events(const char *event_glob) +{ + unsigned int i, type, prev_type = -1, printed = 0, ntypes_printed = 0; + struct event_symbol *syms = event_symbols; + char name[MAX_NAME_LEN]; + + printf("\n"); + printf("List of pre-defined events (to be used in -e):\n"); + + for (i = 0; i < ARRAY_SIZE(event_symbols); i++, syms++) { + type = syms->type; + + if (type != prev_type && printed) { + printf("\n"); + printed = 0; + ntypes_printed++; + } + + if (event_glob != NULL && + !(strglobmatch(syms->symbol, event_glob) || + (syms->alias && strglobmatch(syms->alias, event_glob)))) + continue; + + if (strlen(syms->alias)) + snprintf(name, MAX_NAME_LEN, "%s OR %s", syms->symbol, syms->alias); + else + strncpy(name, syms->symbol, MAX_NAME_LEN); + printf(" %-50s [%s]\n", name, + event_type_descriptors[type]); + + prev_type = type; + ++printed; + } + + if (ntypes_printed) { + printed = 0; + printf("\n"); + } + print_hwcache_events(event_glob); + + if (event_glob != NULL) + return; + + printf("\n"); + printf(" %-50s [%s]\n", + "rNNN (see 'perf list --help' on how to encode it)", + event_type_descriptors[PERF_TYPE_RAW]); + printf("\n"); + + printf(" %-50s [%s]\n", + "mem:<addr>[:access]", + event_type_descriptors[PERF_TYPE_BREAKPOINT]); + printf("\n"); + + print_tracepoint_events(NULL, NULL); + + exit(129); +} diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h new file mode 100644 index 00000000..746d3fcb --- /dev/null +++ b/tools/perf/util/parse-events.h @@ -0,0 +1,41 @@ +#ifndef __PERF_PARSE_EVENTS_H +#define __PERF_PARSE_EVENTS_H +/* + * Parse symbolic events/counts passed in as options: + */ + +#include "../../../include/linux/perf_event.h" + +struct list_head; +struct perf_evsel; + +struct option; + +struct tracepoint_path { + char *system; + char *name; + struct tracepoint_path *next; +}; + +extern struct tracepoint_path *tracepoint_id_to_path(u64 config); +extern bool have_tracepoints(struct list_head *evlist); + +const char *event_type(int type); +const char *event_name(struct perf_evsel *event); +extern const char *__event_name(int type, u64 config); + +extern int parse_events(const struct option *opt, const char *str, int unset); +extern int parse_filter(const struct option *opt, const char *str, int unset); + +#define EVENTS_HELP_MAX (128*1024) + +void print_events(const char *event_glob); +void print_events_type(u8 type); +void print_tracepoint_events(const char *subsys_glob, const char *event_glob); +int print_hwcache_events(const char *event_glob); +extern int is_valid_tracepoint(const char *event_string); + +extern char debugfs_path[]; +extern int valid_debugfs_mount(const char *debugfs); + +#endif /* __PERF_PARSE_EVENTS_H */ diff --git a/tools/perf/util/parse-options.c b/tools/perf/util/parse-options.c new file mode 100644 index 00000000..99d02aa5 --- /dev/null +++ b/tools/perf/util/parse-options.c @@ -0,0 +1,577 @@ +#include "util.h" +#include "parse-options.h" +#include "cache.h" + +#define OPT_SHORT 1 +#define OPT_UNSET 2 + +static int opterror(const struct option *opt, const char *reason, int flags) +{ + if (flags & OPT_SHORT) + return error("switch `%c' %s", opt->short_name, reason); + if (flags & OPT_UNSET) + return error("option `no-%s' %s", opt->long_name, reason); + return error("option `%s' %s", opt->long_name, reason); +} + +static int get_arg(struct parse_opt_ctx_t *p, const struct option *opt, + int flags, const char **arg) +{ + if (p->opt) { + *arg = p->opt; + p->opt = NULL; + } else if ((opt->flags & PARSE_OPT_LASTARG_DEFAULT) && (p->argc == 1 || + **(p->argv + 1) == '-')) { + *arg = (const char *)opt->defval; + } else if (p->argc > 1) { + p->argc--; + *arg = *++p->argv; + } else + return opterror(opt, "requires a value", flags); + return 0; +} + +static int get_value(struct parse_opt_ctx_t *p, + const struct option *opt, int flags) +{ + const char *s, *arg = NULL; + const int unset = flags & OPT_UNSET; + + if (unset && p->opt) + return opterror(opt, "takes no value", flags); + if (unset && (opt->flags & PARSE_OPT_NONEG)) + return opterror(opt, "isn't available", flags); + + if (!(flags & OPT_SHORT) && p->opt) { + switch (opt->type) { + case OPTION_CALLBACK: + if (!(opt->flags & PARSE_OPT_NOARG)) + break; + /* FALLTHROUGH */ + case OPTION_BOOLEAN: + case OPTION_INCR: + case OPTION_BIT: + case OPTION_SET_UINT: + case OPTION_SET_PTR: + return opterror(opt, "takes no value", flags); + case OPTION_END: + case OPTION_ARGUMENT: + case OPTION_GROUP: + case OPTION_STRING: + case OPTION_INTEGER: + case OPTION_UINTEGER: + case OPTION_LONG: + case OPTION_U64: + default: + break; + } + } + + switch (opt->type) { + case OPTION_BIT: + if (unset) + *(int *)opt->value &= ~opt->defval; + else + *(int *)opt->value |= opt->defval; + return 0; + + case OPTION_BOOLEAN: + *(bool *)opt->value = unset ? false : true; + return 0; + + case OPTION_INCR: + *(int *)opt->value = unset ? 0 : *(int *)opt->value + 1; + return 0; + + case OPTION_SET_UINT: + *(unsigned int *)opt->value = unset ? 0 : opt->defval; + return 0; + + case OPTION_SET_PTR: + *(void **)opt->value = unset ? NULL : (void *)opt->defval; + return 0; + + case OPTION_STRING: + if (unset) + *(const char **)opt->value = NULL; + else if (opt->flags & PARSE_OPT_OPTARG && !p->opt) + *(const char **)opt->value = (const char *)opt->defval; + else + return get_arg(p, opt, flags, (const char **)opt->value); + return 0; + + case OPTION_CALLBACK: + if (unset) + return (*opt->callback)(opt, NULL, 1) ? (-1) : 0; + if (opt->flags & PARSE_OPT_NOARG) + return (*opt->callback)(opt, NULL, 0) ? (-1) : 0; + if (opt->flags & PARSE_OPT_OPTARG && !p->opt) + return (*opt->callback)(opt, NULL, 0) ? (-1) : 0; + if (get_arg(p, opt, flags, &arg)) + return -1; + return (*opt->callback)(opt, arg, 0) ? (-1) : 0; + + case OPTION_INTEGER: + if (unset) { + *(int *)opt->value = 0; + return 0; + } + if (opt->flags & PARSE_OPT_OPTARG && !p->opt) { + *(int *)opt->value = opt->defval; + return 0; + } + if (get_arg(p, opt, flags, &arg)) + return -1; + *(int *)opt->value = strtol(arg, (char **)&s, 10); + if (*s) + return opterror(opt, "expects a numerical value", flags); + return 0; + + case OPTION_UINTEGER: + if (unset) { + *(unsigned int *)opt->value = 0; + return 0; + } + if (opt->flags & PARSE_OPT_OPTARG && !p->opt) { + *(unsigned int *)opt->value = opt->defval; + return 0; + } + if (get_arg(p, opt, flags, &arg)) + return -1; + *(unsigned int *)opt->value = strtol(arg, (char **)&s, 10); + if (*s) + return opterror(opt, "expects a numerical value", flags); + return 0; + + case OPTION_LONG: + if (unset) { + *(long *)opt->value = 0; + return 0; + } + if (opt->flags & PARSE_OPT_OPTARG && !p->opt) { + *(long *)opt->value = opt->defval; + return 0; + } + if (get_arg(p, opt, flags, &arg)) + return -1; + *(long *)opt->value = strtol(arg, (char **)&s, 10); + if (*s) + return opterror(opt, "expects a numerical value", flags); + return 0; + + case OPTION_U64: + if (unset) { + *(u64 *)opt->value = 0; + return 0; + } + if (opt->flags & PARSE_OPT_OPTARG && !p->opt) { + *(u64 *)opt->value = opt->defval; + return 0; + } + if (get_arg(p, opt, flags, &arg)) + return -1; + *(u64 *)opt->value = strtoull(arg, (char **)&s, 10); + if (*s) + return opterror(opt, "expects a numerical value", flags); + return 0; + + case OPTION_END: + case OPTION_ARGUMENT: + case OPTION_GROUP: + default: + die("should not happen, someone must be hit on the forehead"); + } +} + +static int parse_short_opt(struct parse_opt_ctx_t *p, const struct option *options) +{ + for (; options->type != OPTION_END; options++) { + if (options->short_name == *p->opt) { + p->opt = p->opt[1] ? p->opt + 1 : NULL; + return get_value(p, options, OPT_SHORT); + } + } + return -2; +} + +static int parse_long_opt(struct parse_opt_ctx_t *p, const char *arg, + const struct option *options) +{ + const char *arg_end = strchr(arg, '='); + const struct option *abbrev_option = NULL, *ambiguous_option = NULL; + int abbrev_flags = 0, ambiguous_flags = 0; + + if (!arg_end) + arg_end = arg + strlen(arg); + + for (; options->type != OPTION_END; options++) { + const char *rest; + int flags = 0; + + if (!options->long_name) + continue; + + rest = skip_prefix(arg, options->long_name); + if (options->type == OPTION_ARGUMENT) { + if (!rest) + continue; + if (*rest == '=') + return opterror(options, "takes no value", flags); + if (*rest) + continue; + p->out[p->cpidx++] = arg - 2; + return 0; + } + if (!rest) { + /* abbreviated? */ + if (!strncmp(options->long_name, arg, arg_end - arg)) { +is_abbreviated: + if (abbrev_option) { + /* + * If this is abbreviated, it is + * ambiguous. So when there is no + * exact match later, we need to + * error out. + */ + ambiguous_option = abbrev_option; + ambiguous_flags = abbrev_flags; + } + if (!(flags & OPT_UNSET) && *arg_end) + p->opt = arg_end + 1; + abbrev_option = options; + abbrev_flags = flags; + continue; + } + /* negated and abbreviated very much? */ + if (!prefixcmp("no-", arg)) { + flags |= OPT_UNSET; + goto is_abbreviated; + } + /* negated? */ + if (strncmp(arg, "no-", 3)) + continue; + flags |= OPT_UNSET; + rest = skip_prefix(arg + 3, options->long_name); + /* abbreviated and negated? */ + if (!rest && !prefixcmp(options->long_name, arg + 3)) + goto is_abbreviated; + if (!rest) + continue; + } + if (*rest) { + if (*rest != '=') + continue; + p->opt = rest + 1; + } + return get_value(p, options, flags); + } + + if (ambiguous_option) + return error("Ambiguous option: %s " + "(could be --%s%s or --%s%s)", + arg, + (ambiguous_flags & OPT_UNSET) ? "no-" : "", + ambiguous_option->long_name, + (abbrev_flags & OPT_UNSET) ? "no-" : "", + abbrev_option->long_name); + if (abbrev_option) + return get_value(p, abbrev_option, abbrev_flags); + return -2; +} + +static void check_typos(const char *arg, const struct option *options) +{ + if (strlen(arg) < 3) + return; + + if (!prefixcmp(arg, "no-")) { + error ("did you mean `--%s` (with two dashes ?)", arg); + exit(129); + } + + for (; options->type != OPTION_END; options++) { + if (!options->long_name) + continue; + if (!prefixcmp(options->long_name, arg)) { + error ("did you mean `--%s` (with two dashes ?)", arg); + exit(129); + } + } +} + +void parse_options_start(struct parse_opt_ctx_t *ctx, + int argc, const char **argv, int flags) +{ + memset(ctx, 0, sizeof(*ctx)); + ctx->argc = argc - 1; + ctx->argv = argv + 1; + ctx->out = argv; + ctx->cpidx = ((flags & PARSE_OPT_KEEP_ARGV0) != 0); + ctx->flags = flags; + if ((flags & PARSE_OPT_KEEP_UNKNOWN) && + (flags & PARSE_OPT_STOP_AT_NON_OPTION)) + die("STOP_AT_NON_OPTION and KEEP_UNKNOWN don't go together"); +} + +static int usage_with_options_internal(const char * const *, + const struct option *, int); + +int parse_options_step(struct parse_opt_ctx_t *ctx, + const struct option *options, + const char * const usagestr[]) +{ + int internal_help = !(ctx->flags & PARSE_OPT_NO_INTERNAL_HELP); + + /* we must reset ->opt, unknown short option leave it dangling */ + ctx->opt = NULL; + + for (; ctx->argc; ctx->argc--, ctx->argv++) { + const char *arg = ctx->argv[0]; + + if (*arg != '-' || !arg[1]) { + if (ctx->flags & PARSE_OPT_STOP_AT_NON_OPTION) + break; + ctx->out[ctx->cpidx++] = ctx->argv[0]; + continue; + } + + if (arg[1] != '-') { + ctx->opt = arg + 1; + if (internal_help && *ctx->opt == 'h') + return parse_options_usage(usagestr, options); + switch (parse_short_opt(ctx, options)) { + case -1: + return parse_options_usage(usagestr, options); + case -2: + goto unknown; + default: + break; + } + if (ctx->opt) + check_typos(arg + 1, options); + while (ctx->opt) { + if (internal_help && *ctx->opt == 'h') + return parse_options_usage(usagestr, options); + switch (parse_short_opt(ctx, options)) { + case -1: + return parse_options_usage(usagestr, options); + case -2: + /* fake a short option thing to hide the fact that we may have + * started to parse aggregated stuff + * + * This is leaky, too bad. + */ + ctx->argv[0] = strdup(ctx->opt - 1); + *(char *)ctx->argv[0] = '-'; + goto unknown; + default: + break; + } + } + continue; + } + + if (!arg[2]) { /* "--" */ + if (!(ctx->flags & PARSE_OPT_KEEP_DASHDASH)) { + ctx->argc--; + ctx->argv++; + } + break; + } + + if (internal_help && !strcmp(arg + 2, "help-all")) + return usage_with_options_internal(usagestr, options, 1); + if (internal_help && !strcmp(arg + 2, "help")) + return parse_options_usage(usagestr, options); + switch (parse_long_opt(ctx, arg + 2, options)) { + case -1: + return parse_options_usage(usagestr, options); + case -2: + goto unknown; + default: + break; + } + continue; +unknown: + if (!(ctx->flags & PARSE_OPT_KEEP_UNKNOWN)) + return PARSE_OPT_UNKNOWN; + ctx->out[ctx->cpidx++] = ctx->argv[0]; + ctx->opt = NULL; + } + return PARSE_OPT_DONE; +} + +int parse_options_end(struct parse_opt_ctx_t *ctx) +{ + memmove(ctx->out + ctx->cpidx, ctx->argv, ctx->argc * sizeof(*ctx->out)); + ctx->out[ctx->cpidx + ctx->argc] = NULL; + return ctx->cpidx + ctx->argc; +} + +int parse_options(int argc, const char **argv, const struct option *options, + const char * const usagestr[], int flags) +{ + struct parse_opt_ctx_t ctx; + + parse_options_start(&ctx, argc, argv, flags); + switch (parse_options_step(&ctx, options, usagestr)) { + case PARSE_OPT_HELP: + exit(129); + case PARSE_OPT_DONE: + break; + default: /* PARSE_OPT_UNKNOWN */ + if (ctx.argv[0][1] == '-') { + error("unknown option `%s'", ctx.argv[0] + 2); + } else { + error("unknown switch `%c'", *ctx.opt); + } + usage_with_options(usagestr, options); + } + + return parse_options_end(&ctx); +} + +#define USAGE_OPTS_WIDTH 24 +#define USAGE_GAP 2 + +int usage_with_options_internal(const char * const *usagestr, + const struct option *opts, int full) +{ + if (!usagestr) + return PARSE_OPT_HELP; + + fprintf(stderr, "\n usage: %s\n", *usagestr++); + while (*usagestr && **usagestr) + fprintf(stderr, " or: %s\n", *usagestr++); + while (*usagestr) { + fprintf(stderr, "%s%s\n", + **usagestr ? " " : "", + *usagestr); + usagestr++; + } + + if (opts->type != OPTION_GROUP) + fputc('\n', stderr); + + for (; opts->type != OPTION_END; opts++) { + size_t pos; + int pad; + + if (opts->type == OPTION_GROUP) { + fputc('\n', stderr); + if (*opts->help) + fprintf(stderr, "%s\n", opts->help); + continue; + } + if (!full && (opts->flags & PARSE_OPT_HIDDEN)) + continue; + + pos = fprintf(stderr, " "); + if (opts->short_name) + pos += fprintf(stderr, "-%c", opts->short_name); + else + pos += fprintf(stderr, " "); + + if (opts->long_name && opts->short_name) + pos += fprintf(stderr, ", "); + if (opts->long_name) + pos += fprintf(stderr, "--%s", opts->long_name); + + switch (opts->type) { + case OPTION_ARGUMENT: + break; + case OPTION_LONG: + case OPTION_U64: + case OPTION_INTEGER: + case OPTION_UINTEGER: + if (opts->flags & PARSE_OPT_OPTARG) + if (opts->long_name) + pos += fprintf(stderr, "[=<n>]"); + else + pos += fprintf(stderr, "[<n>]"); + else + pos += fprintf(stderr, " <n>"); + break; + case OPTION_CALLBACK: + if (opts->flags & PARSE_OPT_NOARG) + break; + /* FALLTHROUGH */ + case OPTION_STRING: + if (opts->argh) { + if (opts->flags & PARSE_OPT_OPTARG) + if (opts->long_name) + pos += fprintf(stderr, "[=<%s>]", opts->argh); + else + pos += fprintf(stderr, "[<%s>]", opts->argh); + else + pos += fprintf(stderr, " <%s>", opts->argh); + } else { + if (opts->flags & PARSE_OPT_OPTARG) + if (opts->long_name) + pos += fprintf(stderr, "[=...]"); + else + pos += fprintf(stderr, "[...]"); + else + pos += fprintf(stderr, " ..."); + } + break; + default: /* OPTION_{BIT,BOOLEAN,SET_UINT,SET_PTR} */ + case OPTION_END: + case OPTION_GROUP: + case OPTION_BIT: + case OPTION_BOOLEAN: + case OPTION_INCR: + case OPTION_SET_UINT: + case OPTION_SET_PTR: + break; + } + + if (pos <= USAGE_OPTS_WIDTH) + pad = USAGE_OPTS_WIDTH - pos; + else { + fputc('\n', stderr); + pad = USAGE_OPTS_WIDTH; + } + fprintf(stderr, "%*s%s\n", pad + USAGE_GAP, "", opts->help); + } + fputc('\n', stderr); + + return PARSE_OPT_HELP; +} + +void usage_with_options(const char * const *usagestr, + const struct option *opts) +{ + exit_browser(false); + usage_with_options_internal(usagestr, opts, 0); + exit(129); +} + +int parse_options_usage(const char * const *usagestr, + const struct option *opts) +{ + return usage_with_options_internal(usagestr, opts, 0); +} + + +int parse_opt_verbosity_cb(const struct option *opt, const char *arg __used, + int unset) +{ + int *target = opt->value; + + if (unset) + /* --no-quiet, --no-verbose */ + *target = 0; + else if (opt->short_name == 'v') { + if (*target >= 0) + (*target)++; + else + *target = 1; + } else { + if (*target <= 0) + (*target)--; + else + *target = -1; + } + return 0; +} diff --git a/tools/perf/util/parse-options.h b/tools/perf/util/parse-options.h new file mode 100644 index 00000000..abc31a1d --- /dev/null +++ b/tools/perf/util/parse-options.h @@ -0,0 +1,192 @@ +#ifndef __PERF_PARSE_OPTIONS_H +#define __PERF_PARSE_OPTIONS_H + +#include <linux/kernel.h> +#include <stdbool.h> + +enum parse_opt_type { + /* special types */ + OPTION_END, + OPTION_ARGUMENT, + OPTION_GROUP, + /* options with no arguments */ + OPTION_BIT, + OPTION_BOOLEAN, + OPTION_INCR, + OPTION_SET_UINT, + OPTION_SET_PTR, + /* options with arguments (usually) */ + OPTION_STRING, + OPTION_INTEGER, + OPTION_LONG, + OPTION_CALLBACK, + OPTION_U64, + OPTION_UINTEGER, +}; + +enum parse_opt_flags { + PARSE_OPT_KEEP_DASHDASH = 1, + PARSE_OPT_STOP_AT_NON_OPTION = 2, + PARSE_OPT_KEEP_ARGV0 = 4, + PARSE_OPT_KEEP_UNKNOWN = 8, + PARSE_OPT_NO_INTERNAL_HELP = 16, +}; + +enum parse_opt_option_flags { + PARSE_OPT_OPTARG = 1, + PARSE_OPT_NOARG = 2, + PARSE_OPT_NONEG = 4, + PARSE_OPT_HIDDEN = 8, + PARSE_OPT_LASTARG_DEFAULT = 16, +}; + +struct option; +typedef int parse_opt_cb(const struct option *, const char *arg, int unset); + +/* + * `type`:: + * holds the type of the option, you must have an OPTION_END last in your + * array. + * + * `short_name`:: + * the character to use as a short option name, '\0' if none. + * + * `long_name`:: + * the long option name, without the leading dashes, NULL if none. + * + * `value`:: + * stores pointers to the values to be filled. + * + * `argh`:: + * token to explain the kind of argument this option wants. Keep it + * homogenous across the repository. + * + * `help`:: + * the short help associated to what the option does. + * Must never be NULL (except for OPTION_END). + * OPTION_GROUP uses this pointer to store the group header. + * + * `flags`:: + * mask of parse_opt_option_flags. + * PARSE_OPT_OPTARG: says that the argument is optionnal (not for BOOLEANs) + * PARSE_OPT_NOARG: says that this option takes no argument, for CALLBACKs + * PARSE_OPT_NONEG: says that this option cannot be negated + * PARSE_OPT_HIDDEN this option is skipped in the default usage, showed in + * the long one. + * + * `callback`:: + * pointer to the callback to use for OPTION_CALLBACK. + * + * `defval`:: + * default value to fill (*->value) with for PARSE_OPT_OPTARG. + * OPTION_{BIT,SET_UINT,SET_PTR} store the {mask,integer,pointer} to put in + * the value when met. + * CALLBACKS can use it like they want. + */ +struct option { + enum parse_opt_type type; + int short_name; + const char *long_name; + void *value; + const char *argh; + const char *help; + + int flags; + parse_opt_cb *callback; + intptr_t defval; +}; + +#define check_vtype(v, type) ( BUILD_BUG_ON_ZERO(!__builtin_types_compatible_p(typeof(v), type)) + v ) + +#define OPT_END() { .type = OPTION_END } +#define OPT_ARGUMENT(l, h) { .type = OPTION_ARGUMENT, .long_name = (l), .help = (h) } +#define OPT_GROUP(h) { .type = OPTION_GROUP, .help = (h) } +#define OPT_BIT(s, l, v, h, b) { .type = OPTION_BIT, .short_name = (s), .long_name = (l), .value = check_vtype(v, int *), .help = (h), .defval = (b) } +#define OPT_BOOLEAN(s, l, v, h) { .type = OPTION_BOOLEAN, .short_name = (s), .long_name = (l), .value = check_vtype(v, bool *), .help = (h) } +#define OPT_INCR(s, l, v, h) { .type = OPTION_INCR, .short_name = (s), .long_name = (l), .value = check_vtype(v, int *), .help = (h) } +#define OPT_SET_UINT(s, l, v, h, i) { .type = OPTION_SET_UINT, .short_name = (s), .long_name = (l), .value = check_vtype(v, unsigned int *), .help = (h), .defval = (i) } +#define OPT_SET_PTR(s, l, v, h, p) { .type = OPTION_SET_PTR, .short_name = (s), .long_name = (l), .value = (v), .help = (h), .defval = (p) } +#define OPT_INTEGER(s, l, v, h) { .type = OPTION_INTEGER, .short_name = (s), .long_name = (l), .value = check_vtype(v, int *), .help = (h) } +#define OPT_UINTEGER(s, l, v, h) { .type = OPTION_UINTEGER, .short_name = (s), .long_name = (l), .value = check_vtype(v, unsigned int *), .help = (h) } +#define OPT_LONG(s, l, v, h) { .type = OPTION_LONG, .short_name = (s), .long_name = (l), .value = check_vtype(v, long *), .help = (h) } +#define OPT_U64(s, l, v, h) { .type = OPTION_U64, .short_name = (s), .long_name = (l), .value = check_vtype(v, u64 *), .help = (h) } +#define OPT_STRING(s, l, v, a, h) { .type = OPTION_STRING, .short_name = (s), .long_name = (l), .value = check_vtype(v, const char **), (a), .help = (h) } +#define OPT_DATE(s, l, v, h) \ + { .type = OPTION_CALLBACK, .short_name = (s), .long_name = (l), .value = (v), .argh = "time", .help = (h), .callback = parse_opt_approxidate_cb } +#define OPT_CALLBACK(s, l, v, a, h, f) \ + { .type = OPTION_CALLBACK, .short_name = (s), .long_name = (l), .value = (v), (a), .help = (h), .callback = (f) } +#define OPT_CALLBACK_NOOPT(s, l, v, a, h, f) \ + { .type = OPTION_CALLBACK, .short_name = (s), .long_name = (l), .value = (v), (a), .help = (h), .callback = (f), .flags = PARSE_OPT_NOARG } +#define OPT_CALLBACK_DEFAULT(s, l, v, a, h, f, d) \ + { .type = OPTION_CALLBACK, .short_name = (s), .long_name = (l), .value = (v), (a), .help = (h), .callback = (f), .defval = (intptr_t)d, .flags = PARSE_OPT_LASTARG_DEFAULT } +#define OPT_CALLBACK_DEFAULT_NOOPT(s, l, v, a, h, f, d) \ + { .type = OPTION_CALLBACK, .short_name = (s), .long_name = (l),\ + .value = (v), (a), .help = (h), .callback = (f), .defval = (intptr_t)d,\ + .flags = PARSE_OPT_LASTARG_DEFAULT | PARSE_OPT_NOARG} + +/* parse_options() will filter out the processed options and leave the + * non-option argments in argv[]. + * Returns the number of arguments left in argv[]. + */ +extern int parse_options(int argc, const char **argv, + const struct option *options, + const char * const usagestr[], int flags); + +extern NORETURN void usage_with_options(const char * const *usagestr, + const struct option *options); + +/*----- incremantal advanced APIs -----*/ + +enum { + PARSE_OPT_HELP = -1, + PARSE_OPT_DONE, + PARSE_OPT_UNKNOWN, +}; + +/* + * It's okay for the caller to consume argv/argc in the usual way. + * Other fields of that structure are private to parse-options and should not + * be modified in any way. + */ +struct parse_opt_ctx_t { + const char **argv; + const char **out; + int argc, cpidx; + const char *opt; + int flags; +}; + +extern int parse_options_usage(const char * const *usagestr, + const struct option *opts); + +extern void parse_options_start(struct parse_opt_ctx_t *ctx, + int argc, const char **argv, int flags); + +extern int parse_options_step(struct parse_opt_ctx_t *ctx, + const struct option *options, + const char * const usagestr[]); + +extern int parse_options_end(struct parse_opt_ctx_t *ctx); + + +/*----- some often used options -----*/ +extern int parse_opt_abbrev_cb(const struct option *, const char *, int); +extern int parse_opt_approxidate_cb(const struct option *, const char *, int); +extern int parse_opt_verbosity_cb(const struct option *, const char *, int); + +#define OPT__VERBOSE(var) OPT_BOOLEAN('v', "verbose", (var), "be verbose") +#define OPT__QUIET(var) OPT_BOOLEAN('q', "quiet", (var), "be quiet") +#define OPT__VERBOSITY(var) \ + { OPTION_CALLBACK, 'v', "verbose", (var), NULL, "be more verbose", \ + PARSE_OPT_NOARG, &parse_opt_verbosity_cb, 0 }, \ + { OPTION_CALLBACK, 'q', "quiet", (var), NULL, "be more quiet", \ + PARSE_OPT_NOARG, &parse_opt_verbosity_cb, 0 } +#define OPT__DRY_RUN(var) OPT_BOOLEAN('n', "dry-run", (var), "dry run") +#define OPT__ABBREV(var) \ + { OPTION_CALLBACK, 0, "abbrev", (var), "n", \ + "use <n> digits to display SHA-1s", \ + PARSE_OPT_OPTARG, &parse_opt_abbrev_cb, 0 } + +extern const char *parse_options_fix_filename(const char *prefix, const char *file); + +#endif /* __PERF_PARSE_OPTIONS_H */ diff --git a/tools/perf/util/path.c b/tools/perf/util/path.c new file mode 100644 index 00000000..bd749771 --- /dev/null +++ b/tools/perf/util/path.c @@ -0,0 +1,157 @@ +/* + * I'm tired of doing "vsnprintf()" etc just to open a + * file, so here's a "return static buffer with printf" + * interface for paths. + * + * It's obviously not thread-safe. Sue me. But it's quite + * useful for doing things like + * + * f = open(mkpath("%s/%s.perf", base, name), O_RDONLY); + * + * which is what it's designed for. + */ +#include "cache.h" + +static char bad_path[] = "/bad-path/"; +/* + * Two hacks: + */ + +static const char *get_perf_dir(void) +{ + return "."; +} + +#ifdef NO_STRLCPY +size_t strlcpy(char *dest, const char *src, size_t size) +{ + size_t ret = strlen(src); + + if (size) { + size_t len = (ret >= size) ? size - 1 : ret; + memcpy(dest, src, len); + dest[len] = '\0'; + } + return ret; +} +#endif + +static char *get_pathname(void) +{ + static char pathname_array[4][PATH_MAX]; + static int idx; + + return pathname_array[3 & ++idx]; +} + +static char *cleanup_path(char *path) +{ + /* Clean it up */ + if (!memcmp(path, "./", 2)) { + path += 2; + while (*path == '/') + path++; + } + return path; +} + +static char *perf_vsnpath(char *buf, size_t n, const char *fmt, va_list args) +{ + const char *perf_dir = get_perf_dir(); + size_t len; + + len = strlen(perf_dir); + if (n < len + 1) + goto bad; + memcpy(buf, perf_dir, len); + if (len && !is_dir_sep(perf_dir[len-1])) + buf[len++] = '/'; + len += vsnprintf(buf + len, n - len, fmt, args); + if (len >= n) + goto bad; + return cleanup_path(buf); +bad: + strlcpy(buf, bad_path, n); + return buf; +} + +char *perf_pathdup(const char *fmt, ...) +{ + char path[PATH_MAX]; + va_list args; + va_start(args, fmt); + (void)perf_vsnpath(path, sizeof(path), fmt, args); + va_end(args); + return xstrdup(path); +} + +char *mkpath(const char *fmt, ...) +{ + va_list args; + unsigned len; + char *pathname = get_pathname(); + + va_start(args, fmt); + len = vsnprintf(pathname, PATH_MAX, fmt, args); + va_end(args); + if (len >= PATH_MAX) + return bad_path; + return cleanup_path(pathname); +} + +char *perf_path(const char *fmt, ...) +{ + const char *perf_dir = get_perf_dir(); + char *pathname = get_pathname(); + va_list args; + unsigned len; + + len = strlen(perf_dir); + if (len > PATH_MAX-100) + return bad_path; + memcpy(pathname, perf_dir, len); + if (len && perf_dir[len-1] != '/') + pathname[len++] = '/'; + va_start(args, fmt); + len += vsnprintf(pathname + len, PATH_MAX - len, fmt, args); + va_end(args); + if (len >= PATH_MAX) + return bad_path; + return cleanup_path(pathname); +} + +/* strip arbitrary amount of directory separators at end of path */ +static inline int chomp_trailing_dir_sep(const char *path, int len) +{ + while (len && is_dir_sep(path[len - 1])) + len--; + return len; +} + +/* + * If path ends with suffix (complete path components), returns the + * part before suffix (sans trailing directory separators). + * Otherwise returns NULL. + */ +char *strip_path_suffix(const char *path, const char *suffix) +{ + int path_len = strlen(path), suffix_len = strlen(suffix); + + while (suffix_len) { + if (!path_len) + return NULL; + + if (is_dir_sep(path[path_len - 1])) { + if (!is_dir_sep(suffix[suffix_len - 1])) + return NULL; + path_len = chomp_trailing_dir_sep(path, path_len); + suffix_len = chomp_trailing_dir_sep(suffix, suffix_len); + } + else if (path[--path_len] != suffix[--suffix_len]) + return NULL; + } + + if (path_len && !is_dir_sep(path[path_len - 1])) + return NULL; + return strndup(path, chomp_trailing_dir_sep(path, path_len)); +} diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c new file mode 100644 index 00000000..8c50da86 --- /dev/null +++ b/tools/perf/util/probe-event.c @@ -0,0 +1,2007 @@ +/* + * probe-event.c : perf-probe definition to probe_events format converter + * + * Written by Masami Hiramatsu <mhiramat@redhat.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + */ + +#define _GNU_SOURCE +#include <sys/utsname.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <fcntl.h> +#include <errno.h> +#include <stdio.h> +#include <unistd.h> +#include <stdlib.h> +#include <string.h> +#include <stdarg.h> +#include <limits.h> +#include <elf.h> + +#undef _GNU_SOURCE +#include "util.h" +#include "event.h" +#include "string.h" +#include "strlist.h" +#include "debug.h" +#include "cache.h" +#include "color.h" +#include "symbol.h" +#include "thread.h" +#include "debugfs.h" +#include "trace-event.h" /* For __unused */ +#include "probe-event.h" +#include "probe-finder.h" + +#define MAX_CMDLEN 256 +#define MAX_PROBE_ARGS 128 +#define PERFPROBE_GROUP "probe" + +bool probe_event_dry_run; /* Dry run flag */ + +#define semantic_error(msg ...) pr_err("Semantic error :" msg) + +/* If there is no space to write, returns -E2BIG. */ +static int e_snprintf(char *str, size_t size, const char *format, ...) + __attribute__((format(printf, 3, 4))); + +static int e_snprintf(char *str, size_t size, const char *format, ...) +{ + int ret; + va_list ap; + va_start(ap, format); + ret = vsnprintf(str, size, format, ap); + va_end(ap); + if (ret >= (int)size) + ret = -E2BIG; + return ret; +} + +static char *synthesize_perf_probe_point(struct perf_probe_point *pp); +static struct machine machine; + +/* Initialize symbol maps and path of vmlinux/modules */ +static int init_vmlinux(void) +{ + int ret; + + symbol_conf.sort_by_name = true; + if (symbol_conf.vmlinux_name == NULL) + symbol_conf.try_vmlinux_path = true; + else + pr_debug("Use vmlinux: %s\n", symbol_conf.vmlinux_name); + ret = symbol__init(); + if (ret < 0) { + pr_debug("Failed to init symbol map.\n"); + goto out; + } + + ret = machine__init(&machine, "", HOST_KERNEL_ID); + if (ret < 0) + goto out; + + if (machine__create_kernel_maps(&machine) < 0) { + pr_debug("machine__create_kernel_maps() failed.\n"); + goto out; + } +out: + if (ret < 0) + pr_warning("Failed to init vmlinux path.\n"); + return ret; +} + +static struct symbol *__find_kernel_function_by_name(const char *name, + struct map **mapp) +{ + return machine__find_kernel_function_by_name(&machine, name, mapp, + NULL); +} + +static struct map *kernel_get_module_map(const char *module) +{ + struct rb_node *nd; + struct map_groups *grp = &machine.kmaps; + + if (!module) + module = "kernel"; + + for (nd = rb_first(&grp->maps[MAP__FUNCTION]); nd; nd = rb_next(nd)) { + struct map *pos = rb_entry(nd, struct map, rb_node); + if (strncmp(pos->dso->short_name + 1, module, + pos->dso->short_name_len - 2) == 0) { + return pos; + } + } + return NULL; +} + +static struct dso *kernel_get_module_dso(const char *module) +{ + struct dso *dso; + struct map *map; + const char *vmlinux_name; + + if (module) { + list_for_each_entry(dso, &machine.kernel_dsos, node) { + if (strncmp(dso->short_name + 1, module, + dso->short_name_len - 2) == 0) + goto found; + } + pr_debug("Failed to find module %s.\n", module); + return NULL; + } + + map = machine.vmlinux_maps[MAP__FUNCTION]; + dso = map->dso; + + vmlinux_name = symbol_conf.vmlinux_name; + if (vmlinux_name) { + if (dso__load_vmlinux(dso, map, vmlinux_name, NULL) <= 0) + return NULL; + } else { + if (dso__load_vmlinux_path(dso, map, NULL) <= 0) { + pr_debug("Failed to load kernel map.\n"); + return NULL; + } + } +found: + return dso; +} + +const char *kernel_get_module_path(const char *module) +{ + struct dso *dso = kernel_get_module_dso(module); + return (dso) ? dso->long_name : NULL; +} + +#ifdef DWARF_SUPPORT +static int open_vmlinux(const char *module) +{ + const char *path = kernel_get_module_path(module); + if (!path) { + pr_err("Failed to find path of %s module.\n", + module ?: "kernel"); + return -ENOENT; + } + pr_debug("Try to open %s\n", path); + return open(path, O_RDONLY); +} + +/* + * Convert trace point to probe point with debuginfo + * Currently only handles kprobes. + */ +static int kprobe_convert_to_perf_probe(struct probe_trace_point *tp, + struct perf_probe_point *pp) +{ + struct symbol *sym; + struct map *map; + u64 addr; + int ret = -ENOENT; + + sym = __find_kernel_function_by_name(tp->symbol, &map); + if (sym) { + addr = map->unmap_ip(map, sym->start + tp->offset); + pr_debug("try to find %s+%ld@%" PRIx64 "\n", tp->symbol, + tp->offset, addr); + ret = find_perf_probe_point((unsigned long)addr, pp); + } + if (ret <= 0) { + pr_debug("Failed to find corresponding probes from " + "debuginfo. Use kprobe event information.\n"); + pp->function = strdup(tp->symbol); + if (pp->function == NULL) + return -ENOMEM; + pp->offset = tp->offset; + } + pp->retprobe = tp->retprobe; + + return 0; +} + +/* Try to find perf_probe_event with debuginfo */ +static int try_to_find_probe_trace_events(struct perf_probe_event *pev, + struct probe_trace_event **tevs, + int max_tevs, const char *module) +{ + bool need_dwarf = perf_probe_event_need_dwarf(pev); + int fd, ntevs; + + fd = open_vmlinux(module); + if (fd < 0) { + if (need_dwarf) { + pr_warning("Failed to open debuginfo file.\n"); + return fd; + } + pr_debug("Could not open vmlinux. Try to use symbols.\n"); + return 0; + } + + /* Searching trace events corresponding to probe event */ + ntevs = find_probe_trace_events(fd, pev, tevs, max_tevs); + + if (ntevs > 0) { /* Succeeded to find trace events */ + pr_debug("find %d probe_trace_events.\n", ntevs); + return ntevs; + } + + if (ntevs == 0) { /* No error but failed to find probe point. */ + pr_warning("Probe point '%s' not found.\n", + synthesize_perf_probe_point(&pev->point)); + return -ENOENT; + } + /* Error path : ntevs < 0 */ + pr_debug("An error occurred in debuginfo analysis (%d).\n", ntevs); + if (ntevs == -EBADF) { + pr_warning("Warning: No dwarf info found in the vmlinux - " + "please rebuild kernel with CONFIG_DEBUG_INFO=y.\n"); + if (!need_dwarf) { + pr_debug("Trying to use symbols.\n"); + return 0; + } + } + return ntevs; +} + +/* + * Find a src file from a DWARF tag path. Prepend optional source path prefix + * and chop off leading directories that do not exist. Result is passed back as + * a newly allocated path on success. + * Return 0 if file was found and readable, -errno otherwise. + */ +static int get_real_path(const char *raw_path, const char *comp_dir, + char **new_path) +{ + const char *prefix = symbol_conf.source_prefix; + + if (!prefix) { + if (raw_path[0] != '/' && comp_dir) + /* If not an absolute path, try to use comp_dir */ + prefix = comp_dir; + else { + if (access(raw_path, R_OK) == 0) { + *new_path = strdup(raw_path); + return 0; + } else + return -errno; + } + } + + *new_path = malloc((strlen(prefix) + strlen(raw_path) + 2)); + if (!*new_path) + return -ENOMEM; + + for (;;) { + sprintf(*new_path, "%s/%s", prefix, raw_path); + + if (access(*new_path, R_OK) == 0) + return 0; + + if (!symbol_conf.source_prefix) + /* In case of searching comp_dir, don't retry */ + return -errno; + + switch (errno) { + case ENAMETOOLONG: + case ENOENT: + case EROFS: + case EFAULT: + raw_path = strchr(++raw_path, '/'); + if (!raw_path) { + free(*new_path); + *new_path = NULL; + return -ENOENT; + } + continue; + + default: + free(*new_path); + *new_path = NULL; + return -errno; + } + } +} + +#define LINEBUF_SIZE 256 +#define NR_ADDITIONAL_LINES 2 + +static int __show_one_line(FILE *fp, int l, bool skip, bool show_num) +{ + char buf[LINEBUF_SIZE]; + const char *color = show_num ? "" : PERF_COLOR_BLUE; + const char *prefix = NULL; + + do { + if (fgets(buf, LINEBUF_SIZE, fp) == NULL) + goto error; + if (skip) + continue; + if (!prefix) { + prefix = show_num ? "%7d " : " "; + color_fprintf(stdout, color, prefix, l); + } + color_fprintf(stdout, color, "%s", buf); + + } while (strchr(buf, '\n') == NULL); + + return 1; +error: + if (ferror(fp)) { + pr_warning("File read error: %s\n", strerror(errno)); + return -1; + } + return 0; +} + +static int _show_one_line(FILE *fp, int l, bool skip, bool show_num) +{ + int rv = __show_one_line(fp, l, skip, show_num); + if (rv == 0) { + pr_warning("Source file is shorter than expected.\n"); + rv = -1; + } + return rv; +} + +#define show_one_line_with_num(f,l) _show_one_line(f,l,false,true) +#define show_one_line(f,l) _show_one_line(f,l,false,false) +#define skip_one_line(f,l) _show_one_line(f,l,true,false) +#define show_one_line_or_eof(f,l) __show_one_line(f,l,false,false) + +/* + * Show line-range always requires debuginfo to find source file and + * line number. + */ +int show_line_range(struct line_range *lr, const char *module) +{ + int l = 1; + struct line_node *ln; + FILE *fp; + int fd, ret; + char *tmp; + + /* Search a line range */ + ret = init_vmlinux(); + if (ret < 0) + return ret; + + fd = open_vmlinux(module); + if (fd < 0) { + pr_warning("Failed to open debuginfo file.\n"); + return fd; + } + + ret = find_line_range(fd, lr); + if (ret == 0) { + pr_warning("Specified source line is not found.\n"); + return -ENOENT; + } else if (ret < 0) { + pr_warning("Debuginfo analysis failed. (%d)\n", ret); + return ret; + } + + /* Convert source file path */ + tmp = lr->path; + ret = get_real_path(tmp, lr->comp_dir, &lr->path); + free(tmp); /* Free old path */ + if (ret < 0) { + pr_warning("Failed to find source file. (%d)\n", ret); + return ret; + } + + setup_pager(); + + if (lr->function) + fprintf(stdout, "<%s@%s:%d>\n", lr->function, lr->path, + lr->start - lr->offset); + else + fprintf(stdout, "<%s:%d>\n", lr->path, lr->start); + + fp = fopen(lr->path, "r"); + if (fp == NULL) { + pr_warning("Failed to open %s: %s\n", lr->path, + strerror(errno)); + return -errno; + } + /* Skip to starting line number */ + while (l < lr->start) { + ret = skip_one_line(fp, l++); + if (ret < 0) + goto end; + } + + list_for_each_entry(ln, &lr->line_list, list) { + for (; ln->line > l; l++) { + ret = show_one_line(fp, l - lr->offset); + if (ret < 0) + goto end; + } + ret = show_one_line_with_num(fp, l++ - lr->offset); + if (ret < 0) + goto end; + } + + if (lr->end == INT_MAX) + lr->end = l + NR_ADDITIONAL_LINES; + while (l <= lr->end) { + ret = show_one_line_or_eof(fp, l++ - lr->offset); + if (ret <= 0) + break; + } +end: + fclose(fp); + return ret; +} + +static int show_available_vars_at(int fd, struct perf_probe_event *pev, + int max_vls, struct strfilter *_filter, + bool externs) +{ + char *buf; + int ret, i, nvars; + struct str_node *node; + struct variable_list *vls = NULL, *vl; + const char *var; + + buf = synthesize_perf_probe_point(&pev->point); + if (!buf) + return -EINVAL; + pr_debug("Searching variables at %s\n", buf); + + ret = find_available_vars_at(fd, pev, &vls, max_vls, externs); + if (ret <= 0) { + pr_err("Failed to find variables at %s (%d)\n", buf, ret); + goto end; + } + /* Some variables are found */ + fprintf(stdout, "Available variables at %s\n", buf); + for (i = 0; i < ret; i++) { + vl = &vls[i]; + /* + * A probe point might be converted to + * several trace points. + */ + fprintf(stdout, "\t@<%s+%lu>\n", vl->point.symbol, + vl->point.offset); + free(vl->point.symbol); + nvars = 0; + if (vl->vars) { + strlist__for_each(node, vl->vars) { + var = strchr(node->s, '\t') + 1; + if (strfilter__compare(_filter, var)) { + fprintf(stdout, "\t\t%s\n", node->s); + nvars++; + } + } + strlist__delete(vl->vars); + } + if (nvars == 0) + fprintf(stdout, "\t\t(No matched variables)\n"); + } + free(vls); +end: + free(buf); + return ret; +} + +/* Show available variables on given probe point */ +int show_available_vars(struct perf_probe_event *pevs, int npevs, + int max_vls, const char *module, + struct strfilter *_filter, bool externs) +{ + int i, fd, ret = 0; + + ret = init_vmlinux(); + if (ret < 0) + return ret; + + setup_pager(); + + for (i = 0; i < npevs && ret >= 0; i++) { + fd = open_vmlinux(module); + if (fd < 0) { + pr_warning("Failed to open debug information file.\n"); + ret = fd; + break; + } + ret = show_available_vars_at(fd, &pevs[i], max_vls, _filter, + externs); + } + return ret; +} + +#else /* !DWARF_SUPPORT */ + +static int kprobe_convert_to_perf_probe(struct probe_trace_point *tp, + struct perf_probe_point *pp) +{ + struct symbol *sym; + + sym = __find_kernel_function_by_name(tp->symbol, NULL); + if (!sym) { + pr_err("Failed to find symbol %s in kernel.\n", tp->symbol); + return -ENOENT; + } + pp->function = strdup(tp->symbol); + if (pp->function == NULL) + return -ENOMEM; + pp->offset = tp->offset; + pp->retprobe = tp->retprobe; + + return 0; +} + +static int try_to_find_probe_trace_events(struct perf_probe_event *pev, + struct probe_trace_event **tevs __unused, + int max_tevs __unused, const char *mod __unused) +{ + if (perf_probe_event_need_dwarf(pev)) { + pr_warning("Debuginfo-analysis is not supported.\n"); + return -ENOSYS; + } + return 0; +} + +int show_line_range(struct line_range *lr __unused, const char *module __unused) +{ + pr_warning("Debuginfo-analysis is not supported.\n"); + return -ENOSYS; +} + +int show_available_vars(struct perf_probe_event *pevs __unused, + int npevs __unused, int max_vls __unused, + const char *module __unused, + struct strfilter *filter __unused, + bool externs __unused) +{ + pr_warning("Debuginfo-analysis is not supported.\n"); + return -ENOSYS; +} +#endif + +static int parse_line_num(char **ptr, int *val, const char *what) +{ + const char *start = *ptr; + + errno = 0; + *val = strtol(*ptr, ptr, 0); + if (errno || *ptr == start) { + semantic_error("'%s' is not a valid number.\n", what); + return -EINVAL; + } + return 0; +} + +/* + * Stuff 'lr' according to the line range described by 'arg'. + * The line range syntax is described by: + * + * SRC[:SLN[+NUM|-ELN]] + * FNC[@SRC][:SLN[+NUM|-ELN]] + */ +int parse_line_range_desc(const char *arg, struct line_range *lr) +{ + char *range, *file, *name = strdup(arg); + int err; + + if (!name) + return -ENOMEM; + + lr->start = 0; + lr->end = INT_MAX; + + range = strchr(name, ':'); + if (range) { + *range++ = '\0'; + + err = parse_line_num(&range, &lr->start, "start line"); + if (err) + goto err; + + if (*range == '+' || *range == '-') { + const char c = *range++; + + err = parse_line_num(&range, &lr->end, "end line"); + if (err) + goto err; + + if (c == '+') { + lr->end += lr->start; + /* + * Adjust the number of lines here. + * If the number of lines == 1, the + * the end of line should be equal to + * the start of line. + */ + lr->end--; + } + } + + pr_debug("Line range is %d to %d\n", lr->start, lr->end); + + err = -EINVAL; + if (lr->start > lr->end) { + semantic_error("Start line must be smaller" + " than end line.\n"); + goto err; + } + if (*range != '\0') { + semantic_error("Tailing with invalid str '%s'.\n", range); + goto err; + } + } + + file = strchr(name, '@'); + if (file) { + *file = '\0'; + lr->file = strdup(++file); + if (lr->file == NULL) { + err = -ENOMEM; + goto err; + } + lr->function = name; + } else if (strchr(name, '.')) + lr->file = name; + else + lr->function = name; + + return 0; +err: + free(name); + return err; +} + +/* Check the name is good for event/group */ +static bool check_event_name(const char *name) +{ + if (!isalpha(*name) && *name != '_') + return false; + while (*++name != '\0') { + if (!isalpha(*name) && !isdigit(*name) && *name != '_') + return false; + } + return true; +} + +/* Parse probepoint definition. */ +static int parse_perf_probe_point(char *arg, struct perf_probe_event *pev) +{ + struct perf_probe_point *pp = &pev->point; + char *ptr, *tmp; + char c, nc = 0; + /* + * <Syntax> + * perf probe [EVENT=]SRC[:LN|;PTN] + * perf probe [EVENT=]FUNC[@SRC][+OFFS|%return|:LN|;PAT] + * + * TODO:Group name support + */ + + ptr = strpbrk(arg, ";=@+%"); + if (ptr && *ptr == '=') { /* Event name */ + *ptr = '\0'; + tmp = ptr + 1; + if (strchr(arg, ':')) { + semantic_error("Group name is not supported yet.\n"); + return -ENOTSUP; + } + if (!check_event_name(arg)) { + semantic_error("%s is bad for event name -it must " + "follow C symbol-naming rule.\n", arg); + return -EINVAL; + } + pev->event = strdup(arg); + if (pev->event == NULL) + return -ENOMEM; + pev->group = NULL; + arg = tmp; + } + + ptr = strpbrk(arg, ";:+@%"); + if (ptr) { + nc = *ptr; + *ptr++ = '\0'; + } + + tmp = strdup(arg); + if (tmp == NULL) + return -ENOMEM; + + /* Check arg is function or file and copy it */ + if (strchr(tmp, '.')) /* File */ + pp->file = tmp; + else /* Function */ + pp->function = tmp; + + /* Parse other options */ + while (ptr) { + arg = ptr; + c = nc; + if (c == ';') { /* Lazy pattern must be the last part */ + pp->lazy_line = strdup(arg); + if (pp->lazy_line == NULL) + return -ENOMEM; + break; + } + ptr = strpbrk(arg, ";:+@%"); + if (ptr) { + nc = *ptr; + *ptr++ = '\0'; + } + switch (c) { + case ':': /* Line number */ + pp->line = strtoul(arg, &tmp, 0); + if (*tmp != '\0') { + semantic_error("There is non-digit char" + " in line number.\n"); + return -EINVAL; + } + break; + case '+': /* Byte offset from a symbol */ + pp->offset = strtoul(arg, &tmp, 0); + if (*tmp != '\0') { + semantic_error("There is non-digit character" + " in offset.\n"); + return -EINVAL; + } + break; + case '@': /* File name */ + if (pp->file) { + semantic_error("SRC@SRC is not allowed.\n"); + return -EINVAL; + } + pp->file = strdup(arg); + if (pp->file == NULL) + return -ENOMEM; + break; + case '%': /* Probe places */ + if (strcmp(arg, "return") == 0) { + pp->retprobe = 1; + } else { /* Others not supported yet */ + semantic_error("%%%s is not supported.\n", arg); + return -ENOTSUP; + } + break; + default: /* Buggy case */ + pr_err("This program has a bug at %s:%d.\n", + __FILE__, __LINE__); + return -ENOTSUP; + break; + } + } + + /* Exclusion check */ + if (pp->lazy_line && pp->line) { + semantic_error("Lazy pattern can't be used with" + " line number.\n"); + return -EINVAL; + } + + if (pp->lazy_line && pp->offset) { + semantic_error("Lazy pattern can't be used with offset.\n"); + return -EINVAL; + } + + if (pp->line && pp->offset) { + semantic_error("Offset can't be used with line number.\n"); + return -EINVAL; + } + + if (!pp->line && !pp->lazy_line && pp->file && !pp->function) { + semantic_error("File always requires line number or " + "lazy pattern.\n"); + return -EINVAL; + } + + if (pp->offset && !pp->function) { + semantic_error("Offset requires an entry function.\n"); + return -EINVAL; + } + + if (pp->retprobe && !pp->function) { + semantic_error("Return probe requires an entry function.\n"); + return -EINVAL; + } + + if ((pp->offset || pp->line || pp->lazy_line) && pp->retprobe) { + semantic_error("Offset/Line/Lazy pattern can't be used with " + "return probe.\n"); + return -EINVAL; + } + + pr_debug("symbol:%s file:%s line:%d offset:%lu return:%d lazy:%s\n", + pp->function, pp->file, pp->line, pp->offset, pp->retprobe, + pp->lazy_line); + return 0; +} + +/* Parse perf-probe event argument */ +static int parse_perf_probe_arg(char *str, struct perf_probe_arg *arg) +{ + char *tmp, *goodname; + struct perf_probe_arg_field **fieldp; + + pr_debug("parsing arg: %s into ", str); + + tmp = strchr(str, '='); + if (tmp) { + arg->name = strndup(str, tmp - str); + if (arg->name == NULL) + return -ENOMEM; + pr_debug("name:%s ", arg->name); + str = tmp + 1; + } + + tmp = strchr(str, ':'); + if (tmp) { /* Type setting */ + *tmp = '\0'; + arg->type = strdup(tmp + 1); + if (arg->type == NULL) + return -ENOMEM; + pr_debug("type:%s ", arg->type); + } + + tmp = strpbrk(str, "-.["); + if (!is_c_varname(str) || !tmp) { + /* A variable, register, symbol or special value */ + arg->var = strdup(str); + if (arg->var == NULL) + return -ENOMEM; + pr_debug("%s\n", arg->var); + return 0; + } + + /* Structure fields or array element */ + arg->var = strndup(str, tmp - str); + if (arg->var == NULL) + return -ENOMEM; + goodname = arg->var; + pr_debug("%s, ", arg->var); + fieldp = &arg->field; + + do { + *fieldp = zalloc(sizeof(struct perf_probe_arg_field)); + if (*fieldp == NULL) + return -ENOMEM; + if (*tmp == '[') { /* Array */ + str = tmp; + (*fieldp)->index = strtol(str + 1, &tmp, 0); + (*fieldp)->ref = true; + if (*tmp != ']' || tmp == str + 1) { + semantic_error("Array index must be a" + " number.\n"); + return -EINVAL; + } + tmp++; + if (*tmp == '\0') + tmp = NULL; + } else { /* Structure */ + if (*tmp == '.') { + str = tmp + 1; + (*fieldp)->ref = false; + } else if (tmp[1] == '>') { + str = tmp + 2; + (*fieldp)->ref = true; + } else { + semantic_error("Argument parse error: %s\n", + str); + return -EINVAL; + } + tmp = strpbrk(str, "-.["); + } + if (tmp) { + (*fieldp)->name = strndup(str, tmp - str); + if ((*fieldp)->name == NULL) + return -ENOMEM; + if (*str != '[') + goodname = (*fieldp)->name; + pr_debug("%s(%d), ", (*fieldp)->name, (*fieldp)->ref); + fieldp = &(*fieldp)->next; + } + } while (tmp); + (*fieldp)->name = strdup(str); + if ((*fieldp)->name == NULL) + return -ENOMEM; + if (*str != '[') + goodname = (*fieldp)->name; + pr_debug("%s(%d)\n", (*fieldp)->name, (*fieldp)->ref); + + /* If no name is specified, set the last field name (not array index)*/ + if (!arg->name) { + arg->name = strdup(goodname); + if (arg->name == NULL) + return -ENOMEM; + } + return 0; +} + +/* Parse perf-probe event command */ +int parse_perf_probe_command(const char *cmd, struct perf_probe_event *pev) +{ + char **argv; + int argc, i, ret = 0; + + argv = argv_split(cmd, &argc); + if (!argv) { + pr_debug("Failed to split arguments.\n"); + return -ENOMEM; + } + if (argc - 1 > MAX_PROBE_ARGS) { + semantic_error("Too many probe arguments (%d).\n", argc - 1); + ret = -ERANGE; + goto out; + } + /* Parse probe point */ + ret = parse_perf_probe_point(argv[0], pev); + if (ret < 0) + goto out; + + /* Copy arguments and ensure return probe has no C argument */ + pev->nargs = argc - 1; + pev->args = zalloc(sizeof(struct perf_probe_arg) * pev->nargs); + if (pev->args == NULL) { + ret = -ENOMEM; + goto out; + } + for (i = 0; i < pev->nargs && ret >= 0; i++) { + ret = parse_perf_probe_arg(argv[i + 1], &pev->args[i]); + if (ret >= 0 && + is_c_varname(pev->args[i].var) && pev->point.retprobe) { + semantic_error("You can't specify local variable for" + " kretprobe.\n"); + ret = -EINVAL; + } + } +out: + argv_free(argv); + + return ret; +} + +/* Return true if this perf_probe_event requires debuginfo */ +bool perf_probe_event_need_dwarf(struct perf_probe_event *pev) +{ + int i; + + if (pev->point.file || pev->point.line || pev->point.lazy_line) + return true; + + for (i = 0; i < pev->nargs; i++) + if (is_c_varname(pev->args[i].var)) + return true; + + return false; +} + +/* Parse probe_events event into struct probe_point */ +static int parse_probe_trace_command(const char *cmd, + struct probe_trace_event *tev) +{ + struct probe_trace_point *tp = &tev->point; + char pr; + char *p; + int ret, i, argc; + char **argv; + + pr_debug("Parsing probe_events: %s\n", cmd); + argv = argv_split(cmd, &argc); + if (!argv) { + pr_debug("Failed to split arguments.\n"); + return -ENOMEM; + } + if (argc < 2) { + semantic_error("Too few probe arguments.\n"); + ret = -ERANGE; + goto out; + } + + /* Scan event and group name. */ + ret = sscanf(argv[0], "%c:%a[^/ \t]/%a[^ \t]", + &pr, (float *)(void *)&tev->group, + (float *)(void *)&tev->event); + if (ret != 3) { + semantic_error("Failed to parse event name: %s\n", argv[0]); + ret = -EINVAL; + goto out; + } + pr_debug("Group:%s Event:%s probe:%c\n", tev->group, tev->event, pr); + + tp->retprobe = (pr == 'r'); + + /* Scan function name and offset */ + ret = sscanf(argv[1], "%a[^+]+%lu", (float *)(void *)&tp->symbol, + &tp->offset); + if (ret == 1) + tp->offset = 0; + + tev->nargs = argc - 2; + tev->args = zalloc(sizeof(struct probe_trace_arg) * tev->nargs); + if (tev->args == NULL) { + ret = -ENOMEM; + goto out; + } + for (i = 0; i < tev->nargs; i++) { + p = strchr(argv[i + 2], '='); + if (p) /* We don't need which register is assigned. */ + *p++ = '\0'; + else + p = argv[i + 2]; + tev->args[i].name = strdup(argv[i + 2]); + /* TODO: parse regs and offset */ + tev->args[i].value = strdup(p); + if (tev->args[i].name == NULL || tev->args[i].value == NULL) { + ret = -ENOMEM; + goto out; + } + } + ret = 0; +out: + argv_free(argv); + return ret; +} + +/* Compose only probe arg */ +int synthesize_perf_probe_arg(struct perf_probe_arg *pa, char *buf, size_t len) +{ + struct perf_probe_arg_field *field = pa->field; + int ret; + char *tmp = buf; + + if (pa->name && pa->var) + ret = e_snprintf(tmp, len, "%s=%s", pa->name, pa->var); + else + ret = e_snprintf(tmp, len, "%s", pa->name ? pa->name : pa->var); + if (ret <= 0) + goto error; + tmp += ret; + len -= ret; + + while (field) { + if (field->name[0] == '[') + ret = e_snprintf(tmp, len, "%s", field->name); + else + ret = e_snprintf(tmp, len, "%s%s", + field->ref ? "->" : ".", field->name); + if (ret <= 0) + goto error; + tmp += ret; + len -= ret; + field = field->next; + } + + if (pa->type) { + ret = e_snprintf(tmp, len, ":%s", pa->type); + if (ret <= 0) + goto error; + tmp += ret; + len -= ret; + } + + return tmp - buf; +error: + pr_debug("Failed to synthesize perf probe argument: %s\n", + strerror(-ret)); + return ret; +} + +/* Compose only probe point (not argument) */ +static char *synthesize_perf_probe_point(struct perf_probe_point *pp) +{ + char *buf, *tmp; + char offs[32] = "", line[32] = "", file[32] = ""; + int ret, len; + + buf = zalloc(MAX_CMDLEN); + if (buf == NULL) { + ret = -ENOMEM; + goto error; + } + if (pp->offset) { + ret = e_snprintf(offs, 32, "+%lu", pp->offset); + if (ret <= 0) + goto error; + } + if (pp->line) { + ret = e_snprintf(line, 32, ":%d", pp->line); + if (ret <= 0) + goto error; + } + if (pp->file) { + tmp = pp->file; + len = strlen(tmp); + if (len > 30) { + tmp = strchr(pp->file + len - 30, '/'); + tmp = tmp ? tmp + 1 : pp->file + len - 30; + } + ret = e_snprintf(file, 32, "@%s", tmp); + if (ret <= 0) + goto error; + } + + if (pp->function) + ret = e_snprintf(buf, MAX_CMDLEN, "%s%s%s%s%s", pp->function, + offs, pp->retprobe ? "%return" : "", line, + file); + else + ret = e_snprintf(buf, MAX_CMDLEN, "%s%s", file, line); + if (ret <= 0) + goto error; + + return buf; +error: + pr_debug("Failed to synthesize perf probe point: %s\n", + strerror(-ret)); + if (buf) + free(buf); + return NULL; +} + +#if 0 +char *synthesize_perf_probe_command(struct perf_probe_event *pev) +{ + char *buf; + int i, len, ret; + + buf = synthesize_perf_probe_point(&pev->point); + if (!buf) + return NULL; + + len = strlen(buf); + for (i = 0; i < pev->nargs; i++) { + ret = e_snprintf(&buf[len], MAX_CMDLEN - len, " %s", + pev->args[i].name); + if (ret <= 0) { + free(buf); + return NULL; + } + len += ret; + } + + return buf; +} +#endif + +static int __synthesize_probe_trace_arg_ref(struct probe_trace_arg_ref *ref, + char **buf, size_t *buflen, + int depth) +{ + int ret; + if (ref->next) { + depth = __synthesize_probe_trace_arg_ref(ref->next, buf, + buflen, depth + 1); + if (depth < 0) + goto out; + } + + ret = e_snprintf(*buf, *buflen, "%+ld(", ref->offset); + if (ret < 0) + depth = ret; + else { + *buf += ret; + *buflen -= ret; + } +out: + return depth; + +} + +static int synthesize_probe_trace_arg(struct probe_trace_arg *arg, + char *buf, size_t buflen) +{ + struct probe_trace_arg_ref *ref = arg->ref; + int ret, depth = 0; + char *tmp = buf; + + /* Argument name or separator */ + if (arg->name) + ret = e_snprintf(buf, buflen, " %s=", arg->name); + else + ret = e_snprintf(buf, buflen, " "); + if (ret < 0) + return ret; + buf += ret; + buflen -= ret; + + /* Special case: @XXX */ + if (arg->value[0] == '@' && arg->ref) + ref = ref->next; + + /* Dereferencing arguments */ + if (ref) { + depth = __synthesize_probe_trace_arg_ref(ref, &buf, + &buflen, 1); + if (depth < 0) + return depth; + } + + /* Print argument value */ + if (arg->value[0] == '@' && arg->ref) + ret = e_snprintf(buf, buflen, "%s%+ld", arg->value, + arg->ref->offset); + else + ret = e_snprintf(buf, buflen, "%s", arg->value); + if (ret < 0) + return ret; + buf += ret; + buflen -= ret; + + /* Closing */ + while (depth--) { + ret = e_snprintf(buf, buflen, ")"); + if (ret < 0) + return ret; + buf += ret; + buflen -= ret; + } + /* Print argument type */ + if (arg->type) { + ret = e_snprintf(buf, buflen, ":%s", arg->type); + if (ret <= 0) + return ret; + buf += ret; + } + + return buf - tmp; +} + +char *synthesize_probe_trace_command(struct probe_trace_event *tev) +{ + struct probe_trace_point *tp = &tev->point; + char *buf; + int i, len, ret; + + buf = zalloc(MAX_CMDLEN); + if (buf == NULL) + return NULL; + + len = e_snprintf(buf, MAX_CMDLEN, "%c:%s/%s %s+%lu", + tp->retprobe ? 'r' : 'p', + tev->group, tev->event, + tp->symbol, tp->offset); + if (len <= 0) + goto error; + + for (i = 0; i < tev->nargs; i++) { + ret = synthesize_probe_trace_arg(&tev->args[i], buf + len, + MAX_CMDLEN - len); + if (ret <= 0) + goto error; + len += ret; + } + + return buf; +error: + free(buf); + return NULL; +} + +static int convert_to_perf_probe_event(struct probe_trace_event *tev, + struct perf_probe_event *pev) +{ + char buf[64] = ""; + int i, ret; + + /* Convert event/group name */ + pev->event = strdup(tev->event); + pev->group = strdup(tev->group); + if (pev->event == NULL || pev->group == NULL) + return -ENOMEM; + + /* Convert trace_point to probe_point */ + ret = kprobe_convert_to_perf_probe(&tev->point, &pev->point); + if (ret < 0) + return ret; + + /* Convert trace_arg to probe_arg */ + pev->nargs = tev->nargs; + pev->args = zalloc(sizeof(struct perf_probe_arg) * pev->nargs); + if (pev->args == NULL) + return -ENOMEM; + for (i = 0; i < tev->nargs && ret >= 0; i++) { + if (tev->args[i].name) + pev->args[i].name = strdup(tev->args[i].name); + else { + ret = synthesize_probe_trace_arg(&tev->args[i], + buf, 64); + pev->args[i].name = strdup(buf); + } + if (pev->args[i].name == NULL && ret >= 0) + ret = -ENOMEM; + } + + if (ret < 0) + clear_perf_probe_event(pev); + + return ret; +} + +void clear_perf_probe_event(struct perf_probe_event *pev) +{ + struct perf_probe_point *pp = &pev->point; + struct perf_probe_arg_field *field, *next; + int i; + + if (pev->event) + free(pev->event); + if (pev->group) + free(pev->group); + if (pp->file) + free(pp->file); + if (pp->function) + free(pp->function); + if (pp->lazy_line) + free(pp->lazy_line); + for (i = 0; i < pev->nargs; i++) { + if (pev->args[i].name) + free(pev->args[i].name); + if (pev->args[i].var) + free(pev->args[i].var); + if (pev->args[i].type) + free(pev->args[i].type); + field = pev->args[i].field; + while (field) { + next = field->next; + if (field->name) + free(field->name); + free(field); + field = next; + } + } + if (pev->args) + free(pev->args); + memset(pev, 0, sizeof(*pev)); +} + +static void clear_probe_trace_event(struct probe_trace_event *tev) +{ + struct probe_trace_arg_ref *ref, *next; + int i; + + if (tev->event) + free(tev->event); + if (tev->group) + free(tev->group); + if (tev->point.symbol) + free(tev->point.symbol); + for (i = 0; i < tev->nargs; i++) { + if (tev->args[i].name) + free(tev->args[i].name); + if (tev->args[i].value) + free(tev->args[i].value); + if (tev->args[i].type) + free(tev->args[i].type); + ref = tev->args[i].ref; + while (ref) { + next = ref->next; + free(ref); + ref = next; + } + } + if (tev->args) + free(tev->args); + memset(tev, 0, sizeof(*tev)); +} + +static int open_kprobe_events(bool readwrite) +{ + char buf[PATH_MAX]; + const char *__debugfs; + int ret; + + __debugfs = debugfs_find_mountpoint(); + if (__debugfs == NULL) { + pr_warning("Debugfs is not mounted.\n"); + return -ENOENT; + } + + ret = e_snprintf(buf, PATH_MAX, "%stracing/kprobe_events", __debugfs); + if (ret >= 0) { + pr_debug("Opening %s write=%d\n", buf, readwrite); + if (readwrite && !probe_event_dry_run) + ret = open(buf, O_RDWR, O_APPEND); + else + ret = open(buf, O_RDONLY, 0); + } + + if (ret < 0) { + if (errno == ENOENT) + pr_warning("kprobe_events file does not exist - please" + " rebuild kernel with CONFIG_KPROBE_EVENT.\n"); + else + pr_warning("Failed to open kprobe_events file: %s\n", + strerror(errno)); + } + return ret; +} + +/* Get raw string list of current kprobe_events */ +static struct strlist *get_probe_trace_command_rawlist(int fd) +{ + int ret, idx; + FILE *fp; + char buf[MAX_CMDLEN]; + char *p; + struct strlist *sl; + + sl = strlist__new(true, NULL); + + fp = fdopen(dup(fd), "r"); + while (!feof(fp)) { + p = fgets(buf, MAX_CMDLEN, fp); + if (!p) + break; + + idx = strlen(p) - 1; + if (p[idx] == '\n') + p[idx] = '\0'; + ret = strlist__add(sl, buf); + if (ret < 0) { + pr_debug("strlist__add failed: %s\n", strerror(-ret)); + strlist__delete(sl); + return NULL; + } + } + fclose(fp); + + return sl; +} + +/* Show an event */ +static int show_perf_probe_event(struct perf_probe_event *pev) +{ + int i, ret; + char buf[128]; + char *place; + + /* Synthesize only event probe point */ + place = synthesize_perf_probe_point(&pev->point); + if (!place) + return -EINVAL; + + ret = e_snprintf(buf, 128, "%s:%s", pev->group, pev->event); + if (ret < 0) + return ret; + + printf(" %-20s (on %s", buf, place); + + if (pev->nargs > 0) { + printf(" with"); + for (i = 0; i < pev->nargs; i++) { + ret = synthesize_perf_probe_arg(&pev->args[i], + buf, 128); + if (ret < 0) + break; + printf(" %s", buf); + } + } + printf(")\n"); + free(place); + return ret; +} + +/* List up current perf-probe events */ +int show_perf_probe_events(void) +{ + int fd, ret; + struct probe_trace_event tev; + struct perf_probe_event pev; + struct strlist *rawlist; + struct str_node *ent; + + setup_pager(); + ret = init_vmlinux(); + if (ret < 0) + return ret; + + memset(&tev, 0, sizeof(tev)); + memset(&pev, 0, sizeof(pev)); + + fd = open_kprobe_events(false); + if (fd < 0) + return fd; + + rawlist = get_probe_trace_command_rawlist(fd); + close(fd); + if (!rawlist) + return -ENOENT; + + strlist__for_each(ent, rawlist) { + ret = parse_probe_trace_command(ent->s, &tev); + if (ret >= 0) { + ret = convert_to_perf_probe_event(&tev, &pev); + if (ret >= 0) + ret = show_perf_probe_event(&pev); + } + clear_perf_probe_event(&pev); + clear_probe_trace_event(&tev); + if (ret < 0) + break; + } + strlist__delete(rawlist); + + return ret; +} + +/* Get current perf-probe event names */ +static struct strlist *get_probe_trace_event_names(int fd, bool include_group) +{ + char buf[128]; + struct strlist *sl, *rawlist; + struct str_node *ent; + struct probe_trace_event tev; + int ret = 0; + + memset(&tev, 0, sizeof(tev)); + rawlist = get_probe_trace_command_rawlist(fd); + sl = strlist__new(true, NULL); + strlist__for_each(ent, rawlist) { + ret = parse_probe_trace_command(ent->s, &tev); + if (ret < 0) + break; + if (include_group) { + ret = e_snprintf(buf, 128, "%s:%s", tev.group, + tev.event); + if (ret >= 0) + ret = strlist__add(sl, buf); + } else + ret = strlist__add(sl, tev.event); + clear_probe_trace_event(&tev); + if (ret < 0) + break; + } + strlist__delete(rawlist); + + if (ret < 0) { + strlist__delete(sl); + return NULL; + } + return sl; +} + +static int write_probe_trace_event(int fd, struct probe_trace_event *tev) +{ + int ret = 0; + char *buf = synthesize_probe_trace_command(tev); + + if (!buf) { + pr_debug("Failed to synthesize probe trace event.\n"); + return -EINVAL; + } + + pr_debug("Writing event: %s\n", buf); + if (!probe_event_dry_run) { + ret = write(fd, buf, strlen(buf)); + if (ret <= 0) + pr_warning("Failed to write event: %s\n", + strerror(errno)); + } + free(buf); + return ret; +} + +static int get_new_event_name(char *buf, size_t len, const char *base, + struct strlist *namelist, bool allow_suffix) +{ + int i, ret; + + /* Try no suffix */ + ret = e_snprintf(buf, len, "%s", base); + if (ret < 0) { + pr_debug("snprintf() failed: %s\n", strerror(-ret)); + return ret; + } + if (!strlist__has_entry(namelist, buf)) + return 0; + + if (!allow_suffix) { + pr_warning("Error: event \"%s\" already exists. " + "(Use -f to force duplicates.)\n", base); + return -EEXIST; + } + + /* Try to add suffix */ + for (i = 1; i < MAX_EVENT_INDEX; i++) { + ret = e_snprintf(buf, len, "%s_%d", base, i); + if (ret < 0) { + pr_debug("snprintf() failed: %s\n", strerror(-ret)); + return ret; + } + if (!strlist__has_entry(namelist, buf)) + break; + } + if (i == MAX_EVENT_INDEX) { + pr_warning("Too many events are on the same function.\n"); + ret = -ERANGE; + } + + return ret; +} + +static int __add_probe_trace_events(struct perf_probe_event *pev, + struct probe_trace_event *tevs, + int ntevs, bool allow_suffix) +{ + int i, fd, ret; + struct probe_trace_event *tev = NULL; + char buf[64]; + const char *event, *group; + struct strlist *namelist; + + fd = open_kprobe_events(true); + if (fd < 0) + return fd; + /* Get current event names */ + namelist = get_probe_trace_event_names(fd, false); + if (!namelist) { + pr_debug("Failed to get current event list.\n"); + return -EIO; + } + + ret = 0; + printf("Add new event%s\n", (ntevs > 1) ? "s:" : ":"); + for (i = 0; i < ntevs; i++) { + tev = &tevs[i]; + if (pev->event) + event = pev->event; + else + if (pev->point.function) + event = pev->point.function; + else + event = tev->point.symbol; + if (pev->group) + group = pev->group; + else + group = PERFPROBE_GROUP; + + /* Get an unused new event name */ + ret = get_new_event_name(buf, 64, event, + namelist, allow_suffix); + if (ret < 0) + break; + event = buf; + + tev->event = strdup(event); + tev->group = strdup(group); + if (tev->event == NULL || tev->group == NULL) { + ret = -ENOMEM; + break; + } + ret = write_probe_trace_event(fd, tev); + if (ret < 0) + break; + /* Add added event name to namelist */ + strlist__add(namelist, event); + + /* Trick here - save current event/group */ + event = pev->event; + group = pev->group; + pev->event = tev->event; + pev->group = tev->group; + show_perf_probe_event(pev); + /* Trick here - restore current event/group */ + pev->event = (char *)event; + pev->group = (char *)group; + + /* + * Probes after the first probe which comes from same + * user input are always allowed to add suffix, because + * there might be several addresses corresponding to + * one code line. + */ + allow_suffix = true; + } + + if (ret >= 0) { + /* Show how to use the event. */ + printf("\nYou can now use it on all perf tools, such as:\n\n"); + printf("\tperf record -e %s:%s -aR sleep 1\n\n", tev->group, + tev->event); + } + + strlist__delete(namelist); + close(fd); + return ret; +} + +static int convert_to_probe_trace_events(struct perf_probe_event *pev, + struct probe_trace_event **tevs, + int max_tevs, const char *module) +{ + struct symbol *sym; + int ret = 0, i; + struct probe_trace_event *tev; + + /* Convert perf_probe_event with debuginfo */ + ret = try_to_find_probe_trace_events(pev, tevs, max_tevs, module); + if (ret != 0) + return ret; + + /* Allocate trace event buffer */ + tev = *tevs = zalloc(sizeof(struct probe_trace_event)); + if (tev == NULL) + return -ENOMEM; + + /* Copy parameters */ + tev->point.symbol = strdup(pev->point.function); + if (tev->point.symbol == NULL) { + ret = -ENOMEM; + goto error; + } + tev->point.offset = pev->point.offset; + tev->point.retprobe = pev->point.retprobe; + tev->nargs = pev->nargs; + if (tev->nargs) { + tev->args = zalloc(sizeof(struct probe_trace_arg) + * tev->nargs); + if (tev->args == NULL) { + ret = -ENOMEM; + goto error; + } + for (i = 0; i < tev->nargs; i++) { + if (pev->args[i].name) { + tev->args[i].name = strdup(pev->args[i].name); + if (tev->args[i].name == NULL) { + ret = -ENOMEM; + goto error; + } + } + tev->args[i].value = strdup(pev->args[i].var); + if (tev->args[i].value == NULL) { + ret = -ENOMEM; + goto error; + } + if (pev->args[i].type) { + tev->args[i].type = strdup(pev->args[i].type); + if (tev->args[i].type == NULL) { + ret = -ENOMEM; + goto error; + } + } + } + } + + /* Currently just checking function name from symbol map */ + sym = __find_kernel_function_by_name(tev->point.symbol, NULL); + if (!sym) { + pr_warning("Kernel symbol \'%s\' not found.\n", + tev->point.symbol); + ret = -ENOENT; + goto error; + } + + return 1; +error: + clear_probe_trace_event(tev); + free(tev); + *tevs = NULL; + return ret; +} + +struct __event_package { + struct perf_probe_event *pev; + struct probe_trace_event *tevs; + int ntevs; +}; + +int add_perf_probe_events(struct perf_probe_event *pevs, int npevs, + int max_tevs, const char *module, bool force_add) +{ + int i, j, ret; + struct __event_package *pkgs; + + pkgs = zalloc(sizeof(struct __event_package) * npevs); + if (pkgs == NULL) + return -ENOMEM; + + /* Init vmlinux path */ + ret = init_vmlinux(); + if (ret < 0) { + free(pkgs); + return ret; + } + + /* Loop 1: convert all events */ + for (i = 0; i < npevs; i++) { + pkgs[i].pev = &pevs[i]; + /* Convert with or without debuginfo */ + ret = convert_to_probe_trace_events(pkgs[i].pev, + &pkgs[i].tevs, + max_tevs, + module); + if (ret < 0) + goto end; + pkgs[i].ntevs = ret; + } + + /* Loop 2: add all events */ + for (i = 0; i < npevs; i++) { + ret = __add_probe_trace_events(pkgs[i].pev, pkgs[i].tevs, + pkgs[i].ntevs, force_add); + if (ret < 0) + break; + } +end: + /* Loop 3: cleanup and free trace events */ + for (i = 0; i < npevs; i++) { + for (j = 0; j < pkgs[i].ntevs; j++) + clear_probe_trace_event(&pkgs[i].tevs[j]); + free(pkgs[i].tevs); + } + free(pkgs); + + return ret; +} + +static int __del_trace_probe_event(int fd, struct str_node *ent) +{ + char *p; + char buf[128]; + int ret; + + /* Convert from perf-probe event to trace-probe event */ + ret = e_snprintf(buf, 128, "-:%s", ent->s); + if (ret < 0) + goto error; + + p = strchr(buf + 2, ':'); + if (!p) { + pr_debug("Internal error: %s should have ':' but not.\n", + ent->s); + ret = -ENOTSUP; + goto error; + } + *p = '/'; + + pr_debug("Writing event: %s\n", buf); + ret = write(fd, buf, strlen(buf)); + if (ret < 0) { + ret = -errno; + goto error; + } + + printf("Remove event: %s\n", ent->s); + return 0; +error: + pr_warning("Failed to delete event: %s\n", strerror(-ret)); + return ret; +} + +static int del_trace_probe_event(int fd, const char *group, + const char *event, struct strlist *namelist) +{ + char buf[128]; + struct str_node *ent, *n; + int found = 0, ret = 0; + + ret = e_snprintf(buf, 128, "%s:%s", group, event); + if (ret < 0) { + pr_err("Failed to copy event.\n"); + return ret; + } + + if (strpbrk(buf, "*?")) { /* Glob-exp */ + strlist__for_each_safe(ent, n, namelist) + if (strglobmatch(ent->s, buf)) { + found++; + ret = __del_trace_probe_event(fd, ent); + if (ret < 0) + break; + strlist__remove(namelist, ent); + } + } else { + ent = strlist__find(namelist, buf); + if (ent) { + found++; + ret = __del_trace_probe_event(fd, ent); + if (ret >= 0) + strlist__remove(namelist, ent); + } + } + if (found == 0 && ret >= 0) + pr_info("Info: Event \"%s\" does not exist.\n", buf); + + return ret; +} + +int del_perf_probe_events(struct strlist *dellist) +{ + int fd, ret = 0; + const char *group, *event; + char *p, *str; + struct str_node *ent; + struct strlist *namelist; + + fd = open_kprobe_events(true); + if (fd < 0) + return fd; + + /* Get current event names */ + namelist = get_probe_trace_event_names(fd, true); + if (namelist == NULL) + return -EINVAL; + + strlist__for_each(ent, dellist) { + str = strdup(ent->s); + if (str == NULL) { + ret = -ENOMEM; + break; + } + pr_debug("Parsing: %s\n", str); + p = strchr(str, ':'); + if (p) { + group = str; + *p = '\0'; + event = p + 1; + } else { + group = "*"; + event = str; + } + pr_debug("Group: %s, Event: %s\n", group, event); + ret = del_trace_probe_event(fd, group, event, namelist); + free(str); + if (ret < 0) + break; + } + strlist__delete(namelist); + close(fd); + + return ret; +} +/* TODO: don't use a global variable for filter ... */ +static struct strfilter *available_func_filter; + +/* + * If a symbol corresponds to a function with global binding and + * matches filter return 0. For all others return 1. + */ +static int filter_available_functions(struct map *map __unused, + struct symbol *sym) +{ + if (sym->binding == STB_GLOBAL && + strfilter__compare(available_func_filter, sym->name)) + return 0; + return 1; +} + +int show_available_funcs(const char *module, struct strfilter *_filter) +{ + struct map *map; + int ret; + + setup_pager(); + + ret = init_vmlinux(); + if (ret < 0) + return ret; + + map = kernel_get_module_map(module); + if (!map) { + pr_err("Failed to find %s map.\n", (module) ? : "kernel"); + return -EINVAL; + } + available_func_filter = _filter; + if (map__load(map, filter_available_functions)) { + pr_err("Failed to load map.\n"); + return -EINVAL; + } + if (!dso__sorted_by_name(map->dso, map->type)) + dso__sort_by_name(map->dso, map->type); + + dso__fprintf_symbols_by_name(map->dso, map->type, stdout); + return 0; +} diff --git a/tools/perf/util/probe-event.h b/tools/perf/util/probe-event.h new file mode 100644 index 00000000..3434fc9d --- /dev/null +++ b/tools/perf/util/probe-event.h @@ -0,0 +1,137 @@ +#ifndef _PROBE_EVENT_H +#define _PROBE_EVENT_H + +#include <stdbool.h> +#include "strlist.h" +#include "strfilter.h" + +extern bool probe_event_dry_run; + +/* kprobe-tracer tracing point */ +struct probe_trace_point { + char *symbol; /* Base symbol */ + unsigned long offset; /* Offset from symbol */ + bool retprobe; /* Return probe flag */ +}; + +/* probe-tracer tracing argument referencing offset */ +struct probe_trace_arg_ref { + struct probe_trace_arg_ref *next; /* Next reference */ + long offset; /* Offset value */ +}; + +/* kprobe-tracer tracing argument */ +struct probe_trace_arg { + char *name; /* Argument name */ + char *value; /* Base value */ + char *type; /* Type name */ + struct probe_trace_arg_ref *ref; /* Referencing offset */ +}; + +/* kprobe-tracer tracing event (point + arg) */ +struct probe_trace_event { + char *event; /* Event name */ + char *group; /* Group name */ + struct probe_trace_point point; /* Trace point */ + int nargs; /* Number of args */ + struct probe_trace_arg *args; /* Arguments */ +}; + +/* Perf probe probing point */ +struct perf_probe_point { + char *file; /* File path */ + char *function; /* Function name */ + int line; /* Line number */ + bool retprobe; /* Return probe flag */ + char *lazy_line; /* Lazy matching pattern */ + unsigned long offset; /* Offset from function entry */ +}; + +/* Perf probe probing argument field chain */ +struct perf_probe_arg_field { + struct perf_probe_arg_field *next; /* Next field */ + char *name; /* Name of the field */ + long index; /* Array index number */ + bool ref; /* Referencing flag */ +}; + +/* Perf probe probing argument */ +struct perf_probe_arg { + char *name; /* Argument name */ + char *var; /* Variable name */ + char *type; /* Type name */ + struct perf_probe_arg_field *field; /* Structure fields */ +}; + +/* Perf probe probing event (point + arg) */ +struct perf_probe_event { + char *event; /* Event name */ + char *group; /* Group name */ + struct perf_probe_point point; /* Probe point */ + int nargs; /* Number of arguments */ + struct perf_probe_arg *args; /* Arguments */ +}; + + +/* Line number container */ +struct line_node { + struct list_head list; + int line; +}; + +/* Line range */ +struct line_range { + char *file; /* File name */ + char *function; /* Function name */ + int start; /* Start line number */ + int end; /* End line number */ + int offset; /* Start line offset */ + char *path; /* Real path name */ + char *comp_dir; /* Compile directory */ + struct list_head line_list; /* Visible lines */ +}; + +/* List of variables */ +struct variable_list { + struct probe_trace_point point; /* Actual probepoint */ + struct strlist *vars; /* Available variables */ +}; + +/* Command string to events */ +extern int parse_perf_probe_command(const char *cmd, + struct perf_probe_event *pev); + +/* Events to command string */ +extern char *synthesize_perf_probe_command(struct perf_probe_event *pev); +extern char *synthesize_probe_trace_command(struct probe_trace_event *tev); +extern int synthesize_perf_probe_arg(struct perf_probe_arg *pa, char *buf, + size_t len); + +/* Check the perf_probe_event needs debuginfo */ +extern bool perf_probe_event_need_dwarf(struct perf_probe_event *pev); + +/* Release event contents */ +extern void clear_perf_probe_event(struct perf_probe_event *pev); + +/* Command string to line-range */ +extern int parse_line_range_desc(const char *cmd, struct line_range *lr); + +/* Internal use: Return kernel/module path */ +extern const char *kernel_get_module_path(const char *module); + +extern int add_perf_probe_events(struct perf_probe_event *pevs, int npevs, + int max_probe_points, const char *module, + bool force_add); +extern int del_perf_probe_events(struct strlist *dellist); +extern int show_perf_probe_events(void); +extern int show_line_range(struct line_range *lr, const char *module); +extern int show_available_vars(struct perf_probe_event *pevs, int npevs, + int max_probe_points, const char *module, + struct strfilter *filter, bool externs); +extern int show_available_funcs(const char *module, struct strfilter *filter); + + +/* Maximum index number of event-name postfix */ +#define MAX_EVENT_INDEX 1024 + +#endif /*_PROBE_EVENT_H */ diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c new file mode 100644 index 00000000..3b9d0b80 --- /dev/null +++ b/tools/perf/util/probe-finder.c @@ -0,0 +1,2064 @@ +/* + * probe-finder.c : C expression to kprobe event converter + * + * Written by Masami Hiramatsu <mhiramat@redhat.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + */ + +#include <sys/utsname.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <fcntl.h> +#include <errno.h> +#include <stdio.h> +#include <unistd.h> +#include <getopt.h> +#include <stdlib.h> +#include <string.h> +#include <stdarg.h> +#include <ctype.h> +#include <dwarf-regs.h> + +#include <linux/bitops.h> +#include "event.h" +#include "debug.h" +#include "util.h" +#include "symbol.h" +#include "probe-finder.h" + +/* Kprobe tracer basic type is up to u64 */ +#define MAX_BASIC_TYPE_BITS 64 + +/* + * Compare the tail of two strings. + * Return 0 if whole of either string is same as another's tail part. + */ +static int strtailcmp(const char *s1, const char *s2) +{ + int i1 = strlen(s1); + int i2 = strlen(s2); + while (--i1 >= 0 && --i2 >= 0) { + if (s1[i1] != s2[i2]) + return s1[i1] - s2[i2]; + } + return 0; +} + +/* Line number list operations */ + +/* Add a line to line number list */ +static int line_list__add_line(struct list_head *head, int line) +{ + struct line_node *ln; + struct list_head *p; + + /* Reverse search, because new line will be the last one */ + list_for_each_entry_reverse(ln, head, list) { + if (ln->line < line) { + p = &ln->list; + goto found; + } else if (ln->line == line) /* Already exist */ + return 1; + } + /* List is empty, or the smallest entry */ + p = head; +found: + pr_debug("line list: add a line %u\n", line); + ln = zalloc(sizeof(struct line_node)); + if (ln == NULL) + return -ENOMEM; + ln->line = line; + INIT_LIST_HEAD(&ln->list); + list_add(&ln->list, p); + return 0; +} + +/* Check if the line in line number list */ +static int line_list__has_line(struct list_head *head, int line) +{ + struct line_node *ln; + + /* Reverse search, because new line will be the last one */ + list_for_each_entry(ln, head, list) + if (ln->line == line) + return 1; + + return 0; +} + +/* Init line number list */ +static void line_list__init(struct list_head *head) +{ + INIT_LIST_HEAD(head); +} + +/* Free line number list */ +static void line_list__free(struct list_head *head) +{ + struct line_node *ln; + while (!list_empty(head)) { + ln = list_first_entry(head, struct line_node, list); + list_del(&ln->list); + free(ln); + } +} + +/* Dwarf FL wrappers */ +static char *debuginfo_path; /* Currently dummy */ + +static const Dwfl_Callbacks offline_callbacks = { + .find_debuginfo = dwfl_standard_find_debuginfo, + .debuginfo_path = &debuginfo_path, + + .section_address = dwfl_offline_section_address, + + /* We use this table for core files too. */ + .find_elf = dwfl_build_id_find_elf, +}; + +/* Get a Dwarf from offline image */ +static Dwarf *dwfl_init_offline_dwarf(int fd, Dwfl **dwflp, Dwarf_Addr *bias) +{ + Dwfl_Module *mod; + Dwarf *dbg = NULL; + + if (!dwflp) + return NULL; + + *dwflp = dwfl_begin(&offline_callbacks); + if (!*dwflp) + return NULL; + + mod = dwfl_report_offline(*dwflp, "", "", fd); + if (!mod) + goto error; + + dbg = dwfl_module_getdwarf(mod, bias); + if (!dbg) { +error: + dwfl_end(*dwflp); + *dwflp = NULL; + } + return dbg; +} + +#if _ELFUTILS_PREREQ(0, 148) +/* This method is buggy if elfutils is older than 0.148 */ +static int __linux_kernel_find_elf(Dwfl_Module *mod, + void **userdata, + const char *module_name, + Dwarf_Addr base, + char **file_name, Elf **elfp) +{ + int fd; + const char *path = kernel_get_module_path(module_name); + + pr_debug2("Use file %s for %s\n", path, module_name); + if (path) { + fd = open(path, O_RDONLY); + if (fd >= 0) { + *file_name = strdup(path); + return fd; + } + } + /* If failed, try to call standard method */ + return dwfl_linux_kernel_find_elf(mod, userdata, module_name, base, + file_name, elfp); +} + +static const Dwfl_Callbacks kernel_callbacks = { + .find_debuginfo = dwfl_standard_find_debuginfo, + .debuginfo_path = &debuginfo_path, + + .find_elf = __linux_kernel_find_elf, + .section_address = dwfl_linux_kernel_module_section_address, +}; + +/* Get a Dwarf from live kernel image */ +static Dwarf *dwfl_init_live_kernel_dwarf(Dwarf_Addr addr, Dwfl **dwflp, + Dwarf_Addr *bias) +{ + Dwarf *dbg; + + if (!dwflp) + return NULL; + + *dwflp = dwfl_begin(&kernel_callbacks); + if (!*dwflp) + return NULL; + + /* Load the kernel dwarves: Don't care the result here */ + dwfl_linux_kernel_report_kernel(*dwflp); + dwfl_linux_kernel_report_modules(*dwflp); + + dbg = dwfl_addrdwarf(*dwflp, addr, bias); + /* Here, check whether we could get a real dwarf */ + if (!dbg) { + pr_debug("Failed to find kernel dwarf at %lx\n", + (unsigned long)addr); + dwfl_end(*dwflp); + *dwflp = NULL; + } + return dbg; +} +#else +/* With older elfutils, this just support kernel module... */ +static Dwarf *dwfl_init_live_kernel_dwarf(Dwarf_Addr addr __used, Dwfl **dwflp, + Dwarf_Addr *bias) +{ + int fd; + const char *path = kernel_get_module_path("kernel"); + + if (!path) { + pr_err("Failed to find vmlinux path\n"); + return NULL; + } + + pr_debug2("Use file %s for debuginfo\n", path); + fd = open(path, O_RDONLY); + if (fd < 0) + return NULL; + + return dwfl_init_offline_dwarf(fd, dwflp, bias); +} +#endif + +/* Dwarf wrappers */ + +/* Find the realpath of the target file. */ +static const char *cu_find_realpath(Dwarf_Die *cu_die, const char *fname) +{ + Dwarf_Files *files; + size_t nfiles, i; + const char *src = NULL; + int ret; + + if (!fname) + return NULL; + + ret = dwarf_getsrcfiles(cu_die, &files, &nfiles); + if (ret != 0) + return NULL; + + for (i = 0; i < nfiles; i++) { + src = dwarf_filesrc(files, i, NULL, NULL); + if (strtailcmp(src, fname) == 0) + break; + } + if (i == nfiles) + return NULL; + return src; +} + +/* Get DW_AT_comp_dir (should be NULL with older gcc) */ +static const char *cu_get_comp_dir(Dwarf_Die *cu_die) +{ + Dwarf_Attribute attr; + if (dwarf_attr(cu_die, DW_AT_comp_dir, &attr) == NULL) + return NULL; + return dwarf_formstring(&attr); +} + +/* Get a line number and file name for given address */ +static int cu_find_lineinfo(Dwarf_Die *cudie, unsigned long addr, + const char **fname, int *lineno) +{ + Dwarf_Line *line; + Dwarf_Addr laddr; + + line = dwarf_getsrc_die(cudie, (Dwarf_Addr)addr); + if (line && dwarf_lineaddr(line, &laddr) == 0 && + addr == (unsigned long)laddr && dwarf_lineno(line, lineno) == 0) { + *fname = dwarf_linesrc(line, NULL, NULL); + if (!*fname) + /* line number is useless without filename */ + *lineno = 0; + } + + return *lineno ?: -ENOENT; +} + +/* Compare diename and tname */ +static bool die_compare_name(Dwarf_Die *dw_die, const char *tname) +{ + const char *name; + name = dwarf_diename(dw_die); + return name ? (strcmp(tname, name) == 0) : false; +} + +/* Get callsite line number of inline-function instance */ +static int die_get_call_lineno(Dwarf_Die *in_die) +{ + Dwarf_Attribute attr; + Dwarf_Word ret; + + if (!dwarf_attr(in_die, DW_AT_call_line, &attr)) + return -ENOENT; + + dwarf_formudata(&attr, &ret); + return (int)ret; +} + +/* Get type die */ +static Dwarf_Die *die_get_type(Dwarf_Die *vr_die, Dwarf_Die *die_mem) +{ + Dwarf_Attribute attr; + + if (dwarf_attr_integrate(vr_die, DW_AT_type, &attr) && + dwarf_formref_die(&attr, die_mem)) + return die_mem; + else + return NULL; +} + +/* Get a type die, but skip qualifiers */ +static Dwarf_Die *__die_get_real_type(Dwarf_Die *vr_die, Dwarf_Die *die_mem) +{ + int tag; + + do { + vr_die = die_get_type(vr_die, die_mem); + if (!vr_die) + break; + tag = dwarf_tag(vr_die); + } while (tag == DW_TAG_const_type || + tag == DW_TAG_restrict_type || + tag == DW_TAG_volatile_type || + tag == DW_TAG_shared_type); + + return vr_die; +} + +/* Get a type die, but skip qualifiers and typedef */ +static Dwarf_Die *die_get_real_type(Dwarf_Die *vr_die, Dwarf_Die *die_mem) +{ + do { + vr_die = __die_get_real_type(vr_die, die_mem); + } while (vr_die && dwarf_tag(vr_die) == DW_TAG_typedef); + + return vr_die; +} + +static int die_get_attr_udata(Dwarf_Die *tp_die, unsigned int attr_name, + Dwarf_Word *result) +{ + Dwarf_Attribute attr; + + if (dwarf_attr(tp_die, attr_name, &attr) == NULL || + dwarf_formudata(&attr, result) != 0) + return -ENOENT; + + return 0; +} + +static bool die_is_signed_type(Dwarf_Die *tp_die) +{ + Dwarf_Word ret; + + if (die_get_attr_udata(tp_die, DW_AT_encoding, &ret)) + return false; + + return (ret == DW_ATE_signed_char || ret == DW_ATE_signed || + ret == DW_ATE_signed_fixed); +} + +static int die_get_byte_size(Dwarf_Die *tp_die) +{ + Dwarf_Word ret; + + if (die_get_attr_udata(tp_die, DW_AT_byte_size, &ret)) + return 0; + + return (int)ret; +} + +static int die_get_bit_size(Dwarf_Die *tp_die) +{ + Dwarf_Word ret; + + if (die_get_attr_udata(tp_die, DW_AT_bit_size, &ret)) + return 0; + + return (int)ret; +} + +static int die_get_bit_offset(Dwarf_Die *tp_die) +{ + Dwarf_Word ret; + + if (die_get_attr_udata(tp_die, DW_AT_bit_offset, &ret)) + return 0; + + return (int)ret; +} + +/* Get data_member_location offset */ +static int die_get_data_member_location(Dwarf_Die *mb_die, Dwarf_Word *offs) +{ + Dwarf_Attribute attr; + Dwarf_Op *expr; + size_t nexpr; + int ret; + + if (dwarf_attr(mb_die, DW_AT_data_member_location, &attr) == NULL) + return -ENOENT; + + if (dwarf_formudata(&attr, offs) != 0) { + /* DW_AT_data_member_location should be DW_OP_plus_uconst */ + ret = dwarf_getlocation(&attr, &expr, &nexpr); + if (ret < 0 || nexpr == 0) + return -ENOENT; + + if (expr[0].atom != DW_OP_plus_uconst || nexpr != 1) { + pr_debug("Unable to get offset:Unexpected OP %x (%zd)\n", + expr[0].atom, nexpr); + return -ENOTSUP; + } + *offs = (Dwarf_Word)expr[0].number; + } + return 0; +} + +/* Return values for die_find callbacks */ +enum { + DIE_FIND_CB_FOUND = 0, /* End of Search */ + DIE_FIND_CB_CHILD = 1, /* Search only children */ + DIE_FIND_CB_SIBLING = 2, /* Search only siblings */ + DIE_FIND_CB_CONTINUE = 3, /* Search children and siblings */ +}; + +/* Search a child die */ +static Dwarf_Die *die_find_child(Dwarf_Die *rt_die, + int (*callback)(Dwarf_Die *, void *), + void *data, Dwarf_Die *die_mem) +{ + Dwarf_Die child_die; + int ret; + + ret = dwarf_child(rt_die, die_mem); + if (ret != 0) + return NULL; + + do { + ret = callback(die_mem, data); + if (ret == DIE_FIND_CB_FOUND) + return die_mem; + + if ((ret & DIE_FIND_CB_CHILD) && + die_find_child(die_mem, callback, data, &child_die)) { + memcpy(die_mem, &child_die, sizeof(Dwarf_Die)); + return die_mem; + } + } while ((ret & DIE_FIND_CB_SIBLING) && + dwarf_siblingof(die_mem, die_mem) == 0); + + return NULL; +} + +struct __addr_die_search_param { + Dwarf_Addr addr; + Dwarf_Die *die_mem; +}; + +static int __die_search_func_cb(Dwarf_Die *fn_die, void *data) +{ + struct __addr_die_search_param *ad = data; + + if (dwarf_tag(fn_die) == DW_TAG_subprogram && + dwarf_haspc(fn_die, ad->addr)) { + memcpy(ad->die_mem, fn_die, sizeof(Dwarf_Die)); + return DWARF_CB_ABORT; + } + return DWARF_CB_OK; +} + +/* Search a real subprogram including this line, */ +static Dwarf_Die *die_find_real_subprogram(Dwarf_Die *cu_die, Dwarf_Addr addr, + Dwarf_Die *die_mem) +{ + struct __addr_die_search_param ad; + ad.addr = addr; + ad.die_mem = die_mem; + /* dwarf_getscopes can't find subprogram. */ + if (!dwarf_getfuncs(cu_die, __die_search_func_cb, &ad, 0)) + return NULL; + else + return die_mem; +} + +/* die_find callback for inline function search */ +static int __die_find_inline_cb(Dwarf_Die *die_mem, void *data) +{ + Dwarf_Addr *addr = data; + + if (dwarf_tag(die_mem) == DW_TAG_inlined_subroutine && + dwarf_haspc(die_mem, *addr)) + return DIE_FIND_CB_FOUND; + + return DIE_FIND_CB_CONTINUE; +} + +/* Similar to dwarf_getfuncs, but returns inlined_subroutine if exists. */ +static Dwarf_Die *die_find_inlinefunc(Dwarf_Die *sp_die, Dwarf_Addr addr, + Dwarf_Die *die_mem) +{ + Dwarf_Die tmp_die; + + sp_die = die_find_child(sp_die, __die_find_inline_cb, &addr, &tmp_die); + if (!sp_die) + return NULL; + + /* Inlined function could be recursive. Trace it until fail */ + while (sp_die) { + memcpy(die_mem, sp_die, sizeof(Dwarf_Die)); + sp_die = die_find_child(sp_die, __die_find_inline_cb, &addr, + &tmp_die); + } + + return die_mem; +} + +/* Walker on lines (Note: line number will not be sorted) */ +typedef int (* line_walk_handler_t) (const char *fname, int lineno, + Dwarf_Addr addr, void *data); + +struct __line_walk_param { + const char *fname; + line_walk_handler_t handler; + void *data; + int retval; +}; + +static int __die_walk_funclines_cb(Dwarf_Die *in_die, void *data) +{ + struct __line_walk_param *lw = data; + Dwarf_Addr addr; + int lineno; + + if (dwarf_tag(in_die) == DW_TAG_inlined_subroutine) { + lineno = die_get_call_lineno(in_die); + if (lineno > 0 && dwarf_entrypc(in_die, &addr) == 0) { + lw->retval = lw->handler(lw->fname, lineno, addr, + lw->data); + if (lw->retval != 0) + return DIE_FIND_CB_FOUND; + } + } + return DIE_FIND_CB_SIBLING; +} + +/* Walk on lines of blocks included in given DIE */ +static int __die_walk_funclines(Dwarf_Die *sp_die, + line_walk_handler_t handler, void *data) +{ + struct __line_walk_param lw = { + .handler = handler, + .data = data, + .retval = 0, + }; + Dwarf_Die die_mem; + Dwarf_Addr addr; + int lineno; + + /* Handle function declaration line */ + lw.fname = dwarf_decl_file(sp_die); + if (lw.fname && dwarf_decl_line(sp_die, &lineno) == 0 && + dwarf_entrypc(sp_die, &addr) == 0) { + lw.retval = handler(lw.fname, lineno, addr, data); + if (lw.retval != 0) + goto done; + } + die_find_child(sp_die, __die_walk_funclines_cb, &lw, &die_mem); +done: + return lw.retval; +} + +static int __die_walk_culines_cb(Dwarf_Die *sp_die, void *data) +{ + struct __line_walk_param *lw = data; + + lw->retval = __die_walk_funclines(sp_die, lw->handler, lw->data); + if (lw->retval != 0) + return DWARF_CB_ABORT; + + return DWARF_CB_OK; +} + +/* + * Walk on lines inside given PDIE. If the PDIE is subprogram, walk only on + * the lines inside the subprogram, otherwise PDIE must be a CU DIE. + */ +static int die_walk_lines(Dwarf_Die *pdie, line_walk_handler_t handler, + void *data) +{ + Dwarf_Lines *lines; + Dwarf_Line *line; + Dwarf_Addr addr; + const char *fname; + int lineno, ret = 0; + Dwarf_Die die_mem, *cu_die; + size_t nlines, i; + + /* Get the CU die */ + if (dwarf_tag(pdie) == DW_TAG_subprogram) + cu_die = dwarf_diecu(pdie, &die_mem, NULL, NULL); + else + cu_die = pdie; + if (!cu_die) { + pr_debug2("Failed to get CU from subprogram\n"); + return -EINVAL; + } + + /* Get lines list in the CU */ + if (dwarf_getsrclines(cu_die, &lines, &nlines) != 0) { + pr_debug2("Failed to get source lines on this CU.\n"); + return -ENOENT; + } + pr_debug2("Get %zd lines from this CU\n", nlines); + + /* Walk on the lines on lines list */ + for (i = 0; i < nlines; i++) { + line = dwarf_onesrcline(lines, i); + if (line == NULL || + dwarf_lineno(line, &lineno) != 0 || + dwarf_lineaddr(line, &addr) != 0) { + pr_debug2("Failed to get line info. " + "Possible error in debuginfo.\n"); + continue; + } + /* Filter lines based on address */ + if (pdie != cu_die) + /* + * Address filtering + * The line is included in given function, and + * no inline block includes it. + */ + if (!dwarf_haspc(pdie, addr) || + die_find_inlinefunc(pdie, addr, &die_mem)) + continue; + /* Get source line */ + fname = dwarf_linesrc(line, NULL, NULL); + + ret = handler(fname, lineno, addr, data); + if (ret != 0) + return ret; + } + + /* + * Dwarf lines doesn't include function declarations and inlined + * subroutines. We have to check functions list or given function. + */ + if (pdie != cu_die) + ret = __die_walk_funclines(pdie, handler, data); + else { + struct __line_walk_param param = { + .handler = handler, + .data = data, + .retval = 0, + }; + dwarf_getfuncs(cu_die, __die_walk_culines_cb, ¶m, 0); + ret = param.retval; + } + + return ret; +} + +struct __find_variable_param { + const char *name; + Dwarf_Addr addr; +}; + +static int __die_find_variable_cb(Dwarf_Die *die_mem, void *data) +{ + struct __find_variable_param *fvp = data; + int tag; + + tag = dwarf_tag(die_mem); + if ((tag == DW_TAG_formal_parameter || + tag == DW_TAG_variable) && + die_compare_name(die_mem, fvp->name)) + return DIE_FIND_CB_FOUND; + + if (dwarf_haspc(die_mem, fvp->addr)) + return DIE_FIND_CB_CONTINUE; + else + return DIE_FIND_CB_SIBLING; +} + +/* Find a variable called 'name' at given address */ +static Dwarf_Die *die_find_variable_at(Dwarf_Die *sp_die, const char *name, + Dwarf_Addr addr, Dwarf_Die *die_mem) +{ + struct __find_variable_param fvp = { .name = name, .addr = addr}; + + return die_find_child(sp_die, __die_find_variable_cb, (void *)&fvp, + die_mem); +} + +static int __die_find_member_cb(Dwarf_Die *die_mem, void *data) +{ + const char *name = data; + + if ((dwarf_tag(die_mem) == DW_TAG_member) && + die_compare_name(die_mem, name)) + return DIE_FIND_CB_FOUND; + + return DIE_FIND_CB_SIBLING; +} + +/* Find a member called 'name' */ +static Dwarf_Die *die_find_member(Dwarf_Die *st_die, const char *name, + Dwarf_Die *die_mem) +{ + return die_find_child(st_die, __die_find_member_cb, (void *)name, + die_mem); +} + +/* Get the name of given variable DIE */ +static int die_get_typename(Dwarf_Die *vr_die, char *buf, int len) +{ + Dwarf_Die type; + int tag, ret, ret2; + const char *tmp = ""; + + if (__die_get_real_type(vr_die, &type) == NULL) + return -ENOENT; + + tag = dwarf_tag(&type); + if (tag == DW_TAG_array_type || tag == DW_TAG_pointer_type) + tmp = "*"; + else if (tag == DW_TAG_subroutine_type) { + /* Function pointer */ + ret = snprintf(buf, len, "(function_type)"); + return (ret >= len) ? -E2BIG : ret; + } else { + if (!dwarf_diename(&type)) + return -ENOENT; + if (tag == DW_TAG_union_type) + tmp = "union "; + else if (tag == DW_TAG_structure_type) + tmp = "struct "; + /* Write a base name */ + ret = snprintf(buf, len, "%s%s", tmp, dwarf_diename(&type)); + return (ret >= len) ? -E2BIG : ret; + } + ret = die_get_typename(&type, buf, len); + if (ret > 0) { + ret2 = snprintf(buf + ret, len - ret, "%s", tmp); + ret = (ret2 >= len - ret) ? -E2BIG : ret2 + ret; + } + return ret; +} + +/* Get the name and type of given variable DIE, stored as "type\tname" */ +static int die_get_varname(Dwarf_Die *vr_die, char *buf, int len) +{ + int ret, ret2; + + ret = die_get_typename(vr_die, buf, len); + if (ret < 0) { + pr_debug("Failed to get type, make it unknown.\n"); + ret = snprintf(buf, len, "(unknown_type)"); + } + if (ret > 0) { + ret2 = snprintf(buf + ret, len - ret, "\t%s", + dwarf_diename(vr_die)); + ret = (ret2 >= len - ret) ? -E2BIG : ret2 + ret; + } + return ret; +} + +/* + * Probe finder related functions + */ + +static struct probe_trace_arg_ref *alloc_trace_arg_ref(long offs) +{ + struct probe_trace_arg_ref *ref; + ref = zalloc(sizeof(struct probe_trace_arg_ref)); + if (ref != NULL) + ref->offset = offs; + return ref; +} + +/* + * Convert a location into trace_arg. + * If tvar == NULL, this just checks variable can be converted. + */ +static int convert_variable_location(Dwarf_Die *vr_die, Dwarf_Addr addr, + Dwarf_Op *fb_ops, + struct probe_trace_arg *tvar) +{ + Dwarf_Attribute attr; + Dwarf_Op *op; + size_t nops; + unsigned int regn; + Dwarf_Word offs = 0; + bool ref = false; + const char *regs; + int ret; + + if (dwarf_attr(vr_die, DW_AT_external, &attr) != NULL) + goto static_var; + + /* TODO: handle more than 1 exprs */ + if (dwarf_attr(vr_die, DW_AT_location, &attr) == NULL || + dwarf_getlocation_addr(&attr, addr, &op, &nops, 1) <= 0 || + nops == 0) { + /* TODO: Support const_value */ + return -ENOENT; + } + + if (op->atom == DW_OP_addr) { +static_var: + if (!tvar) + return 0; + /* Static variables on memory (not stack), make @varname */ + ret = strlen(dwarf_diename(vr_die)); + tvar->value = zalloc(ret + 2); + if (tvar->value == NULL) + return -ENOMEM; + snprintf(tvar->value, ret + 2, "@%s", dwarf_diename(vr_die)); + tvar->ref = alloc_trace_arg_ref((long)offs); + if (tvar->ref == NULL) + return -ENOMEM; + return 0; + } + + /* If this is based on frame buffer, set the offset */ + if (op->atom == DW_OP_fbreg) { + if (fb_ops == NULL) + return -ENOTSUP; + ref = true; + offs = op->number; + op = &fb_ops[0]; + } + + if (op->atom >= DW_OP_breg0 && op->atom <= DW_OP_breg31) { + regn = op->atom - DW_OP_breg0; + offs += op->number; + ref = true; + } else if (op->atom >= DW_OP_reg0 && op->atom <= DW_OP_reg31) { + regn = op->atom - DW_OP_reg0; + } else if (op->atom == DW_OP_bregx) { + regn = op->number; + offs += op->number2; + ref = true; + } else if (op->atom == DW_OP_regx) { + regn = op->number; + } else { + pr_debug("DW_OP %x is not supported.\n", op->atom); + return -ENOTSUP; + } + + if (!tvar) + return 0; + + regs = get_arch_regstr(regn); + if (!regs) { + /* This should be a bug in DWARF or this tool */ + pr_warning("Mapping for the register number %u " + "missing on this architecture.\n", regn); + return -ERANGE; + } + + tvar->value = strdup(regs); + if (tvar->value == NULL) + return -ENOMEM; + + if (ref) { + tvar->ref = alloc_trace_arg_ref((long)offs); + if (tvar->ref == NULL) + return -ENOMEM; + } + return 0; +} + +#define BYTES_TO_BITS(nb) ((nb) * BITS_PER_LONG / sizeof(long)) + +static int convert_variable_type(Dwarf_Die *vr_die, + struct probe_trace_arg *tvar, + const char *cast) +{ + struct probe_trace_arg_ref **ref_ptr = &tvar->ref; + Dwarf_Die type; + char buf[16]; + int ret; + + /* TODO: check all types */ + if (cast && strcmp(cast, "string") != 0) { + /* Non string type is OK */ + tvar->type = strdup(cast); + return (tvar->type == NULL) ? -ENOMEM : 0; + } + + if (die_get_bit_size(vr_die) != 0) { + /* This is a bitfield */ + ret = snprintf(buf, 16, "b%d@%d/%zd", die_get_bit_size(vr_die), + die_get_bit_offset(vr_die), + BYTES_TO_BITS(die_get_byte_size(vr_die))); + goto formatted; + } + + if (die_get_real_type(vr_die, &type) == NULL) { + pr_warning("Failed to get a type information of %s.\n", + dwarf_diename(vr_die)); + return -ENOENT; + } + + pr_debug("%s type is %s.\n", + dwarf_diename(vr_die), dwarf_diename(&type)); + + if (cast && strcmp(cast, "string") == 0) { /* String type */ + ret = dwarf_tag(&type); + if (ret != DW_TAG_pointer_type && + ret != DW_TAG_array_type) { + pr_warning("Failed to cast into string: " + "%s(%s) is not a pointer nor array.\n", + dwarf_diename(vr_die), dwarf_diename(&type)); + return -EINVAL; + } + if (ret == DW_TAG_pointer_type) { + if (die_get_real_type(&type, &type) == NULL) { + pr_warning("Failed to get a type" + " information.\n"); + return -ENOENT; + } + while (*ref_ptr) + ref_ptr = &(*ref_ptr)->next; + /* Add new reference with offset +0 */ + *ref_ptr = zalloc(sizeof(struct probe_trace_arg_ref)); + if (*ref_ptr == NULL) { + pr_warning("Out of memory error\n"); + return -ENOMEM; + } + } + if (!die_compare_name(&type, "char") && + !die_compare_name(&type, "unsigned char")) { + pr_warning("Failed to cast into string: " + "%s is not (unsigned) char *.\n", + dwarf_diename(vr_die)); + return -EINVAL; + } + tvar->type = strdup(cast); + return (tvar->type == NULL) ? -ENOMEM : 0; + } + + ret = BYTES_TO_BITS(die_get_byte_size(&type)); + if (!ret) + /* No size ... try to use default type */ + return 0; + + /* Check the bitwidth */ + if (ret > MAX_BASIC_TYPE_BITS) { + pr_info("%s exceeds max-bitwidth. Cut down to %d bits.\n", + dwarf_diename(&type), MAX_BASIC_TYPE_BITS); + ret = MAX_BASIC_TYPE_BITS; + } + ret = snprintf(buf, 16, "%c%d", + die_is_signed_type(&type) ? 's' : 'u', ret); + +formatted: + if (ret < 0 || ret >= 16) { + if (ret >= 16) + ret = -E2BIG; + pr_warning("Failed to convert variable type: %s\n", + strerror(-ret)); + return ret; + } + tvar->type = strdup(buf); + if (tvar->type == NULL) + return -ENOMEM; + return 0; +} + +static int convert_variable_fields(Dwarf_Die *vr_die, const char *varname, + struct perf_probe_arg_field *field, + struct probe_trace_arg_ref **ref_ptr, + Dwarf_Die *die_mem) +{ + struct probe_trace_arg_ref *ref = *ref_ptr; + Dwarf_Die type; + Dwarf_Word offs; + int ret, tag; + + pr_debug("converting %s in %s\n", field->name, varname); + if (die_get_real_type(vr_die, &type) == NULL) { + pr_warning("Failed to get the type of %s.\n", varname); + return -ENOENT; + } + pr_debug2("Var real type: (%x)\n", (unsigned)dwarf_dieoffset(&type)); + tag = dwarf_tag(&type); + + if (field->name[0] == '[' && + (tag == DW_TAG_array_type || tag == DW_TAG_pointer_type)) { + if (field->next) + /* Save original type for next field */ + memcpy(die_mem, &type, sizeof(*die_mem)); + /* Get the type of this array */ + if (die_get_real_type(&type, &type) == NULL) { + pr_warning("Failed to get the type of %s.\n", varname); + return -ENOENT; + } + pr_debug2("Array real type: (%x)\n", + (unsigned)dwarf_dieoffset(&type)); + if (tag == DW_TAG_pointer_type) { + ref = zalloc(sizeof(struct probe_trace_arg_ref)); + if (ref == NULL) + return -ENOMEM; + if (*ref_ptr) + (*ref_ptr)->next = ref; + else + *ref_ptr = ref; + } + ref->offset += die_get_byte_size(&type) * field->index; + if (!field->next) + /* Save vr_die for converting types */ + memcpy(die_mem, vr_die, sizeof(*die_mem)); + goto next; + } else if (tag == DW_TAG_pointer_type) { + /* Check the pointer and dereference */ + if (!field->ref) { + pr_err("Semantic error: %s must be referred by '->'\n", + field->name); + return -EINVAL; + } + /* Get the type pointed by this pointer */ + if (die_get_real_type(&type, &type) == NULL) { + pr_warning("Failed to get the type of %s.\n", varname); + return -ENOENT; + } + /* Verify it is a data structure */ + if (dwarf_tag(&type) != DW_TAG_structure_type) { + pr_warning("%s is not a data structure.\n", varname); + return -EINVAL; + } + + ref = zalloc(sizeof(struct probe_trace_arg_ref)); + if (ref == NULL) + return -ENOMEM; + if (*ref_ptr) + (*ref_ptr)->next = ref; + else + *ref_ptr = ref; + } else { + /* Verify it is a data structure */ + if (tag != DW_TAG_structure_type) { + pr_warning("%s is not a data structure.\n", varname); + return -EINVAL; + } + if (field->name[0] == '[') { + pr_err("Semantic error: %s is not a pointor" + " nor array.\n", varname); + return -EINVAL; + } + if (field->ref) { + pr_err("Semantic error: %s must be referred by '.'\n", + field->name); + return -EINVAL; + } + if (!ref) { + pr_warning("Structure on a register is not " + "supported yet.\n"); + return -ENOTSUP; + } + } + + if (die_find_member(&type, field->name, die_mem) == NULL) { + pr_warning("%s(tyep:%s) has no member %s.\n", varname, + dwarf_diename(&type), field->name); + return -EINVAL; + } + + /* Get the offset of the field */ + ret = die_get_data_member_location(die_mem, &offs); + if (ret < 0) { + pr_warning("Failed to get the offset of %s.\n", field->name); + return ret; + } + ref->offset += (long)offs; + +next: + /* Converting next field */ + if (field->next) + return convert_variable_fields(die_mem, field->name, + field->next, &ref, die_mem); + else + return 0; +} + +/* Show a variables in kprobe event format */ +static int convert_variable(Dwarf_Die *vr_die, struct probe_finder *pf) +{ + Dwarf_Die die_mem; + int ret; + + pr_debug("Converting variable %s into trace event.\n", + dwarf_diename(vr_die)); + + ret = convert_variable_location(vr_die, pf->addr, pf->fb_ops, + pf->tvar); + if (ret == -ENOENT) + pr_err("Failed to find the location of %s at this address.\n" + " Perhaps, it has been optimized out.\n", pf->pvar->var); + else if (ret == -ENOTSUP) + pr_err("Sorry, we don't support this variable location yet.\n"); + else if (pf->pvar->field) { + ret = convert_variable_fields(vr_die, pf->pvar->var, + pf->pvar->field, &pf->tvar->ref, + &die_mem); + vr_die = &die_mem; + } + if (ret == 0) + ret = convert_variable_type(vr_die, pf->tvar, pf->pvar->type); + /* *expr will be cached in libdw. Don't free it. */ + return ret; +} + +/* Find a variable in a subprogram die */ +static int find_variable(Dwarf_Die *sp_die, struct probe_finder *pf) +{ + Dwarf_Die vr_die, *scopes; + char buf[32], *ptr; + int ret, nscopes; + + if (!is_c_varname(pf->pvar->var)) { + /* Copy raw parameters */ + pf->tvar->value = strdup(pf->pvar->var); + if (pf->tvar->value == NULL) + return -ENOMEM; + if (pf->pvar->type) { + pf->tvar->type = strdup(pf->pvar->type); + if (pf->tvar->type == NULL) + return -ENOMEM; + } + if (pf->pvar->name) { + pf->tvar->name = strdup(pf->pvar->name); + if (pf->tvar->name == NULL) + return -ENOMEM; + } else + pf->tvar->name = NULL; + return 0; + } + + if (pf->pvar->name) + pf->tvar->name = strdup(pf->pvar->name); + else { + ret = synthesize_perf_probe_arg(pf->pvar, buf, 32); + if (ret < 0) + return ret; + ptr = strchr(buf, ':'); /* Change type separator to _ */ + if (ptr) + *ptr = '_'; + pf->tvar->name = strdup(buf); + } + if (pf->tvar->name == NULL) + return -ENOMEM; + + pr_debug("Searching '%s' variable in context.\n", + pf->pvar->var); + /* Search child die for local variables and parameters. */ + if (die_find_variable_at(sp_die, pf->pvar->var, pf->addr, &vr_die)) + ret = convert_variable(&vr_die, pf); + else { + /* Search upper class */ + nscopes = dwarf_getscopes_die(sp_die, &scopes); + while (nscopes-- > 1) { + pr_debug("Searching variables in %s\n", + dwarf_diename(&scopes[nscopes])); + /* We should check this scope, so give dummy address */ + if (die_find_variable_at(&scopes[nscopes], + pf->pvar->var, 0, + &vr_die)) { + ret = convert_variable(&vr_die, pf); + goto found; + } + } + if (scopes) + free(scopes); + ret = -ENOENT; + } +found: + if (ret < 0) + pr_warning("Failed to find '%s' in this function.\n", + pf->pvar->var); + return ret; +} + +/* Convert subprogram DIE to trace point */ +static int convert_to_trace_point(Dwarf_Die *sp_die, Dwarf_Addr paddr, + bool retprobe, struct probe_trace_point *tp) +{ + Dwarf_Addr eaddr; + const char *name; + + /* Copy the name of probe point */ + name = dwarf_diename(sp_die); + if (name) { + if (dwarf_entrypc(sp_die, &eaddr) != 0) { + pr_warning("Failed to get entry address of %s\n", + dwarf_diename(sp_die)); + return -ENOENT; + } + tp->symbol = strdup(name); + if (tp->symbol == NULL) + return -ENOMEM; + tp->offset = (unsigned long)(paddr - eaddr); + } else + /* This function has no name. */ + tp->offset = (unsigned long)paddr; + + /* Return probe must be on the head of a subprogram */ + if (retprobe) { + if (eaddr != paddr) { + pr_warning("Return probe must be on the head of" + " a real function.\n"); + return -EINVAL; + } + tp->retprobe = true; + } + + return 0; +} + +/* Call probe_finder callback with real subprogram DIE */ +static int call_probe_finder(Dwarf_Die *sp_die, struct probe_finder *pf) +{ + Dwarf_Die die_mem; + Dwarf_Attribute fb_attr; + size_t nops; + int ret; + + /* If no real subprogram, find a real one */ + if (!sp_die || dwarf_tag(sp_die) != DW_TAG_subprogram) { + sp_die = die_find_real_subprogram(&pf->cu_die, + pf->addr, &die_mem); + if (!sp_die) { + pr_warning("Failed to find probe point in any " + "functions.\n"); + return -ENOENT; + } + } + + /* Get the frame base attribute/ops */ + dwarf_attr(sp_die, DW_AT_frame_base, &fb_attr); + ret = dwarf_getlocation_addr(&fb_attr, pf->addr, &pf->fb_ops, &nops, 1); + if (ret <= 0 || nops == 0) { + pf->fb_ops = NULL; +#if _ELFUTILS_PREREQ(0, 142) + } else if (nops == 1 && pf->fb_ops[0].atom == DW_OP_call_frame_cfa && + pf->cfi != NULL) { + Dwarf_Frame *frame; + if (dwarf_cfi_addrframe(pf->cfi, pf->addr, &frame) != 0 || + dwarf_frame_cfa(frame, &pf->fb_ops, &nops) != 0) { + pr_warning("Failed to get call frame on 0x%jx\n", + (uintmax_t)pf->addr); + return -ENOENT; + } +#endif + } + + /* Call finder's callback handler */ + ret = pf->callback(sp_die, pf); + + /* *pf->fb_ops will be cached in libdw. Don't free it. */ + pf->fb_ops = NULL; + + return ret; +} + +static int probe_point_line_walker(const char *fname, int lineno, + Dwarf_Addr addr, void *data) +{ + struct probe_finder *pf = data; + int ret; + + if (lineno != pf->lno || strtailcmp(fname, pf->fname) != 0) + return 0; + + pf->addr = addr; + ret = call_probe_finder(NULL, pf); + + /* Continue if no error, because the line will be in inline function */ + return ret < 0 ? ret : 0; +} + +/* Find probe point from its line number */ +static int find_probe_point_by_line(struct probe_finder *pf) +{ + return die_walk_lines(&pf->cu_die, probe_point_line_walker, pf); +} + +/* Find lines which match lazy pattern */ +static int find_lazy_match_lines(struct list_head *head, + const char *fname, const char *pat) +{ + FILE *fp; + char *line = NULL; + size_t line_len; + ssize_t len; + int count = 0, linenum = 1; + + fp = fopen(fname, "r"); + if (!fp) { + pr_warning("Failed to open %s: %s\n", fname, strerror(errno)); + return -errno; + } + + while ((len = getline(&line, &line_len, fp)) > 0) { + + if (line[len - 1] == '\n') + line[len - 1] = '\0'; + + if (strlazymatch(line, pat)) { + line_list__add_line(head, linenum); + count++; + } + linenum++; + } + + if (ferror(fp)) + count = -errno; + free(line); + fclose(fp); + + if (count == 0) + pr_debug("No matched lines found in %s.\n", fname); + return count; +} + +static int probe_point_lazy_walker(const char *fname, int lineno, + Dwarf_Addr addr, void *data) +{ + struct probe_finder *pf = data; + int ret; + + if (!line_list__has_line(&pf->lcache, lineno) || + strtailcmp(fname, pf->fname) != 0) + return 0; + + pr_debug("Probe line found: line:%d addr:0x%llx\n", + lineno, (unsigned long long)addr); + pf->addr = addr; + ret = call_probe_finder(NULL, pf); + + /* + * Continue if no error, because the lazy pattern will match + * to other lines + */ + return ret < 0 ? ret : 0; +} + +/* Find probe points from lazy pattern */ +static int find_probe_point_lazy(Dwarf_Die *sp_die, struct probe_finder *pf) +{ + int ret = 0; + + if (list_empty(&pf->lcache)) { + /* Matching lazy line pattern */ + ret = find_lazy_match_lines(&pf->lcache, pf->fname, + pf->pev->point.lazy_line); + if (ret <= 0) + return ret; + } + + return die_walk_lines(sp_die, probe_point_lazy_walker, pf); +} + +/* Callback parameter with return value */ +struct dwarf_callback_param { + void *data; + int retval; +}; + +static int probe_point_inline_cb(Dwarf_Die *in_die, void *data) +{ + struct dwarf_callback_param *param = data; + struct probe_finder *pf = param->data; + struct perf_probe_point *pp = &pf->pev->point; + Dwarf_Addr addr; + + if (pp->lazy_line) + param->retval = find_probe_point_lazy(in_die, pf); + else { + /* Get probe address */ + if (dwarf_entrypc(in_die, &addr) != 0) { + pr_warning("Failed to get entry address of %s.\n", + dwarf_diename(in_die)); + param->retval = -ENOENT; + return DWARF_CB_ABORT; + } + pf->addr = addr; + pf->addr += pp->offset; + pr_debug("found inline addr: 0x%jx\n", + (uintmax_t)pf->addr); + + param->retval = call_probe_finder(in_die, pf); + if (param->retval < 0) + return DWARF_CB_ABORT; + } + + return DWARF_CB_OK; +} + +/* Search function from function name */ +static int probe_point_search_cb(Dwarf_Die *sp_die, void *data) +{ + struct dwarf_callback_param *param = data; + struct probe_finder *pf = param->data; + struct perf_probe_point *pp = &pf->pev->point; + + /* Check tag and diename */ + if (dwarf_tag(sp_die) != DW_TAG_subprogram || + !die_compare_name(sp_die, pp->function)) + return DWARF_CB_OK; + + /* Check declared file */ + if (pp->file && strtailcmp(pp->file, dwarf_decl_file(sp_die))) + return DWARF_CB_OK; + + pf->fname = dwarf_decl_file(sp_die); + if (pp->line) { /* Function relative line */ + dwarf_decl_line(sp_die, &pf->lno); + pf->lno += pp->line; + param->retval = find_probe_point_by_line(pf); + } else if (!dwarf_func_inline(sp_die)) { + /* Real function */ + if (pp->lazy_line) + param->retval = find_probe_point_lazy(sp_die, pf); + else { + if (dwarf_entrypc(sp_die, &pf->addr) != 0) { + pr_warning("Failed to get entry address of " + "%s.\n", dwarf_diename(sp_die)); + param->retval = -ENOENT; + return DWARF_CB_ABORT; + } + pf->addr += pp->offset; + /* TODO: Check the address in this function */ + param->retval = call_probe_finder(sp_die, pf); + } + } else { + struct dwarf_callback_param _param = {.data = (void *)pf, + .retval = 0}; + /* Inlined function: search instances */ + dwarf_func_inline_instances(sp_die, probe_point_inline_cb, + &_param); + param->retval = _param.retval; + } + + return DWARF_CB_ABORT; /* Exit; no same symbol in this CU. */ +} + +static int find_probe_point_by_func(struct probe_finder *pf) +{ + struct dwarf_callback_param _param = {.data = (void *)pf, + .retval = 0}; + dwarf_getfuncs(&pf->cu_die, probe_point_search_cb, &_param, 0); + return _param.retval; +} + +struct pubname_callback_param { + char *function; + char *file; + Dwarf_Die *cu_die; + Dwarf_Die *sp_die; + int found; +}; + +static int pubname_search_cb(Dwarf *dbg, Dwarf_Global *gl, void *data) +{ + struct pubname_callback_param *param = data; + + if (dwarf_offdie(dbg, gl->die_offset, param->sp_die)) { + if (dwarf_tag(param->sp_die) != DW_TAG_subprogram) + return DWARF_CB_OK; + + if (die_compare_name(param->sp_die, param->function)) { + if (!dwarf_offdie(dbg, gl->cu_offset, param->cu_die)) + return DWARF_CB_OK; + + if (param->file && + strtailcmp(param->file, dwarf_decl_file(param->sp_die))) + return DWARF_CB_OK; + + param->found = 1; + return DWARF_CB_ABORT; + } + } + + return DWARF_CB_OK; +} + +/* Find probe points from debuginfo */ +static int find_probes(int fd, struct probe_finder *pf) +{ + struct perf_probe_point *pp = &pf->pev->point; + Dwarf_Off off, noff; + size_t cuhl; + Dwarf_Die *diep; + Dwarf *dbg = NULL; + Dwfl *dwfl; + Dwarf_Addr bias; /* Currently ignored */ + int ret = 0; + + dbg = dwfl_init_offline_dwarf(fd, &dwfl, &bias); + if (!dbg) { + pr_warning("No debug information found in the vmlinux - " + "please rebuild with CONFIG_DEBUG_INFO=y.\n"); + close(fd); /* Without dwfl_end(), fd isn't closed. */ + return -EBADF; + } + +#if _ELFUTILS_PREREQ(0, 142) + /* Get the call frame information from this dwarf */ + pf->cfi = dwarf_getcfi(dbg); +#endif + + off = 0; + line_list__init(&pf->lcache); + + /* Fastpath: lookup by function name from .debug_pubnames section */ + if (pp->function) { + struct pubname_callback_param pubname_param = { + .function = pp->function, + .file = pp->file, + .cu_die = &pf->cu_die, + .sp_die = &pf->sp_die, + .found = 0, + }; + struct dwarf_callback_param probe_param = { + .data = pf, + }; + + dwarf_getpubnames(dbg, pubname_search_cb, &pubname_param, 0); + if (pubname_param.found) { + ret = probe_point_search_cb(&pf->sp_die, &probe_param); + if (ret) + goto found; + } + } + + /* Loop on CUs (Compilation Unit) */ + while (!dwarf_nextcu(dbg, off, &noff, &cuhl, NULL, NULL, NULL)) { + /* Get the DIE(Debugging Information Entry) of this CU */ + diep = dwarf_offdie(dbg, off + cuhl, &pf->cu_die); + if (!diep) + continue; + + /* Check if target file is included. */ + if (pp->file) + pf->fname = cu_find_realpath(&pf->cu_die, pp->file); + else + pf->fname = NULL; + + if (!pp->file || pf->fname) { + if (pp->function) + ret = find_probe_point_by_func(pf); + else if (pp->lazy_line) + ret = find_probe_point_lazy(NULL, pf); + else { + pf->lno = pp->line; + ret = find_probe_point_by_line(pf); + } + if (ret < 0) + break; + } + off = noff; + } + +found: + line_list__free(&pf->lcache); + if (dwfl) + dwfl_end(dwfl); + + return ret; +} + +/* Add a found probe point into trace event list */ +static int add_probe_trace_event(Dwarf_Die *sp_die, struct probe_finder *pf) +{ + struct trace_event_finder *tf = + container_of(pf, struct trace_event_finder, pf); + struct probe_trace_event *tev; + int ret, i; + + /* Check number of tevs */ + if (tf->ntevs == tf->max_tevs) { + pr_warning("Too many( > %d) probe point found.\n", + tf->max_tevs); + return -ERANGE; + } + tev = &tf->tevs[tf->ntevs++]; + + ret = convert_to_trace_point(sp_die, pf->addr, pf->pev->point.retprobe, + &tev->point); + if (ret < 0) + return ret; + + pr_debug("Probe point found: %s+%lu\n", tev->point.symbol, + tev->point.offset); + + /* Find each argument */ + tev->nargs = pf->pev->nargs; + tev->args = zalloc(sizeof(struct probe_trace_arg) * tev->nargs); + if (tev->args == NULL) + return -ENOMEM; + for (i = 0; i < pf->pev->nargs; i++) { + pf->pvar = &pf->pev->args[i]; + pf->tvar = &tev->args[i]; + ret = find_variable(sp_die, pf); + if (ret != 0) + return ret; + } + + return 0; +} + +/* Find probe_trace_events specified by perf_probe_event from debuginfo */ +int find_probe_trace_events(int fd, struct perf_probe_event *pev, + struct probe_trace_event **tevs, int max_tevs) +{ + struct trace_event_finder tf = { + .pf = {.pev = pev, .callback = add_probe_trace_event}, + .max_tevs = max_tevs}; + int ret; + + /* Allocate result tevs array */ + *tevs = zalloc(sizeof(struct probe_trace_event) * max_tevs); + if (*tevs == NULL) + return -ENOMEM; + + tf.tevs = *tevs; + tf.ntevs = 0; + + ret = find_probes(fd, &tf.pf); + if (ret < 0) { + free(*tevs); + *tevs = NULL; + return ret; + } + + return (ret < 0) ? ret : tf.ntevs; +} + +#define MAX_VAR_LEN 64 + +/* Collect available variables in this scope */ +static int collect_variables_cb(Dwarf_Die *die_mem, void *data) +{ + struct available_var_finder *af = data; + struct variable_list *vl; + char buf[MAX_VAR_LEN]; + int tag, ret; + + vl = &af->vls[af->nvls - 1]; + + tag = dwarf_tag(die_mem); + if (tag == DW_TAG_formal_parameter || + tag == DW_TAG_variable) { + ret = convert_variable_location(die_mem, af->pf.addr, + af->pf.fb_ops, NULL); + if (ret == 0) { + ret = die_get_varname(die_mem, buf, MAX_VAR_LEN); + pr_debug2("Add new var: %s\n", buf); + if (ret > 0) + strlist__add(vl->vars, buf); + } + } + + if (af->child && dwarf_haspc(die_mem, af->pf.addr)) + return DIE_FIND_CB_CONTINUE; + else + return DIE_FIND_CB_SIBLING; +} + +/* Add a found vars into available variables list */ +static int add_available_vars(Dwarf_Die *sp_die, struct probe_finder *pf) +{ + struct available_var_finder *af = + container_of(pf, struct available_var_finder, pf); + struct variable_list *vl; + Dwarf_Die die_mem, *scopes = NULL; + int ret, nscopes; + + /* Check number of tevs */ + if (af->nvls == af->max_vls) { + pr_warning("Too many( > %d) probe point found.\n", af->max_vls); + return -ERANGE; + } + vl = &af->vls[af->nvls++]; + + ret = convert_to_trace_point(sp_die, pf->addr, pf->pev->point.retprobe, + &vl->point); + if (ret < 0) + return ret; + + pr_debug("Probe point found: %s+%lu\n", vl->point.symbol, + vl->point.offset); + + /* Find local variables */ + vl->vars = strlist__new(true, NULL); + if (vl->vars == NULL) + return -ENOMEM; + af->child = true; + die_find_child(sp_die, collect_variables_cb, (void *)af, &die_mem); + + /* Find external variables */ + if (!af->externs) + goto out; + /* Don't need to search child DIE for externs. */ + af->child = false; + nscopes = dwarf_getscopes_die(sp_die, &scopes); + while (nscopes-- > 1) + die_find_child(&scopes[nscopes], collect_variables_cb, + (void *)af, &die_mem); + if (scopes) + free(scopes); + +out: + if (strlist__empty(vl->vars)) { + strlist__delete(vl->vars); + vl->vars = NULL; + } + + return ret; +} + +/* Find available variables at given probe point */ +int find_available_vars_at(int fd, struct perf_probe_event *pev, + struct variable_list **vls, int max_vls, + bool externs) +{ + struct available_var_finder af = { + .pf = {.pev = pev, .callback = add_available_vars}, + .max_vls = max_vls, .externs = externs}; + int ret; + + /* Allocate result vls array */ + *vls = zalloc(sizeof(struct variable_list) * max_vls); + if (*vls == NULL) + return -ENOMEM; + + af.vls = *vls; + af.nvls = 0; + + ret = find_probes(fd, &af.pf); + if (ret < 0) { + /* Free vlist for error */ + while (af.nvls--) { + if (af.vls[af.nvls].point.symbol) + free(af.vls[af.nvls].point.symbol); + if (af.vls[af.nvls].vars) + strlist__delete(af.vls[af.nvls].vars); + } + free(af.vls); + *vls = NULL; + return ret; + } + + return (ret < 0) ? ret : af.nvls; +} + +/* Reverse search */ +int find_perf_probe_point(unsigned long addr, struct perf_probe_point *ppt) +{ + Dwarf_Die cudie, spdie, indie; + Dwarf *dbg = NULL; + Dwfl *dwfl = NULL; + Dwarf_Addr _addr, baseaddr, bias = 0; + const char *fname = NULL, *func = NULL, *tmp; + int baseline = 0, lineno = 0, ret = 0; + + /* Open the live linux kernel */ + dbg = dwfl_init_live_kernel_dwarf(addr, &dwfl, &bias); + if (!dbg) { + pr_warning("No debug information found in the vmlinux - " + "please rebuild with CONFIG_DEBUG_INFO=y.\n"); + ret = -EINVAL; + goto end; + } + + /* Adjust address with bias */ + addr += bias; + /* Find cu die */ + if (!dwarf_addrdie(dbg, (Dwarf_Addr)addr - bias, &cudie)) { + pr_warning("Failed to find debug information for address %lx\n", + addr); + ret = -EINVAL; + goto end; + } + + /* Find a corresponding line (filename and lineno) */ + cu_find_lineinfo(&cudie, addr, &fname, &lineno); + /* Don't care whether it failed or not */ + + /* Find a corresponding function (name, baseline and baseaddr) */ + if (die_find_real_subprogram(&cudie, (Dwarf_Addr)addr, &spdie)) { + /* Get function entry information */ + tmp = dwarf_diename(&spdie); + if (!tmp || + dwarf_entrypc(&spdie, &baseaddr) != 0 || + dwarf_decl_line(&spdie, &baseline) != 0) + goto post; + func = tmp; + + if (addr == (unsigned long)baseaddr) + /* Function entry - Relative line number is 0 */ + lineno = baseline; + else if (die_find_inlinefunc(&spdie, (Dwarf_Addr)addr, + &indie)) { + if (dwarf_entrypc(&indie, &_addr) == 0 && + _addr == addr) + /* + * addr is at an inline function entry. + * In this case, lineno should be the call-site + * line number. + */ + lineno = die_get_call_lineno(&indie); + else { + /* + * addr is in an inline function body. + * Since lineno points one of the lines + * of the inline function, baseline should + * be the entry line of the inline function. + */ + tmp = dwarf_diename(&indie); + if (tmp && + dwarf_decl_line(&spdie, &baseline) == 0) + func = tmp; + } + } + } + +post: + /* Make a relative line number or an offset */ + if (lineno) + ppt->line = lineno - baseline; + else if (func) + ppt->offset = addr - (unsigned long)baseaddr; + + /* Duplicate strings */ + if (func) { + ppt->function = strdup(func); + if (ppt->function == NULL) { + ret = -ENOMEM; + goto end; + } + } + if (fname) { + ppt->file = strdup(fname); + if (ppt->file == NULL) { + if (ppt->function) { + free(ppt->function); + ppt->function = NULL; + } + ret = -ENOMEM; + goto end; + } + } +end: + if (dwfl) + dwfl_end(dwfl); + if (ret == 0 && (fname || func)) + ret = 1; /* Found a point */ + return ret; +} + +/* Add a line and store the src path */ +static int line_range_add_line(const char *src, unsigned int lineno, + struct line_range *lr) +{ + /* Copy source path */ + if (!lr->path) { + lr->path = strdup(src); + if (lr->path == NULL) + return -ENOMEM; + } + return line_list__add_line(&lr->line_list, lineno); +} + +static int line_range_walk_cb(const char *fname, int lineno, + Dwarf_Addr addr __used, + void *data) +{ + struct line_finder *lf = data; + + if ((strtailcmp(fname, lf->fname) != 0) || + (lf->lno_s > lineno || lf->lno_e < lineno)) + return 0; + + if (line_range_add_line(fname, lineno, lf->lr) < 0) + return -EINVAL; + + return 0; +} + +/* Find line range from its line number */ +static int find_line_range_by_line(Dwarf_Die *sp_die, struct line_finder *lf) +{ + int ret; + + ret = die_walk_lines(sp_die ?: &lf->cu_die, line_range_walk_cb, lf); + + /* Update status */ + if (ret >= 0) + if (!list_empty(&lf->lr->line_list)) + ret = lf->found = 1; + else + ret = 0; /* Lines are not found */ + else { + free(lf->lr->path); + lf->lr->path = NULL; + } + return ret; +} + +static int line_range_inline_cb(Dwarf_Die *in_die, void *data) +{ + struct dwarf_callback_param *param = data; + + param->retval = find_line_range_by_line(in_die, param->data); + return DWARF_CB_ABORT; /* No need to find other instances */ +} + +/* Search function from function name */ +static int line_range_search_cb(Dwarf_Die *sp_die, void *data) +{ + struct dwarf_callback_param *param = data; + struct line_finder *lf = param->data; + struct line_range *lr = lf->lr; + + /* Check declared file */ + if (lr->file && strtailcmp(lr->file, dwarf_decl_file(sp_die))) + return DWARF_CB_OK; + + if (dwarf_tag(sp_die) == DW_TAG_subprogram && + die_compare_name(sp_die, lr->function)) { + lf->fname = dwarf_decl_file(sp_die); + dwarf_decl_line(sp_die, &lr->offset); + pr_debug("fname: %s, lineno:%d\n", lf->fname, lr->offset); + lf->lno_s = lr->offset + lr->start; + if (lf->lno_s < 0) /* Overflow */ + lf->lno_s = INT_MAX; + lf->lno_e = lr->offset + lr->end; + if (lf->lno_e < 0) /* Overflow */ + lf->lno_e = INT_MAX; + pr_debug("New line range: %d to %d\n", lf->lno_s, lf->lno_e); + lr->start = lf->lno_s; + lr->end = lf->lno_e; + if (dwarf_func_inline(sp_die)) { + struct dwarf_callback_param _param; + _param.data = (void *)lf; + _param.retval = 0; + dwarf_func_inline_instances(sp_die, + line_range_inline_cb, + &_param); + param->retval = _param.retval; + } else + param->retval = find_line_range_by_line(sp_die, lf); + return DWARF_CB_ABORT; + } + return DWARF_CB_OK; +} + +static int find_line_range_by_func(struct line_finder *lf) +{ + struct dwarf_callback_param param = {.data = (void *)lf, .retval = 0}; + dwarf_getfuncs(&lf->cu_die, line_range_search_cb, ¶m, 0); + return param.retval; +} + +int find_line_range(int fd, struct line_range *lr) +{ + struct line_finder lf = {.lr = lr, .found = 0}; + int ret = 0; + Dwarf_Off off = 0, noff; + size_t cuhl; + Dwarf_Die *diep; + Dwarf *dbg = NULL; + Dwfl *dwfl; + Dwarf_Addr bias; /* Currently ignored */ + const char *comp_dir; + + dbg = dwfl_init_offline_dwarf(fd, &dwfl, &bias); + if (!dbg) { + pr_warning("No debug information found in the vmlinux - " + "please rebuild with CONFIG_DEBUG_INFO=y.\n"); + close(fd); /* Without dwfl_end(), fd isn't closed. */ + return -EBADF; + } + + /* Fastpath: lookup by function name from .debug_pubnames section */ + if (lr->function) { + struct pubname_callback_param pubname_param = { + .function = lr->function, .file = lr->file, + .cu_die = &lf.cu_die, .sp_die = &lf.sp_die, .found = 0}; + struct dwarf_callback_param line_range_param = { + .data = (void *)&lf, .retval = 0}; + + dwarf_getpubnames(dbg, pubname_search_cb, &pubname_param, 0); + if (pubname_param.found) { + line_range_search_cb(&lf.sp_die, &line_range_param); + if (lf.found) + goto found; + } + } + + /* Loop on CUs (Compilation Unit) */ + while (!lf.found && ret >= 0) { + if (dwarf_nextcu(dbg, off, &noff, &cuhl, NULL, NULL, NULL) != 0) + break; + + /* Get the DIE(Debugging Information Entry) of this CU */ + diep = dwarf_offdie(dbg, off + cuhl, &lf.cu_die); + if (!diep) + continue; + + /* Check if target file is included. */ + if (lr->file) + lf.fname = cu_find_realpath(&lf.cu_die, lr->file); + else + lf.fname = 0; + + if (!lr->file || lf.fname) { + if (lr->function) + ret = find_line_range_by_func(&lf); + else { + lf.lno_s = lr->start; + lf.lno_e = lr->end; + ret = find_line_range_by_line(NULL, &lf); + } + } + off = noff; + } + +found: + /* Store comp_dir */ + if (lf.found) { + comp_dir = cu_get_comp_dir(&lf.cu_die); + if (comp_dir) { + lr->comp_dir = strdup(comp_dir); + if (!lr->comp_dir) + ret = -ENOMEM; + } + } + + pr_debug("path: %s\n", lr->path); + dwfl_end(dwfl); + return (ret < 0) ? ret : lf.found; +} + diff --git a/tools/perf/util/probe-finder.h b/tools/perf/util/probe-finder.h new file mode 100644 index 00000000..605730a3 --- /dev/null +++ b/tools/perf/util/probe-finder.h @@ -0,0 +1,93 @@ +#ifndef _PROBE_FINDER_H +#define _PROBE_FINDER_H + +#include <stdbool.h> +#include "util.h" +#include "probe-event.h" + +#define MAX_PATH_LEN 256 +#define MAX_PROBE_BUFFER 1024 +#define MAX_PROBES 128 + +static inline int is_c_varname(const char *name) +{ + /* TODO */ + return isalpha(name[0]) || name[0] == '_'; +} + +#ifdef DWARF_SUPPORT +/* Find probe_trace_events specified by perf_probe_event from debuginfo */ +extern int find_probe_trace_events(int fd, struct perf_probe_event *pev, + struct probe_trace_event **tevs, + int max_tevs); + +/* Find a perf_probe_point from debuginfo */ +extern int find_perf_probe_point(unsigned long addr, + struct perf_probe_point *ppt); + +/* Find a line range */ +extern int find_line_range(int fd, struct line_range *lr); + +/* Find available variables */ +extern int find_available_vars_at(int fd, struct perf_probe_event *pev, + struct variable_list **vls, int max_points, + bool externs); + +#include <dwarf.h> +#include <elfutils/libdw.h> +#include <elfutils/libdwfl.h> +#include <elfutils/version.h> + +struct probe_finder { + struct perf_probe_event *pev; /* Target probe event */ + + /* Callback when a probe point is found */ + int (*callback)(Dwarf_Die *sp_die, struct probe_finder *pf); + + /* For function searching */ + int lno; /* Line number */ + Dwarf_Addr addr; /* Address */ + const char *fname; /* Real file name */ + Dwarf_Die cu_die; /* Current CU */ + Dwarf_Die sp_die; + struct list_head lcache; /* Line cache for lazy match */ + + /* For variable searching */ +#if _ELFUTILS_PREREQ(0, 142) + Dwarf_CFI *cfi; /* Call Frame Information */ +#endif + Dwarf_Op *fb_ops; /* Frame base attribute */ + struct perf_probe_arg *pvar; /* Current target variable */ + struct probe_trace_arg *tvar; /* Current result variable */ +}; + +struct trace_event_finder { + struct probe_finder pf; + struct probe_trace_event *tevs; /* Found trace events */ + int ntevs; /* Number of trace events */ + int max_tevs; /* Max number of trace events */ +}; + +struct available_var_finder { + struct probe_finder pf; + struct variable_list *vls; /* Found variable lists */ + int nvls; /* Number of variable lists */ + int max_vls; /* Max no. of variable lists */ + bool externs; /* Find external vars too */ + bool child; /* Search child scopes */ +}; + +struct line_finder { + struct line_range *lr; /* Target line range */ + + const char *fname; /* File name */ + int lno_s; /* Start line number */ + int lno_e; /* End line number */ + Dwarf_Die cu_die; /* Current CU */ + Dwarf_Die sp_die; + int found; +}; + +#endif /* DWARF_SUPPORT */ + +#endif /*_PROBE_FINDER_H */ diff --git a/tools/perf/util/pstack.c b/tools/perf/util/pstack.c new file mode 100644 index 00000000..13d36faf --- /dev/null +++ b/tools/perf/util/pstack.c @@ -0,0 +1,75 @@ +/* + * Simple pointer stack + * + * (c) 2010 Arnaldo Carvalho de Melo <acme@redhat.com> + */ + +#include "util.h" +#include "pstack.h" +#include <linux/kernel.h> +#include <stdlib.h> + +struct pstack { + unsigned short top; + unsigned short max_nr_entries; + void *entries[0]; +}; + +struct pstack *pstack__new(unsigned short max_nr_entries) +{ + struct pstack *self = zalloc((sizeof(*self) + + max_nr_entries * sizeof(void *))); + if (self != NULL) + self->max_nr_entries = max_nr_entries; + return self; +} + +void pstack__delete(struct pstack *self) +{ + free(self); +} + +bool pstack__empty(const struct pstack *self) +{ + return self->top == 0; +} + +void pstack__remove(struct pstack *self, void *key) +{ + unsigned short i = self->top, last_index = self->top - 1; + + while (i-- != 0) { + if (self->entries[i] == key) { + if (i < last_index) + memmove(self->entries + i, + self->entries + i + 1, + (last_index - i) * sizeof(void *)); + --self->top; + return; + } + } + pr_err("%s: %p not on the pstack!\n", __func__, key); +} + +void pstack__push(struct pstack *self, void *key) +{ + if (self->top == self->max_nr_entries) { + pr_err("%s: top=%d, overflow!\n", __func__, self->top); + return; + } + self->entries[self->top++] = key; +} + +void *pstack__pop(struct pstack *self) +{ + void *ret; + + if (self->top == 0) { + pr_err("%s: underflow!\n", __func__); + return NULL; + } + + ret = self->entries[--self->top]; + self->entries[self->top] = NULL; + return ret; +} diff --git a/tools/perf/util/pstack.h b/tools/perf/util/pstack.h new file mode 100644 index 00000000..4cedea59 --- /dev/null +++ b/tools/perf/util/pstack.h @@ -0,0 +1,14 @@ +#ifndef _PERF_PSTACK_ +#define _PERF_PSTACK_ + +#include <stdbool.h> + +struct pstack; +struct pstack *pstack__new(unsigned short max_nr_entries); +void pstack__delete(struct pstack *self); +bool pstack__empty(const struct pstack *self); +void pstack__remove(struct pstack *self, void *key); +void pstack__push(struct pstack *self, void *key); +void *pstack__pop(struct pstack *self); + +#endif /* _PERF_PSTACK_ */ diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c new file mode 100644 index 00000000..a9ac0504 --- /dev/null +++ b/tools/perf/util/python.c @@ -0,0 +1,905 @@ +#include <Python.h> +#include <structmember.h> +#include <inttypes.h> +#include <poll.h> +#include "evlist.h" +#include "evsel.h" +#include "event.h" +#include "cpumap.h" +#include "thread_map.h" + +/* Define PyVarObject_HEAD_INIT for python 2.5 */ +#ifndef PyVarObject_HEAD_INIT +# define PyVarObject_HEAD_INIT(type, size) PyObject_HEAD_INIT(type) size, +#endif + +struct throttle_event { + struct perf_event_header header; + u64 time; + u64 id; + u64 stream_id; +}; + +PyMODINIT_FUNC initperf(void); + +#define member_def(type, member, ptype, help) \ + { #member, ptype, \ + offsetof(struct pyrf_event, event) + offsetof(struct type, member), \ + 0, help } + +#define sample_member_def(name, member, ptype, help) \ + { #name, ptype, \ + offsetof(struct pyrf_event, sample) + offsetof(struct perf_sample, member), \ + 0, help } + +struct pyrf_event { + PyObject_HEAD + struct perf_sample sample; + union perf_event event; +}; + +#define sample_members \ + sample_member_def(sample_ip, ip, T_ULONGLONG, "event type"), \ + sample_member_def(sample_pid, pid, T_INT, "event pid"), \ + sample_member_def(sample_tid, tid, T_INT, "event tid"), \ + sample_member_def(sample_time, time, T_ULONGLONG, "event timestamp"), \ + sample_member_def(sample_addr, addr, T_ULONGLONG, "event addr"), \ + sample_member_def(sample_id, id, T_ULONGLONG, "event id"), \ + sample_member_def(sample_stream_id, stream_id, T_ULONGLONG, "event stream id"), \ + sample_member_def(sample_period, period, T_ULONGLONG, "event period"), \ + sample_member_def(sample_cpu, cpu, T_UINT, "event cpu"), + +static char pyrf_mmap_event__doc[] = PyDoc_STR("perf mmap event object."); + +static PyMemberDef pyrf_mmap_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(mmap_event, pid, T_UINT, "event pid"), + member_def(mmap_event, tid, T_UINT, "event tid"), + member_def(mmap_event, start, T_ULONGLONG, "start of the map"), + member_def(mmap_event, len, T_ULONGLONG, "map length"), + member_def(mmap_event, pgoff, T_ULONGLONG, "page offset"), + member_def(mmap_event, filename, T_STRING_INPLACE, "backing store"), + { .name = NULL, }, +}; + +static PyObject *pyrf_mmap_event__repr(struct pyrf_event *pevent) +{ + PyObject *ret; + char *s; + + if (asprintf(&s, "{ type: mmap, pid: %u, tid: %u, start: %#" PRIx64 ", " + "length: %#" PRIx64 ", offset: %#" PRIx64 ", " + "filename: %s }", + pevent->event.mmap.pid, pevent->event.mmap.tid, + pevent->event.mmap.start, pevent->event.mmap.len, + pevent->event.mmap.pgoff, pevent->event.mmap.filename) < 0) { + ret = PyErr_NoMemory(); + } else { + ret = PyString_FromString(s); + free(s); + } + return ret; +} + +static PyTypeObject pyrf_mmap_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.mmap_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_mmap_event__doc, + .tp_members = pyrf_mmap_event__members, + .tp_repr = (reprfunc)pyrf_mmap_event__repr, +}; + +static char pyrf_task_event__doc[] = PyDoc_STR("perf task (fork/exit) event object."); + +static PyMemberDef pyrf_task_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(fork_event, pid, T_UINT, "event pid"), + member_def(fork_event, ppid, T_UINT, "event ppid"), + member_def(fork_event, tid, T_UINT, "event tid"), + member_def(fork_event, ptid, T_UINT, "event ptid"), + member_def(fork_event, time, T_ULONGLONG, "timestamp"), + { .name = NULL, }, +}; + +static PyObject *pyrf_task_event__repr(struct pyrf_event *pevent) +{ + return PyString_FromFormat("{ type: %s, pid: %u, ppid: %u, tid: %u, " + "ptid: %u, time: %" PRIu64 "}", + pevent->event.header.type == PERF_RECORD_FORK ? "fork" : "exit", + pevent->event.fork.pid, + pevent->event.fork.ppid, + pevent->event.fork.tid, + pevent->event.fork.ptid, + pevent->event.fork.time); +} + +static PyTypeObject pyrf_task_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.task_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_task_event__doc, + .tp_members = pyrf_task_event__members, + .tp_repr = (reprfunc)pyrf_task_event__repr, +}; + +static char pyrf_comm_event__doc[] = PyDoc_STR("perf comm event object."); + +static PyMemberDef pyrf_comm_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(comm_event, pid, T_UINT, "event pid"), + member_def(comm_event, tid, T_UINT, "event tid"), + member_def(comm_event, comm, T_STRING_INPLACE, "process name"), + { .name = NULL, }, +}; + +static PyObject *pyrf_comm_event__repr(struct pyrf_event *pevent) +{ + return PyString_FromFormat("{ type: comm, pid: %u, tid: %u, comm: %s }", + pevent->event.comm.pid, + pevent->event.comm.tid, + pevent->event.comm.comm); +} + +static PyTypeObject pyrf_comm_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.comm_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_comm_event__doc, + .tp_members = pyrf_comm_event__members, + .tp_repr = (reprfunc)pyrf_comm_event__repr, +}; + +static char pyrf_throttle_event__doc[] = PyDoc_STR("perf throttle event object."); + +static PyMemberDef pyrf_throttle_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(throttle_event, time, T_ULONGLONG, "timestamp"), + member_def(throttle_event, id, T_ULONGLONG, "event id"), + member_def(throttle_event, stream_id, T_ULONGLONG, "event stream id"), + { .name = NULL, }, +}; + +static PyObject *pyrf_throttle_event__repr(struct pyrf_event *pevent) +{ + struct throttle_event *te = (struct throttle_event *)(&pevent->event.header + 1); + + return PyString_FromFormat("{ type: %sthrottle, time: %" PRIu64 ", id: %" PRIu64 + ", stream_id: %" PRIu64 " }", + pevent->event.header.type == PERF_RECORD_THROTTLE ? "" : "un", + te->time, te->id, te->stream_id); +} + +static PyTypeObject pyrf_throttle_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.throttle_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_throttle_event__doc, + .tp_members = pyrf_throttle_event__members, + .tp_repr = (reprfunc)pyrf_throttle_event__repr, +}; + +static int pyrf_event__setup_types(void) +{ + int err; + pyrf_mmap_event__type.tp_new = + pyrf_task_event__type.tp_new = + pyrf_comm_event__type.tp_new = + pyrf_throttle_event__type.tp_new = PyType_GenericNew; + err = PyType_Ready(&pyrf_mmap_event__type); + if (err < 0) + goto out; + err = PyType_Ready(&pyrf_task_event__type); + if (err < 0) + goto out; + err = PyType_Ready(&pyrf_comm_event__type); + if (err < 0) + goto out; + err = PyType_Ready(&pyrf_throttle_event__type); + if (err < 0) + goto out; +out: + return err; +} + +static PyTypeObject *pyrf_event__type[] = { + [PERF_RECORD_MMAP] = &pyrf_mmap_event__type, + [PERF_RECORD_LOST] = &pyrf_mmap_event__type, + [PERF_RECORD_COMM] = &pyrf_comm_event__type, + [PERF_RECORD_EXIT] = &pyrf_task_event__type, + [PERF_RECORD_THROTTLE] = &pyrf_throttle_event__type, + [PERF_RECORD_UNTHROTTLE] = &pyrf_throttle_event__type, + [PERF_RECORD_FORK] = &pyrf_task_event__type, + [PERF_RECORD_READ] = &pyrf_mmap_event__type, + [PERF_RECORD_SAMPLE] = &pyrf_mmap_event__type, +}; + +static PyObject *pyrf_event__new(union perf_event *event) +{ + struct pyrf_event *pevent; + PyTypeObject *ptype; + + if (event->header.type < PERF_RECORD_MMAP || + event->header.type > PERF_RECORD_SAMPLE) + return NULL; + + ptype = pyrf_event__type[event->header.type]; + pevent = PyObject_New(struct pyrf_event, ptype); + if (pevent != NULL) + memcpy(&pevent->event, event, event->header.size); + return (PyObject *)pevent; +} + +struct pyrf_cpu_map { + PyObject_HEAD + + struct cpu_map *cpus; +}; + +static int pyrf_cpu_map__init(struct pyrf_cpu_map *pcpus, + PyObject *args, PyObject *kwargs) +{ + static char *kwlist[] = { "cpustr", NULL, NULL, }; + char *cpustr = NULL; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|s", + kwlist, &cpustr)) + return -1; + + pcpus->cpus = cpu_map__new(cpustr); + if (pcpus->cpus == NULL) + return -1; + return 0; +} + +static void pyrf_cpu_map__delete(struct pyrf_cpu_map *pcpus) +{ + cpu_map__delete(pcpus->cpus); + pcpus->ob_type->tp_free((PyObject*)pcpus); +} + +static Py_ssize_t pyrf_cpu_map__length(PyObject *obj) +{ + struct pyrf_cpu_map *pcpus = (void *)obj; + + return pcpus->cpus->nr; +} + +static PyObject *pyrf_cpu_map__item(PyObject *obj, Py_ssize_t i) +{ + struct pyrf_cpu_map *pcpus = (void *)obj; + + if (i >= pcpus->cpus->nr) + return NULL; + + return Py_BuildValue("i", pcpus->cpus->map[i]); +} + +static PySequenceMethods pyrf_cpu_map__sequence_methods = { + .sq_length = pyrf_cpu_map__length, + .sq_item = pyrf_cpu_map__item, +}; + +static char pyrf_cpu_map__doc[] = PyDoc_STR("cpu map object."); + +static PyTypeObject pyrf_cpu_map__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.cpu_map", + .tp_basicsize = sizeof(struct pyrf_cpu_map), + .tp_dealloc = (destructor)pyrf_cpu_map__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_cpu_map__doc, + .tp_as_sequence = &pyrf_cpu_map__sequence_methods, + .tp_init = (initproc)pyrf_cpu_map__init, +}; + +static int pyrf_cpu_map__setup_types(void) +{ + pyrf_cpu_map__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_cpu_map__type); +} + +struct pyrf_thread_map { + PyObject_HEAD + + struct thread_map *threads; +}; + +static int pyrf_thread_map__init(struct pyrf_thread_map *pthreads, + PyObject *args, PyObject *kwargs) +{ + static char *kwlist[] = { "pid", "tid", NULL, NULL, }; + int pid = -1, tid = -1; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ii", + kwlist, &pid, &tid)) + return -1; + + pthreads->threads = thread_map__new(pid, tid); + if (pthreads->threads == NULL) + return -1; + return 0; +} + +static void pyrf_thread_map__delete(struct pyrf_thread_map *pthreads) +{ + thread_map__delete(pthreads->threads); + pthreads->ob_type->tp_free((PyObject*)pthreads); +} + +static Py_ssize_t pyrf_thread_map__length(PyObject *obj) +{ + struct pyrf_thread_map *pthreads = (void *)obj; + + return pthreads->threads->nr; +} + +static PyObject *pyrf_thread_map__item(PyObject *obj, Py_ssize_t i) +{ + struct pyrf_thread_map *pthreads = (void *)obj; + + if (i >= pthreads->threads->nr) + return NULL; + + return Py_BuildValue("i", pthreads->threads->map[i]); +} + +static PySequenceMethods pyrf_thread_map__sequence_methods = { + .sq_length = pyrf_thread_map__length, + .sq_item = pyrf_thread_map__item, +}; + +static char pyrf_thread_map__doc[] = PyDoc_STR("thread map object."); + +static PyTypeObject pyrf_thread_map__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.thread_map", + .tp_basicsize = sizeof(struct pyrf_thread_map), + .tp_dealloc = (destructor)pyrf_thread_map__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_thread_map__doc, + .tp_as_sequence = &pyrf_thread_map__sequence_methods, + .tp_init = (initproc)pyrf_thread_map__init, +}; + +static int pyrf_thread_map__setup_types(void) +{ + pyrf_thread_map__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_thread_map__type); +} + +struct pyrf_evsel { + PyObject_HEAD + + struct perf_evsel evsel; +}; + +static int pyrf_evsel__init(struct pyrf_evsel *pevsel, + PyObject *args, PyObject *kwargs) +{ + struct perf_event_attr attr = { + .type = PERF_TYPE_HARDWARE, + .config = PERF_COUNT_HW_CPU_CYCLES, + .sample_type = PERF_SAMPLE_PERIOD | PERF_SAMPLE_TID, + }; + static char *kwlist[] = { + "type", + "config", + "sample_freq", + "sample_period", + "sample_type", + "read_format", + "disabled", + "inherit", + "pinned", + "exclusive", + "exclude_user", + "exclude_kernel", + "exclude_hv", + "exclude_idle", + "mmap", + "comm", + "freq", + "inherit_stat", + "enable_on_exec", + "task", + "watermark", + "precise_ip", + "mmap_data", + "sample_id_all", + "wakeup_events", + "bp_type", + "bp_addr", + "bp_len", NULL, NULL, }; + u64 sample_period = 0; + u32 disabled = 0, + inherit = 0, + pinned = 0, + exclusive = 0, + exclude_user = 0, + exclude_kernel = 0, + exclude_hv = 0, + exclude_idle = 0, + mmap = 0, + comm = 0, + freq = 1, + inherit_stat = 0, + enable_on_exec = 0, + task = 0, + watermark = 0, + precise_ip = 0, + mmap_data = 0, + sample_id_all = 1; + int idx = 0; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, + "|iKiKKiiiiiiiiiiiiiiiiiiiiiKK", kwlist, + &attr.type, &attr.config, &attr.sample_freq, + &sample_period, &attr.sample_type, + &attr.read_format, &disabled, &inherit, + &pinned, &exclusive, &exclude_user, + &exclude_kernel, &exclude_hv, &exclude_idle, + &mmap, &comm, &freq, &inherit_stat, + &enable_on_exec, &task, &watermark, + &precise_ip, &mmap_data, &sample_id_all, + &attr.wakeup_events, &attr.bp_type, + &attr.bp_addr, &attr.bp_len, &idx)) + return -1; + + /* union... */ + if (sample_period != 0) { + if (attr.sample_freq != 0) + return -1; /* FIXME: throw right exception */ + attr.sample_period = sample_period; + } + + /* Bitfields */ + attr.disabled = disabled; + attr.inherit = inherit; + attr.pinned = pinned; + attr.exclusive = exclusive; + attr.exclude_user = exclude_user; + attr.exclude_kernel = exclude_kernel; + attr.exclude_hv = exclude_hv; + attr.exclude_idle = exclude_idle; + attr.mmap = mmap; + attr.comm = comm; + attr.freq = freq; + attr.inherit_stat = inherit_stat; + attr.enable_on_exec = enable_on_exec; + attr.task = task; + attr.watermark = watermark; + attr.precise_ip = precise_ip; + attr.mmap_data = mmap_data; + attr.sample_id_all = sample_id_all; + + perf_evsel__init(&pevsel->evsel, &attr, idx); + return 0; +} + +static void pyrf_evsel__delete(struct pyrf_evsel *pevsel) +{ + perf_evsel__exit(&pevsel->evsel); + pevsel->ob_type->tp_free((PyObject*)pevsel); +} + +static PyObject *pyrf_evsel__open(struct pyrf_evsel *pevsel, + PyObject *args, PyObject *kwargs) +{ + struct perf_evsel *evsel = &pevsel->evsel; + struct cpu_map *cpus = NULL; + struct thread_map *threads = NULL; + PyObject *pcpus = NULL, *pthreads = NULL; + int group = 0, inherit = 0; + static char *kwlist[] = {"cpus", "threads", "group", "inherit", NULL, NULL}; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|OOii", kwlist, + &pcpus, &pthreads, &group, &inherit)) + return NULL; + + if (pthreads != NULL) + threads = ((struct pyrf_thread_map *)pthreads)->threads; + + if (pcpus != NULL) + cpus = ((struct pyrf_cpu_map *)pcpus)->cpus; + + evsel->attr.inherit = inherit; + if (perf_evsel__open(evsel, cpus, threads, group) < 0) { + PyErr_SetFromErrno(PyExc_OSError); + return NULL; + } + + Py_INCREF(Py_None); + return Py_None; +} + +static PyMethodDef pyrf_evsel__methods[] = { + { + .ml_name = "open", + .ml_meth = (PyCFunction)pyrf_evsel__open, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("open the event selector file descriptor table.") + }, + { .ml_name = NULL, } +}; + +static char pyrf_evsel__doc[] = PyDoc_STR("perf event selector list object."); + +static PyTypeObject pyrf_evsel__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.evsel", + .tp_basicsize = sizeof(struct pyrf_evsel), + .tp_dealloc = (destructor)pyrf_evsel__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_evsel__doc, + .tp_methods = pyrf_evsel__methods, + .tp_init = (initproc)pyrf_evsel__init, +}; + +static int pyrf_evsel__setup_types(void) +{ + pyrf_evsel__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_evsel__type); +} + +struct pyrf_evlist { + PyObject_HEAD + + struct perf_evlist evlist; +}; + +static int pyrf_evlist__init(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs __used) +{ + PyObject *pcpus = NULL, *pthreads = NULL; + struct cpu_map *cpus; + struct thread_map *threads; + + if (!PyArg_ParseTuple(args, "OO", &pcpus, &pthreads)) + return -1; + + threads = ((struct pyrf_thread_map *)pthreads)->threads; + cpus = ((struct pyrf_cpu_map *)pcpus)->cpus; + perf_evlist__init(&pevlist->evlist, cpus, threads); + return 0; +} + +static void pyrf_evlist__delete(struct pyrf_evlist *pevlist) +{ + perf_evlist__exit(&pevlist->evlist); + pevlist->ob_type->tp_free((PyObject*)pevlist); +} + +static PyObject *pyrf_evlist__mmap(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + static char *kwlist[] = {"pages", "overwrite", + NULL, NULL}; + int pages = 128, overwrite = false; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ii", kwlist, + &pages, &overwrite)) + return NULL; + + if (perf_evlist__mmap(evlist, pages, overwrite) < 0) { + PyErr_SetFromErrno(PyExc_OSError); + return NULL; + } + + Py_INCREF(Py_None); + return Py_None; +} + +static PyObject *pyrf_evlist__poll(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + static char *kwlist[] = {"timeout", NULL, NULL}; + int timeout = -1, n; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|i", kwlist, &timeout)) + return NULL; + + n = poll(evlist->pollfd, evlist->nr_fds, timeout); + if (n < 0) { + PyErr_SetFromErrno(PyExc_OSError); + return NULL; + } + + return Py_BuildValue("i", n); +} + +static PyObject *pyrf_evlist__get_pollfd(struct pyrf_evlist *pevlist, + PyObject *args __used, PyObject *kwargs __used) +{ + struct perf_evlist *evlist = &pevlist->evlist; + PyObject *list = PyList_New(0); + int i; + + for (i = 0; i < evlist->nr_fds; ++i) { + PyObject *file; + FILE *fp = fdopen(evlist->pollfd[i].fd, "r"); + + if (fp == NULL) + goto free_list; + + file = PyFile_FromFile(fp, "perf", "r", NULL); + if (file == NULL) + goto free_list; + + if (PyList_Append(list, file) != 0) { + Py_DECREF(file); + goto free_list; + } + + Py_DECREF(file); + } + + return list; +free_list: + return PyErr_NoMemory(); +} + + +static PyObject *pyrf_evlist__add(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs __used) +{ + struct perf_evlist *evlist = &pevlist->evlist; + PyObject *pevsel; + struct perf_evsel *evsel; + + if (!PyArg_ParseTuple(args, "O", &pevsel)) + return NULL; + + Py_INCREF(pevsel); + evsel = &((struct pyrf_evsel *)pevsel)->evsel; + evsel->idx = evlist->nr_entries; + perf_evlist__add(evlist, evsel); + + return Py_BuildValue("i", evlist->nr_entries); +} + +static PyObject *pyrf_evlist__read_on_cpu(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + union perf_event *event; + int sample_id_all = 1, cpu; + static char *kwlist[] = {"cpu", "sample_id_all", NULL, NULL}; + int err; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "i|i", kwlist, + &cpu, &sample_id_all)) + return NULL; + + event = perf_evlist__mmap_read(evlist, cpu); + if (event != NULL) { + struct perf_evsel *first; + PyObject *pyevent = pyrf_event__new(event); + struct pyrf_event *pevent = (struct pyrf_event *)pyevent; + + if (pyevent == NULL) + return PyErr_NoMemory(); + + first = list_entry(evlist->entries.next, struct perf_evsel, node); + err = perf_event__parse_sample(event, first->attr.sample_type, + perf_evsel__sample_size(first), + sample_id_all, &pevent->sample); + if (err) + return PyErr_Format(PyExc_OSError, + "perf: can't parse sample, err=%d", err); + return pyevent; + } + + Py_INCREF(Py_None); + return Py_None; +} + +static PyMethodDef pyrf_evlist__methods[] = { + { + .ml_name = "mmap", + .ml_meth = (PyCFunction)pyrf_evlist__mmap, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("mmap the file descriptor table.") + }, + { + .ml_name = "poll", + .ml_meth = (PyCFunction)pyrf_evlist__poll, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("poll the file descriptor table.") + }, + { + .ml_name = "get_pollfd", + .ml_meth = (PyCFunction)pyrf_evlist__get_pollfd, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("get the poll file descriptor table.") + }, + { + .ml_name = "add", + .ml_meth = (PyCFunction)pyrf_evlist__add, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("adds an event selector to the list.") + }, + { + .ml_name = "read_on_cpu", + .ml_meth = (PyCFunction)pyrf_evlist__read_on_cpu, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("reads an event.") + }, + { .ml_name = NULL, } +}; + +static Py_ssize_t pyrf_evlist__length(PyObject *obj) +{ + struct pyrf_evlist *pevlist = (void *)obj; + + return pevlist->evlist.nr_entries; +} + +static PyObject *pyrf_evlist__item(PyObject *obj, Py_ssize_t i) +{ + struct pyrf_evlist *pevlist = (void *)obj; + struct perf_evsel *pos; + + if (i >= pevlist->evlist.nr_entries) + return NULL; + + list_for_each_entry(pos, &pevlist->evlist.entries, node) + if (i-- == 0) + break; + + return Py_BuildValue("O", container_of(pos, struct pyrf_evsel, evsel)); +} + +static PySequenceMethods pyrf_evlist__sequence_methods = { + .sq_length = pyrf_evlist__length, + .sq_item = pyrf_evlist__item, +}; + +static char pyrf_evlist__doc[] = PyDoc_STR("perf event selector list object."); + +static PyTypeObject pyrf_evlist__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.evlist", + .tp_basicsize = sizeof(struct pyrf_evlist), + .tp_dealloc = (destructor)pyrf_evlist__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_as_sequence = &pyrf_evlist__sequence_methods, + .tp_doc = pyrf_evlist__doc, + .tp_methods = pyrf_evlist__methods, + .tp_init = (initproc)pyrf_evlist__init, +}; + +static int pyrf_evlist__setup_types(void) +{ + pyrf_evlist__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_evlist__type); +} + +static struct { + const char *name; + int value; +} perf__constants[] = { + { "TYPE_HARDWARE", PERF_TYPE_HARDWARE }, + { "TYPE_SOFTWARE", PERF_TYPE_SOFTWARE }, + { "TYPE_TRACEPOINT", PERF_TYPE_TRACEPOINT }, + { "TYPE_HW_CACHE", PERF_TYPE_HW_CACHE }, + { "TYPE_RAW", PERF_TYPE_RAW }, + { "TYPE_BREAKPOINT", PERF_TYPE_BREAKPOINT }, + + { "COUNT_HW_CPU_CYCLES", PERF_COUNT_HW_CPU_CYCLES }, + { "COUNT_HW_INSTRUCTIONS", PERF_COUNT_HW_INSTRUCTIONS }, + { "COUNT_HW_CACHE_REFERENCES", PERF_COUNT_HW_CACHE_REFERENCES }, + { "COUNT_HW_CACHE_MISSES", PERF_COUNT_HW_CACHE_MISSES }, + { "COUNT_HW_BRANCH_INSTRUCTIONS", PERF_COUNT_HW_BRANCH_INSTRUCTIONS }, + { "COUNT_HW_BRANCH_MISSES", PERF_COUNT_HW_BRANCH_MISSES }, + { "COUNT_HW_BUS_CYCLES", PERF_COUNT_HW_BUS_CYCLES }, + { "COUNT_HW_CACHE_L1D", PERF_COUNT_HW_CACHE_L1D }, + { "COUNT_HW_CACHE_L1I", PERF_COUNT_HW_CACHE_L1I }, + { "COUNT_HW_CACHE_LL", PERF_COUNT_HW_CACHE_LL }, + { "COUNT_HW_CACHE_DTLB", PERF_COUNT_HW_CACHE_DTLB }, + { "COUNT_HW_CACHE_ITLB", PERF_COUNT_HW_CACHE_ITLB }, + { "COUNT_HW_CACHE_BPU", PERF_COUNT_HW_CACHE_BPU }, + { "COUNT_HW_CACHE_OP_READ", PERF_COUNT_HW_CACHE_OP_READ }, + { "COUNT_HW_CACHE_OP_WRITE", PERF_COUNT_HW_CACHE_OP_WRITE }, + { "COUNT_HW_CACHE_OP_PREFETCH", PERF_COUNT_HW_CACHE_OP_PREFETCH }, + { "COUNT_HW_CACHE_RESULT_ACCESS", PERF_COUNT_HW_CACHE_RESULT_ACCESS }, + { "COUNT_HW_CACHE_RESULT_MISS", PERF_COUNT_HW_CACHE_RESULT_MISS }, + + { "COUNT_HW_STALLED_CYCLES_FRONTEND", PERF_COUNT_HW_STALLED_CYCLES_FRONTEND }, + { "COUNT_HW_STALLED_CYCLES_BACKEND", PERF_COUNT_HW_STALLED_CYCLES_BACKEND }, + + { "COUNT_SW_CPU_CLOCK", PERF_COUNT_SW_CPU_CLOCK }, + { "COUNT_SW_TASK_CLOCK", PERF_COUNT_SW_TASK_CLOCK }, + { "COUNT_SW_PAGE_FAULTS", PERF_COUNT_SW_PAGE_FAULTS }, + { "COUNT_SW_CONTEXT_SWITCHES", PERF_COUNT_SW_CONTEXT_SWITCHES }, + { "COUNT_SW_CPU_MIGRATIONS", PERF_COUNT_SW_CPU_MIGRATIONS }, + { "COUNT_SW_PAGE_FAULTS_MIN", PERF_COUNT_SW_PAGE_FAULTS_MIN }, + { "COUNT_SW_PAGE_FAULTS_MAJ", PERF_COUNT_SW_PAGE_FAULTS_MAJ }, + { "COUNT_SW_ALIGNMENT_FAULTS", PERF_COUNT_SW_ALIGNMENT_FAULTS }, + { "COUNT_SW_EMULATION_FAULTS", PERF_COUNT_SW_EMULATION_FAULTS }, + + { "SAMPLE_IP", PERF_SAMPLE_IP }, + { "SAMPLE_TID", PERF_SAMPLE_TID }, + { "SAMPLE_TIME", PERF_SAMPLE_TIME }, + { "SAMPLE_ADDR", PERF_SAMPLE_ADDR }, + { "SAMPLE_READ", PERF_SAMPLE_READ }, + { "SAMPLE_CALLCHAIN", PERF_SAMPLE_CALLCHAIN }, + { "SAMPLE_ID", PERF_SAMPLE_ID }, + { "SAMPLE_CPU", PERF_SAMPLE_CPU }, + { "SAMPLE_PERIOD", PERF_SAMPLE_PERIOD }, + { "SAMPLE_STREAM_ID", PERF_SAMPLE_STREAM_ID }, + { "SAMPLE_RAW", PERF_SAMPLE_RAW }, + + { "FORMAT_TOTAL_TIME_ENABLED", PERF_FORMAT_TOTAL_TIME_ENABLED }, + { "FORMAT_TOTAL_TIME_RUNNING", PERF_FORMAT_TOTAL_TIME_RUNNING }, + { "FORMAT_ID", PERF_FORMAT_ID }, + { "FORMAT_GROUP", PERF_FORMAT_GROUP }, + + { "RECORD_MMAP", PERF_RECORD_MMAP }, + { "RECORD_LOST", PERF_RECORD_LOST }, + { "RECORD_COMM", PERF_RECORD_COMM }, + { "RECORD_EXIT", PERF_RECORD_EXIT }, + { "RECORD_THROTTLE", PERF_RECORD_THROTTLE }, + { "RECORD_UNTHROTTLE", PERF_RECORD_UNTHROTTLE }, + { "RECORD_FORK", PERF_RECORD_FORK }, + { "RECORD_READ", PERF_RECORD_READ }, + { "RECORD_SAMPLE", PERF_RECORD_SAMPLE }, + { .name = NULL, }, +}; + +static PyMethodDef perf__methods[] = { + { .ml_name = NULL, } +}; + +PyMODINIT_FUNC initperf(void) +{ + PyObject *obj; + int i; + PyObject *dict, *module = Py_InitModule("perf", perf__methods); + + if (module == NULL || + pyrf_event__setup_types() < 0 || + pyrf_evlist__setup_types() < 0 || + pyrf_evsel__setup_types() < 0 || + pyrf_thread_map__setup_types() < 0 || + pyrf_cpu_map__setup_types() < 0) + return; + + Py_INCREF(&pyrf_evlist__type); + PyModule_AddObject(module, "evlist", (PyObject*)&pyrf_evlist__type); + + Py_INCREF(&pyrf_evsel__type); + PyModule_AddObject(module, "evsel", (PyObject*)&pyrf_evsel__type); + + Py_INCREF(&pyrf_thread_map__type); + PyModule_AddObject(module, "thread_map", (PyObject*)&pyrf_thread_map__type); + + Py_INCREF(&pyrf_cpu_map__type); + PyModule_AddObject(module, "cpu_map", (PyObject*)&pyrf_cpu_map__type); + + dict = PyModule_GetDict(module); + if (dict == NULL) + goto error; + + for (i = 0; perf__constants[i].name != NULL; i++) { + obj = PyInt_FromLong(perf__constants[i].value); + if (obj == NULL) + goto error; + PyDict_SetItemString(dict, perf__constants[i].name, obj); + Py_DECREF(obj); + } + +error: + if (PyErr_Occurred()) + PyErr_SetString(PyExc_ImportError, "perf: Init failed!"); +} diff --git a/tools/perf/util/quote.c b/tools/perf/util/quote.c new file mode 100644 index 00000000..01f03242 --- /dev/null +++ b/tools/perf/util/quote.c @@ -0,0 +1,54 @@ +#include "cache.h" +#include "quote.h" + +/* Help to copy the thing properly quoted for the shell safety. + * any single quote is replaced with '\'', any exclamation point + * is replaced with '\!', and the whole thing is enclosed in a + * + * E.g. + * original sq_quote result + * name ==> name ==> 'name' + * a b ==> a b ==> 'a b' + * a'b ==> a'\''b ==> 'a'\''b' + * a!b ==> a'\!'b ==> 'a'\!'b' + */ +static inline int need_bs_quote(char c) +{ + return (c == '\'' || c == '!'); +} + +static void sq_quote_buf(struct strbuf *dst, const char *src) +{ + char *to_free = NULL; + + if (dst->buf == src) + to_free = strbuf_detach(dst, NULL); + + strbuf_addch(dst, '\''); + while (*src) { + size_t len = strcspn(src, "'!"); + strbuf_add(dst, src, len); + src += len; + while (need_bs_quote(*src)) { + strbuf_addstr(dst, "'\\"); + strbuf_addch(dst, *src++); + strbuf_addch(dst, '\''); + } + } + strbuf_addch(dst, '\''); + free(to_free); +} + +void sq_quote_argv(struct strbuf *dst, const char** argv, size_t maxlen) +{ + int i; + + /* Copy into destination buffer. */ + strbuf_grow(dst, 255); + for (i = 0; argv[i]; ++i) { + strbuf_addch(dst, ' '); + sq_quote_buf(dst, argv[i]); + if (maxlen && dst->len > maxlen) + die("Too many or long arguments"); + } +} diff --git a/tools/perf/util/quote.h b/tools/perf/util/quote.h new file mode 100644 index 00000000..172889ea --- /dev/null +++ b/tools/perf/util/quote.h @@ -0,0 +1,29 @@ +#ifndef __PERF_QUOTE_H +#define __PERF_QUOTE_H + +#include <stddef.h> +#include <stdio.h> + +/* Help to copy the thing properly quoted for the shell safety. + * any single quote is replaced with '\'', any exclamation point + * is replaced with '\!', and the whole thing is enclosed in a + * single quote pair. + * + * For example, if you are passing the result to system() as an + * argument: + * + * sprintf(cmd, "foobar %s %s", sq_quote(arg0), sq_quote(arg1)) + * + * would be appropriate. If the system() is going to call ssh to + * run the command on the other side: + * + * sprintf(cmd, "git-diff-tree %s %s", sq_quote(arg0), sq_quote(arg1)); + * sprintf(rcmd, "ssh %s %s", sq_util/quote.host), sq_quote(cmd)); + * + * Note that the above examples leak memory! Remember to free result from + * sq_quote() in a real application. + */ + +extern void sq_quote_argv(struct strbuf *, const char **argv, size_t maxlen); + +#endif /* __PERF_QUOTE_H */ diff --git a/tools/perf/util/run-command.c b/tools/perf/util/run-command.c new file mode 100644 index 00000000..da8e9b28 --- /dev/null +++ b/tools/perf/util/run-command.c @@ -0,0 +1,214 @@ +#include "cache.h" +#include "run-command.h" +#include "exec_cmd.h" + +static inline void close_pair(int fd[2]) +{ + close(fd[0]); + close(fd[1]); +} + +static inline void dup_devnull(int to) +{ + int fd = open("/dev/null", O_RDWR); + dup2(fd, to); + close(fd); +} + +int start_command(struct child_process *cmd) +{ + int need_in, need_out, need_err; + int fdin[2], fdout[2], fderr[2]; + + /* + * In case of errors we must keep the promise to close FDs + * that have been passed in via ->in and ->out. + */ + + need_in = !cmd->no_stdin && cmd->in < 0; + if (need_in) { + if (pipe(fdin) < 0) { + if (cmd->out > 0) + close(cmd->out); + return -ERR_RUN_COMMAND_PIPE; + } + cmd->in = fdin[1]; + } + + need_out = !cmd->no_stdout + && !cmd->stdout_to_stderr + && cmd->out < 0; + if (need_out) { + if (pipe(fdout) < 0) { + if (need_in) + close_pair(fdin); + else if (cmd->in) + close(cmd->in); + return -ERR_RUN_COMMAND_PIPE; + } + cmd->out = fdout[0]; + } + + need_err = !cmd->no_stderr && cmd->err < 0; + if (need_err) { + if (pipe(fderr) < 0) { + if (need_in) + close_pair(fdin); + else if (cmd->in) + close(cmd->in); + if (need_out) + close_pair(fdout); + else if (cmd->out) + close(cmd->out); + return -ERR_RUN_COMMAND_PIPE; + } + cmd->err = fderr[0]; + } + + fflush(NULL); + cmd->pid = fork(); + if (!cmd->pid) { + if (cmd->no_stdin) + dup_devnull(0); + else if (need_in) { + dup2(fdin[0], 0); + close_pair(fdin); + } else if (cmd->in) { + dup2(cmd->in, 0); + close(cmd->in); + } + + if (cmd->no_stderr) + dup_devnull(2); + else if (need_err) { + dup2(fderr[1], 2); + close_pair(fderr); + } + + if (cmd->no_stdout) + dup_devnull(1); + else if (cmd->stdout_to_stderr) + dup2(2, 1); + else if (need_out) { + dup2(fdout[1], 1); + close_pair(fdout); + } else if (cmd->out > 1) { + dup2(cmd->out, 1); + close(cmd->out); + } + + if (cmd->dir && chdir(cmd->dir)) + die("exec %s: cd to %s failed (%s)", cmd->argv[0], + cmd->dir, strerror(errno)); + if (cmd->env) { + for (; *cmd->env; cmd->env++) { + if (strchr(*cmd->env, '=')) + putenv((char*)*cmd->env); + else + unsetenv(*cmd->env); + } + } + if (cmd->preexec_cb) + cmd->preexec_cb(); + if (cmd->perf_cmd) { + execv_perf_cmd(cmd->argv); + } else { + execvp(cmd->argv[0], (char *const*) cmd->argv); + } + exit(127); + } + + if (cmd->pid < 0) { + int err = errno; + if (need_in) + close_pair(fdin); + else if (cmd->in) + close(cmd->in); + if (need_out) + close_pair(fdout); + else if (cmd->out) + close(cmd->out); + if (need_err) + close_pair(fderr); + return err == ENOENT ? + -ERR_RUN_COMMAND_EXEC : + -ERR_RUN_COMMAND_FORK; + } + + if (need_in) + close(fdin[0]); + else if (cmd->in) + close(cmd->in); + + if (need_out) + close(fdout[1]); + else if (cmd->out) + close(cmd->out); + + if (need_err) + close(fderr[1]); + + return 0; +} + +static int wait_or_whine(pid_t pid) +{ + for (;;) { + int status, code; + pid_t waiting = waitpid(pid, &status, 0); + + if (waiting < 0) { + if (errno == EINTR) + continue; + error("waitpid failed (%s)", strerror(errno)); + return -ERR_RUN_COMMAND_WAITPID; + } + if (waiting != pid) + return -ERR_RUN_COMMAND_WAITPID_WRONG_PID; + if (WIFSIGNALED(status)) + return -ERR_RUN_COMMAND_WAITPID_SIGNAL; + + if (!WIFEXITED(status)) + return -ERR_RUN_COMMAND_WAITPID_NOEXIT; + code = WEXITSTATUS(status); + switch (code) { + case 127: + return -ERR_RUN_COMMAND_EXEC; + case 0: + return 0; + default: + return -code; + } + } +} + +int finish_command(struct child_process *cmd) +{ + return wait_or_whine(cmd->pid); +} + +int run_command(struct child_process *cmd) +{ + int code = start_command(cmd); + if (code) + return code; + return finish_command(cmd); +} + +static void prepare_run_command_v_opt(struct child_process *cmd, + const char **argv, + int opt) +{ + memset(cmd, 0, sizeof(*cmd)); + cmd->argv = argv; + cmd->no_stdin = opt & RUN_COMMAND_NO_STDIN ? 1 : 0; + cmd->perf_cmd = opt & RUN_PERF_CMD ? 1 : 0; + cmd->stdout_to_stderr = opt & RUN_COMMAND_STDOUT_TO_STDERR ? 1 : 0; +} + +int run_command_v_opt(const char **argv, int opt) +{ + struct child_process cmd; + prepare_run_command_v_opt(&cmd, argv, opt); + return run_command(&cmd); +} diff --git a/tools/perf/util/run-command.h b/tools/perf/util/run-command.h new file mode 100644 index 00000000..1ef264d5 --- /dev/null +++ b/tools/perf/util/run-command.h @@ -0,0 +1,58 @@ +#ifndef __PERF_RUN_COMMAND_H +#define __PERF_RUN_COMMAND_H + +enum { + ERR_RUN_COMMAND_FORK = 10000, + ERR_RUN_COMMAND_EXEC, + ERR_RUN_COMMAND_PIPE, + ERR_RUN_COMMAND_WAITPID, + ERR_RUN_COMMAND_WAITPID_WRONG_PID, + ERR_RUN_COMMAND_WAITPID_SIGNAL, + ERR_RUN_COMMAND_WAITPID_NOEXIT, +}; +#define IS_RUN_COMMAND_ERR(x) (-(x) >= ERR_RUN_COMMAND_FORK) + +struct child_process { + const char **argv; + pid_t pid; + /* + * Using .in, .out, .err: + * - Specify 0 for no redirections (child inherits stdin, stdout, + * stderr from parent). + * - Specify -1 to have a pipe allocated as follows: + * .in: returns the writable pipe end; parent writes to it, + * the readable pipe end becomes child's stdin + * .out, .err: returns the readable pipe end; parent reads from + * it, the writable pipe end becomes child's stdout/stderr + * The caller of start_command() must close the returned FDs + * after it has completed reading from/writing to it! + * - Specify > 0 to set a channel to a particular FD as follows: + * .in: a readable FD, becomes child's stdin + * .out: a writable FD, becomes child's stdout/stderr + * .err > 0 not supported + * The specified FD is closed by start_command(), even in case + * of errors! + */ + int in; + int out; + int err; + const char *dir; + const char *const *env; + unsigned no_stdin:1; + unsigned no_stdout:1; + unsigned no_stderr:1; + unsigned perf_cmd:1; /* if this is to be perf sub-command */ + unsigned stdout_to_stderr:1; + void (*preexec_cb)(void); +}; + +int start_command(struct child_process *); +int finish_command(struct child_process *); +int run_command(struct child_process *); + +#define RUN_COMMAND_NO_STDIN 1 +#define RUN_PERF_CMD 2 /*If this is to be perf sub-command */ +#define RUN_COMMAND_STDOUT_TO_STDERR 4 +int run_command_v_opt(const char **argv, int opt); + +#endif /* __PERF_RUN_COMMAND_H */ diff --git a/tools/perf/util/scripting-engines/trace-event-perl.c b/tools/perf/util/scripting-engines/trace-event-perl.c new file mode 100644 index 00000000..74350ffb --- /dev/null +++ b/tools/perf/util/scripting-engines/trace-event-perl.c @@ -0,0 +1,571 @@ +/* + * trace-event-perl. Feed perf script events to an embedded Perl interpreter. + * + * Copyright (C) 2009 Tom Zanussi <tzanussi@gmail.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <ctype.h> +#include <errno.h> + +#include "../../perf.h" +#include "../util.h" +#include "../trace-event.h" + +#include <EXTERN.h> +#include <perl.h> + +void boot_Perf__Trace__Context(pTHX_ CV *cv); +void boot_DynaLoader(pTHX_ CV *cv); +typedef PerlInterpreter * INTERP; + +void xs_init(pTHX); + +void xs_init(pTHX) +{ + const char *file = __FILE__; + dXSUB_SYS; + + newXS("Perf::Trace::Context::bootstrap", boot_Perf__Trace__Context, + file); + newXS("DynaLoader::boot_DynaLoader", boot_DynaLoader, file); +} + +INTERP my_perl; + +#define FTRACE_MAX_EVENT \ + ((1 << (sizeof(unsigned short) * 8)) - 1) + +struct event *events[FTRACE_MAX_EVENT]; + +extern struct scripting_context *scripting_context; + +static char *cur_field_name; +static int zero_flag_atom; + +static void define_symbolic_value(const char *ev_name, + const char *field_name, + const char *field_value, + const char *field_str) +{ + unsigned long long value; + dSP; + + value = eval_flag(field_value); + + ENTER; + SAVETMPS; + PUSHMARK(SP); + + XPUSHs(sv_2mortal(newSVpv(ev_name, 0))); + XPUSHs(sv_2mortal(newSVpv(field_name, 0))); + XPUSHs(sv_2mortal(newSVuv(value))); + XPUSHs(sv_2mortal(newSVpv(field_str, 0))); + + PUTBACK; + if (get_cv("main::define_symbolic_value", 0)) + call_pv("main::define_symbolic_value", G_SCALAR); + SPAGAIN; + PUTBACK; + FREETMPS; + LEAVE; +} + +static void define_symbolic_values(struct print_flag_sym *field, + const char *ev_name, + const char *field_name) +{ + define_symbolic_value(ev_name, field_name, field->value, field->str); + if (field->next) + define_symbolic_values(field->next, ev_name, field_name); +} + +static void define_symbolic_field(const char *ev_name, + const char *field_name) +{ + dSP; + + ENTER; + SAVETMPS; + PUSHMARK(SP); + + XPUSHs(sv_2mortal(newSVpv(ev_name, 0))); + XPUSHs(sv_2mortal(newSVpv(field_name, 0))); + + PUTBACK; + if (get_cv("main::define_symbolic_field", 0)) + call_pv("main::define_symbolic_field", G_SCALAR); + SPAGAIN; + PUTBACK; + FREETMPS; + LEAVE; +} + +static void define_flag_value(const char *ev_name, + const char *field_name, + const char *field_value, + const char *field_str) +{ + unsigned long long value; + dSP; + + value = eval_flag(field_value); + + ENTER; + SAVETMPS; + PUSHMARK(SP); + + XPUSHs(sv_2mortal(newSVpv(ev_name, 0))); + XPUSHs(sv_2mortal(newSVpv(field_name, 0))); + XPUSHs(sv_2mortal(newSVuv(value))); + XPUSHs(sv_2mortal(newSVpv(field_str, 0))); + + PUTBACK; + if (get_cv("main::define_flag_value", 0)) + call_pv("main::define_flag_value", G_SCALAR); + SPAGAIN; + PUTBACK; + FREETMPS; + LEAVE; +} + +static void define_flag_values(struct print_flag_sym *field, + const char *ev_name, + const char *field_name) +{ + define_flag_value(ev_name, field_name, field->value, field->str); + if (field->next) + define_flag_values(field->next, ev_name, field_name); +} + +static void define_flag_field(const char *ev_name, + const char *field_name, + const char *delim) +{ + dSP; + + ENTER; + SAVETMPS; + PUSHMARK(SP); + + XPUSHs(sv_2mortal(newSVpv(ev_name, 0))); + XPUSHs(sv_2mortal(newSVpv(field_name, 0))); + XPUSHs(sv_2mortal(newSVpv(delim, 0))); + + PUTBACK; + if (get_cv("main::define_flag_field", 0)) + call_pv("main::define_flag_field", G_SCALAR); + SPAGAIN; + PUTBACK; + FREETMPS; + LEAVE; +} + +static void define_event_symbols(struct event *event, + const char *ev_name, + struct print_arg *args) +{ + switch (args->type) { + case PRINT_NULL: + break; + case PRINT_ATOM: + define_flag_value(ev_name, cur_field_name, "0", + args->atom.atom); + zero_flag_atom = 0; + break; + case PRINT_FIELD: + if (cur_field_name) + free(cur_field_name); + cur_field_name = strdup(args->field.name); + break; + case PRINT_FLAGS: + define_event_symbols(event, ev_name, args->flags.field); + define_flag_field(ev_name, cur_field_name, args->flags.delim); + define_flag_values(args->flags.flags, ev_name, cur_field_name); + break; + case PRINT_SYMBOL: + define_event_symbols(event, ev_name, args->symbol.field); + define_symbolic_field(ev_name, cur_field_name); + define_symbolic_values(args->symbol.symbols, ev_name, + cur_field_name); + break; + case PRINT_STRING: + break; + case PRINT_TYPE: + define_event_symbols(event, ev_name, args->typecast.item); + break; + case PRINT_OP: + if (strcmp(args->op.op, ":") == 0) + zero_flag_atom = 1; + define_event_symbols(event, ev_name, args->op.left); + define_event_symbols(event, ev_name, args->op.right); + break; + default: + /* we should warn... */ + return; + } + + if (args->next) + define_event_symbols(event, ev_name, args->next); +} + +static inline struct event *find_cache_event(int type) +{ + static char ev_name[256]; + struct event *event; + + if (events[type]) + return events[type]; + + events[type] = event = trace_find_event(type); + if (!event) + return NULL; + + sprintf(ev_name, "%s::%s", event->system, event->name); + + define_event_symbols(event, ev_name, event->print_fmt.args); + + return event; +} + +static void perl_process_event(union perf_event *pevent __unused, + struct perf_sample *sample, + struct perf_evsel *evsel, + struct perf_session *session __unused, + struct thread *thread) +{ + struct format_field *field; + static char handler[256]; + unsigned long long val; + unsigned long s, ns; + struct event *event; + int type; + int pid; + int cpu = sample->cpu; + void *data = sample->raw_data; + unsigned long long nsecs = sample->time; + char *comm = thread->comm; + + dSP; + + type = trace_parse_common_type(data); + + event = find_cache_event(type); + if (!event) + die("ug! no event found for type %d", type); + + pid = trace_parse_common_pid(data); + + sprintf(handler, "%s::%s", event->system, event->name); + + s = nsecs / NSECS_PER_SEC; + ns = nsecs - s * NSECS_PER_SEC; + + scripting_context->event_data = data; + + ENTER; + SAVETMPS; + PUSHMARK(SP); + + XPUSHs(sv_2mortal(newSVpv(handler, 0))); + XPUSHs(sv_2mortal(newSViv(PTR2IV(scripting_context)))); + XPUSHs(sv_2mortal(newSVuv(cpu))); + XPUSHs(sv_2mortal(newSVuv(s))); + XPUSHs(sv_2mortal(newSVuv(ns))); + XPUSHs(sv_2mortal(newSViv(pid))); + XPUSHs(sv_2mortal(newSVpv(comm, 0))); + + /* common fields other than pid can be accessed via xsub fns */ + + for (field = event->format.fields; field; field = field->next) { + if (field->flags & FIELD_IS_STRING) { + int offset; + if (field->flags & FIELD_IS_DYNAMIC) { + offset = *(int *)(data + field->offset); + offset &= 0xffff; + } else + offset = field->offset; + XPUSHs(sv_2mortal(newSVpv((char *)data + offset, 0))); + } else { /* FIELD_IS_NUMERIC */ + val = read_size(data + field->offset, field->size); + if (field->flags & FIELD_IS_SIGNED) { + XPUSHs(sv_2mortal(newSViv(val))); + } else { + XPUSHs(sv_2mortal(newSVuv(val))); + } + } + } + + PUTBACK; + + if (get_cv(handler, 0)) + call_pv(handler, G_SCALAR); + else if (get_cv("main::trace_unhandled", 0)) { + XPUSHs(sv_2mortal(newSVpv(handler, 0))); + XPUSHs(sv_2mortal(newSViv(PTR2IV(scripting_context)))); + XPUSHs(sv_2mortal(newSVuv(cpu))); + XPUSHs(sv_2mortal(newSVuv(nsecs))); + XPUSHs(sv_2mortal(newSViv(pid))); + XPUSHs(sv_2mortal(newSVpv(comm, 0))); + call_pv("main::trace_unhandled", G_SCALAR); + } + SPAGAIN; + PUTBACK; + FREETMPS; + LEAVE; +} + +static void run_start_sub(void) +{ + dSP; /* access to Perl stack */ + PUSHMARK(SP); + + if (get_cv("main::trace_begin", 0)) + call_pv("main::trace_begin", G_DISCARD | G_NOARGS); +} + +/* + * Start trace script + */ +static int perl_start_script(const char *script, int argc, const char **argv) +{ + const char **command_line; + int i, err = 0; + + command_line = malloc((argc + 2) * sizeof(const char *)); + command_line[0] = ""; + command_line[1] = script; + for (i = 2; i < argc + 2; i++) + command_line[i] = argv[i - 2]; + + my_perl = perl_alloc(); + perl_construct(my_perl); + + if (perl_parse(my_perl, xs_init, argc + 2, (char **)command_line, + (char **)NULL)) { + err = -1; + goto error; + } + + if (perl_run(my_perl)) { + err = -1; + goto error; + } + + if (SvTRUE(ERRSV)) { + err = -1; + goto error; + } + + run_start_sub(); + + free(command_line); + return 0; +error: + perl_free(my_perl); + free(command_line); + + return err; +} + +/* + * Stop trace script + */ +static int perl_stop_script(void) +{ + dSP; /* access to Perl stack */ + PUSHMARK(SP); + + if (get_cv("main::trace_end", 0)) + call_pv("main::trace_end", G_DISCARD | G_NOARGS); + + perl_destruct(my_perl); + perl_free(my_perl); + + return 0; +} + +static int perl_generate_script(const char *outfile) +{ + struct event *event = NULL; + struct format_field *f; + char fname[PATH_MAX]; + int not_first, count; + FILE *ofp; + + sprintf(fname, "%s.pl", outfile); + ofp = fopen(fname, "w"); + if (ofp == NULL) { + fprintf(stderr, "couldn't open %s\n", fname); + return -1; + } + + fprintf(ofp, "# perf script event handlers, " + "generated by perf script -g perl\n"); + + fprintf(ofp, "# Licensed under the terms of the GNU GPL" + " License version 2\n\n"); + + fprintf(ofp, "# The common_* event handler fields are the most useful " + "fields common to\n"); + + fprintf(ofp, "# all events. They don't necessarily correspond to " + "the 'common_*' fields\n"); + + fprintf(ofp, "# in the format files. Those fields not available as " + "handler params can\n"); + + fprintf(ofp, "# be retrieved using Perl functions of the form " + "common_*($context).\n"); + + fprintf(ofp, "# See Context.pm for the list of available " + "functions.\n\n"); + + fprintf(ofp, "use lib \"$ENV{'PERF_EXEC_PATH'}/scripts/perl/" + "Perf-Trace-Util/lib\";\n"); + + fprintf(ofp, "use lib \"./Perf-Trace-Util/lib\";\n"); + fprintf(ofp, "use Perf::Trace::Core;\n"); + fprintf(ofp, "use Perf::Trace::Context;\n"); + fprintf(ofp, "use Perf::Trace::Util;\n\n"); + + fprintf(ofp, "sub trace_begin\n{\n\t# optional\n}\n\n"); + fprintf(ofp, "sub trace_end\n{\n\t# optional\n}\n\n"); + + while ((event = trace_find_next_event(event))) { + fprintf(ofp, "sub %s::%s\n{\n", event->system, event->name); + fprintf(ofp, "\tmy ("); + + fprintf(ofp, "$event_name, "); + fprintf(ofp, "$context, "); + fprintf(ofp, "$common_cpu, "); + fprintf(ofp, "$common_secs, "); + fprintf(ofp, "$common_nsecs,\n"); + fprintf(ofp, "\t $common_pid, "); + fprintf(ofp, "$common_comm,\n\t "); + + not_first = 0; + count = 0; + + for (f = event->format.fields; f; f = f->next) { + if (not_first++) + fprintf(ofp, ", "); + if (++count % 5 == 0) + fprintf(ofp, "\n\t "); + + fprintf(ofp, "$%s", f->name); + } + fprintf(ofp, ") = @_;\n\n"); + + fprintf(ofp, "\tprint_header($event_name, $common_cpu, " + "$common_secs, $common_nsecs,\n\t " + "$common_pid, $common_comm);\n\n"); + + fprintf(ofp, "\tprintf(\""); + + not_first = 0; + count = 0; + + for (f = event->format.fields; f; f = f->next) { + if (not_first++) + fprintf(ofp, ", "); + if (count && count % 4 == 0) { + fprintf(ofp, "\".\n\t \""); + } + count++; + + fprintf(ofp, "%s=", f->name); + if (f->flags & FIELD_IS_STRING || + f->flags & FIELD_IS_FLAG || + f->flags & FIELD_IS_SYMBOLIC) + fprintf(ofp, "%%s"); + else if (f->flags & FIELD_IS_SIGNED) + fprintf(ofp, "%%d"); + else + fprintf(ofp, "%%u"); + } + + fprintf(ofp, "\\n\",\n\t "); + + not_first = 0; + count = 0; + + for (f = event->format.fields; f; f = f->next) { + if (not_first++) + fprintf(ofp, ", "); + + if (++count % 5 == 0) + fprintf(ofp, "\n\t "); + + if (f->flags & FIELD_IS_FLAG) { + if ((count - 1) % 5 != 0) { + fprintf(ofp, "\n\t "); + count = 4; + } + fprintf(ofp, "flag_str(\""); + fprintf(ofp, "%s::%s\", ", event->system, + event->name); + fprintf(ofp, "\"%s\", $%s)", f->name, + f->name); + } else if (f->flags & FIELD_IS_SYMBOLIC) { + if ((count - 1) % 5 != 0) { + fprintf(ofp, "\n\t "); + count = 4; + } + fprintf(ofp, "symbol_str(\""); + fprintf(ofp, "%s::%s\", ", event->system, + event->name); + fprintf(ofp, "\"%s\", $%s)", f->name, + f->name); + } else + fprintf(ofp, "$%s", f->name); + } + + fprintf(ofp, ");\n"); + fprintf(ofp, "}\n\n"); + } + + fprintf(ofp, "sub trace_unhandled\n{\n\tmy ($event_name, $context, " + "$common_cpu, $common_secs, $common_nsecs,\n\t " + "$common_pid, $common_comm) = @_;\n\n"); + + fprintf(ofp, "\tprint_header($event_name, $common_cpu, " + "$common_secs, $common_nsecs,\n\t $common_pid, " + "$common_comm);\n}\n\n"); + + fprintf(ofp, "sub print_header\n{\n" + "\tmy ($event_name, $cpu, $secs, $nsecs, $pid, $comm) = @_;\n\n" + "\tprintf(\"%%-20s %%5u %%05u.%%09u %%8u %%-20s \",\n\t " + "$event_name, $cpu, $secs, $nsecs, $pid, $comm);\n}"); + + fclose(ofp); + + fprintf(stderr, "generated Perl script: %s\n", fname); + + return 0; +} + +struct scripting_ops perl_scripting_ops = { + .name = "Perl", + .start_script = perl_start_script, + .stop_script = perl_stop_script, + .process_event = perl_process_event, + .generate_script = perl_generate_script, +}; diff --git a/tools/perf/util/scripting-engines/trace-event-python.c b/tools/perf/util/scripting-engines/trace-event-python.c new file mode 100644 index 00000000..6ccf70e8 --- /dev/null +++ b/tools/perf/util/scripting-engines/trace-event-python.c @@ -0,0 +1,599 @@ +/* + * trace-event-python. Feed trace events to an embedded Python interpreter. + * + * Copyright (C) 2010 Tom Zanussi <tzanussi@gmail.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#include <Python.h> + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <ctype.h> +#include <errno.h> + +#include "../../perf.h" +#include "../util.h" +#include "../trace-event.h" + +PyMODINIT_FUNC initperf_trace_context(void); + +#define FTRACE_MAX_EVENT \ + ((1 << (sizeof(unsigned short) * 8)) - 1) + +struct event *events[FTRACE_MAX_EVENT]; + +#define MAX_FIELDS 64 +#define N_COMMON_FIELDS 7 + +extern struct scripting_context *scripting_context; + +static char *cur_field_name; +static int zero_flag_atom; + +static PyObject *main_module, *main_dict; + +static void handler_call_die(const char *handler_name) +{ + PyErr_Print(); + Py_FatalError("problem in Python trace event handler"); +} + +static void define_value(enum print_arg_type field_type, + const char *ev_name, + const char *field_name, + const char *field_value, + const char *field_str) +{ + const char *handler_name = "define_flag_value"; + PyObject *handler, *t, *retval; + unsigned long long value; + unsigned n = 0; + + if (field_type == PRINT_SYMBOL) + handler_name = "define_symbolic_value"; + + t = PyTuple_New(4); + if (!t) + Py_FatalError("couldn't create Python tuple"); + + value = eval_flag(field_value); + + PyTuple_SetItem(t, n++, PyString_FromString(ev_name)); + PyTuple_SetItem(t, n++, PyString_FromString(field_name)); + PyTuple_SetItem(t, n++, PyInt_FromLong(value)); + PyTuple_SetItem(t, n++, PyString_FromString(field_str)); + + handler = PyDict_GetItemString(main_dict, handler_name); + if (handler && PyCallable_Check(handler)) { + retval = PyObject_CallObject(handler, t); + if (retval == NULL) + handler_call_die(handler_name); + } + + Py_DECREF(t); +} + +static void define_values(enum print_arg_type field_type, + struct print_flag_sym *field, + const char *ev_name, + const char *field_name) +{ + define_value(field_type, ev_name, field_name, field->value, + field->str); + + if (field->next) + define_values(field_type, field->next, ev_name, field_name); +} + +static void define_field(enum print_arg_type field_type, + const char *ev_name, + const char *field_name, + const char *delim) +{ + const char *handler_name = "define_flag_field"; + PyObject *handler, *t, *retval; + unsigned n = 0; + + if (field_type == PRINT_SYMBOL) + handler_name = "define_symbolic_field"; + + if (field_type == PRINT_FLAGS) + t = PyTuple_New(3); + else + t = PyTuple_New(2); + if (!t) + Py_FatalError("couldn't create Python tuple"); + + PyTuple_SetItem(t, n++, PyString_FromString(ev_name)); + PyTuple_SetItem(t, n++, PyString_FromString(field_name)); + if (field_type == PRINT_FLAGS) + PyTuple_SetItem(t, n++, PyString_FromString(delim)); + + handler = PyDict_GetItemString(main_dict, handler_name); + if (handler && PyCallable_Check(handler)) { + retval = PyObject_CallObject(handler, t); + if (retval == NULL) + handler_call_die(handler_name); + } + + Py_DECREF(t); +} + +static void define_event_symbols(struct event *event, + const char *ev_name, + struct print_arg *args) +{ + switch (args->type) { + case PRINT_NULL: + break; + case PRINT_ATOM: + define_value(PRINT_FLAGS, ev_name, cur_field_name, "0", + args->atom.atom); + zero_flag_atom = 0; + break; + case PRINT_FIELD: + if (cur_field_name) + free(cur_field_name); + cur_field_name = strdup(args->field.name); + break; + case PRINT_FLAGS: + define_event_symbols(event, ev_name, args->flags.field); + define_field(PRINT_FLAGS, ev_name, cur_field_name, + args->flags.delim); + define_values(PRINT_FLAGS, args->flags.flags, ev_name, + cur_field_name); + break; + case PRINT_SYMBOL: + define_event_symbols(event, ev_name, args->symbol.field); + define_field(PRINT_SYMBOL, ev_name, cur_field_name, NULL); + define_values(PRINT_SYMBOL, args->symbol.symbols, ev_name, + cur_field_name); + break; + case PRINT_STRING: + break; + case PRINT_TYPE: + define_event_symbols(event, ev_name, args->typecast.item); + break; + case PRINT_OP: + if (strcmp(args->op.op, ":") == 0) + zero_flag_atom = 1; + define_event_symbols(event, ev_name, args->op.left); + define_event_symbols(event, ev_name, args->op.right); + break; + default: + /* we should warn... */ + return; + } + + if (args->next) + define_event_symbols(event, ev_name, args->next); +} + +static inline struct event *find_cache_event(int type) +{ + static char ev_name[256]; + struct event *event; + + if (events[type]) + return events[type]; + + events[type] = event = trace_find_event(type); + if (!event) + return NULL; + + sprintf(ev_name, "%s__%s", event->system, event->name); + + define_event_symbols(event, ev_name, event->print_fmt.args); + + return event; +} + +static void python_process_event(union perf_event *pevent __unused, + struct perf_sample *sample, + struct perf_evsel *evsel __unused, + struct perf_session *session __unused, + struct thread *thread) +{ + PyObject *handler, *retval, *context, *t, *obj, *dict = NULL; + static char handler_name[256]; + struct format_field *field; + unsigned long long val; + unsigned long s, ns; + struct event *event; + unsigned n = 0; + int type; + int pid; + int cpu = sample->cpu; + void *data = sample->raw_data; + unsigned long long nsecs = sample->time; + char *comm = thread->comm; + + t = PyTuple_New(MAX_FIELDS); + if (!t) + Py_FatalError("couldn't create Python tuple"); + + type = trace_parse_common_type(data); + + event = find_cache_event(type); + if (!event) + die("ug! no event found for type %d", type); + + pid = trace_parse_common_pid(data); + + sprintf(handler_name, "%s__%s", event->system, event->name); + + handler = PyDict_GetItemString(main_dict, handler_name); + if (handler && !PyCallable_Check(handler)) + handler = NULL; + if (!handler) { + dict = PyDict_New(); + if (!dict) + Py_FatalError("couldn't create Python dict"); + } + s = nsecs / NSECS_PER_SEC; + ns = nsecs - s * NSECS_PER_SEC; + + scripting_context->event_data = data; + + context = PyCObject_FromVoidPtr(scripting_context, NULL); + + PyTuple_SetItem(t, n++, PyString_FromString(handler_name)); + PyTuple_SetItem(t, n++, context); + + if (handler) { + PyTuple_SetItem(t, n++, PyInt_FromLong(cpu)); + PyTuple_SetItem(t, n++, PyInt_FromLong(s)); + PyTuple_SetItem(t, n++, PyInt_FromLong(ns)); + PyTuple_SetItem(t, n++, PyInt_FromLong(pid)); + PyTuple_SetItem(t, n++, PyString_FromString(comm)); + } else { + PyDict_SetItemString(dict, "common_cpu", PyInt_FromLong(cpu)); + PyDict_SetItemString(dict, "common_s", PyInt_FromLong(s)); + PyDict_SetItemString(dict, "common_ns", PyInt_FromLong(ns)); + PyDict_SetItemString(dict, "common_pid", PyInt_FromLong(pid)); + PyDict_SetItemString(dict, "common_comm", PyString_FromString(comm)); + } + for (field = event->format.fields; field; field = field->next) { + if (field->flags & FIELD_IS_STRING) { + int offset; + if (field->flags & FIELD_IS_DYNAMIC) { + offset = *(int *)(data + field->offset); + offset &= 0xffff; + } else + offset = field->offset; + obj = PyString_FromString((char *)data + offset); + } else { /* FIELD_IS_NUMERIC */ + val = read_size(data + field->offset, field->size); + if (field->flags & FIELD_IS_SIGNED) { + if ((long long)val >= LONG_MIN && + (long long)val <= LONG_MAX) + obj = PyInt_FromLong(val); + else + obj = PyLong_FromLongLong(val); + } else { + if (val <= LONG_MAX) + obj = PyInt_FromLong(val); + else + obj = PyLong_FromUnsignedLongLong(val); + } + } + if (handler) + PyTuple_SetItem(t, n++, obj); + else + PyDict_SetItemString(dict, field->name, obj); + + } + if (!handler) + PyTuple_SetItem(t, n++, dict); + + if (_PyTuple_Resize(&t, n) == -1) + Py_FatalError("error resizing Python tuple"); + + if (handler) { + retval = PyObject_CallObject(handler, t); + if (retval == NULL) + handler_call_die(handler_name); + } else { + handler = PyDict_GetItemString(main_dict, "trace_unhandled"); + if (handler && PyCallable_Check(handler)) { + + retval = PyObject_CallObject(handler, t); + if (retval == NULL) + handler_call_die("trace_unhandled"); + } + Py_DECREF(dict); + } + + Py_DECREF(t); +} + +static int run_start_sub(void) +{ + PyObject *handler, *retval; + int err = 0; + + main_module = PyImport_AddModule("__main__"); + if (main_module == NULL) + return -1; + Py_INCREF(main_module); + + main_dict = PyModule_GetDict(main_module); + if (main_dict == NULL) { + err = -1; + goto error; + } + Py_INCREF(main_dict); + + handler = PyDict_GetItemString(main_dict, "trace_begin"); + if (handler == NULL || !PyCallable_Check(handler)) + goto out; + + retval = PyObject_CallObject(handler, NULL); + if (retval == NULL) + handler_call_die("trace_begin"); + + Py_DECREF(retval); + return err; +error: + Py_XDECREF(main_dict); + Py_XDECREF(main_module); +out: + return err; +} + +/* + * Start trace script + */ +static int python_start_script(const char *script, int argc, const char **argv) +{ + const char **command_line; + char buf[PATH_MAX]; + int i, err = 0; + FILE *fp; + + command_line = malloc((argc + 1) * sizeof(const char *)); + command_line[0] = script; + for (i = 1; i < argc + 1; i++) + command_line[i] = argv[i - 1]; + + Py_Initialize(); + + initperf_trace_context(); + + PySys_SetArgv(argc + 1, (char **)command_line); + + fp = fopen(script, "r"); + if (!fp) { + sprintf(buf, "Can't open python script \"%s\"", script); + perror(buf); + err = -1; + goto error; + } + + err = PyRun_SimpleFile(fp, script); + if (err) { + fprintf(stderr, "Error running python script %s\n", script); + goto error; + } + + err = run_start_sub(); + if (err) { + fprintf(stderr, "Error starting python script %s\n", script); + goto error; + } + + free(command_line); + + return err; +error: + Py_Finalize(); + free(command_line); + + return err; +} + +/* + * Stop trace script + */ +static int python_stop_script(void) +{ + PyObject *handler, *retval; + int err = 0; + + handler = PyDict_GetItemString(main_dict, "trace_end"); + if (handler == NULL || !PyCallable_Check(handler)) + goto out; + + retval = PyObject_CallObject(handler, NULL); + if (retval == NULL) + handler_call_die("trace_end"); + else + Py_DECREF(retval); +out: + Py_XDECREF(main_dict); + Py_XDECREF(main_module); + Py_Finalize(); + + return err; +} + +static int python_generate_script(const char *outfile) +{ + struct event *event = NULL; + struct format_field *f; + char fname[PATH_MAX]; + int not_first, count; + FILE *ofp; + + sprintf(fname, "%s.py", outfile); + ofp = fopen(fname, "w"); + if (ofp == NULL) { + fprintf(stderr, "couldn't open %s\n", fname); + return -1; + } + fprintf(ofp, "# perf script event handlers, " + "generated by perf script -g python\n"); + + fprintf(ofp, "# Licensed under the terms of the GNU GPL" + " License version 2\n\n"); + + fprintf(ofp, "# The common_* event handler fields are the most useful " + "fields common to\n"); + + fprintf(ofp, "# all events. They don't necessarily correspond to " + "the 'common_*' fields\n"); + + fprintf(ofp, "# in the format files. Those fields not available as " + "handler params can\n"); + + fprintf(ofp, "# be retrieved using Python functions of the form " + "common_*(context).\n"); + + fprintf(ofp, "# See the perf-trace-python Documentation for the list " + "of available functions.\n\n"); + + fprintf(ofp, "import os\n"); + fprintf(ofp, "import sys\n\n"); + + fprintf(ofp, "sys.path.append(os.environ['PERF_EXEC_PATH'] + \\\n"); + fprintf(ofp, "\t'/scripts/python/Perf-Trace-Util/lib/Perf/Trace')\n"); + fprintf(ofp, "\nfrom perf_trace_context import *\n"); + fprintf(ofp, "from Core import *\n\n\n"); + + fprintf(ofp, "def trace_begin():\n"); + fprintf(ofp, "\tprint \"in trace_begin\"\n\n"); + + fprintf(ofp, "def trace_end():\n"); + fprintf(ofp, "\tprint \"in trace_end\"\n\n"); + + while ((event = trace_find_next_event(event))) { + fprintf(ofp, "def %s__%s(", event->system, event->name); + fprintf(ofp, "event_name, "); + fprintf(ofp, "context, "); + fprintf(ofp, "common_cpu,\n"); + fprintf(ofp, "\tcommon_secs, "); + fprintf(ofp, "common_nsecs, "); + fprintf(ofp, "common_pid, "); + fprintf(ofp, "common_comm,\n\t"); + + not_first = 0; + count = 0; + + for (f = event->format.fields; f; f = f->next) { + if (not_first++) + fprintf(ofp, ", "); + if (++count % 5 == 0) + fprintf(ofp, "\n\t"); + + fprintf(ofp, "%s", f->name); + } + fprintf(ofp, "):\n"); + + fprintf(ofp, "\t\tprint_header(event_name, common_cpu, " + "common_secs, common_nsecs,\n\t\t\t" + "common_pid, common_comm)\n\n"); + + fprintf(ofp, "\t\tprint \""); + + not_first = 0; + count = 0; + + for (f = event->format.fields; f; f = f->next) { + if (not_first++) + fprintf(ofp, ", "); + if (count && count % 3 == 0) { + fprintf(ofp, "\" \\\n\t\t\""); + } + count++; + + fprintf(ofp, "%s=", f->name); + if (f->flags & FIELD_IS_STRING || + f->flags & FIELD_IS_FLAG || + f->flags & FIELD_IS_SYMBOLIC) + fprintf(ofp, "%%s"); + else if (f->flags & FIELD_IS_SIGNED) + fprintf(ofp, "%%d"); + else + fprintf(ofp, "%%u"); + } + + fprintf(ofp, "\\n\" %% \\\n\t\t("); + + not_first = 0; + count = 0; + + for (f = event->format.fields; f; f = f->next) { + if (not_first++) + fprintf(ofp, ", "); + + if (++count % 5 == 0) + fprintf(ofp, "\n\t\t"); + + if (f->flags & FIELD_IS_FLAG) { + if ((count - 1) % 5 != 0) { + fprintf(ofp, "\n\t\t"); + count = 4; + } + fprintf(ofp, "flag_str(\""); + fprintf(ofp, "%s__%s\", ", event->system, + event->name); + fprintf(ofp, "\"%s\", %s)", f->name, + f->name); + } else if (f->flags & FIELD_IS_SYMBOLIC) { + if ((count - 1) % 5 != 0) { + fprintf(ofp, "\n\t\t"); + count = 4; + } + fprintf(ofp, "symbol_str(\""); + fprintf(ofp, "%s__%s\", ", event->system, + event->name); + fprintf(ofp, "\"%s\", %s)", f->name, + f->name); + } else + fprintf(ofp, "%s", f->name); + } + + fprintf(ofp, "),\n\n"); + } + + fprintf(ofp, "def trace_unhandled(event_name, context, " + "event_fields_dict):\n"); + + fprintf(ofp, "\t\tprint ' '.join(['%%s=%%s'%%(k,str(v))" + "for k,v in sorted(event_fields_dict.items())])\n\n"); + + fprintf(ofp, "def print_header(" + "event_name, cpu, secs, nsecs, pid, comm):\n" + "\tprint \"%%-20s %%5u %%05u.%%09u %%8u %%-20s \" %% \\\n\t" + "(event_name, cpu, secs, nsecs, pid, comm),\n"); + + fclose(ofp); + + fprintf(stderr, "generated Python script: %s\n", fname); + + return 0; +} + +struct scripting_ops python_scripting_ops = { + .name = "Python", + .start_script = python_start_script, + .stop_script = python_stop_script, + .process_event = python_process_event, + .generate_script = python_generate_script, +}; diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c new file mode 100644 index 00000000..2dbf0abd --- /dev/null +++ b/tools/perf/util/session.c @@ -0,0 +1,1270 @@ +#define _FILE_OFFSET_BITS 64 + +#include <linux/kernel.h> + +#include <byteswap.h> +#include <unistd.h> +#include <sys/types.h> +#include <sys/mman.h> + +#include "evlist.h" +#include "evsel.h" +#include "session.h" +#include "sort.h" +#include "util.h" + +static int perf_session__open(struct perf_session *self, bool force) +{ + struct stat input_stat; + + if (!strcmp(self->filename, "-")) { + self->fd_pipe = true; + self->fd = STDIN_FILENO; + + if (perf_session__read_header(self, self->fd) < 0) + pr_err("incompatible file format"); + + return 0; + } + + self->fd = open(self->filename, O_RDONLY); + if (self->fd < 0) { + int err = errno; + + pr_err("failed to open %s: %s", self->filename, strerror(err)); + if (err == ENOENT && !strcmp(self->filename, "perf.data")) + pr_err(" (try 'perf record' first)"); + pr_err("\n"); + return -errno; + } + + if (fstat(self->fd, &input_stat) < 0) + goto out_close; + + if (!force && input_stat.st_uid && (input_stat.st_uid != geteuid())) { + pr_err("file %s not owned by current user or root\n", + self->filename); + goto out_close; + } + + if (!input_stat.st_size) { + pr_info("zero-sized file (%s), nothing to do!\n", + self->filename); + goto out_close; + } + + if (perf_session__read_header(self, self->fd) < 0) { + pr_err("incompatible file format"); + goto out_close; + } + + if (!perf_evlist__valid_sample_type(self->evlist)) { + pr_err("non matching sample_type"); + goto out_close; + } + + if (!perf_evlist__valid_sample_id_all(self->evlist)) { + pr_err("non matching sample_id_all"); + goto out_close; + } + + self->size = input_stat.st_size; + return 0; + +out_close: + close(self->fd); + self->fd = -1; + return -1; +} + +static void perf_session__id_header_size(struct perf_session *session) +{ + struct perf_sample *data; + u64 sample_type = session->sample_type; + u16 size = 0; + + if (!session->sample_id_all) + goto out; + + if (sample_type & PERF_SAMPLE_TID) + size += sizeof(data->tid) * 2; + + if (sample_type & PERF_SAMPLE_TIME) + size += sizeof(data->time); + + if (sample_type & PERF_SAMPLE_ID) + size += sizeof(data->id); + + if (sample_type & PERF_SAMPLE_STREAM_ID) + size += sizeof(data->stream_id); + + if (sample_type & PERF_SAMPLE_CPU) + size += sizeof(data->cpu) * 2; +out: + session->id_hdr_size = size; +} + +void perf_session__update_sample_type(struct perf_session *self) +{ + self->sample_type = perf_evlist__sample_type(self->evlist); + self->sample_size = __perf_evsel__sample_size(self->sample_type); + self->sample_id_all = perf_evlist__sample_id_all(self->evlist); + perf_session__id_header_size(self); +} + +int perf_session__create_kernel_maps(struct perf_session *self) +{ + int ret = machine__create_kernel_maps(&self->host_machine); + + if (ret >= 0) + ret = machines__create_guest_kernel_maps(&self->machines); + return ret; +} + +static void perf_session__destroy_kernel_maps(struct perf_session *self) +{ + machine__destroy_kernel_maps(&self->host_machine); + machines__destroy_guest_kernel_maps(&self->machines); +} + +struct perf_session *perf_session__new(const char *filename, int mode, + bool force, bool repipe, + struct perf_event_ops *ops) +{ + size_t len = filename ? strlen(filename) + 1 : 0; + struct perf_session *self = zalloc(sizeof(*self) + len); + + if (self == NULL) + goto out; + + memcpy(self->filename, filename, len); + self->threads = RB_ROOT; + INIT_LIST_HEAD(&self->dead_threads); + self->last_match = NULL; + /* + * On 64bit we can mmap the data file in one go. No need for tiny mmap + * slices. On 32bit we use 32MB. + */ +#if BITS_PER_LONG == 64 + self->mmap_window = ULLONG_MAX; +#else + self->mmap_window = 32 * 1024 * 1024ULL; +#endif + self->machines = RB_ROOT; + self->repipe = repipe; + INIT_LIST_HEAD(&self->ordered_samples.samples); + INIT_LIST_HEAD(&self->ordered_samples.sample_cache); + INIT_LIST_HEAD(&self->ordered_samples.to_free); + machine__init(&self->host_machine, "", HOST_KERNEL_ID); + + if (mode == O_RDONLY) { + if (perf_session__open(self, force) < 0) + goto out_delete; + perf_session__update_sample_type(self); + } else if (mode == O_WRONLY) { + /* + * In O_RDONLY mode this will be performed when reading the + * kernel MMAP event, in perf_event__process_mmap(). + */ + if (perf_session__create_kernel_maps(self) < 0) + goto out_delete; + } + + if (ops && ops->ordering_requires_timestamps && + ops->ordered_samples && !self->sample_id_all) { + dump_printf("WARNING: No sample_id_all support, falling back to unordered processing\n"); + ops->ordered_samples = false; + } + +out: + return self; +out_delete: + perf_session__delete(self); + return NULL; +} + +static void perf_session__delete_dead_threads(struct perf_session *self) +{ + struct thread *n, *t; + + list_for_each_entry_safe(t, n, &self->dead_threads, node) { + list_del(&t->node); + thread__delete(t); + } +} + +static void perf_session__delete_threads(struct perf_session *self) +{ + struct rb_node *nd = rb_first(&self->threads); + + while (nd) { + struct thread *t = rb_entry(nd, struct thread, rb_node); + + rb_erase(&t->rb_node, &self->threads); + nd = rb_next(nd); + thread__delete(t); + } +} + +void perf_session__delete(struct perf_session *self) +{ + perf_session__destroy_kernel_maps(self); + perf_session__delete_dead_threads(self); + perf_session__delete_threads(self); + machine__exit(&self->host_machine); + close(self->fd); + free(self); +} + +void perf_session__remove_thread(struct perf_session *self, struct thread *th) +{ + self->last_match = NULL; + rb_erase(&th->rb_node, &self->threads); + /* + * We may have references to this thread, for instance in some hist_entry + * instances, so just move them to a separate list. + */ + list_add_tail(&th->node, &self->dead_threads); +} + +static bool symbol__match_parent_regex(struct symbol *sym) +{ + if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0)) + return 1; + + return 0; +} + +int perf_session__resolve_callchain(struct perf_session *self, + struct thread *thread, + struct ip_callchain *chain, + struct symbol **parent) +{ + u8 cpumode = PERF_RECORD_MISC_USER; + unsigned int i; + int err; + + callchain_cursor_reset(&self->callchain_cursor); + + for (i = 0; i < chain->nr; i++) { + u64 ip = chain->ips[i]; + struct addr_location al; + + if (ip >= PERF_CONTEXT_MAX) { + switch (ip) { + case PERF_CONTEXT_HV: + cpumode = PERF_RECORD_MISC_HYPERVISOR; break; + case PERF_CONTEXT_KERNEL: + cpumode = PERF_RECORD_MISC_KERNEL; break; + case PERF_CONTEXT_USER: + cpumode = PERF_RECORD_MISC_USER; break; + default: + break; + } + continue; + } + + al.filtered = false; + thread__find_addr_location(thread, self, cpumode, + MAP__FUNCTION, thread->pid, ip, &al, NULL); + if (al.sym != NULL) { + if (sort__has_parent && !*parent && + symbol__match_parent_regex(al.sym)) + *parent = al.sym; + if (!symbol_conf.use_callchain) + break; + } + + err = callchain_cursor_append(&self->callchain_cursor, + ip, al.map, al.sym); + if (err) + return err; + } + + return 0; +} + +static int process_event_synth_stub(union perf_event *event __used, + struct perf_session *session __used) +{ + dump_printf(": unhandled!\n"); + return 0; +} + +static int process_event_sample_stub(union perf_event *event __used, + struct perf_sample *sample __used, + struct perf_evsel *evsel __used, + struct perf_session *session __used) +{ + dump_printf(": unhandled!\n"); + return 0; +} + +static int process_event_stub(union perf_event *event __used, + struct perf_sample *sample __used, + struct perf_session *session __used) +{ + dump_printf(": unhandled!\n"); + return 0; +} + +static int process_finished_round_stub(union perf_event *event __used, + struct perf_session *session __used, + struct perf_event_ops *ops __used) +{ + dump_printf(": unhandled!\n"); + return 0; +} + +static int process_finished_round(union perf_event *event, + struct perf_session *session, + struct perf_event_ops *ops); + +static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) +{ + if (handler->sample == NULL) + handler->sample = process_event_sample_stub; + if (handler->mmap == NULL) + handler->mmap = process_event_stub; + if (handler->comm == NULL) + handler->comm = process_event_stub; + if (handler->fork == NULL) + handler->fork = process_event_stub; + if (handler->exit == NULL) + handler->exit = process_event_stub; + if (handler->lost == NULL) + handler->lost = perf_event__process_lost; + if (handler->read == NULL) + handler->read = process_event_stub; + if (handler->throttle == NULL) + handler->throttle = process_event_stub; + if (handler->unthrottle == NULL) + handler->unthrottle = process_event_stub; + if (handler->attr == NULL) + handler->attr = process_event_synth_stub; + if (handler->event_type == NULL) + handler->event_type = process_event_synth_stub; + if (handler->tracing_data == NULL) + handler->tracing_data = process_event_synth_stub; + if (handler->build_id == NULL) + handler->build_id = process_event_synth_stub; + if (handler->finished_round == NULL) { + if (handler->ordered_samples) + handler->finished_round = process_finished_round; + else + handler->finished_round = process_finished_round_stub; + } +} + +void mem_bswap_64(void *src, int byte_size) +{ + u64 *m = src; + + while (byte_size > 0) { + *m = bswap_64(*m); + byte_size -= sizeof(u64); + ++m; + } +} + +static void perf_event__all64_swap(union perf_event *event) +{ + struct perf_event_header *hdr = &event->header; + mem_bswap_64(hdr + 1, event->header.size - sizeof(*hdr)); +} + +static void perf_event__comm_swap(union perf_event *event) +{ + event->comm.pid = bswap_32(event->comm.pid); + event->comm.tid = bswap_32(event->comm.tid); +} + +static void perf_event__mmap_swap(union perf_event *event) +{ + event->mmap.pid = bswap_32(event->mmap.pid); + event->mmap.tid = bswap_32(event->mmap.tid); + event->mmap.start = bswap_64(event->mmap.start); + event->mmap.len = bswap_64(event->mmap.len); + event->mmap.pgoff = bswap_64(event->mmap.pgoff); +} + +static void perf_event__task_swap(union perf_event *event) +{ + event->fork.pid = bswap_32(event->fork.pid); + event->fork.tid = bswap_32(event->fork.tid); + event->fork.ppid = bswap_32(event->fork.ppid); + event->fork.ptid = bswap_32(event->fork.ptid); + event->fork.time = bswap_64(event->fork.time); +} + +static void perf_event__read_swap(union perf_event *event) +{ + event->read.pid = bswap_32(event->read.pid); + event->read.tid = bswap_32(event->read.tid); + event->read.value = bswap_64(event->read.value); + event->read.time_enabled = bswap_64(event->read.time_enabled); + event->read.time_running = bswap_64(event->read.time_running); + event->read.id = bswap_64(event->read.id); +} + +/* exported for swapping attributes in file header */ +void perf_event__attr_swap(struct perf_event_attr *attr) +{ + attr->type = bswap_32(attr->type); + attr->size = bswap_32(attr->size); + attr->config = bswap_64(attr->config); + attr->sample_period = bswap_64(attr->sample_period); + attr->sample_type = bswap_64(attr->sample_type); + attr->read_format = bswap_64(attr->read_format); + attr->wakeup_events = bswap_32(attr->wakeup_events); + attr->bp_type = bswap_32(attr->bp_type); + attr->bp_addr = bswap_64(attr->bp_addr); + attr->bp_len = bswap_64(attr->bp_len); +} + +static void perf_event__hdr_attr_swap(union perf_event *event) +{ + size_t size; + + perf_event__attr_swap(&event->attr.attr); + + size = event->header.size; + size -= (void *)&event->attr.id - (void *)event; + mem_bswap_64(event->attr.id, size); +} + +static void perf_event__event_type_swap(union perf_event *event) +{ + event->event_type.event_type.event_id = + bswap_64(event->event_type.event_type.event_id); +} + +static void perf_event__tracing_data_swap(union perf_event *event) +{ + event->tracing_data.size = bswap_32(event->tracing_data.size); +} + +typedef void (*perf_event__swap_op)(union perf_event *event); + +static perf_event__swap_op perf_event__swap_ops[] = { + [PERF_RECORD_MMAP] = perf_event__mmap_swap, + [PERF_RECORD_COMM] = perf_event__comm_swap, + [PERF_RECORD_FORK] = perf_event__task_swap, + [PERF_RECORD_EXIT] = perf_event__task_swap, + [PERF_RECORD_LOST] = perf_event__all64_swap, + [PERF_RECORD_READ] = perf_event__read_swap, + [PERF_RECORD_SAMPLE] = perf_event__all64_swap, + [PERF_RECORD_HEADER_ATTR] = perf_event__hdr_attr_swap, + [PERF_RECORD_HEADER_EVENT_TYPE] = perf_event__event_type_swap, + [PERF_RECORD_HEADER_TRACING_DATA] = perf_event__tracing_data_swap, + [PERF_RECORD_HEADER_BUILD_ID] = NULL, + [PERF_RECORD_HEADER_MAX] = NULL, +}; + +struct sample_queue { + u64 timestamp; + u64 file_offset; + union perf_event *event; + struct list_head list; +}; + +static void perf_session_free_sample_buffers(struct perf_session *session) +{ + struct ordered_samples *os = &session->ordered_samples; + + while (!list_empty(&os->to_free)) { + struct sample_queue *sq; + + sq = list_entry(os->to_free.next, struct sample_queue, list); + list_del(&sq->list); + free(sq); + } +} + +static int perf_session_deliver_event(struct perf_session *session, + union perf_event *event, + struct perf_sample *sample, + struct perf_event_ops *ops, + u64 file_offset); + +static void flush_sample_queue(struct perf_session *s, + struct perf_event_ops *ops) +{ + struct ordered_samples *os = &s->ordered_samples; + struct list_head *head = &os->samples; + struct sample_queue *tmp, *iter; + struct perf_sample sample; + u64 limit = os->next_flush; + u64 last_ts = os->last_sample ? os->last_sample->timestamp : 0ULL; + int ret; + + if (!ops->ordered_samples || !limit) + return; + + list_for_each_entry_safe(iter, tmp, head, list) { + if (iter->timestamp > limit) + break; + + ret = perf_session__parse_sample(s, iter->event, &sample); + if (ret) + pr_err("Can't parse sample, err = %d\n", ret); + else + perf_session_deliver_event(s, iter->event, &sample, ops, + iter->file_offset); + + os->last_flush = iter->timestamp; + list_del(&iter->list); + list_add(&iter->list, &os->sample_cache); + } + + if (list_empty(head)) { + os->last_sample = NULL; + } else if (last_ts <= limit) { + os->last_sample = + list_entry(head->prev, struct sample_queue, list); + } +} + +/* + * When perf record finishes a pass on every buffers, it records this pseudo + * event. + * We record the max timestamp t found in the pass n. + * Assuming these timestamps are monotonic across cpus, we know that if + * a buffer still has events with timestamps below t, they will be all + * available and then read in the pass n + 1. + * Hence when we start to read the pass n + 2, we can safely flush every + * events with timestamps below t. + * + * ============ PASS n ================= + * CPU 0 | CPU 1 + * | + * cnt1 timestamps | cnt2 timestamps + * 1 | 2 + * 2 | 3 + * - | 4 <--- max recorded + * + * ============ PASS n + 1 ============== + * CPU 0 | CPU 1 + * | + * cnt1 timestamps | cnt2 timestamps + * 3 | 5 + * 4 | 6 + * 5 | 7 <---- max recorded + * + * Flush every events below timestamp 4 + * + * ============ PASS n + 2 ============== + * CPU 0 | CPU 1 + * | + * cnt1 timestamps | cnt2 timestamps + * 6 | 8 + * 7 | 9 + * - | 10 + * + * Flush every events below timestamp 7 + * etc... + */ +static int process_finished_round(union perf_event *event __used, + struct perf_session *session, + struct perf_event_ops *ops) +{ + flush_sample_queue(session, ops); + session->ordered_samples.next_flush = session->ordered_samples.max_timestamp; + + return 0; +} + +/* The queue is ordered by time */ +static void __queue_event(struct sample_queue *new, struct perf_session *s) +{ + struct ordered_samples *os = &s->ordered_samples; + struct sample_queue *sample = os->last_sample; + u64 timestamp = new->timestamp; + struct list_head *p; + + os->last_sample = new; + + if (!sample) { + list_add(&new->list, &os->samples); + os->max_timestamp = timestamp; + return; + } + + /* + * last_sample might point to some random place in the list as it's + * the last queued event. We expect that the new event is close to + * this. + */ + if (sample->timestamp <= timestamp) { + while (sample->timestamp <= timestamp) { + p = sample->list.next; + if (p == &os->samples) { + list_add_tail(&new->list, &os->samples); + os->max_timestamp = timestamp; + return; + } + sample = list_entry(p, struct sample_queue, list); + } + list_add_tail(&new->list, &sample->list); + } else { + while (sample->timestamp > timestamp) { + p = sample->list.prev; + if (p == &os->samples) { + list_add(&new->list, &os->samples); + return; + } + sample = list_entry(p, struct sample_queue, list); + } + list_add(&new->list, &sample->list); + } +} + +#define MAX_SAMPLE_BUFFER (64 * 1024 / sizeof(struct sample_queue)) + +static int perf_session_queue_event(struct perf_session *s, union perf_event *event, + struct perf_sample *sample, u64 file_offset) +{ + struct ordered_samples *os = &s->ordered_samples; + struct list_head *sc = &os->sample_cache; + u64 timestamp = sample->time; + struct sample_queue *new; + + if (!timestamp || timestamp == ~0ULL) + return -ETIME; + + if (timestamp < s->ordered_samples.last_flush) { + printf("Warning: Timestamp below last timeslice flush\n"); + return -EINVAL; + } + + if (!list_empty(sc)) { + new = list_entry(sc->next, struct sample_queue, list); + list_del(&new->list); + } else if (os->sample_buffer) { + new = os->sample_buffer + os->sample_buffer_idx; + if (++os->sample_buffer_idx == MAX_SAMPLE_BUFFER) + os->sample_buffer = NULL; + } else { + os->sample_buffer = malloc(MAX_SAMPLE_BUFFER * sizeof(*new)); + if (!os->sample_buffer) + return -ENOMEM; + list_add(&os->sample_buffer->list, &os->to_free); + os->sample_buffer_idx = 2; + new = os->sample_buffer + 1; + } + + new->timestamp = timestamp; + new->file_offset = file_offset; + new->event = event; + + __queue_event(new, s); + + return 0; +} + +static void callchain__printf(struct perf_sample *sample) +{ + unsigned int i; + + printf("... chain: nr:%" PRIu64 "\n", sample->callchain->nr); + + for (i = 0; i < sample->callchain->nr; i++) + printf("..... %2d: %016" PRIx64 "\n", + i, sample->callchain->ips[i]); +} + +static void perf_session__print_tstamp(struct perf_session *session, + union perf_event *event, + struct perf_sample *sample) +{ + if (event->header.type != PERF_RECORD_SAMPLE && + !session->sample_id_all) { + fputs("-1 -1 ", stdout); + return; + } + + if ((session->sample_type & PERF_SAMPLE_CPU)) + printf("%u ", sample->cpu); + + if (session->sample_type & PERF_SAMPLE_TIME) + printf("%" PRIu64 " ", sample->time); +} + +static void dump_event(struct perf_session *session, union perf_event *event, + u64 file_offset, struct perf_sample *sample) +{ + if (!dump_trace) + return; + + printf("\n%#" PRIx64 " [%#x]: event: %d\n", + file_offset, event->header.size, event->header.type); + + trace_event(event); + + if (sample) + perf_session__print_tstamp(session, event, sample); + + printf("%#" PRIx64 " [%#x]: PERF_RECORD_%s", file_offset, + event->header.size, perf_event__name(event->header.type)); +} + +static void dump_sample(struct perf_session *session, union perf_event *event, + struct perf_sample *sample) +{ + if (!dump_trace) + return; + + printf("(IP, %d): %d/%d: %#" PRIx64 " period: %" PRIu64 "\n", + event->header.misc, sample->pid, sample->tid, sample->ip, + sample->period); + + if (session->sample_type & PERF_SAMPLE_CALLCHAIN) + callchain__printf(sample); +} + +static int perf_session_deliver_event(struct perf_session *session, + union perf_event *event, + struct perf_sample *sample, + struct perf_event_ops *ops, + u64 file_offset) +{ + struct perf_evsel *evsel; + + dump_event(session, event, file_offset, sample); + + switch (event->header.type) { + case PERF_RECORD_SAMPLE: + dump_sample(session, event, sample); + evsel = perf_evlist__id2evsel(session->evlist, sample->id); + if (evsel == NULL) { + ++session->hists.stats.nr_unknown_id; + return -1; + } + return ops->sample(event, sample, evsel, session); + case PERF_RECORD_MMAP: + return ops->mmap(event, sample, session); + case PERF_RECORD_COMM: + return ops->comm(event, sample, session); + case PERF_RECORD_FORK: + return ops->fork(event, sample, session); + case PERF_RECORD_EXIT: + return ops->exit(event, sample, session); + case PERF_RECORD_LOST: + return ops->lost(event, sample, session); + case PERF_RECORD_READ: + return ops->read(event, sample, session); + case PERF_RECORD_THROTTLE: + return ops->throttle(event, sample, session); + case PERF_RECORD_UNTHROTTLE: + return ops->unthrottle(event, sample, session); + default: + ++session->hists.stats.nr_unknown_events; + return -1; + } +} + +static int perf_session__preprocess_sample(struct perf_session *session, + union perf_event *event, struct perf_sample *sample) +{ + if (event->header.type != PERF_RECORD_SAMPLE || + !(session->sample_type & PERF_SAMPLE_CALLCHAIN)) + return 0; + + if (!ip_callchain__valid(sample->callchain, event)) { + pr_debug("call-chain problem with event, skipping it.\n"); + ++session->hists.stats.nr_invalid_chains; + session->hists.stats.total_invalid_chains += sample->period; + return -EINVAL; + } + return 0; +} + +static int perf_session__process_user_event(struct perf_session *session, union perf_event *event, + struct perf_event_ops *ops, u64 file_offset) +{ + dump_event(session, event, file_offset, NULL); + + /* These events are processed right away */ + switch (event->header.type) { + case PERF_RECORD_HEADER_ATTR: + return ops->attr(event, session); + case PERF_RECORD_HEADER_EVENT_TYPE: + return ops->event_type(event, session); + case PERF_RECORD_HEADER_TRACING_DATA: + /* setup for reading amidst mmap */ + lseek(session->fd, file_offset, SEEK_SET); + return ops->tracing_data(event, session); + case PERF_RECORD_HEADER_BUILD_ID: + return ops->build_id(event, session); + case PERF_RECORD_FINISHED_ROUND: + return ops->finished_round(event, session, ops); + default: + return -EINVAL; + } +} + +static int perf_session__process_event(struct perf_session *session, + union perf_event *event, + struct perf_event_ops *ops, + u64 file_offset) +{ + struct perf_sample sample; + int ret; + + if (session->header.needs_swap && + perf_event__swap_ops[event->header.type]) + perf_event__swap_ops[event->header.type](event); + + if (event->header.type >= PERF_RECORD_HEADER_MAX) + return -EINVAL; + + hists__inc_nr_events(&session->hists, event->header.type); + + if (event->header.type >= PERF_RECORD_USER_TYPE_START) + return perf_session__process_user_event(session, event, ops, file_offset); + + /* + * For all kernel events we get the sample data + */ + ret = perf_session__parse_sample(session, event, &sample); + if (ret) + return ret; + + /* Preprocess sample records - precheck callchains */ + if (perf_session__preprocess_sample(session, event, &sample)) + return 0; + + if (ops->ordered_samples) { + ret = perf_session_queue_event(session, event, &sample, + file_offset); + if (ret != -ETIME) + return ret; + } + + return perf_session_deliver_event(session, event, &sample, ops, + file_offset); +} + +void perf_event_header__bswap(struct perf_event_header *self) +{ + self->type = bswap_32(self->type); + self->misc = bswap_16(self->misc); + self->size = bswap_16(self->size); +} + +static struct thread *perf_session__register_idle_thread(struct perf_session *self) +{ + struct thread *thread = perf_session__findnew(self, 0); + + if (thread == NULL || thread__set_comm(thread, "swapper")) { + pr_err("problem inserting idle task.\n"); + thread = NULL; + } + + return thread; +} + +static void perf_session__warn_about_errors(const struct perf_session *session, + const struct perf_event_ops *ops) +{ + if (ops->lost == perf_event__process_lost && + session->hists.stats.total_lost != 0) { + ui__warning("Processed %" PRIu64 " events and LOST %" PRIu64 + "!\n\nCheck IO/CPU overload!\n\n", + session->hists.stats.total_period, + session->hists.stats.total_lost); + } + + if (session->hists.stats.nr_unknown_events != 0) { + ui__warning("Found %u unknown events!\n\n" + "Is this an older tool processing a perf.data " + "file generated by a more recent tool?\n\n" + "If that is not the case, consider " + "reporting to linux-kernel@vger.kernel.org.\n\n", + session->hists.stats.nr_unknown_events); + } + + if (session->hists.stats.nr_unknown_id != 0) { + ui__warning("%u samples with id not present in the header\n", + session->hists.stats.nr_unknown_id); + } + + if (session->hists.stats.nr_invalid_chains != 0) { + ui__warning("Found invalid callchains!\n\n" + "%u out of %u events were discarded for this reason.\n\n" + "Consider reporting to linux-kernel@vger.kernel.org.\n\n", + session->hists.stats.nr_invalid_chains, + session->hists.stats.nr_events[PERF_RECORD_SAMPLE]); + } +} + +#define session_done() (*(volatile int *)(&session_done)) +volatile int session_done; + +static int __perf_session__process_pipe_events(struct perf_session *self, + struct perf_event_ops *ops) +{ + union perf_event event; + uint32_t size; + int skip = 0; + u64 head; + int err; + void *p; + + perf_event_ops__fill_defaults(ops); + + head = 0; +more: + err = readn(self->fd, &event, sizeof(struct perf_event_header)); + if (err <= 0) { + if (err == 0) + goto done; + + pr_err("failed to read event header\n"); + goto out_err; + } + + if (self->header.needs_swap) + perf_event_header__bswap(&event.header); + + size = event.header.size; + if (size == 0) + size = 8; + + p = &event; + p += sizeof(struct perf_event_header); + + if (size - sizeof(struct perf_event_header)) { + err = readn(self->fd, p, size - sizeof(struct perf_event_header)); + if (err <= 0) { + if (err == 0) { + pr_err("unexpected end of event stream\n"); + goto done; + } + + pr_err("failed to read event data\n"); + goto out_err; + } + } + + if (size == 0 || + (skip = perf_session__process_event(self, &event, ops, head)) < 0) { + dump_printf("%#" PRIx64 " [%#x]: skipping unknown header type: %d\n", + head, event.header.size, event.header.type); + /* + * assume we lost track of the stream, check alignment, and + * increment a single u64 in the hope to catch on again 'soon'. + */ + if (unlikely(head & 7)) + head &= ~7ULL; + + size = 8; + } + + head += size; + + if (skip > 0) + head += skip; + + if (!session_done()) + goto more; +done: + err = 0; +out_err: + perf_session__warn_about_errors(self, ops); + perf_session_free_sample_buffers(self); + return err; +} + +static union perf_event * +fetch_mmaped_event(struct perf_session *session, + u64 head, size_t mmap_size, char *buf) +{ + union perf_event *event; + + /* + * Ensure we have enough space remaining to read + * the size of the event in the headers. + */ + if (head + sizeof(event->header) > mmap_size) + return NULL; + + event = (union perf_event *)(buf + head); + + if (session->header.needs_swap) + perf_event_header__bswap(&event->header); + + if (head + event->header.size > mmap_size) + return NULL; + + return event; +} + +int __perf_session__process_events(struct perf_session *session, + u64 data_offset, u64 data_size, + u64 file_size, struct perf_event_ops *ops) +{ + u64 head, page_offset, file_offset, file_pos, progress_next; + int err, mmap_prot, mmap_flags, map_idx = 0; + struct ui_progress *progress; + size_t page_size, mmap_size; + char *buf, *mmaps[8]; + union perf_event *event; + uint32_t size; + + perf_event_ops__fill_defaults(ops); + + page_size = sysconf(_SC_PAGESIZE); + + page_offset = page_size * (data_offset / page_size); + file_offset = page_offset; + head = data_offset - page_offset; + + if (data_offset + data_size < file_size) + file_size = data_offset + data_size; + + progress_next = file_size / 16; + progress = ui_progress__new("Processing events...", file_size); + if (progress == NULL) + return -1; + + mmap_size = session->mmap_window; + if (mmap_size > file_size) + mmap_size = file_size; + + memset(mmaps, 0, sizeof(mmaps)); + + mmap_prot = PROT_READ; + mmap_flags = MAP_SHARED; + + if (session->header.needs_swap) { + mmap_prot |= PROT_WRITE; + mmap_flags = MAP_PRIVATE; + } +remap: + buf = mmap(NULL, mmap_size, mmap_prot, mmap_flags, session->fd, + file_offset); + if (buf == MAP_FAILED) { + pr_err("failed to mmap file\n"); + err = -errno; + goto out_err; + } + mmaps[map_idx] = buf; + map_idx = (map_idx + 1) & (ARRAY_SIZE(mmaps) - 1); + file_pos = file_offset + head; + +more: + event = fetch_mmaped_event(session, head, mmap_size, buf); + if (!event) { + if (mmaps[map_idx]) { + munmap(mmaps[map_idx], mmap_size); + mmaps[map_idx] = NULL; + } + + page_offset = page_size * (head / page_size); + file_offset += page_offset; + head -= page_offset; + goto remap; + } + + size = event->header.size; + + if (size == 0 || + perf_session__process_event(session, event, ops, file_pos) < 0) { + dump_printf("%#" PRIx64 " [%#x]: skipping unknown header type: %d\n", + file_offset + head, event->header.size, + event->header.type); + /* + * assume we lost track of the stream, check alignment, and + * increment a single u64 in the hope to catch on again 'soon'. + */ + if (unlikely(head & 7)) + head &= ~7ULL; + + size = 8; + } + + head += size; + file_pos += size; + + if (file_pos >= progress_next) { + progress_next += file_size / 16; + ui_progress__update(progress, file_pos); + } + + if (file_pos < file_size) + goto more; + + err = 0; + /* do the final flush for ordered samples */ + session->ordered_samples.next_flush = ULLONG_MAX; + flush_sample_queue(session, ops); +out_err: + ui_progress__delete(progress); + perf_session__warn_about_errors(session, ops); + perf_session_free_sample_buffers(session); + return err; +} + +int perf_session__process_events(struct perf_session *self, + struct perf_event_ops *ops) +{ + int err; + + if (perf_session__register_idle_thread(self) == NULL) + return -ENOMEM; + + if (!self->fd_pipe) + err = __perf_session__process_events(self, + self->header.data_offset, + self->header.data_size, + self->size, ops); + else + err = __perf_session__process_pipe_events(self, ops); + + return err; +} + +bool perf_session__has_traces(struct perf_session *self, const char *msg) +{ + if (!(self->sample_type & PERF_SAMPLE_RAW)) { + pr_err("No trace sample to read. Did you call 'perf %s'?\n", msg); + return false; + } + + return true; +} + +int perf_session__set_kallsyms_ref_reloc_sym(struct map **maps, + const char *symbol_name, + u64 addr) +{ + char *bracket; + enum map_type i; + struct ref_reloc_sym *ref; + + ref = zalloc(sizeof(struct ref_reloc_sym)); + if (ref == NULL) + return -ENOMEM; + + ref->name = strdup(symbol_name); + if (ref->name == NULL) { + free(ref); + return -ENOMEM; + } + + bracket = strchr(ref->name, ']'); + if (bracket) + *bracket = '\0'; + + ref->addr = addr; + + for (i = 0; i < MAP__NR_TYPES; ++i) { + struct kmap *kmap = map__kmap(maps[i]); + kmap->ref_reloc_sym = ref; + } + + return 0; +} + +size_t perf_session__fprintf_dsos(struct perf_session *self, FILE *fp) +{ + return __dsos__fprintf(&self->host_machine.kernel_dsos, fp) + + __dsos__fprintf(&self->host_machine.user_dsos, fp) + + machines__fprintf_dsos(&self->machines, fp); +} + +size_t perf_session__fprintf_dsos_buildid(struct perf_session *self, FILE *fp, + bool with_hits) +{ + size_t ret = machine__fprintf_dsos_buildid(&self->host_machine, fp, with_hits); + return ret + machines__fprintf_dsos_buildid(&self->machines, fp, with_hits); +} + +size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp) +{ + struct perf_evsel *pos; + size_t ret = fprintf(fp, "Aggregated stats:\n"); + + ret += hists__fprintf_nr_events(&session->hists, fp); + + list_for_each_entry(pos, &session->evlist->entries, node) { + ret += fprintf(fp, "%s stats:\n", event_name(pos)); + ret += hists__fprintf_nr_events(&pos->hists, fp); + } + + return ret; +} + +struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session, + unsigned int type) +{ + struct perf_evsel *pos; + + list_for_each_entry(pos, &session->evlist->entries, node) { + if (pos->attr.type == type) + return pos; + } + return NULL; +} + +void perf_session__print_symbols(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) +{ + struct addr_location al; + const char *symname, *dsoname; + struct callchain_cursor *cursor = &session->callchain_cursor; + struct callchain_cursor_node *node; + + if (perf_event__preprocess_sample(event, session, &al, sample, + NULL) < 0) { + error("problem processing %d event, skipping it.\n", + event->header.type); + return; + } + + if (symbol_conf.use_callchain && sample->callchain) { + + if (perf_session__resolve_callchain(session, al.thread, + sample->callchain, NULL) != 0) { + if (verbose) + error("Failed to resolve callchain. Skipping\n"); + return; + } + callchain_cursor_commit(cursor); + + while (1) { + node = callchain_cursor_current(cursor); + if (!node) + break; + + if (node->sym && node->sym->name) + symname = node->sym->name; + else + symname = ""; + + if (node->map && node->map->dso && node->map->dso->name) + dsoname = node->map->dso->name; + else + dsoname = ""; + + printf("\t%16" PRIx64 " %s (%s)\n", node->ip, symname, dsoname); + + callchain_cursor_advance(cursor); + } + + } else { + if (al.sym && al.sym->name) + symname = al.sym->name; + else + symname = ""; + + if (al.map && al.map->dso && al.map->dso->name) + dsoname = al.map->dso->name; + else + dsoname = ""; + + printf("%16" PRIx64 " %s (%s)", al.addr, symname, dsoname); + } +} diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h new file mode 100644 index 00000000..b84c0031 --- /dev/null +++ b/tools/perf/util/session.h @@ -0,0 +1,175 @@ +#ifndef __PERF_SESSION_H +#define __PERF_SESSION_H + +#include "hist.h" +#include "event.h" +#include "header.h" +#include "symbol.h" +#include "thread.h" +#include <linux/rbtree.h> +#include "../../../include/linux/perf_event.h" + +struct sample_queue; +struct ip_callchain; +struct thread; + +struct ordered_samples { + u64 last_flush; + u64 next_flush; + u64 max_timestamp; + struct list_head samples; + struct list_head sample_cache; + struct list_head to_free; + struct sample_queue *sample_buffer; + struct sample_queue *last_sample; + int sample_buffer_idx; +}; + +struct perf_session { + struct perf_header header; + unsigned long size; + unsigned long mmap_window; + struct rb_root threads; + struct list_head dead_threads; + struct thread *last_match; + struct machine host_machine; + struct rb_root machines; + struct perf_evlist *evlist; + /* + * FIXME: Need to split this up further, we need global + * stats + per event stats. 'perf diff' also needs + * to properly support multiple events in a single + * perf.data file. + */ + struct hists hists; + u64 sample_type; + int sample_size; + int fd; + bool fd_pipe; + bool repipe; + bool sample_id_all; + u16 id_hdr_size; + int cwdlen; + char *cwd; + struct ordered_samples ordered_samples; + struct callchain_cursor callchain_cursor; + char filename[0]; +}; + +struct perf_evsel; +struct perf_event_ops; + +typedef int (*event_sample)(union perf_event *event, struct perf_sample *sample, + struct perf_evsel *evsel, struct perf_session *session); +typedef int (*event_op)(union perf_event *self, struct perf_sample *sample, + struct perf_session *session); +typedef int (*event_synth_op)(union perf_event *self, + struct perf_session *session); +typedef int (*event_op2)(union perf_event *self, struct perf_session *session, + struct perf_event_ops *ops); + +struct perf_event_ops { + event_sample sample; + event_op mmap, + comm, + fork, + exit, + lost, + read, + throttle, + unthrottle; + event_synth_op attr, + event_type, + tracing_data, + build_id; + event_op2 finished_round; + bool ordered_samples; + bool ordering_requires_timestamps; +}; + +struct perf_session *perf_session__new(const char *filename, int mode, + bool force, bool repipe, + struct perf_event_ops *ops); +void perf_session__delete(struct perf_session *self); + +void perf_event_header__bswap(struct perf_event_header *self); + +int __perf_session__process_events(struct perf_session *self, + u64 data_offset, u64 data_size, u64 size, + struct perf_event_ops *ops); +int perf_session__process_events(struct perf_session *self, + struct perf_event_ops *event_ops); + +int perf_session__resolve_callchain(struct perf_session *self, + struct thread *thread, + struct ip_callchain *chain, + struct symbol **parent); + +bool perf_session__has_traces(struct perf_session *self, const char *msg); + +int perf_session__set_kallsyms_ref_reloc_sym(struct map **maps, + const char *symbol_name, + u64 addr); + +void mem_bswap_64(void *src, int byte_size); +void perf_event__attr_swap(struct perf_event_attr *attr); + +int perf_session__create_kernel_maps(struct perf_session *self); + +void perf_session__update_sample_type(struct perf_session *self); +void perf_session__remove_thread(struct perf_session *self, struct thread *th); + +static inline +struct machine *perf_session__find_host_machine(struct perf_session *self) +{ + return &self->host_machine; +} + +static inline +struct machine *perf_session__find_machine(struct perf_session *self, pid_t pid) +{ + if (pid == HOST_KERNEL_ID) + return &self->host_machine; + return machines__find(&self->machines, pid); +} + +static inline +struct machine *perf_session__findnew_machine(struct perf_session *self, pid_t pid) +{ + if (pid == HOST_KERNEL_ID) + return &self->host_machine; + return machines__findnew(&self->machines, pid); +} + +static inline +void perf_session__process_machines(struct perf_session *self, + machine__process_t process) +{ + process(&self->host_machine, self); + return machines__process(&self->machines, process, self); +} + +size_t perf_session__fprintf_dsos(struct perf_session *self, FILE *fp); + +size_t perf_session__fprintf_dsos_buildid(struct perf_session *self, + FILE *fp, bool with_hits); + +size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp); + +static inline int perf_session__parse_sample(struct perf_session *session, + const union perf_event *event, + struct perf_sample *sample) +{ + return perf_event__parse_sample(event, session->sample_type, + session->sample_size, + session->sample_id_all, sample); +} + +struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session, + unsigned int type); + +void perf_session__print_symbols(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session); + +#endif /* __PERF_SESSION_H */ diff --git a/tools/perf/util/setup.py b/tools/perf/util/setup.py new file mode 100644 index 00000000..bbc982f5 --- /dev/null +++ b/tools/perf/util/setup.py @@ -0,0 +1,24 @@ +#!/usr/bin/python2 + +from distutils.core import setup, Extension +from os import getenv + +cflags = ['-fno-strict-aliasing', '-Wno-write-strings'] +cflags += getenv('CFLAGS', '').split() + +perf = Extension('perf', + sources = ['util/python.c', 'util/ctype.c', 'util/evlist.c', + 'util/evsel.c', 'util/cpumap.c', 'util/thread_map.c', + 'util/util.c', 'util/xyarray.c', 'util/cgroup.c'], + include_dirs = ['util/include'], + extra_compile_args = cflags, + ) + +setup(name='perf', + version='0.1', + description='Interface with the Linux profiling infrastructure', + author='Arnaldo Carvalho de Melo', + author_email='acme@redhat.com', + license='GPLv2', + url='http://perf.wiki.kernel.org', + ext_modules=[perf]) diff --git a/tools/perf/util/sigchain.c b/tools/perf/util/sigchain.c new file mode 100644 index 00000000..ba785e9b --- /dev/null +++ b/tools/perf/util/sigchain.c @@ -0,0 +1,52 @@ +#include "sigchain.h" +#include "cache.h" + +#define SIGCHAIN_MAX_SIGNALS 32 + +struct sigchain_signal { + sigchain_fun *old; + int n; + int alloc; +}; +static struct sigchain_signal signals[SIGCHAIN_MAX_SIGNALS]; + +static void check_signum(int sig) +{ + if (sig < 1 || sig >= SIGCHAIN_MAX_SIGNALS) + die("BUG: signal out of range: %d", sig); +} + +static int sigchain_push(int sig, sigchain_fun f) +{ + struct sigchain_signal *s = signals + sig; + check_signum(sig); + + ALLOC_GROW(s->old, s->n + 1, s->alloc); + s->old[s->n] = signal(sig, f); + if (s->old[s->n] == SIG_ERR) + return -1; + s->n++; + return 0; +} + +int sigchain_pop(int sig) +{ + struct sigchain_signal *s = signals + sig; + check_signum(sig); + if (s->n < 1) + return 0; + + if (signal(sig, s->old[s->n - 1]) == SIG_ERR) + return -1; + s->n--; + return 0; +} + +void sigchain_push_common(sigchain_fun f) +{ + sigchain_push(SIGINT, f); + sigchain_push(SIGHUP, f); + sigchain_push(SIGTERM, f); + sigchain_push(SIGQUIT, f); + sigchain_push(SIGPIPE, f); +} diff --git a/tools/perf/util/sigchain.h b/tools/perf/util/sigchain.h new file mode 100644 index 00000000..959d64eb --- /dev/null +++ b/tools/perf/util/sigchain.h @@ -0,0 +1,10 @@ +#ifndef __PERF_SIGCHAIN_H +#define __PERF_SIGCHAIN_H + +typedef void (*sigchain_fun)(int); + +int sigchain_pop(int sig); + +void sigchain_push_common(sigchain_fun f); + +#endif /* __PERF_SIGCHAIN_H */ diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c new file mode 100644 index 00000000..0ffbc8e5 --- /dev/null +++ b/tools/perf/util/sort.c @@ -0,0 +1,348 @@ +#include "sort.h" +#include "hist.h" + +regex_t parent_regex; +const char default_parent_pattern[] = "^sys_|^do_page_fault"; +const char *parent_pattern = default_parent_pattern; +const char default_sort_order[] = "comm,dso,symbol"; +const char *sort_order = default_sort_order; +int sort__need_collapse = 0; +int sort__has_parent = 0; + +enum sort_type sort__first_dimension; + +char * field_sep; + +LIST_HEAD(hist_entry__sort_list); + +static int hist_entry__thread_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width); +static int hist_entry__comm_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width); +static int hist_entry__dso_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width); +static int hist_entry__sym_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width); +static int hist_entry__parent_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width); +static int hist_entry__cpu_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width); + +struct sort_entry sort_thread = { + .se_header = "Command: Pid", + .se_cmp = sort__thread_cmp, + .se_snprintf = hist_entry__thread_snprintf, + .se_width_idx = HISTC_THREAD, +}; + +struct sort_entry sort_comm = { + .se_header = "Command", + .se_cmp = sort__comm_cmp, + .se_collapse = sort__comm_collapse, + .se_snprintf = hist_entry__comm_snprintf, + .se_width_idx = HISTC_COMM, +}; + +struct sort_entry sort_dso = { + .se_header = "Shared Object", + .se_cmp = sort__dso_cmp, + .se_snprintf = hist_entry__dso_snprintf, + .se_width_idx = HISTC_DSO, +}; + +struct sort_entry sort_sym = { + .se_header = "Symbol", + .se_cmp = sort__sym_cmp, + .se_snprintf = hist_entry__sym_snprintf, + .se_width_idx = HISTC_SYMBOL, +}; + +struct sort_entry sort_parent = { + .se_header = "Parent symbol", + .se_cmp = sort__parent_cmp, + .se_snprintf = hist_entry__parent_snprintf, + .se_width_idx = HISTC_PARENT, +}; + +struct sort_entry sort_cpu = { + .se_header = "CPU", + .se_cmp = sort__cpu_cmp, + .se_snprintf = hist_entry__cpu_snprintf, + .se_width_idx = HISTC_CPU, +}; + +struct sort_dimension { + const char *name; + struct sort_entry *entry; + int taken; +}; + +static struct sort_dimension sort_dimensions[] = { + { .name = "pid", .entry = &sort_thread, }, + { .name = "comm", .entry = &sort_comm, }, + { .name = "dso", .entry = &sort_dso, }, + { .name = "symbol", .entry = &sort_sym, }, + { .name = "parent", .entry = &sort_parent, }, + { .name = "cpu", .entry = &sort_cpu, }, +}; + +int64_t cmp_null(void *l, void *r) +{ + if (!l && !r) + return 0; + else if (!l) + return -1; + else + return 1; +} + +/* --sort pid */ + +int64_t +sort__thread_cmp(struct hist_entry *left, struct hist_entry *right) +{ + return right->thread->pid - left->thread->pid; +} + +static int repsep_snprintf(char *bf, size_t size, const char *fmt, ...) +{ + int n; + va_list ap; + + va_start(ap, fmt); + n = vsnprintf(bf, size, fmt, ap); + if (field_sep && n > 0) { + char *sep = bf; + + while (1) { + sep = strchr(sep, *field_sep); + if (sep == NULL) + break; + *sep = '.'; + } + } + va_end(ap); + + if (n >= (int)size) + return size - 1; + return n; +} + +static int hist_entry__thread_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width) +{ + return repsep_snprintf(bf, size, "%*s:%5d", width, + self->thread->comm ?: "", self->thread->pid); +} + +static int hist_entry__comm_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width) +{ + return repsep_snprintf(bf, size, "%*s", width, self->thread->comm); +} + +/* --sort dso */ + +int64_t +sort__dso_cmp(struct hist_entry *left, struct hist_entry *right) +{ + struct dso *dso_l = left->ms.map ? left->ms.map->dso : NULL; + struct dso *dso_r = right->ms.map ? right->ms.map->dso : NULL; + const char *dso_name_l, *dso_name_r; + + if (!dso_l || !dso_r) + return cmp_null(dso_l, dso_r); + + if (verbose) { + dso_name_l = dso_l->long_name; + dso_name_r = dso_r->long_name; + } else { + dso_name_l = dso_l->short_name; + dso_name_r = dso_r->short_name; + } + + return strcmp(dso_name_l, dso_name_r); +} + +static int hist_entry__dso_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width) +{ + if (self->ms.map && self->ms.map->dso) { + const char *dso_name = !verbose ? self->ms.map->dso->short_name : + self->ms.map->dso->long_name; + return repsep_snprintf(bf, size, "%-*s", width, dso_name); + } + + return repsep_snprintf(bf, size, "%-*s", width, "[unknown]"); +} + +/* --sort symbol */ + +int64_t +sort__sym_cmp(struct hist_entry *left, struct hist_entry *right) +{ + u64 ip_l, ip_r; + + if (left->ms.sym == right->ms.sym) + return 0; + + ip_l = left->ms.sym ? left->ms.sym->start : left->ip; + ip_r = right->ms.sym ? right->ms.sym->start : right->ip; + + return (int64_t)(ip_r - ip_l); +} + +static int hist_entry__sym_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width __used) +{ + size_t ret = 0; + + if (verbose) { + char o = self->ms.map ? dso__symtab_origin(self->ms.map->dso) : '!'; + ret += repsep_snprintf(bf, size, "%-#*llx %c ", + BITS_PER_LONG / 4, self->ip, o); + } + + ret += repsep_snprintf(bf + ret, size - ret, "[%c] ", self->level); + if (self->ms.sym) + ret += repsep_snprintf(bf + ret, size - ret, "%s", + self->ms.sym->name); + else + ret += repsep_snprintf(bf + ret, size - ret, "%-#*llx", + BITS_PER_LONG / 4, self->ip); + + return ret; +} + +/* --sort comm */ + +int64_t +sort__comm_cmp(struct hist_entry *left, struct hist_entry *right) +{ + return right->thread->pid - left->thread->pid; +} + +int64_t +sort__comm_collapse(struct hist_entry *left, struct hist_entry *right) +{ + char *comm_l = left->thread->comm; + char *comm_r = right->thread->comm; + + if (!comm_l || !comm_r) + return cmp_null(comm_l, comm_r); + + return strcmp(comm_l, comm_r); +} + +/* --sort parent */ + +int64_t +sort__parent_cmp(struct hist_entry *left, struct hist_entry *right) +{ + struct symbol *sym_l = left->parent; + struct symbol *sym_r = right->parent; + + if (!sym_l || !sym_r) + return cmp_null(sym_l, sym_r); + + return strcmp(sym_l->name, sym_r->name); +} + +static int hist_entry__parent_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width) +{ + return repsep_snprintf(bf, size, "%-*s", width, + self->parent ? self->parent->name : "[other]"); +} + +/* --sort cpu */ + +int64_t +sort__cpu_cmp(struct hist_entry *left, struct hist_entry *right) +{ + return right->cpu - left->cpu; +} + +static int hist_entry__cpu_snprintf(struct hist_entry *self, char *bf, + size_t size, unsigned int width) +{ + return repsep_snprintf(bf, size, "%-*d", width, self->cpu); +} + +int sort_dimension__add(const char *tok) +{ + unsigned int i; + + for (i = 0; i < ARRAY_SIZE(sort_dimensions); i++) { + struct sort_dimension *sd = &sort_dimensions[i]; + + if (sd->taken) + continue; + + if (strncasecmp(tok, sd->name, strlen(tok))) + continue; + + if (sd->entry->se_collapse) + sort__need_collapse = 1; + + if (sd->entry == &sort_parent) { + int ret = regcomp(&parent_regex, parent_pattern, REG_EXTENDED); + if (ret) { + char err[BUFSIZ]; + + regerror(ret, &parent_regex, err, sizeof(err)); + pr_err("Invalid regex: %s\n%s", parent_pattern, err); + return -EINVAL; + } + sort__has_parent = 1; + } + + if (list_empty(&hist_entry__sort_list)) { + if (!strcmp(sd->name, "pid")) + sort__first_dimension = SORT_PID; + else if (!strcmp(sd->name, "comm")) + sort__first_dimension = SORT_COMM; + else if (!strcmp(sd->name, "dso")) + sort__first_dimension = SORT_DSO; + else if (!strcmp(sd->name, "symbol")) + sort__first_dimension = SORT_SYM; + else if (!strcmp(sd->name, "parent")) + sort__first_dimension = SORT_PARENT; + else if (!strcmp(sd->name, "cpu")) + sort__first_dimension = SORT_CPU; + } + + list_add_tail(&sd->entry->list, &hist_entry__sort_list); + sd->taken = 1; + + return 0; + } + + return -ESRCH; +} + +void setup_sorting(const char * const usagestr[], const struct option *opts) +{ + char *tmp, *tok, *str = strdup(sort_order); + + for (tok = strtok_r(str, ", ", &tmp); + tok; tok = strtok_r(NULL, ", ", &tmp)) { + if (sort_dimension__add(tok) < 0) { + error("Unknown --sort key: `%s'", tok); + usage_with_options(usagestr, opts); + } + } + + free(str); +} + +void sort_entry__setup_elide(struct sort_entry *self, struct strlist *list, + const char *list_name, FILE *fp) +{ + if (list && strlist__nr_entries(list) == 1) { + if (fp != NULL) + fprintf(fp, "# %s: %s\n", list_name, + strlist__entry(list, 0)->s); + self->elide = true; + } +} diff --git a/tools/perf/util/sort.h b/tools/perf/util/sort.h new file mode 100644 index 00000000..0b91053a --- /dev/null +++ b/tools/perf/util/sort.h @@ -0,0 +1,124 @@ +#ifndef __PERF_SORT_H +#define __PERF_SORT_H +#include "../builtin.h" + +#include "util.h" + +#include "color.h" +#include <linux/list.h> +#include "cache.h" +#include <linux/rbtree.h> +#include "symbol.h" +#include "string.h" +#include "callchain.h" +#include "strlist.h" +#include "values.h" + +#include "../perf.h" +#include "debug.h" +#include "header.h" + +#include "parse-options.h" +#include "parse-events.h" + +#include "thread.h" +#include "sort.h" + +extern regex_t parent_regex; +extern const char *sort_order; +extern const char default_parent_pattern[]; +extern const char *parent_pattern; +extern const char default_sort_order[]; +extern int sort__need_collapse; +extern int sort__has_parent; +extern char *field_sep; +extern struct sort_entry sort_comm; +extern struct sort_entry sort_dso; +extern struct sort_entry sort_sym; +extern struct sort_entry sort_parent; +extern enum sort_type sort__first_dimension; + +/** + * struct hist_entry - histogram entry + * + * @row_offset - offset from the first callchain expanded to appear on screen + * @nr_rows - rows expanded in callchain, recalculated on folding/unfolding + */ +struct hist_entry { + struct rb_node rb_node; + u64 period; + u64 period_sys; + u64 period_us; + u64 period_guest_sys; + u64 period_guest_us; + struct map_symbol ms; + struct thread *thread; + u64 ip; + s32 cpu; + u32 nr_events; + + /* XXX These two should move to some tree widget lib */ + u16 row_offset; + u16 nr_rows; + + bool init_have_children; + char level; + u8 filtered; + struct symbol *parent; + union { + unsigned long position; + struct hist_entry *pair; + struct rb_root sorted_chain; + }; + struct callchain_root callchain[0]; +}; + +enum sort_type { + SORT_PID, + SORT_COMM, + SORT_DSO, + SORT_SYM, + SORT_PARENT, + SORT_CPU, +}; + +/* + * configurable sorting bits + */ + +struct sort_entry { + struct list_head list; + + const char *se_header; + + int64_t (*se_cmp)(struct hist_entry *, struct hist_entry *); + int64_t (*se_collapse)(struct hist_entry *, struct hist_entry *); + int (*se_snprintf)(struct hist_entry *self, char *bf, size_t size, + unsigned int width); + u8 se_width_idx; + bool elide; +}; + +extern struct sort_entry sort_thread; +extern struct list_head hist_entry__sort_list; + +void setup_sorting(const char * const usagestr[], const struct option *opts); + +extern size_t sort__thread_print(FILE *, struct hist_entry *, unsigned int); +extern size_t sort__comm_print(FILE *, struct hist_entry *, unsigned int); +extern size_t sort__dso_print(FILE *, struct hist_entry *, unsigned int); +extern size_t sort__sym_print(FILE *, struct hist_entry *, unsigned int __used); +extern int64_t cmp_null(void *, void *); +extern int64_t sort__thread_cmp(struct hist_entry *, struct hist_entry *); +extern int64_t sort__comm_cmp(struct hist_entry *, struct hist_entry *); +extern int64_t sort__comm_collapse(struct hist_entry *, struct hist_entry *); +extern int64_t sort__dso_cmp(struct hist_entry *, struct hist_entry *); +extern int64_t sort__sym_cmp(struct hist_entry *, struct hist_entry *); +extern int64_t sort__parent_cmp(struct hist_entry *, struct hist_entry *); +int64_t sort__cpu_cmp(struct hist_entry *left, struct hist_entry *right); +extern size_t sort__parent_print(FILE *, struct hist_entry *, unsigned int); +extern int sort_dimension__add(const char *); +void sort_entry__setup_elide(struct sort_entry *self, struct strlist *list, + const char *list_name, FILE *fp); + +#endif /* __PERF_SORT_H */ diff --git a/tools/perf/util/strbuf.c b/tools/perf/util/strbuf.c new file mode 100644 index 00000000..92e06851 --- /dev/null +++ b/tools/perf/util/strbuf.c @@ -0,0 +1,133 @@ +#include "cache.h" + +int prefixcmp(const char *str, const char *prefix) +{ + for (; ; str++, prefix++) + if (!*prefix) + return 0; + else if (*str != *prefix) + return (unsigned char)*prefix - (unsigned char)*str; +} + +/* + * Used as the default ->buf value, so that people can always assume + * buf is non NULL and ->buf is NUL terminated even for a freshly + * initialized strbuf. + */ +char strbuf_slopbuf[1]; + +void strbuf_init(struct strbuf *sb, ssize_t hint) +{ + sb->alloc = sb->len = 0; + sb->buf = strbuf_slopbuf; + if (hint) + strbuf_grow(sb, hint); +} + +void strbuf_release(struct strbuf *sb) +{ + if (sb->alloc) { + free(sb->buf); + strbuf_init(sb, 0); + } +} + +char *strbuf_detach(struct strbuf *sb, size_t *sz) +{ + char *res = sb->alloc ? sb->buf : NULL; + if (sz) + *sz = sb->len; + strbuf_init(sb, 0); + return res; +} + +void strbuf_grow(struct strbuf *sb, size_t extra) +{ + if (sb->len + extra + 1 <= sb->len) + die("you want to use way too much memory"); + if (!sb->alloc) + sb->buf = NULL; + ALLOC_GROW(sb->buf, sb->len + extra + 1, sb->alloc); +} + +static void strbuf_splice(struct strbuf *sb, size_t pos, size_t len, + const void *data, size_t dlen) +{ + if (pos + len < pos) + die("you want to use way too much memory"); + if (pos > sb->len) + die("`pos' is too far after the end of the buffer"); + if (pos + len > sb->len) + die("`pos + len' is too far after the end of the buffer"); + + if (dlen >= len) + strbuf_grow(sb, dlen - len); + memmove(sb->buf + pos + dlen, + sb->buf + pos + len, + sb->len - pos - len); + memcpy(sb->buf + pos, data, dlen); + strbuf_setlen(sb, sb->len + dlen - len); +} + +void strbuf_remove(struct strbuf *sb, size_t pos, size_t len) +{ + strbuf_splice(sb, pos, len, NULL, 0); +} + +void strbuf_add(struct strbuf *sb, const void *data, size_t len) +{ + strbuf_grow(sb, len); + memcpy(sb->buf + sb->len, data, len); + strbuf_setlen(sb, sb->len + len); +} + +void strbuf_addf(struct strbuf *sb, const char *fmt, ...) +{ + int len; + va_list ap; + + if (!strbuf_avail(sb)) + strbuf_grow(sb, 64); + va_start(ap, fmt); + len = vsnprintf(sb->buf + sb->len, sb->alloc - sb->len, fmt, ap); + va_end(ap); + if (len < 0) + die("your vsnprintf is broken"); + if (len > strbuf_avail(sb)) { + strbuf_grow(sb, len); + va_start(ap, fmt); + len = vsnprintf(sb->buf + sb->len, sb->alloc - sb->len, fmt, ap); + va_end(ap); + if (len > strbuf_avail(sb)) { + die("this should not happen, your snprintf is broken"); + } + } + strbuf_setlen(sb, sb->len + len); +} + +ssize_t strbuf_read(struct strbuf *sb, int fd, ssize_t hint) +{ + size_t oldlen = sb->len; + size_t oldalloc = sb->alloc; + + strbuf_grow(sb, hint ? hint : 8192); + for (;;) { + ssize_t cnt; + + cnt = read(fd, sb->buf + sb->len, sb->alloc - sb->len - 1); + if (cnt < 0) { + if (oldalloc == 0) + strbuf_release(sb); + else + strbuf_setlen(sb, oldlen); + return -1; + } + if (!cnt) + break; + sb->len += cnt; + strbuf_grow(sb, 8192); + } + + sb->buf[sb->len] = '\0'; + return sb->len - oldlen; +} diff --git a/tools/perf/util/strbuf.h b/tools/perf/util/strbuf.h new file mode 100644 index 00000000..436ac319 --- /dev/null +++ b/tools/perf/util/strbuf.h @@ -0,0 +1,92 @@ +#ifndef __PERF_STRBUF_H +#define __PERF_STRBUF_H + +/* + * Strbuf's can be use in many ways: as a byte array, or to store arbitrary + * long, overflow safe strings. + * + * Strbufs has some invariants that are very important to keep in mind: + * + * 1. the ->buf member is always malloc-ed, hence strbuf's can be used to + * build complex strings/buffers whose final size isn't easily known. + * + * It is NOT legal to copy the ->buf pointer away. + * `strbuf_detach' is the operation that detachs a buffer from its shell + * while keeping the shell valid wrt its invariants. + * + * 2. the ->buf member is a byte array that has at least ->len + 1 bytes + * allocated. The extra byte is used to store a '\0', allowing the ->buf + * member to be a valid C-string. Every strbuf function ensure this + * invariant is preserved. + * + * Note that it is OK to "play" with the buffer directly if you work it + * that way: + * + * strbuf_grow(sb, SOME_SIZE); + * ... Here, the memory array starting at sb->buf, and of length + * ... strbuf_avail(sb) is all yours, and you are sure that + * ... strbuf_avail(sb) is at least SOME_SIZE. + * strbuf_setlen(sb, sb->len + SOME_OTHER_SIZE); + * + * Of course, SOME_OTHER_SIZE must be smaller or equal to strbuf_avail(sb). + * + * Doing so is safe, though if it has to be done in many places, adding the + * missing API to the strbuf module is the way to go. + * + * XXX: do _not_ assume that the area that is yours is of size ->alloc - 1 + * even if it's true in the current implementation. Alloc is somehow a + * "private" member that should not be messed with. + */ + +#include <assert.h> + +extern char strbuf_slopbuf[]; +struct strbuf { + size_t alloc; + size_t len; + char *buf; +}; + +#define STRBUF_INIT { 0, 0, strbuf_slopbuf } + +/*----- strbuf life cycle -----*/ +extern void strbuf_init(struct strbuf *buf, ssize_t hint); +extern void strbuf_release(struct strbuf *); +extern char *strbuf_detach(struct strbuf *, size_t *); + +/*----- strbuf size related -----*/ +static inline ssize_t strbuf_avail(const struct strbuf *sb) { + return sb->alloc ? sb->alloc - sb->len - 1 : 0; +} + +extern void strbuf_grow(struct strbuf *, size_t); + +static inline void strbuf_setlen(struct strbuf *sb, size_t len) { + if (!sb->alloc) + strbuf_grow(sb, 0); + assert(len < sb->alloc); + sb->len = len; + sb->buf[len] = '\0'; +} + +/*----- add data in your buffer -----*/ +static inline void strbuf_addch(struct strbuf *sb, int c) { + strbuf_grow(sb, 1); + sb->buf[sb->len++] = c; + sb->buf[sb->len] = '\0'; +} + +extern void strbuf_remove(struct strbuf *, size_t pos, size_t len); + +extern void strbuf_add(struct strbuf *, const void *, size_t); +static inline void strbuf_addstr(struct strbuf *sb, const char *s) { + strbuf_add(sb, s, strlen(s)); +} + +__attribute__((format(printf,2,3))) +extern void strbuf_addf(struct strbuf *sb, const char *fmt, ...); + +/* XXX: if read fails, any partial read is undone */ +extern ssize_t strbuf_read(struct strbuf *, int fd, ssize_t hint); + +#endif /* __PERF_STRBUF_H */ diff --git a/tools/perf/util/strfilter.c b/tools/perf/util/strfilter.c new file mode 100644 index 00000000..834c8ebf --- /dev/null +++ b/tools/perf/util/strfilter.c @@ -0,0 +1,199 @@ +#include "util.h" +#include "string.h" +#include "strfilter.h" + +/* Operators */ +static const char *OP_and = "&"; /* Logical AND */ +static const char *OP_or = "|"; /* Logical OR */ +static const char *OP_not = "!"; /* Logical NOT */ + +#define is_operator(c) ((c) == '|' || (c) == '&' || (c) == '!') +#define is_separator(c) (is_operator(c) || (c) == '(' || (c) == ')') + +static void strfilter_node__delete(struct strfilter_node *self) +{ + if (self) { + if (self->p && !is_operator(*self->p)) + free((char *)self->p); + strfilter_node__delete(self->l); + strfilter_node__delete(self->r); + free(self); + } +} + +void strfilter__delete(struct strfilter *self) +{ + if (self) { + strfilter_node__delete(self->root); + free(self); + } +} + +static const char *get_token(const char *s, const char **e) +{ + const char *p; + + while (isspace(*s)) /* Skip spaces */ + s++; + + if (*s == '\0') { + p = s; + goto end; + } + + p = s + 1; + if (!is_separator(*s)) { + /* End search */ +retry: + while (*p && !is_separator(*p) && !isspace(*p)) + p++; + /* Escape and special case: '!' is also used in glob pattern */ + if (*(p - 1) == '\\' || (*p == '!' && *(p - 1) == '[')) { + p++; + goto retry; + } + } +end: + *e = p; + return s; +} + +static struct strfilter_node *strfilter_node__alloc(const char *op, + struct strfilter_node *l, + struct strfilter_node *r) +{ + struct strfilter_node *ret = zalloc(sizeof(struct strfilter_node)); + + if (ret) { + ret->p = op; + ret->l = l; + ret->r = r; + } + + return ret; +} + +static struct strfilter_node *strfilter_node__new(const char *s, + const char **ep) +{ + struct strfilter_node root, *cur, *last_op; + const char *e; + + if (!s) + return NULL; + + memset(&root, 0, sizeof(root)); + last_op = cur = &root; + + s = get_token(s, &e); + while (*s != '\0' && *s != ')') { + switch (*s) { + case '&': /* Exchg last OP->r with AND */ + if (!cur->r || !last_op->r) + goto error; + cur = strfilter_node__alloc(OP_and, last_op->r, NULL); + if (!cur) + goto nomem; + last_op->r = cur; + last_op = cur; + break; + case '|': /* Exchg the root with OR */ + if (!cur->r || !root.r) + goto error; + cur = strfilter_node__alloc(OP_or, root.r, NULL); + if (!cur) + goto nomem; + root.r = cur; + last_op = cur; + break; + case '!': /* Add NOT as a leaf node */ + if (cur->r) + goto error; + cur->r = strfilter_node__alloc(OP_not, NULL, NULL); + if (!cur->r) + goto nomem; + cur = cur->r; + break; + case '(': /* Recursively parses inside the parenthesis */ + if (cur->r) + goto error; + cur->r = strfilter_node__new(s + 1, &s); + if (!s) + goto nomem; + if (!cur->r || *s != ')') + goto error; + e = s + 1; + break; + default: + if (cur->r) + goto error; + cur->r = strfilter_node__alloc(NULL, NULL, NULL); + if (!cur->r) + goto nomem; + cur->r->p = strndup(s, e - s); + if (!cur->r->p) + goto nomem; + } + s = get_token(e, &e); + } + if (!cur->r) + goto error; + *ep = s; + return root.r; +nomem: + s = NULL; +error: + *ep = s; + strfilter_node__delete(root.r); + return NULL; +} + +/* + * Parse filter rule and return new strfilter. + * Return NULL if fail, and *ep == NULL if memory allocation failed. + */ +struct strfilter *strfilter__new(const char *rules, const char **err) +{ + struct strfilter *ret = zalloc(sizeof(struct strfilter)); + const char *ep = NULL; + + if (ret) + ret->root = strfilter_node__new(rules, &ep); + + if (!ret || !ret->root || *ep != '\0') { + if (err) + *err = ep; + strfilter__delete(ret); + ret = NULL; + } + + return ret; +} + +static bool strfilter_node__compare(struct strfilter_node *self, + const char *str) +{ + if (!self || !self->p) + return false; + + switch (*self->p) { + case '|': /* OR */ + return strfilter_node__compare(self->l, str) || + strfilter_node__compare(self->r, str); + case '&': /* AND */ + return strfilter_node__compare(self->l, str) && + strfilter_node__compare(self->r, str); + case '!': /* NOT */ + return !strfilter_node__compare(self->r, str); + default: + return strglobmatch(str, self->p); + } +} + +/* Return true if STR matches the filter rules */ +bool strfilter__compare(struct strfilter *self, const char *str) +{ + if (!self) + return false; + return strfilter_node__compare(self->root, str); +} diff --git a/tools/perf/util/strfilter.h b/tools/perf/util/strfilter.h new file mode 100644 index 00000000..00f58a75 --- /dev/null +++ b/tools/perf/util/strfilter.h @@ -0,0 +1,48 @@ +#ifndef __PERF_STRFILTER_H +#define __PERF_STRFILTER_H +/* General purpose glob matching filter */ + +#include <linux/list.h> +#include <stdbool.h> + +/* A node of string filter */ +struct strfilter_node { + struct strfilter_node *l; /* Tree left branche (for &,|) */ + struct strfilter_node *r; /* Tree right branche (for !,&,|) */ + const char *p; /* Operator or rule */ +}; + +/* String filter */ +struct strfilter { + struct strfilter_node *root; +}; + +/** + * strfilter__new - Create a new string filter + * @rules: Filter rule, which is a combination of glob expressions. + * @err: Pointer which points an error detected on @rules + * + * Parse @rules and return new strfilter. Return NULL if an error detected. + * In that case, *@err will indicate where it is detected, and *@err is NULL + * if a memory allocation is failed. + */ +struct strfilter *strfilter__new(const char *rules, const char **err); + +/** + * strfilter__compare - compare given string and a string filter + * @self: String filter + * @str: target string + * + * Compare @str and @self. Return true if the str match the rule + */ +bool strfilter__compare(struct strfilter *self, const char *str); + +/** + * strfilter__delete - delete a string filter + * @self: String filter to delete + * + * Delete @self. + */ +void strfilter__delete(struct strfilter *self); + +#endif diff --git a/tools/perf/util/string.c b/tools/perf/util/string.c new file mode 100644 index 00000000..b9a985da --- /dev/null +++ b/tools/perf/util/string.c @@ -0,0 +1,296 @@ +#include "util.h" +#include "string.h" + +#define K 1024LL +/* + * perf_atoll() + * Parse (\d+)(b|B|kb|KB|mb|MB|gb|GB|tb|TB) (e.g. "256MB") + * and return its numeric value + */ +s64 perf_atoll(const char *str) +{ + unsigned int i; + s64 length = -1, unit = 1; + + if (!isdigit(str[0])) + goto out_err; + + for (i = 1; i < strlen(str); i++) { + switch (str[i]) { + case 'B': + case 'b': + break; + case 'K': + if (str[i + 1] != 'B') + goto out_err; + else + goto kilo; + case 'k': + if (str[i + 1] != 'b') + goto out_err; +kilo: + unit = K; + break; + case 'M': + if (str[i + 1] != 'B') + goto out_err; + else + goto mega; + case 'm': + if (str[i + 1] != 'b') + goto out_err; +mega: + unit = K * K; + break; + case 'G': + if (str[i + 1] != 'B') + goto out_err; + else + goto giga; + case 'g': + if (str[i + 1] != 'b') + goto out_err; +giga: + unit = K * K * K; + break; + case 'T': + if (str[i + 1] != 'B') + goto out_err; + else + goto tera; + case 't': + if (str[i + 1] != 'b') + goto out_err; +tera: + unit = K * K * K * K; + break; + case '\0': /* only specified figures */ + unit = 1; + break; + default: + if (!isdigit(str[i])) + goto out_err; + break; + } + } + + length = atoll(str) * unit; + goto out; + +out_err: + length = -1; +out: + return length; +} + +/* + * Helper function for splitting a string into an argv-like array. + * originally copied from lib/argv_split.c + */ +static const char *skip_sep(const char *cp) +{ + while (*cp && isspace(*cp)) + cp++; + + return cp; +} + +static const char *skip_arg(const char *cp) +{ + while (*cp && !isspace(*cp)) + cp++; + + return cp; +} + +static int count_argc(const char *str) +{ + int count = 0; + + while (*str) { + str = skip_sep(str); + if (*str) { + count++; + str = skip_arg(str); + } + } + + return count; +} + +/** + * argv_free - free an argv + * @argv - the argument vector to be freed + * + * Frees an argv and the strings it points to. + */ +void argv_free(char **argv) +{ + char **p; + for (p = argv; *p; p++) + free(*p); + + free(argv); +} + +/** + * argv_split - split a string at whitespace, returning an argv + * @str: the string to be split + * @argcp: returned argument count + * + * Returns an array of pointers to strings which are split out from + * @str. This is performed by strictly splitting on white-space; no + * quote processing is performed. Multiple whitespace characters are + * considered to be a single argument separator. The returned array + * is always NULL-terminated. Returns NULL on memory allocation + * failure. + */ +char **argv_split(const char *str, int *argcp) +{ + int argc = count_argc(str); + char **argv = zalloc(sizeof(*argv) * (argc+1)); + char **argvp; + + if (argv == NULL) + goto out; + + if (argcp) + *argcp = argc; + + argvp = argv; + + while (*str) { + str = skip_sep(str); + + if (*str) { + const char *p = str; + char *t; + + str = skip_arg(str); + + t = strndup(p, str-p); + if (t == NULL) + goto fail; + *argvp++ = t; + } + } + *argvp = NULL; + +out: + return argv; + +fail: + argv_free(argv); + return NULL; +} + +/* Character class matching */ +static bool __match_charclass(const char *pat, char c, const char **npat) +{ + bool complement = false, ret = true; + + if (*pat == '!') { + complement = true; + pat++; + } + if (*pat++ == c) /* First character is special */ + goto end; + + while (*pat && *pat != ']') { /* Matching */ + if (*pat == '-' && *(pat + 1) != ']') { /* Range */ + if (*(pat - 1) <= c && c <= *(pat + 1)) + goto end; + if (*(pat - 1) > *(pat + 1)) + goto error; + pat += 2; + } else if (*pat++ == c) + goto end; + } + if (!*pat) + goto error; + ret = false; + +end: + while (*pat && *pat != ']') /* Searching closing */ + pat++; + if (!*pat) + goto error; + *npat = pat + 1; + return complement ? !ret : ret; + +error: + return false; +} + +/* Glob/lazy pattern matching */ +static bool __match_glob(const char *str, const char *pat, bool ignore_space) +{ + while (*str && *pat && *pat != '*') { + if (ignore_space) { + /* Ignore spaces for lazy matching */ + if (isspace(*str)) { + str++; + continue; + } + if (isspace(*pat)) { + pat++; + continue; + } + } + if (*pat == '?') { /* Matches any single character */ + str++; + pat++; + continue; + } else if (*pat == '[') /* Character classes/Ranges */ + if (__match_charclass(pat + 1, *str, &pat)) { + str++; + continue; + } else + return false; + else if (*pat == '\\') /* Escaped char match as normal char */ + pat++; + if (*str++ != *pat++) + return false; + } + /* Check wild card */ + if (*pat == '*') { + while (*pat == '*') + pat++; + if (!*pat) /* Tail wild card matches all */ + return true; + while (*str) + if (__match_glob(str++, pat, ignore_space)) + return true; + } + return !*str && !*pat; +} + +/** + * strglobmatch - glob expression pattern matching + * @str: the target string to match + * @pat: the pattern string to match + * + * This returns true if the @str matches @pat. @pat can includes wildcards + * ('*','?') and character classes ([CHARS], complementation and ranges are + * also supported). Also, this supports escape character ('\') to use special + * characters as normal character. + * + * Note: if @pat syntax is broken, this always returns false. + */ +bool strglobmatch(const char *str, const char *pat) +{ + return __match_glob(str, pat, false); +} + +/** + * strlazymatch - matching pattern strings lazily with glob pattern + * @str: the target string to match + * @pat: the pattern string to match + * + * This is similar to strglobmatch, except this ignores spaces in + * the target string. + */ +bool strlazymatch(const char *str, const char *pat) +{ + return __match_glob(str, pat, true); +} diff --git a/tools/perf/util/strlist.c b/tools/perf/util/strlist.c new file mode 100644 index 00000000..6783a204 --- /dev/null +++ b/tools/perf/util/strlist.c @@ -0,0 +1,200 @@ +/* + * (c) 2009 Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Licensed under the GPLv2. + */ + +#include "strlist.h" +#include <errno.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +static struct str_node *str_node__new(const char *s, bool dupstr) +{ + struct str_node *self = malloc(sizeof(*self)); + + if (self != NULL) { + if (dupstr) { + s = strdup(s); + if (s == NULL) + goto out_delete; + } + self->s = s; + } + + return self; + +out_delete: + free(self); + return NULL; +} + +static void str_node__delete(struct str_node *self, bool dupstr) +{ + if (dupstr) + free((void *)self->s); + free(self); +} + +int strlist__add(struct strlist *self, const char *new_entry) +{ + struct rb_node **p = &self->entries.rb_node; + struct rb_node *parent = NULL; + struct str_node *sn; + + while (*p != NULL) { + int rc; + + parent = *p; + sn = rb_entry(parent, struct str_node, rb_node); + rc = strcmp(sn->s, new_entry); + + if (rc > 0) + p = &(*p)->rb_left; + else if (rc < 0) + p = &(*p)->rb_right; + else + return -EEXIST; + } + + sn = str_node__new(new_entry, self->dupstr); + if (sn == NULL) + return -ENOMEM; + + rb_link_node(&sn->rb_node, parent, p); + rb_insert_color(&sn->rb_node, &self->entries); + ++self->nr_entries; + + return 0; +} + +int strlist__load(struct strlist *self, const char *filename) +{ + char entry[1024]; + int err; + FILE *fp = fopen(filename, "r"); + + if (fp == NULL) + return errno; + + while (fgets(entry, sizeof(entry), fp) != NULL) { + const size_t len = strlen(entry); + + if (len == 0) + continue; + entry[len - 1] = '\0'; + + err = strlist__add(self, entry); + if (err != 0) + goto out; + } + + err = 0; +out: + fclose(fp); + return err; +} + +void strlist__remove(struct strlist *self, struct str_node *sn) +{ + rb_erase(&sn->rb_node, &self->entries); + str_node__delete(sn, self->dupstr); +} + +struct str_node *strlist__find(struct strlist *self, const char *entry) +{ + struct rb_node **p = &self->entries.rb_node; + struct rb_node *parent = NULL; + + while (*p != NULL) { + struct str_node *sn; + int rc; + + parent = *p; + sn = rb_entry(parent, struct str_node, rb_node); + rc = strcmp(sn->s, entry); + + if (rc > 0) + p = &(*p)->rb_left; + else if (rc < 0) + p = &(*p)->rb_right; + else + return sn; + } + + return NULL; +} + +static int strlist__parse_list_entry(struct strlist *self, const char *s) +{ + if (strncmp(s, "file://", 7) == 0) + return strlist__load(self, s + 7); + + return strlist__add(self, s); +} + +int strlist__parse_list(struct strlist *self, const char *s) +{ + char *sep; + int err; + + while ((sep = strchr(s, ',')) != NULL) { + *sep = '\0'; + err = strlist__parse_list_entry(self, s); + *sep = ','; + if (err != 0) + return err; + s = sep + 1; + } + + return *s ? strlist__parse_list_entry(self, s) : 0; +} + +struct strlist *strlist__new(bool dupstr, const char *slist) +{ + struct strlist *self = malloc(sizeof(*self)); + + if (self != NULL) { + self->entries = RB_ROOT; + self->dupstr = dupstr; + self->nr_entries = 0; + if (slist && strlist__parse_list(self, slist) != 0) + goto out_error; + } + + return self; +out_error: + free(self); + return NULL; +} + +void strlist__delete(struct strlist *self) +{ + if (self != NULL) { + struct str_node *pos; + struct rb_node *next = rb_first(&self->entries); + + while (next) { + pos = rb_entry(next, struct str_node, rb_node); + next = rb_next(&pos->rb_node); + strlist__remove(self, pos); + } + self->entries = RB_ROOT; + free(self); + } +} + +struct str_node *strlist__entry(const struct strlist *self, unsigned int idx) +{ + struct rb_node *nd; + + for (nd = rb_first(&self->entries); nd; nd = rb_next(nd)) { + struct str_node *pos = rb_entry(nd, struct str_node, rb_node); + + if (!idx--) + return pos; + } + + return NULL; +} diff --git a/tools/perf/util/strlist.h b/tools/perf/util/strlist.h new file mode 100644 index 00000000..3ba83900 --- /dev/null +++ b/tools/perf/util/strlist.h @@ -0,0 +1,78 @@ +#ifndef __PERF_STRLIST_H +#define __PERF_STRLIST_H + +#include <linux/rbtree.h> +#include <stdbool.h> + +struct str_node { + struct rb_node rb_node; + const char *s; +}; + +struct strlist { + struct rb_root entries; + unsigned int nr_entries; + bool dupstr; +}; + +struct strlist *strlist__new(bool dupstr, const char *slist); +void strlist__delete(struct strlist *self); + +void strlist__remove(struct strlist *self, struct str_node *sn); +int strlist__load(struct strlist *self, const char *filename); +int strlist__add(struct strlist *self, const char *str); + +struct str_node *strlist__entry(const struct strlist *self, unsigned int idx); +struct str_node *strlist__find(struct strlist *self, const char *entry); + +static inline bool strlist__has_entry(struct strlist *self, const char *entry) +{ + return strlist__find(self, entry) != NULL; +} + +static inline bool strlist__empty(const struct strlist *self) +{ + return self->nr_entries == 0; +} + +static inline unsigned int strlist__nr_entries(const struct strlist *self) +{ + return self->nr_entries; +} + +/* For strlist iteration */ +static inline struct str_node *strlist__first(struct strlist *self) +{ + struct rb_node *rn = rb_first(&self->entries); + return rn ? rb_entry(rn, struct str_node, rb_node) : NULL; +} +static inline struct str_node *strlist__next(struct str_node *sn) +{ + struct rb_node *rn; + if (!sn) + return NULL; + rn = rb_next(&sn->rb_node); + return rn ? rb_entry(rn, struct str_node, rb_node) : NULL; +} + +/** + * strlist_for_each - iterate over a strlist + * @pos: the &struct str_node to use as a loop cursor. + * @self: the &struct strlist for loop. + */ +#define strlist__for_each(pos, self) \ + for (pos = strlist__first(self); pos; pos = strlist__next(pos)) + +/** + * strlist_for_each_safe - iterate over a strlist safe against removal of + * str_node + * @pos: the &struct str_node to use as a loop cursor. + * @n: another &struct str_node to use as temporary storage. + * @self: the &struct strlist for loop. + */ +#define strlist__for_each_safe(pos, n, self) \ + for (pos = strlist__first(self), n = strlist__next(pos); pos;\ + pos = n, n = strlist__next(n)) + +int strlist__parse_list(struct strlist *self, const char *s); +#endif /* __PERF_STRLIST_H */ diff --git a/tools/perf/util/svghelper.c b/tools/perf/util/svghelper.c new file mode 100644 index 00000000..96c86604 --- /dev/null +++ b/tools/perf/util/svghelper.c @@ -0,0 +1,501 @@ +/* + * svghelper.c - helper functions for outputting svg + * + * (C) Copyright 2009 Intel Corporation + * + * Authors: + * Arjan van de Ven <arjan@linux.intel.com> + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; version 2 + * of the License. + */ + +#include <inttypes.h> +#include <stdio.h> +#include <stdlib.h> +#include <unistd.h> +#include <string.h> + +#include "svghelper.h" + +static u64 first_time, last_time; +static u64 turbo_frequency, max_freq; + + +#define SLOT_MULT 30.0 +#define SLOT_HEIGHT 25.0 + +int svg_page_width = 1000; + +#define MIN_TEXT_SIZE 0.01 + +static u64 total_height; +static FILE *svgfile; + +static double cpu2slot(int cpu) +{ + return 2 * cpu + 1; +} + +static double cpu2y(int cpu) +{ + return cpu2slot(cpu) * SLOT_MULT; +} + +static double time2pixels(u64 __time) +{ + double X; + + X = 1.0 * svg_page_width * (__time - first_time) / (last_time - first_time); + return X; +} + +/* + * Round text sizes so that the svg viewer only needs a discrete + * number of renderings of the font + */ +static double round_text_size(double size) +{ + int loop = 100; + double target = 10.0; + + if (size >= 10.0) + return size; + while (loop--) { + if (size >= target) + return target; + target = target / 2.0; + } + return size; +} + +void open_svg(const char *filename, int cpus, int rows, u64 start, u64 end) +{ + int new_width; + + svgfile = fopen(filename, "w"); + if (!svgfile) { + fprintf(stderr, "Cannot open %s for output\n", filename); + return; + } + first_time = start; + first_time = first_time / 100000000 * 100000000; + last_time = end; + + /* + * if the recording is short, we default to a width of 1000, but + * for longer recordings we want at least 200 units of width per second + */ + new_width = (last_time - first_time) / 5000000; + + if (new_width > svg_page_width) + svg_page_width = new_width; + + total_height = (1 + rows + cpu2slot(cpus)) * SLOT_MULT; + fprintf(svgfile, "<?xml version=\"1.0\" standalone=\"no\"?> \n"); + fprintf(svgfile, "<svg width=\"%i\" height=\"%" PRIu64 "\" version=\"1.1\" xmlns=\"http://www.w3.org/2000/svg\">\n", svg_page_width, total_height); + + fprintf(svgfile, "<defs>\n <style type=\"text/css\">\n <![CDATA[\n"); + + fprintf(svgfile, " rect { stroke-width: 1; }\n"); + fprintf(svgfile, " rect.process { fill:rgb(180,180,180); fill-opacity:0.9; stroke-width:1; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.process2 { fill:rgb(180,180,180); fill-opacity:0.9; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.sample { fill:rgb( 0, 0,255); fill-opacity:0.8; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.blocked { fill:rgb(255, 0, 0); fill-opacity:0.5; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.waiting { fill:rgb(224,214, 0); fill-opacity:0.8; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.WAITING { fill:rgb(255,214, 48); fill-opacity:0.6; stroke-width:0; stroke:rgb( 0, 0, 0); } \n"); + fprintf(svgfile, " rect.cpu { fill:rgb(192,192,192); fill-opacity:0.2; stroke-width:0.5; stroke:rgb(128,128,128); } \n"); + fprintf(svgfile, " rect.pstate { fill:rgb(128,128,128); fill-opacity:0.8; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c1 { fill:rgb(255,214,214); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c2 { fill:rgb(255,172,172); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c3 { fill:rgb(255,130,130); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c4 { fill:rgb(255, 88, 88); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c5 { fill:rgb(255, 44, 44); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " rect.c6 { fill:rgb(255, 0, 0); fill-opacity:0.5; stroke-width:0; } \n"); + fprintf(svgfile, " line.pstate { stroke:rgb(255,255, 0); stroke-opacity:0.8; stroke-width:2; } \n"); + + fprintf(svgfile, " ]]>\n </style>\n</defs>\n"); +} + +void svg_box(int Yslot, u64 start, u64 end, const char *type) +{ + if (!svgfile) + return; + + fprintf(svgfile, "<rect x=\"%4.8f\" width=\"%4.8f\" y=\"%4.1f\" height=\"%4.1f\" class=\"%s\"/>\n", + time2pixels(start), time2pixels(end)-time2pixels(start), Yslot * SLOT_MULT, SLOT_HEIGHT, type); +} + +void svg_sample(int Yslot, int cpu, u64 start, u64 end) +{ + double text_size; + if (!svgfile) + return; + + fprintf(svgfile, "<rect x=\"%4.8f\" width=\"%4.8f\" y=\"%4.1f\" height=\"%4.1f\" class=\"sample\"/>\n", + time2pixels(start), time2pixels(end)-time2pixels(start), Yslot * SLOT_MULT, SLOT_HEIGHT); + + text_size = (time2pixels(end)-time2pixels(start)); + if (cpu > 9) + text_size = text_size/2; + if (text_size > 1.25) + text_size = 1.25; + text_size = round_text_size(text_size); + + if (text_size > MIN_TEXT_SIZE) + fprintf(svgfile, "<text x=\"%1.8f\" y=\"%1.8f\" font-size=\"%1.8fpt\">%i</text>\n", + time2pixels(start), Yslot * SLOT_MULT + SLOT_HEIGHT - 1, text_size, cpu + 1); + +} + +static char *time_to_string(u64 duration) +{ + static char text[80]; + + text[0] = 0; + + if (duration < 1000) /* less than 1 usec */ + return text; + + if (duration < 1000 * 1000) { /* less than 1 msec */ + sprintf(text, "%4.1f us", duration / 1000.0); + return text; + } + sprintf(text, "%4.1f ms", duration / 1000.0 / 1000); + + return text; +} + +void svg_waiting(int Yslot, u64 start, u64 end) +{ + char *text; + const char *style; + double font_size; + + if (!svgfile) + return; + + style = "waiting"; + + if (end-start > 10 * 1000000) /* 10 msec */ + style = "WAITING"; + + text = time_to_string(end-start); + + font_size = 1.0 * (time2pixels(end)-time2pixels(start)); + + if (font_size > 3) + font_size = 3; + + font_size = round_text_size(font_size); + + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\">\n", time2pixels(start), Yslot * SLOT_MULT); + fprintf(svgfile, "<rect x=\"0\" width=\"%4.8f\" y=\"0\" height=\"%4.1f\" class=\"%s\"/>\n", + time2pixels(end)-time2pixels(start), SLOT_HEIGHT, style); + if (font_size > MIN_TEXT_SIZE) + fprintf(svgfile, "<text transform=\"rotate(90)\" font-size=\"%1.8fpt\"> %s</text>\n", + font_size, text); + fprintf(svgfile, "</g>\n"); +} + +static char *cpu_model(void) +{ + static char cpu_m[255]; + char buf[256]; + FILE *file; + + cpu_m[0] = 0; + /* CPU type */ + file = fopen("/proc/cpuinfo", "r"); + if (file) { + while (fgets(buf, 255, file)) { + if (strstr(buf, "model name")) { + strncpy(cpu_m, &buf[13], 255); + break; + } + } + fclose(file); + } + + /* CPU type */ + file = fopen("/sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies", "r"); + if (file) { + while (fgets(buf, 255, file)) { + unsigned int freq; + freq = strtoull(buf, NULL, 10); + if (freq > max_freq) + max_freq = freq; + } + fclose(file); + } + return cpu_m; +} + +void svg_cpu_box(int cpu, u64 __max_freq, u64 __turbo_freq) +{ + char cpu_string[80]; + if (!svgfile) + return; + + max_freq = __max_freq; + turbo_frequency = __turbo_freq; + + fprintf(svgfile, "<rect x=\"%4.8f\" width=\"%4.8f\" y=\"%4.1f\" height=\"%4.1f\" class=\"cpu\"/>\n", + time2pixels(first_time), + time2pixels(last_time)-time2pixels(first_time), + cpu2y(cpu), SLOT_MULT+SLOT_HEIGHT); + + sprintf(cpu_string, "CPU %i", (int)cpu+1); + fprintf(svgfile, "<text x=\"%4.8f\" y=\"%4.8f\">%s</text>\n", + 10+time2pixels(first_time), cpu2y(cpu) + SLOT_HEIGHT/2, cpu_string); + + fprintf(svgfile, "<text transform=\"translate(%4.8f,%4.8f)\" font-size=\"1.25pt\">%s</text>\n", + 10+time2pixels(first_time), cpu2y(cpu) + SLOT_MULT + SLOT_HEIGHT - 4, cpu_model()); +} + +void svg_process(int cpu, u64 start, u64 end, const char *type, const char *name) +{ + double width; + + if (!svgfile) + return; + + + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\">\n", time2pixels(start), cpu2y(cpu)); + fprintf(svgfile, "<rect x=\"0\" width=\"%4.8f\" y=\"0\" height=\"%4.1f\" class=\"%s\"/>\n", + time2pixels(end)-time2pixels(start), SLOT_MULT+SLOT_HEIGHT, type); + width = time2pixels(end)-time2pixels(start); + if (width > 6) + width = 6; + + width = round_text_size(width); + + if (width > MIN_TEXT_SIZE) + fprintf(svgfile, "<text transform=\"rotate(90)\" font-size=\"%3.8fpt\">%s</text>\n", + width, name); + + fprintf(svgfile, "</g>\n"); +} + +void svg_cstate(int cpu, u64 start, u64 end, int type) +{ + double width; + char style[128]; + + if (!svgfile) + return; + + + if (type > 6) + type = 6; + sprintf(style, "c%i", type); + + fprintf(svgfile, "<rect class=\"%s\" x=\"%4.8f\" width=\"%4.8f\" y=\"%4.1f\" height=\"%4.1f\"/>\n", + style, + time2pixels(start), time2pixels(end)-time2pixels(start), + cpu2y(cpu), SLOT_MULT+SLOT_HEIGHT); + + width = (time2pixels(end)-time2pixels(start))/2.0; + if (width > 6) + width = 6; + + width = round_text_size(width); + + if (width > MIN_TEXT_SIZE) + fprintf(svgfile, "<text x=\"%4.8f\" y=\"%4.8f\" font-size=\"%3.8fpt\">C%i</text>\n", + time2pixels(start), cpu2y(cpu)+width, width, type); +} + +static char *HzToHuman(unsigned long hz) +{ + static char buffer[1024]; + unsigned long long Hz; + + memset(buffer, 0, 1024); + + Hz = hz; + + /* default: just put the Number in */ + sprintf(buffer, "%9lli", Hz); + + if (Hz > 1000) + sprintf(buffer, " %6lli Mhz", (Hz+500)/1000); + + if (Hz > 1500000) + sprintf(buffer, " %6.2f Ghz", (Hz+5000.0)/1000000); + + if (Hz == turbo_frequency) + sprintf(buffer, "Turbo"); + + return buffer; +} + +void svg_pstate(int cpu, u64 start, u64 end, u64 freq) +{ + double height = 0; + + if (!svgfile) + return; + + if (max_freq) + height = freq * 1.0 / max_freq * (SLOT_HEIGHT + SLOT_MULT); + height = 1 + cpu2y(cpu) + SLOT_MULT + SLOT_HEIGHT - height; + fprintf(svgfile, "<line x1=\"%4.8f\" x2=\"%4.8f\" y1=\"%4.1f\" y2=\"%4.1f\" class=\"pstate\"/>\n", + time2pixels(start), time2pixels(end), height, height); + fprintf(svgfile, "<text x=\"%4.8f\" y=\"%4.8f\" font-size=\"0.25pt\">%s</text>\n", + time2pixels(start), height+0.9, HzToHuman(freq)); + +} + + +void svg_partial_wakeline(u64 start, int row1, char *desc1, int row2, char *desc2) +{ + double height; + + if (!svgfile) + return; + + + if (row1 < row2) { + if (row1) { + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row1 * SLOT_MULT + SLOT_HEIGHT, time2pixels(start), row1 * SLOT_MULT + SLOT_HEIGHT + SLOT_MULT/32); + if (desc2) + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\"><text transform=\"rotate(90)\" font-size=\"0.02pt\">%s ></text></g>\n", + time2pixels(start), row1 * SLOT_MULT + SLOT_HEIGHT + SLOT_HEIGHT/48, desc2); + } + if (row2) { + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row2 * SLOT_MULT - SLOT_MULT/32, time2pixels(start), row2 * SLOT_MULT); + if (desc1) + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\"><text transform=\"rotate(90)\" font-size=\"0.02pt\">%s ></text></g>\n", + time2pixels(start), row2 * SLOT_MULT - SLOT_MULT/32, desc1); + } + } else { + if (row2) { + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row2 * SLOT_MULT + SLOT_HEIGHT, time2pixels(start), row2 * SLOT_MULT + SLOT_HEIGHT + SLOT_MULT/32); + if (desc1) + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\"><text transform=\"rotate(90)\" font-size=\"0.02pt\">%s <</text></g>\n", + time2pixels(start), row2 * SLOT_MULT + SLOT_HEIGHT + SLOT_MULT/48, desc1); + } + if (row1) { + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row1 * SLOT_MULT - SLOT_MULT/32, time2pixels(start), row1 * SLOT_MULT); + if (desc2) + fprintf(svgfile, "<g transform=\"translate(%4.8f,%4.8f)\"><text transform=\"rotate(90)\" font-size=\"0.02pt\">%s <</text></g>\n", + time2pixels(start), row1 * SLOT_MULT - SLOT_HEIGHT/32, desc2); + } + } + height = row1 * SLOT_MULT; + if (row2 > row1) + height += SLOT_HEIGHT; + if (row1) + fprintf(svgfile, "<circle cx=\"%4.8f\" cy=\"%4.2f\" r = \"0.01\" style=\"fill:rgb(32,255,32)\"/>\n", + time2pixels(start), height); +} + +void svg_wakeline(u64 start, int row1, int row2) +{ + double height; + + if (!svgfile) + return; + + + if (row1 < row2) + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row1 * SLOT_MULT + SLOT_HEIGHT, time2pixels(start), row2 * SLOT_MULT); + else + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%4.2f\" style=\"stroke:rgb(32,255,32);stroke-width:0.009\"/>\n", + time2pixels(start), row2 * SLOT_MULT + SLOT_HEIGHT, time2pixels(start), row1 * SLOT_MULT); + + height = row1 * SLOT_MULT; + if (row2 > row1) + height += SLOT_HEIGHT; + fprintf(svgfile, "<circle cx=\"%4.8f\" cy=\"%4.2f\" r = \"0.01\" style=\"fill:rgb(32,255,32)\"/>\n", + time2pixels(start), height); +} + +void svg_interrupt(u64 start, int row) +{ + if (!svgfile) + return; + + fprintf(svgfile, "<circle cx=\"%4.8f\" cy=\"%4.2f\" r = \"0.01\" style=\"fill:rgb(255,128,128)\"/>\n", + time2pixels(start), row * SLOT_MULT); + fprintf(svgfile, "<circle cx=\"%4.8f\" cy=\"%4.2f\" r = \"0.01\" style=\"fill:rgb(255,128,128)\"/>\n", + time2pixels(start), row * SLOT_MULT + SLOT_HEIGHT); +} + +void svg_text(int Yslot, u64 start, const char *text) +{ + if (!svgfile) + return; + + fprintf(svgfile, "<text x=\"%4.8f\" y=\"%4.8f\">%s</text>\n", + time2pixels(start), Yslot * SLOT_MULT+SLOT_HEIGHT/2, text); +} + +static void svg_legenda_box(int X, const char *text, const char *style) +{ + double boxsize; + boxsize = SLOT_HEIGHT / 2; + + fprintf(svgfile, "<rect x=\"%i\" width=\"%4.8f\" y=\"0\" height=\"%4.1f\" class=\"%s\"/>\n", + X, boxsize, boxsize, style); + fprintf(svgfile, "<text transform=\"translate(%4.8f, %4.8f)\" font-size=\"%4.8fpt\">%s</text>\n", + X + boxsize + 5, boxsize, 0.8 * boxsize, text); +} + +void svg_legenda(void) +{ + if (!svgfile) + return; + + svg_legenda_box(0, "Running", "sample"); + svg_legenda_box(100, "Idle","c1"); + svg_legenda_box(200, "Deeper Idle", "c3"); + svg_legenda_box(350, "Deepest Idle", "c6"); + svg_legenda_box(550, "Sleeping", "process2"); + svg_legenda_box(650, "Waiting for cpu", "waiting"); + svg_legenda_box(800, "Blocked on IO", "blocked"); +} + +void svg_time_grid(void) +{ + u64 i; + + if (!svgfile) + return; + + i = first_time; + while (i < last_time) { + int color = 220; + double thickness = 0.075; + if ((i % 100000000) == 0) { + thickness = 0.5; + color = 192; + } + if ((i % 1000000000) == 0) { + thickness = 2.0; + color = 128; + } + + fprintf(svgfile, "<line x1=\"%4.8f\" y1=\"%4.2f\" x2=\"%4.8f\" y2=\"%" PRIu64 "\" style=\"stroke:rgb(%i,%i,%i);stroke-width:%1.3f\"/>\n", + time2pixels(i), SLOT_MULT/2, time2pixels(i), total_height, color, color, color, thickness); + + i += 10000000; + } +} + +void svg_close(void) +{ + if (svgfile) { + fprintf(svgfile, "</svg>\n"); + fclose(svgfile); + svgfile = NULL; + } +} diff --git a/tools/perf/util/svghelper.h b/tools/perf/util/svghelper.h new file mode 100644 index 00000000..e0781989 --- /dev/null +++ b/tools/perf/util/svghelper.h @@ -0,0 +1,28 @@ +#ifndef __PERF_SVGHELPER_H +#define __PERF_SVGHELPER_H + +#include "types.h" + +extern void open_svg(const char *filename, int cpus, int rows, u64 start, u64 end); +extern void svg_box(int Yslot, u64 start, u64 end, const char *type); +extern void svg_sample(int Yslot, int cpu, u64 start, u64 end); +extern void svg_waiting(int Yslot, u64 start, u64 end); +extern void svg_cpu_box(int cpu, u64 max_frequency, u64 turbo_frequency); + + +extern void svg_process(int cpu, u64 start, u64 end, const char *type, const char *name); +extern void svg_cstate(int cpu, u64 start, u64 end, int type); +extern void svg_pstate(int cpu, u64 start, u64 end, u64 freq); + + +extern void svg_time_grid(void); +extern void svg_legenda(void); +extern void svg_wakeline(u64 start, int row1, int row2); +extern void svg_partial_wakeline(u64 start, int row1, char *desc1, int row2, char *desc2); +extern void svg_interrupt(u64 start, int row); +extern void svg_text(int Yslot, u64 start, const char *text); +extern void svg_close(void); + +extern int svg_page_width; + +#endif /* __PERF_SVGHELPER_H */ diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c new file mode 100644 index 00000000..40fd1c7c --- /dev/null +++ b/tools/perf/util/symbol.c @@ -0,0 +1,2669 @@ +#define _GNU_SOURCE +#include <ctype.h> +#include <dirent.h> +#include <errno.h> +#include <libgen.h> +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <sys/param.h> +#include <fcntl.h> +#include <unistd.h> +#include <inttypes.h> +#include "build-id.h" +#include "debug.h" +#include "symbol.h" +#include "strlist.h" + +#include <libelf.h> +#include <gelf.h> +#include <elf.h> +#include <limits.h> +#include <sys/utsname.h> + +#ifndef KSYM_NAME_LEN +#define KSYM_NAME_LEN 128 +#endif + +#ifndef NT_GNU_BUILD_ID +#define NT_GNU_BUILD_ID 3 +#endif + +static bool dso__build_id_equal(const struct dso *dso, u8 *build_id); +static int elf_read_build_id(Elf *elf, void *bf, size_t size); +static void dsos__add(struct list_head *head, struct dso *dso); +static struct map *map__new2(u64 start, struct dso *dso, enum map_type type); +static int dso__load_kernel_sym(struct dso *dso, struct map *map, + symbol_filter_t filter); +static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map, + symbol_filter_t filter); +static int vmlinux_path__nr_entries; +static char **vmlinux_path; + +struct symbol_conf symbol_conf = { + .exclude_other = true, + .use_modules = true, + .try_vmlinux_path = true, + .symfs = "", +}; + +int dso__name_len(const struct dso *dso) +{ + if (verbose) + return dso->long_name_len; + + return dso->short_name_len; +} + +bool dso__loaded(const struct dso *dso, enum map_type type) +{ + return dso->loaded & (1 << type); +} + +bool dso__sorted_by_name(const struct dso *dso, enum map_type type) +{ + return dso->sorted_by_name & (1 << type); +} + +static void dso__set_sorted_by_name(struct dso *dso, enum map_type type) +{ + dso->sorted_by_name |= (1 << type); +} + +bool symbol_type__is_a(char symbol_type, enum map_type map_type) +{ + switch (map_type) { + case MAP__FUNCTION: + return symbol_type == 'T' || symbol_type == 'W'; + case MAP__VARIABLE: + return symbol_type == 'D' || symbol_type == 'd'; + default: + return false; + } +} + +static void symbols__fixup_end(struct rb_root *symbols) +{ + struct rb_node *nd, *prevnd = rb_first(symbols); + struct symbol *curr, *prev; + + if (prevnd == NULL) + return; + + curr = rb_entry(prevnd, struct symbol, rb_node); + + for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) { + prev = curr; + curr = rb_entry(nd, struct symbol, rb_node); + + if (prev->end == prev->start && prev->end != curr->start) + prev->end = curr->start - 1; + } + + /* Last entry */ + if (curr->end == curr->start) + curr->end = roundup(curr->start, 4096); +} + +static void __map_groups__fixup_end(struct map_groups *mg, enum map_type type) +{ + struct map *prev, *curr; + struct rb_node *nd, *prevnd = rb_first(&mg->maps[type]); + + if (prevnd == NULL) + return; + + curr = rb_entry(prevnd, struct map, rb_node); + + for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) { + prev = curr; + curr = rb_entry(nd, struct map, rb_node); + prev->end = curr->start - 1; + } + + /* + * We still haven't the actual symbols, so guess the + * last map final address. + */ + curr->end = ~0ULL; +} + +static void map_groups__fixup_end(struct map_groups *mg) +{ + int i; + for (i = 0; i < MAP__NR_TYPES; ++i) + __map_groups__fixup_end(mg, i); +} + +static struct symbol *symbol__new(u64 start, u64 len, u8 binding, + const char *name) +{ + size_t namelen = strlen(name) + 1; + struct symbol *sym = calloc(1, (symbol_conf.priv_size + + sizeof(*sym) + namelen)); + if (sym == NULL) + return NULL; + + if (symbol_conf.priv_size) + sym = ((void *)sym) + symbol_conf.priv_size; + + sym->start = start; + sym->end = len ? start + len - 1 : start; + sym->binding = binding; + sym->namelen = namelen - 1; + + pr_debug4("%s: %s %#" PRIx64 "-%#" PRIx64 "\n", + __func__, name, start, sym->end); + memcpy(sym->name, name, namelen); + + return sym; +} + +void symbol__delete(struct symbol *sym) +{ + free(((void *)sym) - symbol_conf.priv_size); +} + +static size_t symbol__fprintf(struct symbol *sym, FILE *fp) +{ + return fprintf(fp, " %" PRIx64 "-%" PRIx64 " %c %s\n", + sym->start, sym->end, + sym->binding == STB_GLOBAL ? 'g' : + sym->binding == STB_LOCAL ? 'l' : 'w', + sym->name); +} + +void dso__set_long_name(struct dso *dso, char *name) +{ + if (name == NULL) + return; + dso->long_name = name; + dso->long_name_len = strlen(name); +} + +static void dso__set_short_name(struct dso *dso, const char *name) +{ + if (name == NULL) + return; + dso->short_name = name; + dso->short_name_len = strlen(name); +} + +static void dso__set_basename(struct dso *dso) +{ + dso__set_short_name(dso, basename(dso->long_name)); +} + +struct dso *dso__new(const char *name) +{ + struct dso *dso = calloc(1, sizeof(*dso) + strlen(name) + 1); + + if (dso != NULL) { + int i; + strcpy(dso->name, name); + dso__set_long_name(dso, dso->name); + dso__set_short_name(dso, dso->name); + for (i = 0; i < MAP__NR_TYPES; ++i) + dso->symbols[i] = dso->symbol_names[i] = RB_ROOT; + dso->symtab_type = SYMTAB__NOT_FOUND; + dso->loaded = 0; + dso->sorted_by_name = 0; + dso->has_build_id = 0; + dso->kernel = DSO_TYPE_USER; + INIT_LIST_HEAD(&dso->node); + } + + return dso; +} + +static void symbols__delete(struct rb_root *symbols) +{ + struct symbol *pos; + struct rb_node *next = rb_first(symbols); + + while (next) { + pos = rb_entry(next, struct symbol, rb_node); + next = rb_next(&pos->rb_node); + rb_erase(&pos->rb_node, symbols); + symbol__delete(pos); + } +} + +void dso__delete(struct dso *dso) +{ + int i; + for (i = 0; i < MAP__NR_TYPES; ++i) + symbols__delete(&dso->symbols[i]); + if (dso->sname_alloc) + free((char *)dso->short_name); + if (dso->lname_alloc) + free(dso->long_name); + free(dso); +} + +void dso__set_build_id(struct dso *dso, void *build_id) +{ + memcpy(dso->build_id, build_id, sizeof(dso->build_id)); + dso->has_build_id = 1; +} + +static void symbols__insert(struct rb_root *symbols, struct symbol *sym) +{ + struct rb_node **p = &symbols->rb_node; + struct rb_node *parent = NULL; + const u64 ip = sym->start; + struct symbol *s; + + while (*p != NULL) { + parent = *p; + s = rb_entry(parent, struct symbol, rb_node); + if (ip < s->start) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + rb_link_node(&sym->rb_node, parent, p); + rb_insert_color(&sym->rb_node, symbols); +} + +static struct symbol *symbols__find(struct rb_root *symbols, u64 ip) +{ + struct rb_node *n; + + if (symbols == NULL) + return NULL; + + n = symbols->rb_node; + + while (n) { + struct symbol *s = rb_entry(n, struct symbol, rb_node); + + if (ip < s->start) + n = n->rb_left; + else if (ip > s->end) + n = n->rb_right; + else + return s; + } + + return NULL; +} + +struct symbol_name_rb_node { + struct rb_node rb_node; + struct symbol sym; +}; + +static void symbols__insert_by_name(struct rb_root *symbols, struct symbol *sym) +{ + struct rb_node **p = &symbols->rb_node; + struct rb_node *parent = NULL; + struct symbol_name_rb_node *symn, *s; + + symn = container_of(sym, struct symbol_name_rb_node, sym); + + while (*p != NULL) { + parent = *p; + s = rb_entry(parent, struct symbol_name_rb_node, rb_node); + if (strcmp(sym->name, s->sym.name) < 0) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + rb_link_node(&symn->rb_node, parent, p); + rb_insert_color(&symn->rb_node, symbols); +} + +static void symbols__sort_by_name(struct rb_root *symbols, + struct rb_root *source) +{ + struct rb_node *nd; + + for (nd = rb_first(source); nd; nd = rb_next(nd)) { + struct symbol *pos = rb_entry(nd, struct symbol, rb_node); + symbols__insert_by_name(symbols, pos); + } +} + +static struct symbol *symbols__find_by_name(struct rb_root *symbols, + const char *name) +{ + struct rb_node *n; + + if (symbols == NULL) + return NULL; + + n = symbols->rb_node; + + while (n) { + struct symbol_name_rb_node *s; + int cmp; + + s = rb_entry(n, struct symbol_name_rb_node, rb_node); + cmp = strcmp(name, s->sym.name); + + if (cmp < 0) + n = n->rb_left; + else if (cmp > 0) + n = n->rb_right; + else + return &s->sym; + } + + return NULL; +} + +struct symbol *dso__find_symbol(struct dso *dso, + enum map_type type, u64 addr) +{ + return symbols__find(&dso->symbols[type], addr); +} + +struct symbol *dso__find_symbol_by_name(struct dso *dso, enum map_type type, + const char *name) +{ + return symbols__find_by_name(&dso->symbol_names[type], name); +} + +void dso__sort_by_name(struct dso *dso, enum map_type type) +{ + dso__set_sorted_by_name(dso, type); + return symbols__sort_by_name(&dso->symbol_names[type], + &dso->symbols[type]); +} + +int build_id__sprintf(const u8 *build_id, int len, char *bf) +{ + char *bid = bf; + const u8 *raw = build_id; + int i; + + for (i = 0; i < len; ++i) { + sprintf(bid, "%02x", *raw); + ++raw; + bid += 2; + } + + return raw - build_id; +} + +size_t dso__fprintf_buildid(struct dso *dso, FILE *fp) +{ + char sbuild_id[BUILD_ID_SIZE * 2 + 1]; + + build_id__sprintf(dso->build_id, sizeof(dso->build_id), sbuild_id); + return fprintf(fp, "%s", sbuild_id); +} + +size_t dso__fprintf_symbols_by_name(struct dso *dso, + enum map_type type, FILE *fp) +{ + size_t ret = 0; + struct rb_node *nd; + struct symbol_name_rb_node *pos; + + for (nd = rb_first(&dso->symbol_names[type]); nd; nd = rb_next(nd)) { + pos = rb_entry(nd, struct symbol_name_rb_node, rb_node); + fprintf(fp, "%s\n", pos->sym.name); + } + + return ret; +} + +size_t dso__fprintf(struct dso *dso, enum map_type type, FILE *fp) +{ + struct rb_node *nd; + size_t ret = fprintf(fp, "dso: %s (", dso->short_name); + + if (dso->short_name != dso->long_name) + ret += fprintf(fp, "%s, ", dso->long_name); + ret += fprintf(fp, "%s, %sloaded, ", map_type__name[type], + dso->loaded ? "" : "NOT "); + ret += dso__fprintf_buildid(dso, fp); + ret += fprintf(fp, ")\n"); + for (nd = rb_first(&dso->symbols[type]); nd; nd = rb_next(nd)) { + struct symbol *pos = rb_entry(nd, struct symbol, rb_node); + ret += symbol__fprintf(pos, fp); + } + + return ret; +} + +int kallsyms__parse(const char *filename, void *arg, + int (*process_symbol)(void *arg, const char *name, + char type, u64 start, u64 end)) +{ + char *line = NULL; + size_t n; + int err = -1; + u64 prev_start = 0; + char prev_symbol_type = 0; + char *prev_symbol_name; + FILE *file = fopen(filename, "r"); + + if (file == NULL) + goto out_failure; + + prev_symbol_name = malloc(KSYM_NAME_LEN); + if (prev_symbol_name == NULL) + goto out_close; + + err = 0; + + while (!feof(file)) { + u64 start; + int line_len, len; + char symbol_type; + char *symbol_name; + + line_len = getline(&line, &n, file); + if (line_len < 0 || !line) + break; + + line[--line_len] = '\0'; /* \n */ + + len = hex2u64(line, &start); + + len++; + if (len + 2 >= line_len) + continue; + + symbol_type = toupper(line[len]); + len += 2; + symbol_name = line + len; + len = line_len - len; + + if (len >= KSYM_NAME_LEN) { + err = -1; + break; + } + + if (prev_symbol_type) { + u64 end = start; + if (end != prev_start) + --end; + err = process_symbol(arg, prev_symbol_name, + prev_symbol_type, prev_start, end); + if (err) + break; + } + + memcpy(prev_symbol_name, symbol_name, len + 1); + prev_symbol_type = symbol_type; + prev_start = start; + } + + free(prev_symbol_name); + free(line); +out_close: + fclose(file); + return err; + +out_failure: + return -1; +} + +struct process_kallsyms_args { + struct map *map; + struct dso *dso; +}; + +static u8 kallsyms2elf_type(char type) +{ + if (type == 'W') + return STB_WEAK; + + return isupper(type) ? STB_GLOBAL : STB_LOCAL; +} + +static int map__process_kallsym_symbol(void *arg, const char *name, + char type, u64 start, u64 end) +{ + struct symbol *sym; + struct process_kallsyms_args *a = arg; + struct rb_root *root = &a->dso->symbols[a->map->type]; + + if (!symbol_type__is_a(type, a->map->type)) + return 0; + + sym = symbol__new(start, end - start + 1, + kallsyms2elf_type(type), name); + if (sym == NULL) + return -ENOMEM; + /* + * We will pass the symbols to the filter later, in + * map__split_kallsyms, when we have split the maps per module + */ + symbols__insert(root, sym); + + return 0; +} + +/* + * Loads the function entries in /proc/kallsyms into kernel_map->dso, + * so that we can in the next step set the symbol ->end address and then + * call kernel_maps__split_kallsyms. + */ +static int dso__load_all_kallsyms(struct dso *dso, const char *filename, + struct map *map) +{ + struct process_kallsyms_args args = { .map = map, .dso = dso, }; + return kallsyms__parse(filename, &args, map__process_kallsym_symbol); +} + +/* + * Split the symbols into maps, making sure there are no overlaps, i.e. the + * kernel range is broken in several maps, named [kernel].N, as we don't have + * the original ELF section names vmlinux have. + */ +static int dso__split_kallsyms(struct dso *dso, struct map *map, + symbol_filter_t filter) +{ + struct map_groups *kmaps = map__kmap(map)->kmaps; + struct machine *machine = kmaps->machine; + struct map *curr_map = map; + struct symbol *pos; + int count = 0, moved = 0; + struct rb_root *root = &dso->symbols[map->type]; + struct rb_node *next = rb_first(root); + int kernel_range = 0; + + while (next) { + char *module; + + pos = rb_entry(next, struct symbol, rb_node); + next = rb_next(&pos->rb_node); + + module = strchr(pos->name, '\t'); + if (module) { + if (!symbol_conf.use_modules) + goto discard_symbol; + + *module++ = '\0'; + + if (strcmp(curr_map->dso->short_name, module)) { + if (curr_map != map && + dso->kernel == DSO_TYPE_GUEST_KERNEL && + machine__is_default_guest(machine)) { + /* + * We assume all symbols of a module are + * continuous in * kallsyms, so curr_map + * points to a module and all its + * symbols are in its kmap. Mark it as + * loaded. + */ + dso__set_loaded(curr_map->dso, + curr_map->type); + } + + curr_map = map_groups__find_by_name(kmaps, + map->type, module); + if (curr_map == NULL) { + pr_debug("%s/proc/{kallsyms,modules} " + "inconsistency while looking " + "for \"%s\" module!\n", + machine->root_dir, module); + curr_map = map; + goto discard_symbol; + } + + if (curr_map->dso->loaded && + !machine__is_default_guest(machine)) + goto discard_symbol; + } + /* + * So that we look just like we get from .ko files, + * i.e. not prelinked, relative to map->start. + */ + pos->start = curr_map->map_ip(curr_map, pos->start); + pos->end = curr_map->map_ip(curr_map, pos->end); + } else if (curr_map != map) { + char dso_name[PATH_MAX]; + struct dso *ndso; + + if (count == 0) { + curr_map = map; + goto filter_symbol; + } + + if (dso->kernel == DSO_TYPE_GUEST_KERNEL) + snprintf(dso_name, sizeof(dso_name), + "[guest.kernel].%d", + kernel_range++); + else + snprintf(dso_name, sizeof(dso_name), + "[kernel].%d", + kernel_range++); + + ndso = dso__new(dso_name); + if (ndso == NULL) + return -1; + + ndso->kernel = dso->kernel; + + curr_map = map__new2(pos->start, ndso, map->type); + if (curr_map == NULL) { + dso__delete(ndso); + return -1; + } + + curr_map->map_ip = curr_map->unmap_ip = identity__map_ip; + map_groups__insert(kmaps, curr_map); + ++kernel_range; + } +filter_symbol: + if (filter && filter(curr_map, pos)) { +discard_symbol: rb_erase(&pos->rb_node, root); + symbol__delete(pos); + } else { + if (curr_map != map) { + rb_erase(&pos->rb_node, root); + symbols__insert(&curr_map->dso->symbols[curr_map->type], pos); + ++moved; + } else + ++count; + } + } + + if (curr_map != map && + dso->kernel == DSO_TYPE_GUEST_KERNEL && + machine__is_default_guest(kmaps->machine)) { + dso__set_loaded(curr_map->dso, curr_map->type); + } + + return count + moved; +} + +static bool symbol__restricted_filename(const char *filename, + const char *restricted_filename) +{ + bool restricted = false; + + if (symbol_conf.kptr_restrict) { + char *r = realpath(filename, NULL); + + if (r != NULL) { + restricted = strcmp(r, restricted_filename) == 0; + free(r); + return restricted; + } + } + + return restricted; +} + +int dso__load_kallsyms(struct dso *dso, const char *filename, + struct map *map, symbol_filter_t filter) +{ + if (symbol__restricted_filename(filename, "/proc/kallsyms")) + return -1; + + if (dso__load_all_kallsyms(dso, filename, map) < 0) + return -1; + + if (dso->kernel == DSO_TYPE_GUEST_KERNEL) + dso->symtab_type = SYMTAB__GUEST_KALLSYMS; + else + dso->symtab_type = SYMTAB__KALLSYMS; + + return dso__split_kallsyms(dso, map, filter); +} + +static int dso__load_perf_map(struct dso *dso, struct map *map, + symbol_filter_t filter) +{ + char *line = NULL; + size_t n; + FILE *file; + int nr_syms = 0; + + file = fopen(dso->long_name, "r"); + if (file == NULL) + goto out_failure; + + while (!feof(file)) { + u64 start, size; + struct symbol *sym; + int line_len, len; + + line_len = getline(&line, &n, file); + if (line_len < 0) + break; + + if (!line) + goto out_failure; + + line[--line_len] = '\0'; /* \n */ + + len = hex2u64(line, &start); + + len++; + if (len + 2 >= line_len) + continue; + + len += hex2u64(line + len, &size); + + len++; + if (len + 2 >= line_len) + continue; + + sym = symbol__new(start, size, STB_GLOBAL, line + len); + + if (sym == NULL) + goto out_delete_line; + + if (filter && filter(map, sym)) + symbol__delete(sym); + else { + symbols__insert(&dso->symbols[map->type], sym); + nr_syms++; + } + } + + free(line); + fclose(file); + + return nr_syms; + +out_delete_line: + free(line); +out_failure: + return -1; +} + +/** + * elf_symtab__for_each_symbol - iterate thru all the symbols + * + * @syms: struct elf_symtab instance to iterate + * @idx: uint32_t idx + * @sym: GElf_Sym iterator + */ +#define elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) \ + for (idx = 0, gelf_getsym(syms, idx, &sym);\ + idx < nr_syms; \ + idx++, gelf_getsym(syms, idx, &sym)) + +static inline uint8_t elf_sym__type(const GElf_Sym *sym) +{ + return GELF_ST_TYPE(sym->st_info); +} + +static inline int elf_sym__is_function(const GElf_Sym *sym) +{ + return elf_sym__type(sym) == STT_FUNC && + sym->st_name != 0 && + sym->st_shndx != SHN_UNDEF; +} + +static inline bool elf_sym__is_object(const GElf_Sym *sym) +{ + return elf_sym__type(sym) == STT_OBJECT && + sym->st_name != 0 && + sym->st_shndx != SHN_UNDEF; +} + +static inline int elf_sym__is_label(const GElf_Sym *sym) +{ + return elf_sym__type(sym) == STT_NOTYPE && + sym->st_name != 0 && + sym->st_shndx != SHN_UNDEF && + sym->st_shndx != SHN_ABS; +} + +static inline const char *elf_sec__name(const GElf_Shdr *shdr, + const Elf_Data *secstrs) +{ + return secstrs->d_buf + shdr->sh_name; +} + +static inline int elf_sec__is_text(const GElf_Shdr *shdr, + const Elf_Data *secstrs) +{ + return strstr(elf_sec__name(shdr, secstrs), "text") != NULL; +} + +static inline bool elf_sec__is_data(const GElf_Shdr *shdr, + const Elf_Data *secstrs) +{ + return strstr(elf_sec__name(shdr, secstrs), "data") != NULL; +} + +static inline const char *elf_sym__name(const GElf_Sym *sym, + const Elf_Data *symstrs) +{ + return symstrs->d_buf + sym->st_name; +} + +static Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep, + GElf_Shdr *shp, const char *name, + size_t *idx) +{ + Elf_Scn *sec = NULL; + size_t cnt = 1; + + while ((sec = elf_nextscn(elf, sec)) != NULL) { + char *str; + + gelf_getshdr(sec, shp); + str = elf_strptr(elf, ep->e_shstrndx, shp->sh_name); + if (!strcmp(name, str)) { + if (idx) + *idx = cnt; + break; + } + ++cnt; + } + + return sec; +} + +#define elf_section__for_each_rel(reldata, pos, pos_mem, idx, nr_entries) \ + for (idx = 0, pos = gelf_getrel(reldata, 0, &pos_mem); \ + idx < nr_entries; \ + ++idx, pos = gelf_getrel(reldata, idx, &pos_mem)) + +#define elf_section__for_each_rela(reldata, pos, pos_mem, idx, nr_entries) \ + for (idx = 0, pos = gelf_getrela(reldata, 0, &pos_mem); \ + idx < nr_entries; \ + ++idx, pos = gelf_getrela(reldata, idx, &pos_mem)) + +/* + * We need to check if we have a .dynsym, so that we can handle the + * .plt, synthesizing its symbols, that aren't on the symtabs (be it + * .dynsym or .symtab). + * And always look at the original dso, not at debuginfo packages, that + * have the PLT data stripped out (shdr_rel_plt.sh_type == SHT_NOBITS). + */ +static int dso__synthesize_plt_symbols(struct dso *dso, struct map *map, + symbol_filter_t filter) +{ + uint32_t nr_rel_entries, idx; + GElf_Sym sym; + u64 plt_offset; + GElf_Shdr shdr_plt; + struct symbol *f; + GElf_Shdr shdr_rel_plt, shdr_dynsym; + Elf_Data *reldata, *syms, *symstrs; + Elf_Scn *scn_plt_rel, *scn_symstrs, *scn_dynsym; + size_t dynsym_idx; + GElf_Ehdr ehdr; + char sympltname[1024]; + Elf *elf; + int nr = 0, symidx, fd, err = 0; + char name[PATH_MAX]; + + snprintf(name, sizeof(name), "%s%s", + symbol_conf.symfs, dso->long_name); + fd = open(name, O_RDONLY); + if (fd < 0) + goto out; + + elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL); + if (elf == NULL) + goto out_close; + + if (gelf_getehdr(elf, &ehdr) == NULL) + goto out_elf_end; + + scn_dynsym = elf_section_by_name(elf, &ehdr, &shdr_dynsym, + ".dynsym", &dynsym_idx); + if (scn_dynsym == NULL) + goto out_elf_end; + + scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt, + ".rela.plt", NULL); + if (scn_plt_rel == NULL) { + scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt, + ".rel.plt", NULL); + if (scn_plt_rel == NULL) + goto out_elf_end; + } + + err = -1; + + if (shdr_rel_plt.sh_link != dynsym_idx) + goto out_elf_end; + + if (elf_section_by_name(elf, &ehdr, &shdr_plt, ".plt", NULL) == NULL) + goto out_elf_end; + + /* + * Fetch the relocation section to find the idxes to the GOT + * and the symbols in the .dynsym they refer to. + */ + reldata = elf_getdata(scn_plt_rel, NULL); + if (reldata == NULL) + goto out_elf_end; + + syms = elf_getdata(scn_dynsym, NULL); + if (syms == NULL) + goto out_elf_end; + + scn_symstrs = elf_getscn(elf, shdr_dynsym.sh_link); + if (scn_symstrs == NULL) + goto out_elf_end; + + symstrs = elf_getdata(scn_symstrs, NULL); + if (symstrs == NULL) + goto out_elf_end; + + nr_rel_entries = shdr_rel_plt.sh_size / shdr_rel_plt.sh_entsize; + plt_offset = shdr_plt.sh_offset; + + if (shdr_rel_plt.sh_type == SHT_RELA) { + GElf_Rela pos_mem, *pos; + + elf_section__for_each_rela(reldata, pos, pos_mem, idx, + nr_rel_entries) { + symidx = GELF_R_SYM(pos->r_info); + plt_offset += shdr_plt.sh_entsize; + gelf_getsym(syms, symidx, &sym); + snprintf(sympltname, sizeof(sympltname), + "%s@plt", elf_sym__name(&sym, symstrs)); + + f = symbol__new(plt_offset, shdr_plt.sh_entsize, + STB_GLOBAL, sympltname); + if (!f) + goto out_elf_end; + + if (filter && filter(map, f)) + symbol__delete(f); + else { + symbols__insert(&dso->symbols[map->type], f); + ++nr; + } + } + } else if (shdr_rel_plt.sh_type == SHT_REL) { + GElf_Rel pos_mem, *pos; + elf_section__for_each_rel(reldata, pos, pos_mem, idx, + nr_rel_entries) { + symidx = GELF_R_SYM(pos->r_info); + plt_offset += shdr_plt.sh_entsize; + gelf_getsym(syms, symidx, &sym); + snprintf(sympltname, sizeof(sympltname), + "%s@plt", elf_sym__name(&sym, symstrs)); + + f = symbol__new(plt_offset, shdr_plt.sh_entsize, + STB_GLOBAL, sympltname); + if (!f) + goto out_elf_end; + + if (filter && filter(map, f)) + symbol__delete(f); + else { + symbols__insert(&dso->symbols[map->type], f); + ++nr; + } + } + } + + err = 0; +out_elf_end: + elf_end(elf); +out_close: + close(fd); + + if (err == 0) + return nr; +out: + pr_debug("%s: problems reading %s PLT info.\n", + __func__, dso->long_name); + return 0; +} + +static bool elf_sym__is_a(GElf_Sym *sym, enum map_type type) +{ + switch (type) { + case MAP__FUNCTION: + return elf_sym__is_function(sym); + case MAP__VARIABLE: + return elf_sym__is_object(sym); + default: + return false; + } +} + +static bool elf_sec__is_a(GElf_Shdr *shdr, Elf_Data *secstrs, + enum map_type type) +{ + switch (type) { + case MAP__FUNCTION: + return elf_sec__is_text(shdr, secstrs); + case MAP__VARIABLE: + return elf_sec__is_data(shdr, secstrs); + default: + return false; + } +} + +static size_t elf_addr_to_index(Elf *elf, GElf_Addr addr) +{ + Elf_Scn *sec = NULL; + GElf_Shdr shdr; + size_t cnt = 1; + + while ((sec = elf_nextscn(elf, sec)) != NULL) { + gelf_getshdr(sec, &shdr); + + if ((addr >= shdr.sh_addr) && + (addr < (shdr.sh_addr + shdr.sh_size))) + return cnt; + + ++cnt; + } + + return -1; +} + +static int dso__load_sym(struct dso *dso, struct map *map, const char *name, + int fd, symbol_filter_t filter, int kmodule, + int want_symtab) +{ + struct kmap *kmap = dso->kernel ? map__kmap(map) : NULL; + struct map *curr_map = map; + struct dso *curr_dso = dso; + Elf_Data *symstrs, *secstrs; + uint32_t nr_syms; + int err = -1; + uint32_t idx; + GElf_Ehdr ehdr; + GElf_Shdr shdr, opdshdr; + Elf_Data *syms, *opddata = NULL; + GElf_Sym sym; + Elf_Scn *sec, *sec_strndx, *opdsec; + Elf *elf; + int nr = 0; + size_t opdidx = 0; + + elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL); + if (elf == NULL) { + pr_debug("%s: cannot read %s ELF file.\n", __func__, name); + goto out_close; + } + + if (gelf_getehdr(elf, &ehdr) == NULL) { + pr_debug("%s: cannot get elf header.\n", __func__); + goto out_elf_end; + } + + /* Always reject images with a mismatched build-id: */ + if (dso->has_build_id) { + u8 build_id[BUILD_ID_SIZE]; + + if (elf_read_build_id(elf, build_id, + BUILD_ID_SIZE) != BUILD_ID_SIZE) + goto out_elf_end; + + if (!dso__build_id_equal(dso, build_id)) + goto out_elf_end; + } + + sec = elf_section_by_name(elf, &ehdr, &shdr, ".symtab", NULL); + if (sec == NULL) { + if (want_symtab) + goto out_elf_end; + + sec = elf_section_by_name(elf, &ehdr, &shdr, ".dynsym", NULL); + if (sec == NULL) + goto out_elf_end; + } + + opdsec = elf_section_by_name(elf, &ehdr, &opdshdr, ".opd", &opdidx); + if (opdshdr.sh_type != SHT_PROGBITS) + opdsec = NULL; + if (opdsec) + opddata = elf_rawdata(opdsec, NULL); + + syms = elf_getdata(sec, NULL); + if (syms == NULL) + goto out_elf_end; + + sec = elf_getscn(elf, shdr.sh_link); + if (sec == NULL) + goto out_elf_end; + + symstrs = elf_getdata(sec, NULL); + if (symstrs == NULL) + goto out_elf_end; + + sec_strndx = elf_getscn(elf, ehdr.e_shstrndx); + if (sec_strndx == NULL) + goto out_elf_end; + + secstrs = elf_getdata(sec_strndx, NULL); + if (secstrs == NULL) + goto out_elf_end; + + nr_syms = shdr.sh_size / shdr.sh_entsize; + + memset(&sym, 0, sizeof(sym)); + if (dso->kernel == DSO_TYPE_USER) { + dso->adjust_symbols = (ehdr.e_type == ET_EXEC || + elf_section_by_name(elf, &ehdr, &shdr, + ".gnu.prelink_undo", + NULL) != NULL); + } else { + dso->adjust_symbols = 0; + } + elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) { + struct symbol *f; + const char *elf_name = elf_sym__name(&sym, symstrs); + char *demangled = NULL; + int is_label = elf_sym__is_label(&sym); + const char *section_name; + + if (kmap && kmap->ref_reloc_sym && kmap->ref_reloc_sym->name && + strcmp(elf_name, kmap->ref_reloc_sym->name) == 0) + kmap->ref_reloc_sym->unrelocated_addr = sym.st_value; + + if (!is_label && !elf_sym__is_a(&sym, map->type)) + continue; + + /* Reject ARM ELF "mapping symbols": these aren't unique and + * don't identify functions, so will confuse the profile + * output: */ + if (ehdr.e_machine == EM_ARM) { + if (!strcmp(elf_name, "$a") || + !strcmp(elf_name, "$d") || + !strcmp(elf_name, "$t")) + continue; + } + + if (opdsec && sym.st_shndx == opdidx) { + u32 offset = sym.st_value - opdshdr.sh_addr; + u64 *opd = opddata->d_buf + offset; + sym.st_value = *opd; + sym.st_shndx = elf_addr_to_index(elf, sym.st_value); + } + + sec = elf_getscn(elf, sym.st_shndx); + if (!sec) + goto out_elf_end; + + gelf_getshdr(sec, &shdr); + + if (is_label && !elf_sec__is_a(&shdr, secstrs, map->type)) + continue; + + section_name = elf_sec__name(&shdr, secstrs); + + /* On ARM, symbols for thumb functions have 1 added to + * the symbol address as a flag - remove it */ + if ((ehdr.e_machine == EM_ARM) && + (map->type == MAP__FUNCTION) && + (sym.st_value & 1)) + --sym.st_value; + + if (dso->kernel != DSO_TYPE_USER || kmodule) { + char dso_name[PATH_MAX]; + + if (strcmp(section_name, + (curr_dso->short_name + + dso->short_name_len)) == 0) + goto new_symbol; + + if (strcmp(section_name, ".text") == 0) { + curr_map = map; + curr_dso = dso; + goto new_symbol; + } + + snprintf(dso_name, sizeof(dso_name), + "%s%s", dso->short_name, section_name); + + curr_map = map_groups__find_by_name(kmap->kmaps, map->type, dso_name); + if (curr_map == NULL) { + u64 start = sym.st_value; + + if (kmodule) + start += map->start + shdr.sh_offset; + + curr_dso = dso__new(dso_name); + if (curr_dso == NULL) + goto out_elf_end; + curr_dso->kernel = dso->kernel; + curr_dso->long_name = dso->long_name; + curr_dso->long_name_len = dso->long_name_len; + curr_map = map__new2(start, curr_dso, + map->type); + if (curr_map == NULL) { + dso__delete(curr_dso); + goto out_elf_end; + } + curr_map->map_ip = identity__map_ip; + curr_map->unmap_ip = identity__map_ip; + curr_dso->symtab_type = dso->symtab_type; + map_groups__insert(kmap->kmaps, curr_map); + dsos__add(&dso->node, curr_dso); + dso__set_loaded(curr_dso, map->type); + } else + curr_dso = curr_map->dso; + + goto new_symbol; + } + + if (curr_dso->adjust_symbols) { + pr_debug4("%s: adjusting symbol: st_value: %#" PRIx64 " " + "sh_addr: %#" PRIx64 " sh_offset: %#" PRIx64 "\n", __func__, + (u64)sym.st_value, (u64)shdr.sh_addr, + (u64)shdr.sh_offset); + sym.st_value -= shdr.sh_addr - shdr.sh_offset; + } + /* + * We need to figure out if the object was created from C++ sources + * DWARF DW_compile_unit has this, but we don't always have access + * to it... + */ + demangled = bfd_demangle(NULL, elf_name, DMGL_PARAMS | DMGL_ANSI); + if (demangled != NULL) + elf_name = demangled; +new_symbol: + f = symbol__new(sym.st_value, sym.st_size, + GELF_ST_BIND(sym.st_info), elf_name); + free(demangled); + if (!f) + goto out_elf_end; + + if (filter && filter(curr_map, f)) + symbol__delete(f); + else { + symbols__insert(&curr_dso->symbols[curr_map->type], f); + nr++; + } + } + + /* + * For misannotated, zeroed, ASM function sizes. + */ + if (nr > 0) { + symbols__fixup_end(&dso->symbols[map->type]); + if (kmap) { + /* + * We need to fixup this here too because we create new + * maps here, for things like vsyscall sections. + */ + __map_groups__fixup_end(kmap->kmaps, map->type); + } + } + err = nr; +out_elf_end: + elf_end(elf); +out_close: + return err; +} + +static bool dso__build_id_equal(const struct dso *dso, u8 *build_id) +{ + return memcmp(dso->build_id, build_id, sizeof(dso->build_id)) == 0; +} + +bool __dsos__read_build_ids(struct list_head *head, bool with_hits) +{ + bool have_build_id = false; + struct dso *pos; + + list_for_each_entry(pos, head, node) { + if (with_hits && !pos->hit) + continue; + if (pos->has_build_id) { + have_build_id = true; + continue; + } + if (filename__read_build_id(pos->long_name, pos->build_id, + sizeof(pos->build_id)) > 0) { + have_build_id = true; + pos->has_build_id = true; + } + } + + return have_build_id; +} + +/* + * Align offset to 4 bytes as needed for note name and descriptor data. + */ +#define NOTE_ALIGN(n) (((n) + 3) & -4U) + +static int elf_read_build_id(Elf *elf, void *bf, size_t size) +{ + int err = -1; + GElf_Ehdr ehdr; + GElf_Shdr shdr; + Elf_Data *data; + Elf_Scn *sec; + Elf_Kind ek; + void *ptr; + + if (size < BUILD_ID_SIZE) + goto out; + + ek = elf_kind(elf); + if (ek != ELF_K_ELF) + goto out; + + if (gelf_getehdr(elf, &ehdr) == NULL) { + pr_err("%s: cannot get elf header.\n", __func__); + goto out; + } + + sec = elf_section_by_name(elf, &ehdr, &shdr, + ".note.gnu.build-id", NULL); + if (sec == NULL) { + sec = elf_section_by_name(elf, &ehdr, &shdr, + ".notes", NULL); + if (sec == NULL) + goto out; + } + + data = elf_getdata(sec, NULL); + if (data == NULL) + goto out; + + ptr = data->d_buf; + while (ptr < (data->d_buf + data->d_size)) { + GElf_Nhdr *nhdr = ptr; + int namesz = NOTE_ALIGN(nhdr->n_namesz), + descsz = NOTE_ALIGN(nhdr->n_descsz); + const char *name; + + ptr += sizeof(*nhdr); + name = ptr; + ptr += namesz; + if (nhdr->n_type == NT_GNU_BUILD_ID && + nhdr->n_namesz == sizeof("GNU")) { + if (memcmp(name, "GNU", sizeof("GNU")) == 0) { + memcpy(bf, ptr, BUILD_ID_SIZE); + err = BUILD_ID_SIZE; + break; + } + } + ptr += descsz; + } + +out: + return err; +} + +int filename__read_build_id(const char *filename, void *bf, size_t size) +{ + int fd, err = -1; + Elf *elf; + + if (size < BUILD_ID_SIZE) + goto out; + + fd = open(filename, O_RDONLY); + if (fd < 0) + goto out; + + elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL); + if (elf == NULL) { + pr_debug2("%s: cannot read %s ELF file.\n", __func__, filename); + goto out_close; + } + + err = elf_read_build_id(elf, bf, size); + + elf_end(elf); +out_close: + close(fd); +out: + return err; +} + +int sysfs__read_build_id(const char *filename, void *build_id, size_t size) +{ + int fd, err = -1; + + if (size < BUILD_ID_SIZE) + goto out; + + fd = open(filename, O_RDONLY); + if (fd < 0) + goto out; + + while (1) { + char bf[BUFSIZ]; + GElf_Nhdr nhdr; + int namesz, descsz; + + if (read(fd, &nhdr, sizeof(nhdr)) != sizeof(nhdr)) + break; + + namesz = NOTE_ALIGN(nhdr.n_namesz); + descsz = NOTE_ALIGN(nhdr.n_descsz); + if (nhdr.n_type == NT_GNU_BUILD_ID && + nhdr.n_namesz == sizeof("GNU")) { + if (read(fd, bf, namesz) != namesz) + break; + if (memcmp(bf, "GNU", sizeof("GNU")) == 0) { + if (read(fd, build_id, + BUILD_ID_SIZE) == BUILD_ID_SIZE) { + err = 0; + break; + } + } else if (read(fd, bf, descsz) != descsz) + break; + } else { + int n = namesz + descsz; + if (read(fd, bf, n) != n) + break; + } + } + close(fd); +out: + return err; +} + +char dso__symtab_origin(const struct dso *dso) +{ + static const char origin[] = { + [SYMTAB__KALLSYMS] = 'k', + [SYMTAB__JAVA_JIT] = 'j', + [SYMTAB__BUILD_ID_CACHE] = 'B', + [SYMTAB__FEDORA_DEBUGINFO] = 'f', + [SYMTAB__UBUNTU_DEBUGINFO] = 'u', + [SYMTAB__BUILDID_DEBUGINFO] = 'b', + [SYMTAB__SYSTEM_PATH_DSO] = 'd', + [SYMTAB__SYSTEM_PATH_KMODULE] = 'K', + [SYMTAB__GUEST_KALLSYMS] = 'g', + [SYMTAB__GUEST_KMODULE] = 'G', + }; + + if (dso == NULL || dso->symtab_type == SYMTAB__NOT_FOUND) + return '!'; + return origin[dso->symtab_type]; +} + +int dso__load(struct dso *dso, struct map *map, symbol_filter_t filter) +{ + int size = PATH_MAX; + char *name; + int ret = -1; + int fd; + struct machine *machine; + const char *root_dir; + int want_symtab; + + dso__set_loaded(dso, map->type); + + if (dso->kernel == DSO_TYPE_KERNEL) + return dso__load_kernel_sym(dso, map, filter); + else if (dso->kernel == DSO_TYPE_GUEST_KERNEL) + return dso__load_guest_kernel_sym(dso, map, filter); + + if (map->groups && map->groups->machine) + machine = map->groups->machine; + else + machine = NULL; + + name = malloc(size); + if (!name) + return -1; + + dso->adjust_symbols = 0; + + if (strncmp(dso->name, "/tmp/perf-", 10) == 0) { + ret = dso__load_perf_map(dso, map, filter); + dso->symtab_type = ret > 0 ? SYMTAB__JAVA_JIT : + SYMTAB__NOT_FOUND; + return ret; + } + + /* Iterate over candidate debug images. + * On the first pass, only load images if they have a full symtab. + * Failing that, do a second pass where we accept .dynsym also + */ + want_symtab = 1; +restart: + for (dso->symtab_type = SYMTAB__BUILD_ID_CACHE; + dso->symtab_type != SYMTAB__NOT_FOUND; + dso->symtab_type++) { + switch (dso->symtab_type) { + case SYMTAB__BUILD_ID_CACHE: + /* skip the locally configured cache if a symfs is given */ + if (symbol_conf.symfs[0] || + (dso__build_id_filename(dso, name, size) == NULL)) { + continue; + } + break; + case SYMTAB__FEDORA_DEBUGINFO: + snprintf(name, size, "%s/usr/lib/debug%s.debug", + symbol_conf.symfs, dso->long_name); + break; + case SYMTAB__UBUNTU_DEBUGINFO: + snprintf(name, size, "%s/usr/lib/debug%s", + symbol_conf.symfs, dso->long_name); + break; + case SYMTAB__BUILDID_DEBUGINFO: { + char build_id_hex[BUILD_ID_SIZE * 2 + 1]; + + if (!dso->has_build_id) + continue; + + build_id__sprintf(dso->build_id, + sizeof(dso->build_id), + build_id_hex); + snprintf(name, size, + "%s/usr/lib/debug/.build-id/%.2s/%s.debug", + symbol_conf.symfs, build_id_hex, build_id_hex + 2); + } + break; + case SYMTAB__SYSTEM_PATH_DSO: + snprintf(name, size, "%s%s", + symbol_conf.symfs, dso->long_name); + break; + case SYMTAB__GUEST_KMODULE: + if (map->groups && machine) + root_dir = machine->root_dir; + else + root_dir = ""; + snprintf(name, size, "%s%s%s", symbol_conf.symfs, + root_dir, dso->long_name); + break; + + case SYMTAB__SYSTEM_PATH_KMODULE: + snprintf(name, size, "%s%s", symbol_conf.symfs, + dso->long_name); + break; + default:; + } + + /* Name is now the name of the next image to try */ + fd = open(name, O_RDONLY); + if (fd < 0) + continue; + + ret = dso__load_sym(dso, map, name, fd, filter, 0, + want_symtab); + close(fd); + + /* + * Some people seem to have debuginfo files _WITHOUT_ debug + * info!?!? + */ + if (!ret) + continue; + + if (ret > 0) { + int nr_plt = dso__synthesize_plt_symbols(dso, map, + filter); + if (nr_plt > 0) + ret += nr_plt; + break; + } + } + + /* + * If we wanted a full symtab but no image had one, + * relax our requirements and repeat the search. + */ + if (ret <= 0 && want_symtab) { + want_symtab = 0; + goto restart; + } + + free(name); + if (ret < 0 && strstr(dso->name, " (deleted)") != NULL) + return 0; + return ret; +} + +struct map *map_groups__find_by_name(struct map_groups *mg, + enum map_type type, const char *name) +{ + struct rb_node *nd; + + for (nd = rb_first(&mg->maps[type]); nd; nd = rb_next(nd)) { + struct map *map = rb_entry(nd, struct map, rb_node); + + if (map->dso && strcmp(map->dso->short_name, name) == 0) + return map; + } + + return NULL; +} + +static int dso__kernel_module_get_build_id(struct dso *dso, + const char *root_dir) +{ + char filename[PATH_MAX]; + /* + * kernel module short names are of the form "[module]" and + * we need just "module" here. + */ + const char *name = dso->short_name + 1; + + snprintf(filename, sizeof(filename), + "%s/sys/module/%.*s/notes/.note.gnu.build-id", + root_dir, (int)strlen(name) - 1, name); + + if (sysfs__read_build_id(filename, dso->build_id, + sizeof(dso->build_id)) == 0) + dso->has_build_id = true; + + return 0; +} + +static int map_groups__set_modules_path_dir(struct map_groups *mg, + const char *dir_name) +{ + struct dirent *dent; + DIR *dir = opendir(dir_name); + int ret = 0; + + if (!dir) { + pr_debug("%s: cannot open %s dir\n", __func__, dir_name); + return -1; + } + + while ((dent = readdir(dir)) != NULL) { + char path[PATH_MAX]; + struct stat st; + + /*sshfs might return bad dent->d_type, so we have to stat*/ + sprintf(path, "%s/%s", dir_name, dent->d_name); + if (stat(path, &st)) + continue; + + if (S_ISDIR(st.st_mode)) { + if (!strcmp(dent->d_name, ".") || + !strcmp(dent->d_name, "..")) + continue; + + snprintf(path, sizeof(path), "%s/%s", + dir_name, dent->d_name); + ret = map_groups__set_modules_path_dir(mg, path); + if (ret < 0) + goto out; + } else { + char *dot = strrchr(dent->d_name, '.'), + dso_name[PATH_MAX]; + struct map *map; + char *long_name; + + if (dot == NULL || strcmp(dot, ".ko")) + continue; + snprintf(dso_name, sizeof(dso_name), "[%.*s]", + (int)(dot - dent->d_name), dent->d_name); + + strxfrchar(dso_name, '-', '_'); + map = map_groups__find_by_name(mg, MAP__FUNCTION, + dso_name); + if (map == NULL) + continue; + + snprintf(path, sizeof(path), "%s/%s", + dir_name, dent->d_name); + + long_name = strdup(path); + if (long_name == NULL) { + ret = -1; + goto out; + } + dso__set_long_name(map->dso, long_name); + map->dso->lname_alloc = 1; + dso__kernel_module_get_build_id(map->dso, ""); + } + } + +out: + closedir(dir); + return ret; +} + +static char *get_kernel_version(const char *root_dir) +{ + char version[PATH_MAX]; + FILE *file; + char *name, *tmp; + const char *prefix = "Linux version "; + + sprintf(version, "%s/proc/version", root_dir); + file = fopen(version, "r"); + if (!file) + return NULL; + + version[0] = '\0'; + tmp = fgets(version, sizeof(version), file); + fclose(file); + + name = strstr(version, prefix); + if (!name) + return NULL; + name += strlen(prefix); + tmp = strchr(name, ' '); + if (tmp) + *tmp = '\0'; + + return strdup(name); +} + +static int machine__set_modules_path(struct machine *machine) +{ + char *version; + char modules_path[PATH_MAX]; + + version = get_kernel_version(machine->root_dir); + if (!version) + return -1; + + snprintf(modules_path, sizeof(modules_path), "%s/lib/modules/%s/kernel", + machine->root_dir, version); + free(version); + + return map_groups__set_modules_path_dir(&machine->kmaps, modules_path); +} + +/* + * Constructor variant for modules (where we know from /proc/modules where + * they are loaded) and for vmlinux, where only after we load all the + * symbols we'll know where it starts and ends. + */ +static struct map *map__new2(u64 start, struct dso *dso, enum map_type type) +{ + struct map *map = calloc(1, (sizeof(*map) + + (dso->kernel ? sizeof(struct kmap) : 0))); + if (map != NULL) { + /* + * ->end will be filled after we load all the symbols + */ + map__init(map, type, start, 0, 0, dso); + } + + return map; +} + +struct map *machine__new_module(struct machine *machine, u64 start, + const char *filename) +{ + struct map *map; + struct dso *dso = __dsos__findnew(&machine->kernel_dsos, filename); + + if (dso == NULL) + return NULL; + + map = map__new2(start, dso, MAP__FUNCTION); + if (map == NULL) + return NULL; + + if (machine__is_host(machine)) + dso->symtab_type = SYMTAB__SYSTEM_PATH_KMODULE; + else + dso->symtab_type = SYMTAB__GUEST_KMODULE; + map_groups__insert(&machine->kmaps, map); + return map; +} + +static int machine__create_modules(struct machine *machine) +{ + char *line = NULL; + size_t n; + FILE *file; + struct map *map; + const char *modules; + char path[PATH_MAX]; + + if (machine__is_default_guest(machine)) + modules = symbol_conf.default_guest_modules; + else { + sprintf(path, "%s/proc/modules", machine->root_dir); + modules = path; + } + + if (symbol__restricted_filename(path, "/proc/modules")) + return -1; + + file = fopen(modules, "r"); + if (file == NULL) + return -1; + + while (!feof(file)) { + char name[PATH_MAX]; + u64 start; + char *sep; + int line_len; + + line_len = getline(&line, &n, file); + if (line_len < 0) + break; + + if (!line) + goto out_failure; + + line[--line_len] = '\0'; /* \n */ + + sep = strrchr(line, 'x'); + if (sep == NULL) + continue; + + hex2u64(sep + 1, &start); + + sep = strchr(line, ' '); + if (sep == NULL) + continue; + + *sep = '\0'; + + snprintf(name, sizeof(name), "[%s]", line); + map = machine__new_module(machine, start, name); + if (map == NULL) + goto out_delete_line; + dso__kernel_module_get_build_id(map->dso, machine->root_dir); + } + + free(line); + fclose(file); + + return machine__set_modules_path(machine); + +out_delete_line: + free(line); +out_failure: + return -1; +} + +int dso__load_vmlinux(struct dso *dso, struct map *map, + const char *vmlinux, symbol_filter_t filter) +{ + int err = -1, fd; + char symfs_vmlinux[PATH_MAX]; + + snprintf(symfs_vmlinux, sizeof(symfs_vmlinux), "%s%s", + symbol_conf.symfs, vmlinux); + fd = open(symfs_vmlinux, O_RDONLY); + if (fd < 0) + return -1; + + dso__set_long_name(dso, (char *)vmlinux); + dso__set_loaded(dso, map->type); + err = dso__load_sym(dso, map, symfs_vmlinux, fd, filter, 0, 0); + close(fd); + + if (err > 0) + pr_debug("Using %s for symbols\n", symfs_vmlinux); + + return err; +} + +int dso__load_vmlinux_path(struct dso *dso, struct map *map, + symbol_filter_t filter) +{ + int i, err = 0; + char *filename; + + pr_debug("Looking at the vmlinux_path (%d entries long)\n", + vmlinux_path__nr_entries + 1); + + filename = dso__build_id_filename(dso, NULL, 0); + if (filename != NULL) { + err = dso__load_vmlinux(dso, map, filename, filter); + if (err > 0) { + dso__set_long_name(dso, filename); + goto out; + } + free(filename); + } + + for (i = 0; i < vmlinux_path__nr_entries; ++i) { + err = dso__load_vmlinux(dso, map, vmlinux_path[i], filter); + if (err > 0) { + dso__set_long_name(dso, strdup(vmlinux_path[i])); + break; + } + } +out: + return err; +} + +static int dso__load_kernel_sym(struct dso *dso, struct map *map, + symbol_filter_t filter) +{ + int err; + const char *kallsyms_filename = NULL; + char *kallsyms_allocated_filename = NULL; + /* + * Step 1: if the user specified a kallsyms or vmlinux filename, use + * it and only it, reporting errors to the user if it cannot be used. + * + * For instance, try to analyse an ARM perf.data file _without_ a + * build-id, or if the user specifies the wrong path to the right + * vmlinux file, obviously we can't fallback to another vmlinux (a + * x86_86 one, on the machine where analysis is being performed, say), + * or worse, /proc/kallsyms. + * + * If the specified file _has_ a build-id and there is a build-id + * section in the perf.data file, we will still do the expected + * validation in dso__load_vmlinux and will bail out if they don't + * match. + */ + if (symbol_conf.kallsyms_name != NULL) { + kallsyms_filename = symbol_conf.kallsyms_name; + goto do_kallsyms; + } + + if (symbol_conf.vmlinux_name != NULL) { + err = dso__load_vmlinux(dso, map, + symbol_conf.vmlinux_name, filter); + if (err > 0) { + dso__set_long_name(dso, + strdup(symbol_conf.vmlinux_name)); + goto out_fixup; + } + return err; + } + + if (vmlinux_path != NULL) { + err = dso__load_vmlinux_path(dso, map, filter); + if (err > 0) + goto out_fixup; + } + + /* do not try local files if a symfs was given */ + if (symbol_conf.symfs[0] != 0) + return -1; + + /* + * Say the kernel DSO was created when processing the build-id header table, + * we have a build-id, so check if it is the same as the running kernel, + * using it if it is. + */ + if (dso->has_build_id) { + u8 kallsyms_build_id[BUILD_ID_SIZE]; + char sbuild_id[BUILD_ID_SIZE * 2 + 1]; + + if (sysfs__read_build_id("/sys/kernel/notes", kallsyms_build_id, + sizeof(kallsyms_build_id)) == 0) { + if (dso__build_id_equal(dso, kallsyms_build_id)) { + kallsyms_filename = "/proc/kallsyms"; + goto do_kallsyms; + } + } + /* + * Now look if we have it on the build-id cache in + * $HOME/.debug/[kernel.kallsyms]. + */ + build_id__sprintf(dso->build_id, sizeof(dso->build_id), + sbuild_id); + + if (asprintf(&kallsyms_allocated_filename, + "%s/.debug/[kernel.kallsyms]/%s", + getenv("HOME"), sbuild_id) == -1) { + pr_err("Not enough memory for kallsyms file lookup\n"); + return -1; + } + + kallsyms_filename = kallsyms_allocated_filename; + + if (access(kallsyms_filename, F_OK)) { + pr_err("No kallsyms or vmlinux with build-id %s " + "was found\n", sbuild_id); + free(kallsyms_allocated_filename); + return -1; + } + } else { + /* + * Last resort, if we don't have a build-id and couldn't find + * any vmlinux file, try the running kernel kallsyms table. + */ + kallsyms_filename = "/proc/kallsyms"; + } + +do_kallsyms: + err = dso__load_kallsyms(dso, kallsyms_filename, map, filter); + if (err > 0) + pr_debug("Using %s for symbols\n", kallsyms_filename); + free(kallsyms_allocated_filename); + + if (err > 0) { +out_fixup: + if (kallsyms_filename != NULL) + dso__set_long_name(dso, strdup("[kernel.kallsyms]")); + map__fixup_start(map); + map__fixup_end(map); + } + + return err; +} + +static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map, + symbol_filter_t filter) +{ + int err; + const char *kallsyms_filename = NULL; + struct machine *machine; + char path[PATH_MAX]; + + if (!map->groups) { + pr_debug("Guest kernel map hasn't the point to groups\n"); + return -1; + } + machine = map->groups->machine; + + if (machine__is_default_guest(machine)) { + /* + * if the user specified a vmlinux filename, use it and only + * it, reporting errors to the user if it cannot be used. + * Or use file guest_kallsyms inputted by user on commandline + */ + if (symbol_conf.default_guest_vmlinux_name != NULL) { + err = dso__load_vmlinux(dso, map, + symbol_conf.default_guest_vmlinux_name, filter); + goto out_try_fixup; + } + + kallsyms_filename = symbol_conf.default_guest_kallsyms; + if (!kallsyms_filename) + return -1; + } else { + sprintf(path, "%s/proc/kallsyms", machine->root_dir); + kallsyms_filename = path; + } + + err = dso__load_kallsyms(dso, kallsyms_filename, map, filter); + if (err > 0) + pr_debug("Using %s for symbols\n", kallsyms_filename); + +out_try_fixup: + if (err > 0) { + if (kallsyms_filename != NULL) { + machine__mmap_name(machine, path, sizeof(path)); + dso__set_long_name(dso, strdup(path)); + } + map__fixup_start(map); + map__fixup_end(map); + } + + return err; +} + +static void dsos__add(struct list_head *head, struct dso *dso) +{ + list_add_tail(&dso->node, head); +} + +static struct dso *dsos__find(struct list_head *head, const char *name) +{ + struct dso *pos; + + list_for_each_entry(pos, head, node) + if (strcmp(pos->long_name, name) == 0) + return pos; + return NULL; +} + +struct dso *__dsos__findnew(struct list_head *head, const char *name) +{ + struct dso *dso = dsos__find(head, name); + + if (!dso) { + dso = dso__new(name); + if (dso != NULL) { + dsos__add(head, dso); + dso__set_basename(dso); + } + } + + return dso; +} + +size_t __dsos__fprintf(struct list_head *head, FILE *fp) +{ + struct dso *pos; + size_t ret = 0; + + list_for_each_entry(pos, head, node) { + int i; + for (i = 0; i < MAP__NR_TYPES; ++i) + ret += dso__fprintf(pos, i, fp); + } + + return ret; +} + +size_t machines__fprintf_dsos(struct rb_root *machines, FILE *fp) +{ + struct rb_node *nd; + size_t ret = 0; + + for (nd = rb_first(machines); nd; nd = rb_next(nd)) { + struct machine *pos = rb_entry(nd, struct machine, rb_node); + ret += __dsos__fprintf(&pos->kernel_dsos, fp); + ret += __dsos__fprintf(&pos->user_dsos, fp); + } + + return ret; +} + +static size_t __dsos__fprintf_buildid(struct list_head *head, FILE *fp, + bool with_hits) +{ + struct dso *pos; + size_t ret = 0; + + list_for_each_entry(pos, head, node) { + if (with_hits && !pos->hit) + continue; + ret += dso__fprintf_buildid(pos, fp); + ret += fprintf(fp, " %s\n", pos->long_name); + } + return ret; +} + +size_t machine__fprintf_dsos_buildid(struct machine *machine, FILE *fp, + bool with_hits) +{ + return __dsos__fprintf_buildid(&machine->kernel_dsos, fp, with_hits) + + __dsos__fprintf_buildid(&machine->user_dsos, fp, with_hits); +} + +size_t machines__fprintf_dsos_buildid(struct rb_root *machines, + FILE *fp, bool with_hits) +{ + struct rb_node *nd; + size_t ret = 0; + + for (nd = rb_first(machines); nd; nd = rb_next(nd)) { + struct machine *pos = rb_entry(nd, struct machine, rb_node); + ret += machine__fprintf_dsos_buildid(pos, fp, with_hits); + } + return ret; +} + +struct dso *dso__new_kernel(const char *name) +{ + struct dso *dso = dso__new(name ?: "[kernel.kallsyms]"); + + if (dso != NULL) { + dso__set_short_name(dso, "[kernel]"); + dso->kernel = DSO_TYPE_KERNEL; + } + + return dso; +} + +static struct dso *dso__new_guest_kernel(struct machine *machine, + const char *name) +{ + char bf[PATH_MAX]; + struct dso *dso = dso__new(name ?: machine__mmap_name(machine, bf, + sizeof(bf))); + if (dso != NULL) { + dso__set_short_name(dso, "[guest.kernel]"); + dso->kernel = DSO_TYPE_GUEST_KERNEL; + } + + return dso; +} + +void dso__read_running_kernel_build_id(struct dso *dso, struct machine *machine) +{ + char path[PATH_MAX]; + + if (machine__is_default_guest(machine)) + return; + sprintf(path, "%s/sys/kernel/notes", machine->root_dir); + if (sysfs__read_build_id(path, dso->build_id, + sizeof(dso->build_id)) == 0) + dso->has_build_id = true; +} + +static struct dso *machine__create_kernel(struct machine *machine) +{ + const char *vmlinux_name = NULL; + struct dso *kernel; + + if (machine__is_host(machine)) { + vmlinux_name = symbol_conf.vmlinux_name; + kernel = dso__new_kernel(vmlinux_name); + } else { + if (machine__is_default_guest(machine)) + vmlinux_name = symbol_conf.default_guest_vmlinux_name; + kernel = dso__new_guest_kernel(machine, vmlinux_name); + } + + if (kernel != NULL) { + dso__read_running_kernel_build_id(kernel, machine); + dsos__add(&machine->kernel_dsos, kernel); + } + return kernel; +} + +struct process_args { + u64 start; +}; + +static int symbol__in_kernel(void *arg, const char *name, + char type __used, u64 start, u64 end __used) +{ + struct process_args *args = arg; + + if (strchr(name, '[')) + return 0; + + args->start = start; + return 1; +} + +/* Figure out the start address of kernel map from /proc/kallsyms */ +static u64 machine__get_kernel_start_addr(struct machine *machine) +{ + const char *filename; + char path[PATH_MAX]; + struct process_args args; + + if (machine__is_host(machine)) { + filename = "/proc/kallsyms"; + } else { + if (machine__is_default_guest(machine)) + filename = (char *)symbol_conf.default_guest_kallsyms; + else { + sprintf(path, "%s/proc/kallsyms", machine->root_dir); + filename = path; + } + } + + if (symbol__restricted_filename(filename, "/proc/kallsyms")) + return 0; + + if (kallsyms__parse(filename, &args, symbol__in_kernel) <= 0) + return 0; + + return args.start; +} + +int __machine__create_kernel_maps(struct machine *machine, struct dso *kernel) +{ + enum map_type type; + u64 start = machine__get_kernel_start_addr(machine); + + for (type = 0; type < MAP__NR_TYPES; ++type) { + struct kmap *kmap; + + machine->vmlinux_maps[type] = map__new2(start, kernel, type); + if (machine->vmlinux_maps[type] == NULL) + return -1; + + machine->vmlinux_maps[type]->map_ip = + machine->vmlinux_maps[type]->unmap_ip = + identity__map_ip; + kmap = map__kmap(machine->vmlinux_maps[type]); + kmap->kmaps = &machine->kmaps; + map_groups__insert(&machine->kmaps, + machine->vmlinux_maps[type]); + } + + return 0; +} + +void machine__destroy_kernel_maps(struct machine *machine) +{ + enum map_type type; + + for (type = 0; type < MAP__NR_TYPES; ++type) { + struct kmap *kmap; + + if (machine->vmlinux_maps[type] == NULL) + continue; + + kmap = map__kmap(machine->vmlinux_maps[type]); + map_groups__remove(&machine->kmaps, + machine->vmlinux_maps[type]); + if (kmap->ref_reloc_sym) { + /* + * ref_reloc_sym is shared among all maps, so free just + * on one of them. + */ + if (type == MAP__FUNCTION) { + free((char *)kmap->ref_reloc_sym->name); + kmap->ref_reloc_sym->name = NULL; + free(kmap->ref_reloc_sym); + } + kmap->ref_reloc_sym = NULL; + } + + map__delete(machine->vmlinux_maps[type]); + machine->vmlinux_maps[type] = NULL; + } +} + +int machine__create_kernel_maps(struct machine *machine) +{ + struct dso *kernel = machine__create_kernel(machine); + + if (kernel == NULL || + __machine__create_kernel_maps(machine, kernel) < 0) + return -1; + + if (symbol_conf.use_modules && machine__create_modules(machine) < 0) + pr_debug("Problems creating module maps, continuing anyway...\n"); + /* + * Now that we have all the maps created, just set the ->end of them: + */ + map_groups__fixup_end(&machine->kmaps); + return 0; +} + +static void vmlinux_path__exit(void) +{ + while (--vmlinux_path__nr_entries >= 0) { + free(vmlinux_path[vmlinux_path__nr_entries]); + vmlinux_path[vmlinux_path__nr_entries] = NULL; + } + + free(vmlinux_path); + vmlinux_path = NULL; +} + +static int vmlinux_path__init(void) +{ + struct utsname uts; + char bf[PATH_MAX]; + + vmlinux_path = malloc(sizeof(char *) * 5); + if (vmlinux_path == NULL) + return -1; + + vmlinux_path[vmlinux_path__nr_entries] = strdup("vmlinux"); + if (vmlinux_path[vmlinux_path__nr_entries] == NULL) + goto out_fail; + ++vmlinux_path__nr_entries; + vmlinux_path[vmlinux_path__nr_entries] = strdup("/boot/vmlinux"); + if (vmlinux_path[vmlinux_path__nr_entries] == NULL) + goto out_fail; + ++vmlinux_path__nr_entries; + + /* only try running kernel version if no symfs was given */ + if (symbol_conf.symfs[0] != 0) + return 0; + + if (uname(&uts) < 0) + return -1; + + snprintf(bf, sizeof(bf), "/boot/vmlinux-%s", uts.release); + vmlinux_path[vmlinux_path__nr_entries] = strdup(bf); + if (vmlinux_path[vmlinux_path__nr_entries] == NULL) + goto out_fail; + ++vmlinux_path__nr_entries; + snprintf(bf, sizeof(bf), "/lib/modules/%s/build/vmlinux", uts.release); + vmlinux_path[vmlinux_path__nr_entries] = strdup(bf); + if (vmlinux_path[vmlinux_path__nr_entries] == NULL) + goto out_fail; + ++vmlinux_path__nr_entries; + snprintf(bf, sizeof(bf), "/usr/lib/debug/lib/modules/%s/vmlinux", + uts.release); + vmlinux_path[vmlinux_path__nr_entries] = strdup(bf); + if (vmlinux_path[vmlinux_path__nr_entries] == NULL) + goto out_fail; + ++vmlinux_path__nr_entries; + + return 0; + +out_fail: + vmlinux_path__exit(); + return -1; +} + +size_t machine__fprintf_vmlinux_path(struct machine *machine, FILE *fp) +{ + int i; + size_t printed = 0; + struct dso *kdso = machine->vmlinux_maps[MAP__FUNCTION]->dso; + + if (kdso->has_build_id) { + char filename[PATH_MAX]; + if (dso__build_id_filename(kdso, filename, sizeof(filename))) + printed += fprintf(fp, "[0] %s\n", filename); + } + + for (i = 0; i < vmlinux_path__nr_entries; ++i) + printed += fprintf(fp, "[%d] %s\n", + i + kdso->has_build_id, vmlinux_path[i]); + + return printed; +} + +static int setup_list(struct strlist **list, const char *list_str, + const char *list_name) +{ + if (list_str == NULL) + return 0; + + *list = strlist__new(true, list_str); + if (!*list) { + pr_err("problems parsing %s list\n", list_name); + return -1; + } + return 0; +} + +static bool symbol__read_kptr_restrict(void) +{ + bool value = false; + + if (geteuid() != 0) { + FILE *fp = fopen("/proc/sys/kernel/kptr_restrict", "r"); + if (fp != NULL) { + char line[8]; + + if (fgets(line, sizeof(line), fp) != NULL) + value = atoi(line) != 0; + + fclose(fp); + } + } + + return value; +} + +int symbol__init(void) +{ + const char *symfs; + + if (symbol_conf.initialized) + return 0; + + symbol_conf.priv_size = ALIGN(symbol_conf.priv_size, sizeof(u64)); + + elf_version(EV_CURRENT); + if (symbol_conf.sort_by_name) + symbol_conf.priv_size += (sizeof(struct symbol_name_rb_node) - + sizeof(struct symbol)); + + if (symbol_conf.try_vmlinux_path && vmlinux_path__init() < 0) + return -1; + + if (symbol_conf.field_sep && *symbol_conf.field_sep == '.') { + pr_err("'.' is the only non valid --field-separator argument\n"); + return -1; + } + + if (setup_list(&symbol_conf.dso_list, + symbol_conf.dso_list_str, "dso") < 0) + return -1; + + if (setup_list(&symbol_conf.comm_list, + symbol_conf.comm_list_str, "comm") < 0) + goto out_free_dso_list; + + if (setup_list(&symbol_conf.sym_list, + symbol_conf.sym_list_str, "symbol") < 0) + goto out_free_comm_list; + + /* + * A path to symbols of "/" is identical to "" + * reset here for simplicity. + */ + symfs = realpath(symbol_conf.symfs, NULL); + if (symfs == NULL) + symfs = symbol_conf.symfs; + if (strcmp(symfs, "/") == 0) + symbol_conf.symfs = ""; + if (symfs != symbol_conf.symfs) + free((void *)symfs); + + symbol_conf.kptr_restrict = symbol__read_kptr_restrict(); + + symbol_conf.initialized = true; + return 0; + +out_free_dso_list: + strlist__delete(symbol_conf.dso_list); +out_free_comm_list: + strlist__delete(symbol_conf.comm_list); + return -1; +} + +void symbol__exit(void) +{ + if (!symbol_conf.initialized) + return; + strlist__delete(symbol_conf.sym_list); + strlist__delete(symbol_conf.dso_list); + strlist__delete(symbol_conf.comm_list); + vmlinux_path__exit(); + symbol_conf.sym_list = symbol_conf.dso_list = symbol_conf.comm_list = NULL; + symbol_conf.initialized = false; +} + +int machines__create_kernel_maps(struct rb_root *machines, pid_t pid) +{ + struct machine *machine = machines__findnew(machines, pid); + + if (machine == NULL) + return -1; + + return machine__create_kernel_maps(machine); +} + +static int hex(char ch) +{ + if ((ch >= '0') && (ch <= '9')) + return ch - '0'; + if ((ch >= 'a') && (ch <= 'f')) + return ch - 'a' + 10; + if ((ch >= 'A') && (ch <= 'F')) + return ch - 'A' + 10; + return -1; +} + +/* + * While we find nice hex chars, build a long_val. + * Return number of chars processed. + */ +int hex2u64(const char *ptr, u64 *long_val) +{ + const char *p = ptr; + *long_val = 0; + + while (*p) { + const int hex_val = hex(*p); + + if (hex_val < 0) + break; + + *long_val = (*long_val << 4) | hex_val; + p++; + } + + return p - ptr; +} + +char *strxfrchar(char *s, char from, char to) +{ + char *p = s; + + while ((p = strchr(p, from)) != NULL) + *p++ = to; + + return s; +} + +int machines__create_guest_kernel_maps(struct rb_root *machines) +{ + int ret = 0; + struct dirent **namelist = NULL; + int i, items = 0; + char path[PATH_MAX]; + pid_t pid; + + if (symbol_conf.default_guest_vmlinux_name || + symbol_conf.default_guest_modules || + symbol_conf.default_guest_kallsyms) { + machines__create_kernel_maps(machines, DEFAULT_GUEST_KERNEL_ID); + } + + if (symbol_conf.guestmount) { + items = scandir(symbol_conf.guestmount, &namelist, NULL, NULL); + if (items <= 0) + return -ENOENT; + for (i = 0; i < items; i++) { + if (!isdigit(namelist[i]->d_name[0])) { + /* Filter out . and .. */ + continue; + } + pid = atoi(namelist[i]->d_name); + sprintf(path, "%s/%s/proc/kallsyms", + symbol_conf.guestmount, + namelist[i]->d_name); + ret = access(path, R_OK); + if (ret) { + pr_debug("Can't access file %s\n", path); + goto failure; + } + machines__create_kernel_maps(machines, pid); + } +failure: + free(namelist); + } + + return ret; +} + +void machines__destroy_guest_kernel_maps(struct rb_root *machines) +{ + struct rb_node *next = rb_first(machines); + + while (next) { + struct machine *pos = rb_entry(next, struct machine, rb_node); + + next = rb_next(&pos->rb_node); + rb_erase(&pos->rb_node, machines); + machine__delete(pos); + } +} + +int machine__load_kallsyms(struct machine *machine, const char *filename, + enum map_type type, symbol_filter_t filter) +{ + struct map *map = machine->vmlinux_maps[type]; + int ret = dso__load_kallsyms(map->dso, filename, map, filter); + + if (ret > 0) { + dso__set_loaded(map->dso, type); + /* + * Since /proc/kallsyms will have multiple sessions for the + * kernel, with modules between them, fixup the end of all + * sections. + */ + __map_groups__fixup_end(&machine->kmaps, type); + } + + return ret; +} + +int machine__load_vmlinux_path(struct machine *machine, enum map_type type, + symbol_filter_t filter) +{ + struct map *map = machine->vmlinux_maps[type]; + int ret = dso__load_vmlinux_path(map->dso, map, filter); + + if (ret > 0) { + dso__set_loaded(map->dso, type); + map__reloc_vmlinux(map); + } + + return ret; +} diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h new file mode 100644 index 00000000..325ee36a --- /dev/null +++ b/tools/perf/util/symbol.h @@ -0,0 +1,245 @@ +#ifndef __PERF_SYMBOL +#define __PERF_SYMBOL 1 + +#include <linux/types.h> +#include <stdbool.h> +#include <stdint.h> +#include "map.h" +#include <linux/list.h> +#include <linux/rbtree.h> +#include <stdio.h> + +#ifdef HAVE_CPLUS_DEMANGLE +extern char *cplus_demangle(const char *, int); + +static inline char *bfd_demangle(void __used *v, const char *c, int i) +{ + return cplus_demangle(c, i); +} +#else +#ifdef NO_DEMANGLE +static inline char *bfd_demangle(void __used *v, const char __used *c, + int __used i) +{ + return NULL; +} +#else +#include <bfd.h> +#endif +#endif + +int hex2u64(const char *ptr, u64 *val); +char *strxfrchar(char *s, char from, char to); + +/* + * libelf 0.8.x and earlier do not support ELF_C_READ_MMAP; + * for newer versions we can use mmap to reduce memory usage: + */ +#ifdef LIBELF_NO_MMAP +# define PERF_ELF_C_READ_MMAP ELF_C_READ +#else +# define PERF_ELF_C_READ_MMAP ELF_C_READ_MMAP +#endif + +#ifndef DMGL_PARAMS +#define DMGL_PARAMS (1 << 0) /* Include function args */ +#define DMGL_ANSI (1 << 1) /* Include const, volatile, etc */ +#endif + +#define BUILD_ID_SIZE 20 + +/** struct symbol - symtab entry + * + * @ignore - resolvable but tools ignore it (e.g. idle routines) + */ +struct symbol { + struct rb_node rb_node; + u64 start; + u64 end; + u16 namelen; + u8 binding; + bool ignore; + char name[0]; +}; + +void symbol__delete(struct symbol *sym); + +struct strlist; + +struct symbol_conf { + unsigned short priv_size; + bool try_vmlinux_path, + use_modules, + sort_by_name, + show_nr_samples, + use_callchain, + exclude_other, + show_cpu_utilization, + initialized, + kptr_restrict; + const char *vmlinux_name, + *kallsyms_name, + *source_prefix, + *field_sep; + const char *default_guest_vmlinux_name, + *default_guest_kallsyms, + *default_guest_modules; + const char *guestmount; + const char *dso_list_str, + *comm_list_str, + *sym_list_str, + *col_width_list_str; + struct strlist *dso_list, + *comm_list, + *sym_list; + const char *symfs; +}; + +extern struct symbol_conf symbol_conf; + +static inline void *symbol__priv(struct symbol *sym) +{ + return ((void *)sym) - symbol_conf.priv_size; +} + +struct ref_reloc_sym { + const char *name; + u64 addr; + u64 unrelocated_addr; +}; + +struct map_symbol { + struct map *map; + struct symbol *sym; + bool unfolded; + bool has_children; +}; + +struct addr_location { + struct thread *thread; + struct map *map; + struct symbol *sym; + u64 addr; + char level; + bool filtered; + u8 cpumode; + s32 cpu; +}; + +enum dso_kernel_type { + DSO_TYPE_USER = 0, + DSO_TYPE_KERNEL, + DSO_TYPE_GUEST_KERNEL +}; + +struct dso { + struct list_head node; + struct rb_root symbols[MAP__NR_TYPES]; + struct rb_root symbol_names[MAP__NR_TYPES]; + enum dso_kernel_type kernel; + u8 adjust_symbols:1; + u8 has_build_id:1; + u8 hit:1; + u8 annotate_warned:1; + u8 sname_alloc:1; + u8 lname_alloc:1; + unsigned char symtab_type; + u8 sorted_by_name; + u8 loaded; + u8 build_id[BUILD_ID_SIZE]; + const char *short_name; + char *long_name; + u16 long_name_len; + u16 short_name_len; + char name[0]; +}; + +struct dso *dso__new(const char *name); +struct dso *dso__new_kernel(const char *name); +void dso__delete(struct dso *dso); + +int dso__name_len(const struct dso *dso); + +bool dso__loaded(const struct dso *dso, enum map_type type); +bool dso__sorted_by_name(const struct dso *dso, enum map_type type); + +static inline void dso__set_loaded(struct dso *dso, enum map_type type) +{ + dso->loaded |= (1 << type); +} + +void dso__sort_by_name(struct dso *dso, enum map_type type); + +struct dso *__dsos__findnew(struct list_head *head, const char *name); + +int dso__load(struct dso *dso, struct map *map, symbol_filter_t filter); +int dso__load_vmlinux(struct dso *dso, struct map *map, + const char *vmlinux, symbol_filter_t filter); +int dso__load_vmlinux_path(struct dso *dso, struct map *map, + symbol_filter_t filter); +int dso__load_kallsyms(struct dso *dso, const char *filename, struct map *map, + symbol_filter_t filter); +int machine__load_kallsyms(struct machine *machine, const char *filename, + enum map_type type, symbol_filter_t filter); +int machine__load_vmlinux_path(struct machine *machine, enum map_type type, + symbol_filter_t filter); + +size_t __dsos__fprintf(struct list_head *head, FILE *fp); + +size_t machine__fprintf_dsos_buildid(struct machine *machine, + FILE *fp, bool with_hits); +size_t machines__fprintf_dsos(struct rb_root *machines, FILE *fp); +size_t machines__fprintf_dsos_buildid(struct rb_root *machines, + FILE *fp, bool with_hits); +size_t dso__fprintf_buildid(struct dso *dso, FILE *fp); +size_t dso__fprintf_symbols_by_name(struct dso *dso, + enum map_type type, FILE *fp); +size_t dso__fprintf(struct dso *dso, enum map_type type, FILE *fp); + +enum symtab_type { + SYMTAB__KALLSYMS = 0, + SYMTAB__GUEST_KALLSYMS, + SYMTAB__JAVA_JIT, + SYMTAB__BUILD_ID_CACHE, + SYMTAB__FEDORA_DEBUGINFO, + SYMTAB__UBUNTU_DEBUGINFO, + SYMTAB__BUILDID_DEBUGINFO, + SYMTAB__SYSTEM_PATH_DSO, + SYMTAB__GUEST_KMODULE, + SYMTAB__SYSTEM_PATH_KMODULE, + SYMTAB__NOT_FOUND, +}; + +char dso__symtab_origin(const struct dso *dso); +void dso__set_long_name(struct dso *dso, char *name); +void dso__set_build_id(struct dso *dso, void *build_id); +void dso__read_running_kernel_build_id(struct dso *dso, + struct machine *machine); +struct symbol *dso__find_symbol(struct dso *dso, enum map_type type, + u64 addr); +struct symbol *dso__find_symbol_by_name(struct dso *dso, enum map_type type, + const char *name); + +int filename__read_build_id(const char *filename, void *bf, size_t size); +int sysfs__read_build_id(const char *filename, void *bf, size_t size); +bool __dsos__read_build_ids(struct list_head *head, bool with_hits); +int build_id__sprintf(const u8 *build_id, int len, char *bf); +int kallsyms__parse(const char *filename, void *arg, + int (*process_symbol)(void *arg, const char *name, + char type, u64 start, u64 end)); + +void machine__destroy_kernel_maps(struct machine *machine); +int __machine__create_kernel_maps(struct machine *machine, struct dso *kernel); +int machine__create_kernel_maps(struct machine *machine); + +int machines__create_kernel_maps(struct rb_root *machines, pid_t pid); +int machines__create_guest_kernel_maps(struct rb_root *machines); +void machines__destroy_guest_kernel_maps(struct rb_root *machines); + +int symbol__init(void); +void symbol__exit(void); +bool symbol_type__is_a(char symbol_type, enum map_type map_type); + +size_t machine__fprintf_vmlinux_path(struct machine *machine, FILE *fp); + +#endif /* __PERF_SYMBOL */ diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c new file mode 100644 index 00000000..d5d3b222 --- /dev/null +++ b/tools/perf/util/thread.c @@ -0,0 +1,140 @@ +#include "../perf.h" +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include "session.h" +#include "thread.h" +#include "util.h" +#include "debug.h" + +static struct thread *thread__new(pid_t pid) +{ + struct thread *self = zalloc(sizeof(*self)); + + if (self != NULL) { + map_groups__init(&self->mg); + self->pid = pid; + self->comm = malloc(32); + if (self->comm) + snprintf(self->comm, 32, ":%d", self->pid); + } + + return self; +} + +void thread__delete(struct thread *self) +{ + map_groups__exit(&self->mg); + free(self->comm); + free(self); +} + +int thread__set_comm(struct thread *self, const char *comm) +{ + int err; + + if (self->comm) + free(self->comm); + self->comm = strdup(comm); + err = self->comm == NULL ? -ENOMEM : 0; + if (!err) { + self->comm_set = true; + map_groups__flush(&self->mg); + } + return err; +} + +int thread__comm_len(struct thread *self) +{ + if (!self->comm_len) { + if (!self->comm) + return 0; + self->comm_len = strlen(self->comm); + } + + return self->comm_len; +} + +static size_t thread__fprintf(struct thread *self, FILE *fp) +{ + return fprintf(fp, "Thread %d %s\n", self->pid, self->comm) + + map_groups__fprintf(&self->mg, verbose, fp); +} + +struct thread *perf_session__findnew(struct perf_session *self, pid_t pid) +{ + struct rb_node **p = &self->threads.rb_node; + struct rb_node *parent = NULL; + struct thread *th; + + /* + * Font-end cache - PID lookups come in blocks, + * so most of the time we dont have to look up + * the full rbtree: + */ + if (self->last_match && self->last_match->pid == pid) + return self->last_match; + + while (*p != NULL) { + parent = *p; + th = rb_entry(parent, struct thread, rb_node); + + if (th->pid == pid) { + self->last_match = th; + return th; + } + + if (pid < th->pid) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + th = thread__new(pid); + if (th != NULL) { + rb_link_node(&th->rb_node, parent, p); + rb_insert_color(&th->rb_node, &self->threads); + self->last_match = th; + } + + return th; +} + +void thread__insert_map(struct thread *self, struct map *map) +{ + map_groups__fixup_overlappings(&self->mg, map, verbose, stderr); + map_groups__insert(&self->mg, map); +} + +int thread__fork(struct thread *self, struct thread *parent) +{ + int i; + + if (parent->comm_set) { + if (self->comm) + free(self->comm); + self->comm = strdup(parent->comm); + if (!self->comm) + return -ENOMEM; + self->comm_set = true; + } + + for (i = 0; i < MAP__NR_TYPES; ++i) + if (map_groups__clone(&self->mg, &parent->mg, i) < 0) + return -ENOMEM; + return 0; +} + +size_t perf_session__fprintf(struct perf_session *self, FILE *fp) +{ + size_t ret = 0; + struct rb_node *nd; + + for (nd = rb_first(&self->threads); nd; nd = rb_next(nd)) { + struct thread *pos = rb_entry(nd, struct thread, rb_node); + + ret += thread__fprintf(pos, fp); + } + + return ret; +} diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h new file mode 100644 index 00000000..e5f2401c --- /dev/null +++ b/tools/perf/util/thread.h @@ -0,0 +1,48 @@ +#ifndef __PERF_THREAD_H +#define __PERF_THREAD_H + +#include <linux/rbtree.h> +#include <unistd.h> +#include "symbol.h" + +struct thread { + union { + struct rb_node rb_node; + struct list_head node; + }; + struct map_groups mg; + pid_t pid; + char shortname[3]; + bool comm_set; + char *comm; + int comm_len; +}; + +struct perf_session; + +void thread__delete(struct thread *self); + +int thread__set_comm(struct thread *self, const char *comm); +int thread__comm_len(struct thread *self); +struct thread *perf_session__findnew(struct perf_session *self, pid_t pid); +void thread__insert_map(struct thread *self, struct map *map); +int thread__fork(struct thread *self, struct thread *parent); +size_t perf_session__fprintf(struct perf_session *self, FILE *fp); + +static inline struct map *thread__find_map(struct thread *self, + enum map_type type, u64 addr) +{ + return self ? map_groups__find(&self->mg, type, addr) : NULL; +} + +void thread__find_addr_map(struct thread *self, + struct perf_session *session, u8 cpumode, + enum map_type type, pid_t pid, u64 addr, + struct addr_location *al); + +void thread__find_addr_location(struct thread *self, + struct perf_session *session, u8 cpumode, + enum map_type type, pid_t pid, u64 addr, + struct addr_location *al, + symbol_filter_t filter); +#endif /* __PERF_THREAD_H */ diff --git a/tools/perf/util/thread_map.c b/tools/perf/util/thread_map.c new file mode 100644 index 00000000..a5df131b --- /dev/null +++ b/tools/perf/util/thread_map.c @@ -0,0 +1,64 @@ +#include <dirent.h> +#include <stdlib.h> +#include <stdio.h> +#include "thread_map.h" + +/* Skip "." and ".." directories */ +static int filter(const struct dirent *dir) +{ + if (dir->d_name[0] == '.') + return 0; + else + return 1; +} + +struct thread_map *thread_map__new_by_pid(pid_t pid) +{ + struct thread_map *threads; + char name[256]; + int items; + struct dirent **namelist = NULL; + int i; + + sprintf(name, "/proc/%d/task", pid); + items = scandir(name, &namelist, filter, NULL); + if (items <= 0) + return NULL; + + threads = malloc(sizeof(*threads) + sizeof(pid_t) * items); + if (threads != NULL) { + for (i = 0; i < items; i++) + threads->map[i] = atoi(namelist[i]->d_name); + threads->nr = items; + } + + for (i=0; i<items; i++) + free(namelist[i]); + free(namelist); + + return threads; +} + +struct thread_map *thread_map__new_by_tid(pid_t tid) +{ + struct thread_map *threads = malloc(sizeof(*threads) + sizeof(pid_t)); + + if (threads != NULL) { + threads->map[0] = tid; + threads->nr = 1; + } + + return threads; +} + +struct thread_map *thread_map__new(pid_t pid, pid_t tid) +{ + if (pid != -1) + return thread_map__new_by_pid(pid); + return thread_map__new_by_tid(tid); +} + +void thread_map__delete(struct thread_map *threads) +{ + free(threads); +} diff --git a/tools/perf/util/thread_map.h b/tools/perf/util/thread_map.h new file mode 100644 index 00000000..3cb90731 --- /dev/null +++ b/tools/perf/util/thread_map.h @@ -0,0 +1,15 @@ +#ifndef __PERF_THREAD_MAP_H +#define __PERF_THREAD_MAP_H + +#include <sys/types.h> + +struct thread_map { + int nr; + int map[]; +}; + +struct thread_map *thread_map__new_by_pid(pid_t pid); +struct thread_map *thread_map__new_by_tid(pid_t tid); +struct thread_map *thread_map__new(pid_t pid, pid_t tid); +void thread_map__delete(struct thread_map *threads); +#endif /* __PERF_THREAD_MAP_H */ diff --git a/tools/perf/util/top.c b/tools/perf/util/top.c new file mode 100644 index 00000000..a11f6073 --- /dev/null +++ b/tools/perf/util/top.c @@ -0,0 +1,238 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Refactored from builtin-top.c, see that files for further copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ + +#include "cpumap.h" +#include "event.h" +#include "evlist.h" +#include "evsel.h" +#include "parse-events.h" +#include "symbol.h" +#include "top.h" +#include <inttypes.h> + +/* + * Ordering weight: count-1 * count-2 * ... / count-n + */ +static double sym_weight(const struct sym_entry *sym, struct perf_top *top) +{ + double weight = sym->snap_count; + int counter; + + if (!top->display_weighted) + return weight; + + for (counter = 1; counter < top->evlist->nr_entries - 1; counter++) + weight *= sym->count[counter]; + + weight /= (sym->count[counter] + 1); + + return weight; +} + +static void perf_top__remove_active_sym(struct perf_top *top, struct sym_entry *syme) +{ + pthread_mutex_lock(&top->active_symbols_lock); + list_del_init(&syme->node); + pthread_mutex_unlock(&top->active_symbols_lock); +} + +static void rb_insert_active_sym(struct rb_root *tree, struct sym_entry *se) +{ + struct rb_node **p = &tree->rb_node; + struct rb_node *parent = NULL; + struct sym_entry *iter; + + while (*p != NULL) { + parent = *p; + iter = rb_entry(parent, struct sym_entry, rb_node); + + if (se->weight > iter->weight) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&se->rb_node, parent, p); + rb_insert_color(&se->rb_node, tree); +} + +#define SNPRINTF(buf, size, fmt, args...) \ +({ \ + size_t r = snprintf(buf, size, fmt, ## args); \ + r > size ? size : r; \ +}) + +size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size) +{ + struct perf_evsel *counter; + float samples_per_sec = top->samples / top->delay_secs; + float ksamples_per_sec = top->kernel_samples / top->delay_secs; + float esamples_percent = (100.0 * top->exact_samples) / top->samples; + size_t ret = 0; + + if (!perf_guest) { + ret = SNPRINTF(bf, size, + " PerfTop:%8.0f irqs/sec kernel:%4.1f%%" + " exact: %4.1f%% [", samples_per_sec, + 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) / + samples_per_sec)), + esamples_percent); + } else { + float us_samples_per_sec = top->us_samples / top->delay_secs; + float guest_kernel_samples_per_sec = top->guest_kernel_samples / top->delay_secs; + float guest_us_samples_per_sec = top->guest_us_samples / top->delay_secs; + + ret = SNPRINTF(bf, size, + " PerfTop:%8.0f irqs/sec kernel:%4.1f%% us:%4.1f%%" + " guest kernel:%4.1f%% guest us:%4.1f%%" + " exact: %4.1f%% [", samples_per_sec, + 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) / + samples_per_sec)), + 100.0 - (100.0 * ((samples_per_sec - us_samples_per_sec) / + samples_per_sec)), + 100.0 - (100.0 * ((samples_per_sec - + guest_kernel_samples_per_sec) / + samples_per_sec)), + 100.0 - (100.0 * ((samples_per_sec - + guest_us_samples_per_sec) / + samples_per_sec)), + esamples_percent); + } + + if (top->evlist->nr_entries == 1 || !top->display_weighted) { + struct perf_evsel *first; + first = list_entry(top->evlist->entries.next, struct perf_evsel, node); + ret += SNPRINTF(bf + ret, size - ret, "%" PRIu64 "%s ", + (uint64_t)first->attr.sample_period, + top->freq ? "Hz" : ""); + } + + if (!top->display_weighted) { + ret += SNPRINTF(bf + ret, size - ret, "%s", + event_name(top->sym_evsel)); + } else { + /* + * Don't let events eat all the space. Leaving 30 bytes + * for the rest should be enough. + */ + size_t last_pos = size - 30; + + list_for_each_entry(counter, &top->evlist->entries, node) { + ret += SNPRINTF(bf + ret, size - ret, "%s%s", + counter->idx ? "/" : "", + event_name(counter)); + if (ret > last_pos) { + sprintf(bf + last_pos - 3, ".."); + ret = last_pos - 1; + break; + } + } + } + + ret += SNPRINTF(bf + ret, size - ret, "], "); + + if (top->target_pid != -1) + ret += SNPRINTF(bf + ret, size - ret, " (target_pid: %d", + top->target_pid); + else if (top->target_tid != -1) + ret += SNPRINTF(bf + ret, size - ret, " (target_tid: %d", + top->target_tid); + else + ret += SNPRINTF(bf + ret, size - ret, " (all"); + + if (top->cpu_list) + ret += SNPRINTF(bf + ret, size - ret, ", CPU%s: %s)", + top->evlist->cpus->nr > 1 ? "s" : "", top->cpu_list); + else { + if (top->target_tid != -1) + ret += SNPRINTF(bf + ret, size - ret, ")"); + else + ret += SNPRINTF(bf + ret, size - ret, ", %d CPU%s)", + top->evlist->cpus->nr, + top->evlist->cpus->nr > 1 ? "s" : ""); + } + + return ret; +} + +void perf_top__reset_sample_counters(struct perf_top *top) +{ + top->samples = top->us_samples = top->kernel_samples = + top->exact_samples = top->guest_kernel_samples = + top->guest_us_samples = 0; +} + +float perf_top__decay_samples(struct perf_top *top, struct rb_root *root) +{ + struct sym_entry *syme, *n; + float sum_ksamples = 0.0; + int snap = !top->display_weighted ? top->sym_evsel->idx : 0, j; + + /* Sort the active symbols */ + pthread_mutex_lock(&top->active_symbols_lock); + syme = list_entry(top->active_symbols.next, struct sym_entry, node); + pthread_mutex_unlock(&top->active_symbols_lock); + + top->rb_entries = 0; + list_for_each_entry_safe_from(syme, n, &top->active_symbols, node) { + syme->snap_count = syme->count[snap]; + if (syme->snap_count != 0) { + + if ((top->hide_user_symbols && + syme->map->dso->kernel == DSO_TYPE_USER) || + (top->hide_kernel_symbols && + syme->map->dso->kernel == DSO_TYPE_KERNEL)) { + perf_top__remove_active_sym(top, syme); + continue; + } + syme->weight = sym_weight(syme, top); + + if ((int)syme->snap_count >= top->count_filter) { + rb_insert_active_sym(root, syme); + ++top->rb_entries; + } + sum_ksamples += syme->snap_count; + + for (j = 0; j < top->evlist->nr_entries; j++) + syme->count[j] = top->zero ? 0 : syme->count[j] * 7 / 8; + } else + perf_top__remove_active_sym(top, syme); + } + + return sum_ksamples; +} + +/* + * Find the longest symbol name that will be displayed + */ +void perf_top__find_widths(struct perf_top *top, struct rb_root *root, + int *dso_width, int *dso_short_width, int *sym_width) +{ + struct rb_node *nd; + int printed = 0; + + *sym_width = *dso_width = *dso_short_width = 0; + + for (nd = rb_first(root); nd; nd = rb_next(nd)) { + struct sym_entry *syme = rb_entry(nd, struct sym_entry, rb_node); + struct symbol *sym = sym_entry__symbol(syme); + + if (++printed > top->print_entries || + (int)syme->snap_count < top->count_filter) + continue; + + if (syme->map->dso->long_name_len > *dso_width) + *dso_width = syme->map->dso->long_name_len; + + if (syme->map->dso->short_name_len > *dso_short_width) + *dso_short_width = syme->map->dso->short_name_len; + + if (sym->namelen > *sym_width) + *sym_width = sym->namelen; + } +} diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h new file mode 100644 index 00000000..bfbf95bc --- /dev/null +++ b/tools/perf/util/top.h @@ -0,0 +1,64 @@ +#ifndef __PERF_TOP_H +#define __PERF_TOP_H 1 + +#include "types.h" +#include "../perf.h" +#include <stddef.h> +#include <pthread.h> +#include <linux/list.h> +#include <linux/rbtree.h> + +struct perf_evlist; +struct perf_evsel; + +struct sym_entry { + struct rb_node rb_node; + struct list_head node; + unsigned long snap_count; + double weight; + struct map *map; + unsigned long count[0]; +}; + +static inline struct symbol *sym_entry__symbol(struct sym_entry *self) +{ + return ((void *)self) + symbol_conf.priv_size; +} + +struct perf_top { + struct perf_evlist *evlist; + /* + * Symbols will be added here in perf_event__process_sample and will + * get out after decayed. + */ + struct list_head active_symbols; + pthread_mutex_t active_symbols_lock; + pthread_cond_t active_symbols_cond; + u64 samples; + u64 kernel_samples, us_samples; + u64 exact_samples; + u64 guest_us_samples, guest_kernel_samples; + int print_entries, count_filter, delay_secs; + int display_weighted, freq, rb_entries; + pid_t target_pid, target_tid; + bool hide_kernel_symbols, hide_user_symbols, zero; + const char *cpu_list; + struct sym_entry *sym_filter_entry; + struct perf_evsel *sym_evsel; +}; + +size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size); +void perf_top__reset_sample_counters(struct perf_top *top); +float perf_top__decay_samples(struct perf_top *top, struct rb_root *root); +void perf_top__find_widths(struct perf_top *top, struct rb_root *root, + int *dso_width, int *dso_short_width, int *sym_width); + +#ifdef NO_NEWT_SUPPORT +static inline int perf_top__tui_browser(struct perf_top *top __used) +{ + return 0; +} +#else +int perf_top__tui_browser(struct perf_top *top); +#endif +#endif /* __PERF_TOP_H */ diff --git a/tools/perf/util/trace-event-info.c b/tools/perf/util/trace-event-info.c new file mode 100644 index 00000000..35729f4c --- /dev/null +++ b/tools/perf/util/trace-event-info.c @@ -0,0 +1,565 @@ +/* + * Copyright (C) 2008,2009, Steven Rostedt <srostedt@redhat.com> + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; version 2 of the License (not later!) + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + */ +#define _GNU_SOURCE +#include <dirent.h> +#include <mntent.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <stdarg.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <sys/wait.h> +#include <pthread.h> +#include <fcntl.h> +#include <unistd.h> +#include <ctype.h> +#include <errno.h> +#include <stdbool.h> +#include <linux/list.h> +#include <linux/kernel.h> + +#include "../perf.h" +#include "trace-event.h" +#include "debugfs.h" +#include "evsel.h" + +#define VERSION "0.5" + +#define _STR(x) #x +#define STR(x) _STR(x) +#define MAX_PATH 256 + +#define TRACE_CTRL "tracing_on" +#define TRACE "trace" +#define AVAILABLE "available_tracers" +#define CURRENT "current_tracer" +#define ITER_CTRL "trace_options" +#define MAX_LATENCY "tracing_max_latency" + +unsigned int page_size; + +static const char *output_file = "trace.info"; +static int output_fd; + +struct event_list { + struct event_list *next; + const char *event; +}; + +struct events { + struct events *sibling; + struct events *children; + struct events *next; + char *name; +}; + + + +static void die(const char *fmt, ...) +{ + va_list ap; + int ret = errno; + + if (errno) + perror("trace-cmd"); + else + ret = -1; + + va_start(ap, fmt); + fprintf(stderr, " "); + vfprintf(stderr, fmt, ap); + va_end(ap); + + fprintf(stderr, "\n"); + exit(ret); +} + +void *malloc_or_die(unsigned int size) +{ + void *data; + + data = malloc(size); + if (!data) + die("malloc"); + return data; +} + +static const char *find_debugfs(void) +{ + const char *path = debugfs_mount(NULL); + + if (!path) + die("Your kernel not support debugfs filesystem"); + + return path; +} + +/* + * Finds the path to the debugfs/tracing + * Allocates the string and stores it. + */ +static const char *find_tracing_dir(void) +{ + static char *tracing; + static int tracing_found; + const char *debugfs; + + if (tracing_found) + return tracing; + + debugfs = find_debugfs(); + + tracing = malloc_or_die(strlen(debugfs) + 9); + + sprintf(tracing, "%s/tracing", debugfs); + + tracing_found = 1; + return tracing; +} + +static char *get_tracing_file(const char *name) +{ + const char *tracing; + char *file; + + tracing = find_tracing_dir(); + if (!tracing) + return NULL; + + file = malloc_or_die(strlen(tracing) + strlen(name) + 2); + + sprintf(file, "%s/%s", tracing, name); + return file; +} + +static void put_tracing_file(char *file) +{ + free(file); +} + +static ssize_t calc_data_size; + +static ssize_t write_or_die(const void *buf, size_t len) +{ + int ret; + + if (calc_data_size) { + calc_data_size += len; + return len; + } + + ret = write(output_fd, buf, len); + if (ret < 0) + die("writing to '%s'", output_file); + + return ret; +} + +int bigendian(void) +{ + unsigned char str[] = { 0x1, 0x2, 0x3, 0x4, 0x0, 0x0, 0x0, 0x0}; + unsigned int *ptr; + + ptr = (unsigned int *)(void *)str; + return *ptr == 0x01020304; +} + +static unsigned long long copy_file_fd(int fd) +{ + unsigned long long size = 0; + char buf[BUFSIZ]; + int r; + + do { + r = read(fd, buf, BUFSIZ); + if (r > 0) { + size += r; + write_or_die(buf, r); + } + } while (r > 0); + + return size; +} + +static unsigned long long copy_file(const char *file) +{ + unsigned long long size = 0; + int fd; + + fd = open(file, O_RDONLY); + if (fd < 0) + die("Can't read '%s'", file); + size = copy_file_fd(fd); + close(fd); + + return size; +} + +static unsigned long get_size_fd(int fd) +{ + unsigned long long size = 0; + char buf[BUFSIZ]; + int r; + + do { + r = read(fd, buf, BUFSIZ); + if (r > 0) + size += r; + } while (r > 0); + + lseek(fd, 0, SEEK_SET); + + return size; +} + +static unsigned long get_size(const char *file) +{ + unsigned long long size = 0; + int fd; + + fd = open(file, O_RDONLY); + if (fd < 0) + die("Can't read '%s'", file); + size = get_size_fd(fd); + close(fd); + + return size; +} + +static void read_header_files(void) +{ + unsigned long long size, check_size; + char *path; + int fd; + + path = get_tracing_file("events/header_page"); + fd = open(path, O_RDONLY); + if (fd < 0) + die("can't read '%s'", path); + + /* unfortunately, you can not stat debugfs files for size */ + size = get_size_fd(fd); + + write_or_die("header_page", 12); + write_or_die(&size, 8); + check_size = copy_file_fd(fd); + close(fd); + + if (size != check_size) + die("wrong size for '%s' size=%lld read=%lld", + path, size, check_size); + put_tracing_file(path); + + path = get_tracing_file("events/header_event"); + fd = open(path, O_RDONLY); + if (fd < 0) + die("can't read '%s'", path); + + size = get_size_fd(fd); + + write_or_die("header_event", 13); + write_or_die(&size, 8); + check_size = copy_file_fd(fd); + if (size != check_size) + die("wrong size for '%s'", path); + put_tracing_file(path); + close(fd); +} + +static bool name_in_tp_list(char *sys, struct tracepoint_path *tps) +{ + while (tps) { + if (!strcmp(sys, tps->name)) + return true; + tps = tps->next; + } + + return false; +} + +static void copy_event_system(const char *sys, struct tracepoint_path *tps) +{ + unsigned long long size, check_size; + struct dirent *dent; + struct stat st; + char *format; + DIR *dir; + int count = 0; + int ret; + + dir = opendir(sys); + if (!dir) + die("can't read directory '%s'", sys); + + while ((dent = readdir(dir))) { + if (dent->d_type != DT_DIR || + strcmp(dent->d_name, ".") == 0 || + strcmp(dent->d_name, "..") == 0 || + !name_in_tp_list(dent->d_name, tps)) + continue; + format = malloc_or_die(strlen(sys) + strlen(dent->d_name) + 10); + sprintf(format, "%s/%s/format", sys, dent->d_name); + ret = stat(format, &st); + free(format); + if (ret < 0) + continue; + count++; + } + + write_or_die(&count, 4); + + rewinddir(dir); + while ((dent = readdir(dir))) { + if (dent->d_type != DT_DIR || + strcmp(dent->d_name, ".") == 0 || + strcmp(dent->d_name, "..") == 0 || + !name_in_tp_list(dent->d_name, tps)) + continue; + format = malloc_or_die(strlen(sys) + strlen(dent->d_name) + 10); + sprintf(format, "%s/%s/format", sys, dent->d_name); + ret = stat(format, &st); + + if (ret >= 0) { + /* unfortunately, you can not stat debugfs files for size */ + size = get_size(format); + write_or_die(&size, 8); + check_size = copy_file(format); + if (size != check_size) + die("error in size of file '%s'", format); + } + + free(format); + } + closedir(dir); +} + +static void read_ftrace_files(struct tracepoint_path *tps) +{ + char *path; + + path = get_tracing_file("events/ftrace"); + + copy_event_system(path, tps); + + put_tracing_file(path); +} + +static bool system_in_tp_list(char *sys, struct tracepoint_path *tps) +{ + while (tps) { + if (!strcmp(sys, tps->system)) + return true; + tps = tps->next; + } + + return false; +} + +static void read_event_files(struct tracepoint_path *tps) +{ + struct dirent *dent; + struct stat st; + char *path; + char *sys; + DIR *dir; + int count = 0; + int ret; + + path = get_tracing_file("events"); + + dir = opendir(path); + if (!dir) + die("can't read directory '%s'", path); + + while ((dent = readdir(dir))) { + if (dent->d_type != DT_DIR || + strcmp(dent->d_name, ".") == 0 || + strcmp(dent->d_name, "..") == 0 || + strcmp(dent->d_name, "ftrace") == 0 || + !system_in_tp_list(dent->d_name, tps)) + continue; + count++; + } + + write_or_die(&count, 4); + + rewinddir(dir); + while ((dent = readdir(dir))) { + if (dent->d_type != DT_DIR || + strcmp(dent->d_name, ".") == 0 || + strcmp(dent->d_name, "..") == 0 || + strcmp(dent->d_name, "ftrace") == 0 || + !system_in_tp_list(dent->d_name, tps)) + continue; + sys = malloc_or_die(strlen(path) + strlen(dent->d_name) + 2); + sprintf(sys, "%s/%s", path, dent->d_name); + ret = stat(sys, &st); + if (ret >= 0) { + write_or_die(dent->d_name, strlen(dent->d_name) + 1); + copy_event_system(sys, tps); + } + free(sys); + } + + closedir(dir); + put_tracing_file(path); +} + +static void read_proc_kallsyms(void) +{ + unsigned int size, check_size; + const char *path = "/proc/kallsyms"; + struct stat st; + int ret; + + ret = stat(path, &st); + if (ret < 0) { + /* not found */ + size = 0; + write_or_die(&size, 4); + return; + } + size = get_size(path); + write_or_die(&size, 4); + check_size = copy_file(path); + if (size != check_size) + die("error in size of file '%s'", path); + +} + +static void read_ftrace_printk(void) +{ + unsigned int size, check_size; + char *path; + struct stat st; + int ret; + + path = get_tracing_file("printk_formats"); + ret = stat(path, &st); + if (ret < 0) { + /* not found */ + size = 0; + write_or_die(&size, 4); + goto out; + } + size = get_size(path); + write_or_die(&size, 4); + check_size = copy_file(path); + if (size != check_size) + die("error in size of file '%s'", path); +out: + put_tracing_file(path); +} + +static struct tracepoint_path * +get_tracepoints_path(struct list_head *pattrs) +{ + struct tracepoint_path path, *ppath = &path; + struct perf_evsel *pos; + int nr_tracepoints = 0; + + list_for_each_entry(pos, pattrs, node) { + if (pos->attr.type != PERF_TYPE_TRACEPOINT) + continue; + ++nr_tracepoints; + ppath->next = tracepoint_id_to_path(pos->attr.config); + if (!ppath->next) + die("%s\n", "No memory to alloc tracepoints list"); + ppath = ppath->next; + } + + return nr_tracepoints > 0 ? path.next : NULL; +} + +bool have_tracepoints(struct list_head *pattrs) +{ + struct perf_evsel *pos; + + list_for_each_entry(pos, pattrs, node) + if (pos->attr.type == PERF_TYPE_TRACEPOINT) + return true; + + return false; +} + +int read_tracing_data(int fd, struct list_head *pattrs) +{ + char buf[BUFSIZ]; + struct tracepoint_path *tps = get_tracepoints_path(pattrs); + + /* + * What? No tracepoints? No sense writing anything here, bail out. + */ + if (tps == NULL) + return -1; + + output_fd = fd; + + buf[0] = 23; + buf[1] = 8; + buf[2] = 68; + memcpy(buf + 3, "tracing", 7); + + write_or_die(buf, 10); + + write_or_die(VERSION, strlen(VERSION) + 1); + + /* save endian */ + if (bigendian()) + buf[0] = 1; + else + buf[0] = 0; + + write_or_die(buf, 1); + + /* save size of long */ + buf[0] = sizeof(long); + write_or_die(buf, 1); + + /* save page_size */ + page_size = sysconf(_SC_PAGESIZE); + write_or_die(&page_size, 4); + + read_header_files(); + read_ftrace_files(tps); + read_event_files(tps); + read_proc_kallsyms(); + read_ftrace_printk(); + + return 0; +} + +ssize_t read_tracing_data_size(int fd, struct list_head *pattrs) +{ + ssize_t size; + int err = 0; + + calc_data_size = 1; + err = read_tracing_data(fd, pattrs); + size = calc_data_size - 1; + calc_data_size = 0; + + if (err < 0) + return err; + + return size; +} diff --git a/tools/perf/util/trace-event-parse.c b/tools/perf/util/trace-event-parse.c new file mode 100644 index 00000000..bf54c488 --- /dev/null +++ b/tools/perf/util/trace-event-parse.c @@ -0,0 +1,3147 @@ +/* + * Copyright (C) 2009, Steven Rostedt <srostedt@redhat.com> + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; version 2 of the License (not later!) + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + * + * The parts for function graph printing was taken and modified from the + * Linux Kernel that were written by Frederic Weisbecker. + */ +#define _GNU_SOURCE +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <ctype.h> +#include <errno.h> + +#undef _GNU_SOURCE +#include "../perf.h" +#include "util.h" +#include "trace-event.h" + +int header_page_ts_offset; +int header_page_ts_size; +int header_page_size_offset; +int header_page_size_size; +int header_page_overwrite_offset; +int header_page_overwrite_size; +int header_page_data_offset; +int header_page_data_size; + +bool latency_format; + +static char *input_buf; +static unsigned long long input_buf_ptr; +static unsigned long long input_buf_siz; + +static int cpus; +static int long_size; +static int is_flag_field; +static int is_symbolic_field; + +static struct format_field * +find_any_field(struct event *event, const char *name); + +static void init_input_buf(char *buf, unsigned long long size) +{ + input_buf = buf; + input_buf_siz = size; + input_buf_ptr = 0; +} + +struct cmdline { + char *comm; + int pid; +}; + +static struct cmdline *cmdlines; +static int cmdline_count; + +static int cmdline_cmp(const void *a, const void *b) +{ + const struct cmdline *ca = a; + const struct cmdline *cb = b; + + if (ca->pid < cb->pid) + return -1; + if (ca->pid > cb->pid) + return 1; + + return 0; +} + +void parse_cmdlines(char *file, int size __unused) +{ + struct cmdline_list { + struct cmdline_list *next; + char *comm; + int pid; + } *list = NULL, *item; + char *line; + char *next = NULL; + int i; + + line = strtok_r(file, "\n", &next); + while (line) { + item = malloc_or_die(sizeof(*item)); + sscanf(line, "%d %as", &item->pid, + (float *)(void *)&item->comm); /* workaround gcc warning */ + item->next = list; + list = item; + line = strtok_r(NULL, "\n", &next); + cmdline_count++; + } + + cmdlines = malloc_or_die(sizeof(*cmdlines) * cmdline_count); + + i = 0; + while (list) { + cmdlines[i].pid = list->pid; + cmdlines[i].comm = list->comm; + i++; + item = list; + list = list->next; + free(item); + } + + qsort(cmdlines, cmdline_count, sizeof(*cmdlines), cmdline_cmp); +} + +static struct func_map { + unsigned long long addr; + char *func; + char *mod; +} *func_list; +static unsigned int func_count; + +static int func_cmp(const void *a, const void *b) +{ + const struct func_map *fa = a; + const struct func_map *fb = b; + + if (fa->addr < fb->addr) + return -1; + if (fa->addr > fb->addr) + return 1; + + return 0; +} + +void parse_proc_kallsyms(char *file, unsigned int size __unused) +{ + struct func_list { + struct func_list *next; + unsigned long long addr; + char *func; + char *mod; + } *list = NULL, *item; + char *line; + char *next = NULL; + char *addr_str; + char ch; + int ret __used; + int i; + + line = strtok_r(file, "\n", &next); + while (line) { + item = malloc_or_die(sizeof(*item)); + item->mod = NULL; + ret = sscanf(line, "%as %c %as\t[%as", + (float *)(void *)&addr_str, /* workaround gcc warning */ + &ch, + (float *)(void *)&item->func, + (float *)(void *)&item->mod); + item->addr = strtoull(addr_str, NULL, 16); + free(addr_str); + + /* truncate the extra ']' */ + if (item->mod) + item->mod[strlen(item->mod) - 1] = 0; + + + item->next = list; + list = item; + line = strtok_r(NULL, "\n", &next); + func_count++; + } + + func_list = malloc_or_die(sizeof(*func_list) * (func_count + 1)); + + i = 0; + while (list) { + func_list[i].func = list->func; + func_list[i].addr = list->addr; + func_list[i].mod = list->mod; + i++; + item = list; + list = list->next; + free(item); + } + + qsort(func_list, func_count, sizeof(*func_list), func_cmp); + + /* + * Add a special record at the end. + */ + func_list[func_count].func = NULL; + func_list[func_count].addr = 0; + func_list[func_count].mod = NULL; +} + +/* + * We are searching for a record in between, not an exact + * match. + */ +static int func_bcmp(const void *a, const void *b) +{ + const struct func_map *fa = a; + const struct func_map *fb = b; + + if ((fa->addr == fb->addr) || + + (fa->addr > fb->addr && + fa->addr < (fb+1)->addr)) + return 0; + + if (fa->addr < fb->addr) + return -1; + + return 1; +} + +static struct func_map *find_func(unsigned long long addr) +{ + struct func_map *func; + struct func_map key; + + key.addr = addr; + + func = bsearch(&key, func_list, func_count, sizeof(*func_list), + func_bcmp); + + return func; +} + +void print_funcs(void) +{ + int i; + + for (i = 0; i < (int)func_count; i++) { + printf("%016llx %s", + func_list[i].addr, + func_list[i].func); + if (func_list[i].mod) + printf(" [%s]\n", func_list[i].mod); + else + printf("\n"); + } +} + +static struct printk_map { + unsigned long long addr; + char *printk; +} *printk_list; +static unsigned int printk_count; + +static int printk_cmp(const void *a, const void *b) +{ + const struct func_map *fa = a; + const struct func_map *fb = b; + + if (fa->addr < fb->addr) + return -1; + if (fa->addr > fb->addr) + return 1; + + return 0; +} + +static struct printk_map *find_printk(unsigned long long addr) +{ + struct printk_map *printk; + struct printk_map key; + + key.addr = addr; + + printk = bsearch(&key, printk_list, printk_count, sizeof(*printk_list), + printk_cmp); + + return printk; +} + +void parse_ftrace_printk(char *file, unsigned int size __unused) +{ + struct printk_list { + struct printk_list *next; + unsigned long long addr; + char *printk; + } *list = NULL, *item; + char *line; + char *next = NULL; + char *addr_str; + int i; + + line = strtok_r(file, "\n", &next); + while (line) { + addr_str = strsep(&line, ":"); + if (!line) { + warning("error parsing print strings"); + break; + } + item = malloc_or_die(sizeof(*item)); + item->addr = strtoull(addr_str, NULL, 16); + /* fmt still has a space, skip it */ + item->printk = strdup(line+1); + item->next = list; + list = item; + line = strtok_r(NULL, "\n", &next); + printk_count++; + } + + printk_list = malloc_or_die(sizeof(*printk_list) * printk_count + 1); + + i = 0; + while (list) { + printk_list[i].printk = list->printk; + printk_list[i].addr = list->addr; + i++; + item = list; + list = list->next; + free(item); + } + + qsort(printk_list, printk_count, sizeof(*printk_list), printk_cmp); +} + +void print_printk(void) +{ + int i; + + for (i = 0; i < (int)printk_count; i++) { + printf("%016llx %s\n", + printk_list[i].addr, + printk_list[i].printk); + } +} + +static struct event *alloc_event(void) +{ + struct event *event; + + event = malloc_or_die(sizeof(*event)); + memset(event, 0, sizeof(*event)); + + return event; +} + +enum event_type { + EVENT_ERROR, + EVENT_NONE, + EVENT_SPACE, + EVENT_NEWLINE, + EVENT_OP, + EVENT_DELIM, + EVENT_ITEM, + EVENT_DQUOTE, + EVENT_SQUOTE, +}; + +static struct event *event_list; + +static void add_event(struct event *event) +{ + event->next = event_list; + event_list = event; +} + +static int event_item_type(enum event_type type) +{ + switch (type) { + case EVENT_ITEM ... EVENT_SQUOTE: + return 1; + case EVENT_ERROR ... EVENT_DELIM: + default: + return 0; + } +} + +static void free_arg(struct print_arg *arg) +{ + if (!arg) + return; + + switch (arg->type) { + case PRINT_ATOM: + if (arg->atom.atom) + free(arg->atom.atom); + break; + case PRINT_NULL: + case PRINT_FIELD ... PRINT_OP: + default: + /* todo */ + break; + } + + free(arg); +} + +static enum event_type get_type(int ch) +{ + if (ch == '\n') + return EVENT_NEWLINE; + if (isspace(ch)) + return EVENT_SPACE; + if (isalnum(ch) || ch == '_') + return EVENT_ITEM; + if (ch == '\'') + return EVENT_SQUOTE; + if (ch == '"') + return EVENT_DQUOTE; + if (!isprint(ch)) + return EVENT_NONE; + if (ch == '(' || ch == ')' || ch == ',') + return EVENT_DELIM; + + return EVENT_OP; +} + +static int __read_char(void) +{ + if (input_buf_ptr >= input_buf_siz) + return -1; + + return input_buf[input_buf_ptr++]; +} + +static int __peek_char(void) +{ + if (input_buf_ptr >= input_buf_siz) + return -1; + + return input_buf[input_buf_ptr]; +} + +static enum event_type __read_token(char **tok) +{ + char buf[BUFSIZ]; + int ch, last_ch, quote_ch, next_ch; + int i = 0; + int tok_size = 0; + enum event_type type; + + *tok = NULL; + + + ch = __read_char(); + if (ch < 0) + return EVENT_NONE; + + type = get_type(ch); + if (type == EVENT_NONE) + return type; + + buf[i++] = ch; + + switch (type) { + case EVENT_NEWLINE: + case EVENT_DELIM: + *tok = malloc_or_die(2); + (*tok)[0] = ch; + (*tok)[1] = 0; + return type; + + case EVENT_OP: + switch (ch) { + case '-': + next_ch = __peek_char(); + if (next_ch == '>') { + buf[i++] = __read_char(); + break; + } + /* fall through */ + case '+': + case '|': + case '&': + case '>': + case '<': + last_ch = ch; + ch = __peek_char(); + if (ch != last_ch) + goto test_equal; + buf[i++] = __read_char(); + switch (last_ch) { + case '>': + case '<': + goto test_equal; + default: + break; + } + break; + case '!': + case '=': + goto test_equal; + default: /* what should we do instead? */ + break; + } + buf[i] = 0; + *tok = strdup(buf); + return type; + + test_equal: + ch = __peek_char(); + if (ch == '=') + buf[i++] = __read_char(); + break; + + case EVENT_DQUOTE: + case EVENT_SQUOTE: + /* don't keep quotes */ + i--; + quote_ch = ch; + last_ch = 0; + do { + if (i == (BUFSIZ - 1)) { + buf[i] = 0; + if (*tok) { + *tok = realloc(*tok, tok_size + BUFSIZ); + if (!*tok) + return EVENT_NONE; + strcat(*tok, buf); + } else + *tok = strdup(buf); + + if (!*tok) + return EVENT_NONE; + tok_size += BUFSIZ; + i = 0; + } + last_ch = ch; + ch = __read_char(); + buf[i++] = ch; + /* the '\' '\' will cancel itself */ + if (ch == '\\' && last_ch == '\\') + last_ch = 0; + } while (ch != quote_ch || last_ch == '\\'); + /* remove the last quote */ + i--; + goto out; + + case EVENT_ERROR ... EVENT_SPACE: + case EVENT_ITEM: + default: + break; + } + + while (get_type(__peek_char()) == type) { + if (i == (BUFSIZ - 1)) { + buf[i] = 0; + if (*tok) { + *tok = realloc(*tok, tok_size + BUFSIZ); + if (!*tok) + return EVENT_NONE; + strcat(*tok, buf); + } else + *tok = strdup(buf); + + if (!*tok) + return EVENT_NONE; + tok_size += BUFSIZ; + i = 0; + } + ch = __read_char(); + buf[i++] = ch; + } + + out: + buf[i] = 0; + if (*tok) { + *tok = realloc(*tok, tok_size + i); + if (!*tok) + return EVENT_NONE; + strcat(*tok, buf); + } else + *tok = strdup(buf); + if (!*tok) + return EVENT_NONE; + + return type; +} + +static void free_token(char *tok) +{ + if (tok) + free(tok); +} + +static enum event_type read_token(char **tok) +{ + enum event_type type; + + for (;;) { + type = __read_token(tok); + if (type != EVENT_SPACE) + return type; + + free_token(*tok); + } + + /* not reached */ + return EVENT_NONE; +} + +/* no newline */ +static enum event_type read_token_item(char **tok) +{ + enum event_type type; + + for (;;) { + type = __read_token(tok); + if (type != EVENT_SPACE && type != EVENT_NEWLINE) + return type; + + free_token(*tok); + } + + /* not reached */ + return EVENT_NONE; +} + +static int test_type(enum event_type type, enum event_type expect) +{ + if (type != expect) { + warning("Error: expected type %d but read %d", + expect, type); + return -1; + } + return 0; +} + +static int __test_type_token(enum event_type type, char *token, + enum event_type expect, const char *expect_tok, + bool warn) +{ + if (type != expect) { + if (warn) + warning("Error: expected type %d but read %d", + expect, type); + return -1; + } + + if (strcmp(token, expect_tok) != 0) { + if (warn) + warning("Error: expected '%s' but read '%s'", + expect_tok, token); + return -1; + } + return 0; +} + +static int test_type_token(enum event_type type, char *token, + enum event_type expect, const char *expect_tok) +{ + return __test_type_token(type, token, expect, expect_tok, true); +} + +static int __read_expect_type(enum event_type expect, char **tok, int newline_ok) +{ + enum event_type type; + + if (newline_ok) + type = read_token(tok); + else + type = read_token_item(tok); + return test_type(type, expect); +} + +static int read_expect_type(enum event_type expect, char **tok) +{ + return __read_expect_type(expect, tok, 1); +} + +static int __read_expected(enum event_type expect, const char *str, + int newline_ok, bool warn) +{ + enum event_type type; + char *token; + int ret; + + if (newline_ok) + type = read_token(&token); + else + type = read_token_item(&token); + + ret = __test_type_token(type, token, expect, str, warn); + + free_token(token); + + return ret; +} + +static int read_expected(enum event_type expect, const char *str) +{ + return __read_expected(expect, str, 1, true); +} + +static int read_expected_item(enum event_type expect, const char *str) +{ + return __read_expected(expect, str, 0, true); +} + +static char *event_read_name(void) +{ + char *token; + + if (read_expected(EVENT_ITEM, "name") < 0) + return NULL; + + if (read_expected(EVENT_OP, ":") < 0) + return NULL; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + + return token; + + fail: + free_token(token); + return NULL; +} + +static int event_read_id(void) +{ + char *token; + int id; + + if (read_expected_item(EVENT_ITEM, "ID") < 0) + return -1; + + if (read_expected(EVENT_OP, ":") < 0) + return -1; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + + id = strtoul(token, NULL, 0); + free_token(token); + return id; + + fail: + free_token(token); + return -1; +} + +static int field_is_string(struct format_field *field) +{ + if ((field->flags & FIELD_IS_ARRAY) && + (!strstr(field->type, "char") || !strstr(field->type, "u8") || + !strstr(field->type, "s8"))) + return 1; + + return 0; +} + +static int field_is_dynamic(struct format_field *field) +{ + if (!strncmp(field->type, "__data_loc", 10)) + return 1; + + return 0; +} + +static int event_read_fields(struct event *event, struct format_field **fields) +{ + struct format_field *field = NULL; + enum event_type type; + char *token; + char *last_token; + int count = 0; + + do { + type = read_token(&token); + if (type == EVENT_NEWLINE) { + free_token(token); + return count; + } + + count++; + + if (test_type_token(type, token, EVENT_ITEM, "field")) + goto fail; + free_token(token); + + type = read_token(&token); + /* + * The ftrace fields may still use the "special" name. + * Just ignore it. + */ + if (event->flags & EVENT_FL_ISFTRACE && + type == EVENT_ITEM && strcmp(token, "special") == 0) { + free_token(token); + type = read_token(&token); + } + + if (test_type_token(type, token, EVENT_OP, ":") < 0) + return -1; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + + last_token = token; + + field = malloc_or_die(sizeof(*field)); + memset(field, 0, sizeof(*field)); + + /* read the rest of the type */ + for (;;) { + type = read_token(&token); + if (type == EVENT_ITEM || + (type == EVENT_OP && strcmp(token, "*") == 0) || + /* + * Some of the ftrace fields are broken and have + * an illegal "." in them. + */ + (event->flags & EVENT_FL_ISFTRACE && + type == EVENT_OP && strcmp(token, ".") == 0)) { + + if (strcmp(token, "*") == 0) + field->flags |= FIELD_IS_POINTER; + + if (field->type) { + field->type = realloc(field->type, + strlen(field->type) + + strlen(last_token) + 2); + strcat(field->type, " "); + strcat(field->type, last_token); + } else + field->type = last_token; + last_token = token; + continue; + } + + break; + } + + if (!field->type) { + die("no type found"); + goto fail; + } + field->name = last_token; + + if (test_type(type, EVENT_OP)) + goto fail; + + if (strcmp(token, "[") == 0) { + enum event_type last_type = type; + char *brackets = token; + int len; + + field->flags |= FIELD_IS_ARRAY; + + type = read_token(&token); + while (strcmp(token, "]") != 0) { + if (last_type == EVENT_ITEM && + type == EVENT_ITEM) + len = 2; + else + len = 1; + last_type = type; + + brackets = realloc(brackets, + strlen(brackets) + + strlen(token) + len); + if (len == 2) + strcat(brackets, " "); + strcat(brackets, token); + free_token(token); + type = read_token(&token); + if (type == EVENT_NONE) { + die("failed to find token"); + goto fail; + } + } + + free_token(token); + + brackets = realloc(brackets, strlen(brackets) + 2); + strcat(brackets, "]"); + + /* add brackets to type */ + + type = read_token(&token); + /* + * If the next token is not an OP, then it is of + * the format: type [] item; + */ + if (type == EVENT_ITEM) { + field->type = realloc(field->type, + strlen(field->type) + + strlen(field->name) + + strlen(brackets) + 2); + strcat(field->type, " "); + strcat(field->type, field->name); + free_token(field->name); + strcat(field->type, brackets); + field->name = token; + type = read_token(&token); + } else { + field->type = realloc(field->type, + strlen(field->type) + + strlen(brackets) + 1); + strcat(field->type, brackets); + } + free(brackets); + } + + if (field_is_string(field)) { + field->flags |= FIELD_IS_STRING; + if (field_is_dynamic(field)) + field->flags |= FIELD_IS_DYNAMIC; + } + + if (test_type_token(type, token, EVENT_OP, ";")) + goto fail; + free_token(token); + + if (read_expected(EVENT_ITEM, "offset") < 0) + goto fail_expect; + + if (read_expected(EVENT_OP, ":") < 0) + goto fail_expect; + + if (read_expect_type(EVENT_ITEM, &token)) + goto fail; + field->offset = strtoul(token, NULL, 0); + free_token(token); + + if (read_expected(EVENT_OP, ";") < 0) + goto fail_expect; + + if (read_expected(EVENT_ITEM, "size") < 0) + goto fail_expect; + + if (read_expected(EVENT_OP, ":") < 0) + goto fail_expect; + + if (read_expect_type(EVENT_ITEM, &token)) + goto fail; + field->size = strtoul(token, NULL, 0); + free_token(token); + + if (read_expected(EVENT_OP, ";") < 0) + goto fail_expect; + + type = read_token(&token); + if (type != EVENT_NEWLINE) { + /* newer versions of the kernel have a "signed" type */ + if (test_type_token(type, token, EVENT_ITEM, "signed")) + goto fail; + + free_token(token); + + if (read_expected(EVENT_OP, ":") < 0) + goto fail_expect; + + if (read_expect_type(EVENT_ITEM, &token)) + goto fail; + + if (strtoul(token, NULL, 0)) + field->flags |= FIELD_IS_SIGNED; + + free_token(token); + if (read_expected(EVENT_OP, ";") < 0) + goto fail_expect; + + if (read_expect_type(EVENT_NEWLINE, &token)) + goto fail; + } + + free_token(token); + + *fields = field; + fields = &field->next; + + } while (1); + + return 0; + +fail: + free_token(token); +fail_expect: + if (field) + free(field); + return -1; +} + +static int event_read_format(struct event *event) +{ + char *token; + int ret; + + if (read_expected_item(EVENT_ITEM, "format") < 0) + return -1; + + if (read_expected(EVENT_OP, ":") < 0) + return -1; + + if (read_expect_type(EVENT_NEWLINE, &token)) + goto fail; + free_token(token); + + ret = event_read_fields(event, &event->format.common_fields); + if (ret < 0) + return ret; + event->format.nr_common = ret; + + ret = event_read_fields(event, &event->format.fields); + if (ret < 0) + return ret; + event->format.nr_fields = ret; + + return 0; + + fail: + free_token(token); + return -1; +} + +enum event_type +process_arg_token(struct event *event, struct print_arg *arg, + char **tok, enum event_type type); + +static enum event_type +process_arg(struct event *event, struct print_arg *arg, char **tok) +{ + enum event_type type; + char *token; + + type = read_token(&token); + *tok = token; + + return process_arg_token(event, arg, tok, type); +} + +static enum event_type +process_cond(struct event *event, struct print_arg *top, char **tok) +{ + struct print_arg *arg, *left, *right; + enum event_type type; + char *token = NULL; + + arg = malloc_or_die(sizeof(*arg)); + memset(arg, 0, sizeof(*arg)); + + left = malloc_or_die(sizeof(*left)); + + right = malloc_or_die(sizeof(*right)); + + arg->type = PRINT_OP; + arg->op.left = left; + arg->op.right = right; + + *tok = NULL; + type = process_arg(event, left, &token); + if (test_type_token(type, token, EVENT_OP, ":")) + goto out_free; + + arg->op.op = token; + + type = process_arg(event, right, &token); + + top->op.right = arg; + + *tok = token; + return type; + +out_free: + free_token(*tok); + free(right); + free(left); + free_arg(arg); + return EVENT_ERROR; +} + +static enum event_type +process_array(struct event *event, struct print_arg *top, char **tok) +{ + struct print_arg *arg; + enum event_type type; + char *token = NULL; + + arg = malloc_or_die(sizeof(*arg)); + memset(arg, 0, sizeof(*arg)); + + *tok = NULL; + type = process_arg(event, arg, &token); + if (test_type_token(type, token, EVENT_OP, "]")) + goto out_free; + + top->op.right = arg; + + free_token(token); + type = read_token_item(&token); + *tok = token; + + return type; + +out_free: + free_token(*tok); + free_arg(arg); + return EVENT_ERROR; +} + +static int get_op_prio(char *op) +{ + if (!op[1]) { + switch (op[0]) { + case '*': + case '/': + case '%': + return 6; + case '+': + case '-': + return 7; + /* '>>' and '<<' are 8 */ + case '<': + case '>': + return 9; + /* '==' and '!=' are 10 */ + case '&': + return 11; + case '^': + return 12; + case '|': + return 13; + case '?': + return 16; + default: + die("unknown op '%c'", op[0]); + return -1; + } + } else { + if (strcmp(op, "++") == 0 || + strcmp(op, "--") == 0) { + return 3; + } else if (strcmp(op, ">>") == 0 || + strcmp(op, "<<") == 0) { + return 8; + } else if (strcmp(op, ">=") == 0 || + strcmp(op, "<=") == 0) { + return 9; + } else if (strcmp(op, "==") == 0 || + strcmp(op, "!=") == 0) { + return 10; + } else if (strcmp(op, "&&") == 0) { + return 14; + } else if (strcmp(op, "||") == 0) { + return 15; + } else { + die("unknown op '%s'", op); + return -1; + } + } +} + +static void set_op_prio(struct print_arg *arg) +{ + + /* single ops are the greatest */ + if (!arg->op.left || arg->op.left->type == PRINT_NULL) { + arg->op.prio = 0; + return; + } + + arg->op.prio = get_op_prio(arg->op.op); +} + +static enum event_type +process_op(struct event *event, struct print_arg *arg, char **tok) +{ + struct print_arg *left, *right = NULL; + enum event_type type; + char *token; + + /* the op is passed in via tok */ + token = *tok; + + if (arg->type == PRINT_OP && !arg->op.left) { + /* handle single op */ + if (token[1]) { + die("bad op token %s", token); + return EVENT_ERROR; + } + switch (token[0]) { + case '!': + case '+': + case '-': + break; + default: + die("bad op token %s", token); + return EVENT_ERROR; + } + + /* make an empty left */ + left = malloc_or_die(sizeof(*left)); + left->type = PRINT_NULL; + arg->op.left = left; + + right = malloc_or_die(sizeof(*right)); + arg->op.right = right; + + type = process_arg(event, right, tok); + + } else if (strcmp(token, "?") == 0) { + + left = malloc_or_die(sizeof(*left)); + /* copy the top arg to the left */ + *left = *arg; + + arg->type = PRINT_OP; + arg->op.op = token; + arg->op.left = left; + arg->op.prio = 0; + + type = process_cond(event, arg, tok); + + } else if (strcmp(token, ">>") == 0 || + strcmp(token, "<<") == 0 || + strcmp(token, "&") == 0 || + strcmp(token, "|") == 0 || + strcmp(token, "&&") == 0 || + strcmp(token, "||") == 0 || + strcmp(token, "-") == 0 || + strcmp(token, "+") == 0 || + strcmp(token, "*") == 0 || + strcmp(token, "^") == 0 || + strcmp(token, "/") == 0 || + strcmp(token, "<") == 0 || + strcmp(token, ">") == 0 || + strcmp(token, "==") == 0 || + strcmp(token, "!=") == 0) { + + left = malloc_or_die(sizeof(*left)); + + /* copy the top arg to the left */ + *left = *arg; + + arg->type = PRINT_OP; + arg->op.op = token; + arg->op.left = left; + + set_op_prio(arg); + + right = malloc_or_die(sizeof(*right)); + + type = read_token_item(&token); + *tok = token; + + /* could just be a type pointer */ + if ((strcmp(arg->op.op, "*") == 0) && + type == EVENT_DELIM && (strcmp(token, ")") == 0)) { + if (left->type != PRINT_ATOM) + die("bad pointer type"); + left->atom.atom = realloc(left->atom.atom, + sizeof(left->atom.atom) + 3); + strcat(left->atom.atom, " *"); + *arg = *left; + free(arg); + + return type; + } + + type = process_arg_token(event, right, tok, type); + + arg->op.right = right; + + } else if (strcmp(token, "[") == 0) { + + left = malloc_or_die(sizeof(*left)); + *left = *arg; + + arg->type = PRINT_OP; + arg->op.op = token; + arg->op.left = left; + + arg->op.prio = 0; + type = process_array(event, arg, tok); + + } else { + warning("unknown op '%s'", token); + event->flags |= EVENT_FL_FAILED; + /* the arg is now the left side */ + return EVENT_NONE; + } + + if (type == EVENT_OP) { + int prio; + + /* higher prios need to be closer to the root */ + prio = get_op_prio(*tok); + + if (prio > arg->op.prio) + return process_op(event, arg, tok); + + return process_op(event, right, tok); + } + + return type; +} + +static enum event_type +process_entry(struct event *event __unused, struct print_arg *arg, + char **tok) +{ + enum event_type type; + char *field; + char *token; + + if (read_expected(EVENT_OP, "->") < 0) + return EVENT_ERROR; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + field = token; + + arg->type = PRINT_FIELD; + arg->field.name = field; + + if (is_flag_field) { + arg->field.field = find_any_field(event, arg->field.name); + arg->field.field->flags |= FIELD_IS_FLAG; + is_flag_field = 0; + } else if (is_symbolic_field) { + arg->field.field = find_any_field(event, arg->field.name); + arg->field.field->flags |= FIELD_IS_SYMBOLIC; + is_symbolic_field = 0; + } + + type = read_token(&token); + *tok = token; + + return type; + +fail: + free_token(token); + return EVENT_ERROR; +} + +static char *arg_eval (struct print_arg *arg); + +static long long arg_num_eval(struct print_arg *arg) +{ + long long left, right; + long long val = 0; + + switch (arg->type) { + case PRINT_ATOM: + val = strtoll(arg->atom.atom, NULL, 0); + break; + case PRINT_TYPE: + val = arg_num_eval(arg->typecast.item); + break; + case PRINT_OP: + switch (arg->op.op[0]) { + case '|': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + if (arg->op.op[1]) + val = left || right; + else + val = left | right; + break; + case '&': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + if (arg->op.op[1]) + val = left && right; + else + val = left & right; + break; + case '<': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + switch (arg->op.op[1]) { + case 0: + val = left < right; + break; + case '<': + val = left << right; + break; + case '=': + val = left <= right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + case '>': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + switch (arg->op.op[1]) { + case 0: + val = left > right; + break; + case '>': + val = left >> right; + break; + case '=': + val = left >= right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + case '=': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + + if (arg->op.op[1] != '=') + die("unknown op '%s'", arg->op.op); + + val = left == right; + break; + case '!': + left = arg_num_eval(arg->op.left); + right = arg_num_eval(arg->op.right); + + switch (arg->op.op[1]) { + case '=': + val = left != right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + + case PRINT_NULL: + case PRINT_FIELD ... PRINT_SYMBOL: + case PRINT_STRING: + default: + die("invalid eval type %d", arg->type); + + } + return val; +} + +static char *arg_eval (struct print_arg *arg) +{ + long long val; + static char buf[20]; + + switch (arg->type) { + case PRINT_ATOM: + return arg->atom.atom; + case PRINT_TYPE: + return arg_eval(arg->typecast.item); + case PRINT_OP: + val = arg_num_eval(arg); + sprintf(buf, "%lld", val); + return buf; + + case PRINT_NULL: + case PRINT_FIELD ... PRINT_SYMBOL: + case PRINT_STRING: + default: + die("invalid eval type %d", arg->type); + break; + } + + return NULL; +} + +static enum event_type +process_fields(struct event *event, struct print_flag_sym **list, char **tok) +{ + enum event_type type; + struct print_arg *arg = NULL; + struct print_flag_sym *field; + char *token = NULL; + char *value; + + do { + free_token(token); + type = read_token_item(&token); + if (test_type_token(type, token, EVENT_OP, "{")) + break; + + arg = malloc_or_die(sizeof(*arg)); + + free_token(token); + type = process_arg(event, arg, &token); + if (test_type_token(type, token, EVENT_DELIM, ",")) + goto out_free; + + field = malloc_or_die(sizeof(*field)); + memset(field, 0, sizeof(*field)); + + value = arg_eval(arg); + field->value = strdup(value); + + free_token(token); + type = process_arg(event, arg, &token); + if (test_type_token(type, token, EVENT_OP, "}")) + goto out_free; + + value = arg_eval(arg); + field->str = strdup(value); + free_arg(arg); + arg = NULL; + + *list = field; + list = &field->next; + + free_token(token); + type = read_token_item(&token); + } while (type == EVENT_DELIM && strcmp(token, ",") == 0); + + *tok = token; + return type; + +out_free: + free_arg(arg); + free_token(token); + + return EVENT_ERROR; +} + +static enum event_type +process_flags(struct event *event, struct print_arg *arg, char **tok) +{ + struct print_arg *field; + enum event_type type; + char *token; + + memset(arg, 0, sizeof(*arg)); + arg->type = PRINT_FLAGS; + + if (read_expected_item(EVENT_DELIM, "(") < 0) + return EVENT_ERROR; + + field = malloc_or_die(sizeof(*field)); + + type = process_arg(event, field, &token); + while (type == EVENT_OP) + type = process_op(event, field, &token); + if (test_type_token(type, token, EVENT_DELIM, ",")) + goto out_free; + + arg->flags.field = field; + + type = read_token_item(&token); + if (event_item_type(type)) { + arg->flags.delim = token; + type = read_token_item(&token); + } + + if (test_type_token(type, token, EVENT_DELIM, ",")) + goto out_free; + + type = process_fields(event, &arg->flags.flags, &token); + if (test_type_token(type, token, EVENT_DELIM, ")")) + goto out_free; + + free_token(token); + type = read_token_item(tok); + return type; + +out_free: + free_token(token); + return EVENT_ERROR; +} + +static enum event_type +process_symbols(struct event *event, struct print_arg *arg, char **tok) +{ + struct print_arg *field; + enum event_type type; + char *token; + + memset(arg, 0, sizeof(*arg)); + arg->type = PRINT_SYMBOL; + + if (read_expected_item(EVENT_DELIM, "(") < 0) + return EVENT_ERROR; + + field = malloc_or_die(sizeof(*field)); + + type = process_arg(event, field, &token); + while (type == EVENT_OP) + type = process_op(event, field, &token); + if (test_type_token(type, token, EVENT_DELIM, ",")) + goto out_free; + + arg->symbol.field = field; + + type = process_fields(event, &arg->symbol.symbols, &token); + if (test_type_token(type, token, EVENT_DELIM, ")")) + goto out_free; + + free_token(token); + type = read_token_item(tok); + return type; + +out_free: + free_token(token); + return EVENT_ERROR; +} + +static enum event_type +process_paren(struct event *event, struct print_arg *arg, char **tok) +{ + struct print_arg *item_arg; + enum event_type type; + char *token; + + type = process_arg(event, arg, &token); + + if (type == EVENT_ERROR) + return EVENT_ERROR; + + if (type == EVENT_OP) + type = process_op(event, arg, &token); + + if (type == EVENT_ERROR) + return EVENT_ERROR; + + if (test_type_token(type, token, EVENT_DELIM, ")")) { + free_token(token); + return EVENT_ERROR; + } + + free_token(token); + type = read_token_item(&token); + + /* + * If the next token is an item or another open paren, then + * this was a typecast. + */ + if (event_item_type(type) || + (type == EVENT_DELIM && strcmp(token, "(") == 0)) { + + /* make this a typecast and contine */ + + /* prevous must be an atom */ + if (arg->type != PRINT_ATOM) + die("previous needed to be PRINT_ATOM"); + + item_arg = malloc_or_die(sizeof(*item_arg)); + + arg->type = PRINT_TYPE; + arg->typecast.type = arg->atom.atom; + arg->typecast.item = item_arg; + type = process_arg_token(event, item_arg, &token, type); + + } + + *tok = token; + return type; +} + + +static enum event_type +process_str(struct event *event __unused, struct print_arg *arg, char **tok) +{ + enum event_type type; + char *token; + + if (read_expected(EVENT_DELIM, "(") < 0) + return EVENT_ERROR; + + if (read_expect_type(EVENT_ITEM, &token) < 0) + goto fail; + + arg->type = PRINT_STRING; + arg->string.string = token; + arg->string.offset = -1; + + if (read_expected(EVENT_DELIM, ")") < 0) + return EVENT_ERROR; + + type = read_token(&token); + *tok = token; + + return type; +fail: + free_token(token); + return EVENT_ERROR; +} + +enum event_type +process_arg_token(struct event *event, struct print_arg *arg, + char **tok, enum event_type type) +{ + char *token; + char *atom; + + token = *tok; + + switch (type) { + case EVENT_ITEM: + if (strcmp(token, "REC") == 0) { + free_token(token); + type = process_entry(event, arg, &token); + } else if (strcmp(token, "__print_flags") == 0) { + free_token(token); + is_flag_field = 1; + type = process_flags(event, arg, &token); + } else if (strcmp(token, "__print_symbolic") == 0) { + free_token(token); + is_symbolic_field = 1; + type = process_symbols(event, arg, &token); + } else if (strcmp(token, "__get_str") == 0) { + free_token(token); + type = process_str(event, arg, &token); + } else { + atom = token; + /* test the next token */ + type = read_token_item(&token); + + /* atoms can be more than one token long */ + while (type == EVENT_ITEM) { + atom = realloc(atom, strlen(atom) + strlen(token) + 2); + strcat(atom, " "); + strcat(atom, token); + free_token(token); + type = read_token_item(&token); + } + + /* todo, test for function */ + + arg->type = PRINT_ATOM; + arg->atom.atom = atom; + } + break; + case EVENT_DQUOTE: + case EVENT_SQUOTE: + arg->type = PRINT_ATOM; + arg->atom.atom = token; + type = read_token_item(&token); + break; + case EVENT_DELIM: + if (strcmp(token, "(") == 0) { + free_token(token); + type = process_paren(event, arg, &token); + break; + } + case EVENT_OP: + /* handle single ops */ + arg->type = PRINT_OP; + arg->op.op = token; + arg->op.left = NULL; + type = process_op(event, arg, &token); + + break; + + case EVENT_ERROR ... EVENT_NEWLINE: + default: + die("unexpected type %d", type); + } + *tok = token; + + return type; +} + +static int event_read_print_args(struct event *event, struct print_arg **list) +{ + enum event_type type = EVENT_ERROR; + struct print_arg *arg; + char *token; + int args = 0; + + do { + if (type == EVENT_NEWLINE) { + free_token(token); + type = read_token_item(&token); + continue; + } + + arg = malloc_or_die(sizeof(*arg)); + memset(arg, 0, sizeof(*arg)); + + type = process_arg(event, arg, &token); + + if (type == EVENT_ERROR) { + free_arg(arg); + return -1; + } + + *list = arg; + args++; + + if (type == EVENT_OP) { + type = process_op(event, arg, &token); + list = &arg->next; + continue; + } + + if (type == EVENT_DELIM && strcmp(token, ",") == 0) { + free_token(token); + *list = arg; + list = &arg->next; + continue; + } + break; + } while (type != EVENT_NONE); + + if (type != EVENT_NONE) + free_token(token); + + return args; +} + +static int event_read_print(struct event *event) +{ + enum event_type type; + char *token; + int ret; + + if (read_expected_item(EVENT_ITEM, "print") < 0) + return -1; + + if (read_expected(EVENT_ITEM, "fmt") < 0) + return -1; + + if (read_expected(EVENT_OP, ":") < 0) + return -1; + + if (read_expect_type(EVENT_DQUOTE, &token) < 0) + goto fail; + + concat: + event->print_fmt.format = token; + event->print_fmt.args = NULL; + + /* ok to have no arg */ + type = read_token_item(&token); + + if (type == EVENT_NONE) + return 0; + + /* Handle concatination of print lines */ + if (type == EVENT_DQUOTE) { + char *cat; + + cat = malloc_or_die(strlen(event->print_fmt.format) + + strlen(token) + 1); + strcpy(cat, event->print_fmt.format); + strcat(cat, token); + free_token(token); + free_token(event->print_fmt.format); + event->print_fmt.format = NULL; + token = cat; + goto concat; + } + + if (test_type_token(type, token, EVENT_DELIM, ",")) + goto fail; + + free_token(token); + + ret = event_read_print_args(event, &event->print_fmt.args); + if (ret < 0) + return -1; + + return ret; + + fail: + free_token(token); + return -1; +} + +static struct format_field * +find_common_field(struct event *event, const char *name) +{ + struct format_field *format; + + for (format = event->format.common_fields; + format; format = format->next) { + if (strcmp(format->name, name) == 0) + break; + } + + return format; +} + +static struct format_field * +find_field(struct event *event, const char *name) +{ + struct format_field *format; + + for (format = event->format.fields; + format; format = format->next) { + if (strcmp(format->name, name) == 0) + break; + } + + return format; +} + +static struct format_field * +find_any_field(struct event *event, const char *name) +{ + struct format_field *format; + + format = find_common_field(event, name); + if (format) + return format; + return find_field(event, name); +} + +unsigned long long read_size(void *ptr, int size) +{ + switch (size) { + case 1: + return *(unsigned char *)ptr; + case 2: + return data2host2(ptr); + case 4: + return data2host4(ptr); + case 8: + return data2host8(ptr); + default: + /* BUG! */ + return 0; + } +} + +unsigned long long +raw_field_value(struct event *event, const char *name, void *data) +{ + struct format_field *field; + + field = find_any_field(event, name); + if (!field) + return 0ULL; + + return read_size(data + field->offset, field->size); +} + +void *raw_field_ptr(struct event *event, const char *name, void *data) +{ + struct format_field *field; + + field = find_any_field(event, name); + if (!field) + return NULL; + + if (field->flags & FIELD_IS_DYNAMIC) { + int offset; + + offset = *(int *)(data + field->offset); + offset &= 0xffff; + + return data + offset; + } + + return data + field->offset; +} + +static int get_common_info(const char *type, int *offset, int *size) +{ + struct event *event; + struct format_field *field; + + /* + * All events should have the same common elements. + * Pick any event to find where the type is; + */ + if (!event_list) + die("no event_list!"); + + event = event_list; + field = find_common_field(event, type); + if (!field) + die("field '%s' not found", type); + + *offset = field->offset; + *size = field->size; + + return 0; +} + +static int __parse_common(void *data, int *size, int *offset, + const char *name) +{ + int ret; + + if (!*size) { + ret = get_common_info(name, offset, size); + if (ret < 0) + return ret; + } + return read_size(data + *offset, *size); +} + +int trace_parse_common_type(void *data) +{ + static int type_offset; + static int type_size; + + return __parse_common(data, &type_size, &type_offset, + "common_type"); +} + +int trace_parse_common_pid(void *data) +{ + static int pid_offset; + static int pid_size; + + return __parse_common(data, &pid_size, &pid_offset, + "common_pid"); +} + +int parse_common_pc(void *data) +{ + static int pc_offset; + static int pc_size; + + return __parse_common(data, &pc_size, &pc_offset, + "common_preempt_count"); +} + +int parse_common_flags(void *data) +{ + static int flags_offset; + static int flags_size; + + return __parse_common(data, &flags_size, &flags_offset, + "common_flags"); +} + +int parse_common_lock_depth(void *data) +{ + static int ld_offset; + static int ld_size; + int ret; + + ret = __parse_common(data, &ld_size, &ld_offset, + "common_lock_depth"); + if (ret < 0) + return -1; + + return ret; +} + +struct event *trace_find_event(int id) +{ + struct event *event; + + for (event = event_list; event; event = event->next) { + if (event->id == id) + break; + } + return event; +} + +struct event *trace_find_next_event(struct event *event) +{ + if (!event) + return event_list; + + return event->next; +} + +static unsigned long long eval_num_arg(void *data, int size, + struct event *event, struct print_arg *arg) +{ + unsigned long long val = 0; + unsigned long long left, right; + struct print_arg *larg; + + switch (arg->type) { + case PRINT_NULL: + /* ?? */ + return 0; + case PRINT_ATOM: + return strtoull(arg->atom.atom, NULL, 0); + case PRINT_FIELD: + if (!arg->field.field) { + arg->field.field = find_any_field(event, arg->field.name); + if (!arg->field.field) + die("field %s not found", arg->field.name); + } + /* must be a number */ + val = read_size(data + arg->field.field->offset, + arg->field.field->size); + break; + case PRINT_FLAGS: + case PRINT_SYMBOL: + break; + case PRINT_TYPE: + return eval_num_arg(data, size, event, arg->typecast.item); + case PRINT_STRING: + return 0; + break; + case PRINT_OP: + if (strcmp(arg->op.op, "[") == 0) { + /* + * Arrays are special, since we don't want + * to read the arg as is. + */ + if (arg->op.left->type != PRINT_FIELD) + goto default_op; /* oops, all bets off */ + larg = arg->op.left; + if (!larg->field.field) { + larg->field.field = + find_any_field(event, larg->field.name); + if (!larg->field.field) + die("field %s not found", larg->field.name); + } + right = eval_num_arg(data, size, event, arg->op.right); + val = read_size(data + larg->field.field->offset + + right * long_size, long_size); + break; + } + default_op: + left = eval_num_arg(data, size, event, arg->op.left); + right = eval_num_arg(data, size, event, arg->op.right); + switch (arg->op.op[0]) { + case '|': + if (arg->op.op[1]) + val = left || right; + else + val = left | right; + break; + case '&': + if (arg->op.op[1]) + val = left && right; + else + val = left & right; + break; + case '<': + switch (arg->op.op[1]) { + case 0: + val = left < right; + break; + case '<': + val = left << right; + break; + case '=': + val = left <= right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + case '>': + switch (arg->op.op[1]) { + case 0: + val = left > right; + break; + case '>': + val = left >> right; + break; + case '=': + val = left >= right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + case '=': + if (arg->op.op[1] != '=') + die("unknown op '%s'", arg->op.op); + val = left == right; + break; + case '-': + val = left - right; + break; + case '+': + val = left + right; + break; + default: + die("unknown op '%s'", arg->op.op); + } + break; + default: /* not sure what to do there */ + return 0; + } + return val; +} + +struct flag { + const char *name; + unsigned long long value; +}; + +static const struct flag flags[] = { + { "HI_SOFTIRQ", 0 }, + { "TIMER_SOFTIRQ", 1 }, + { "NET_TX_SOFTIRQ", 2 }, + { "NET_RX_SOFTIRQ", 3 }, + { "BLOCK_SOFTIRQ", 4 }, + { "BLOCK_IOPOLL_SOFTIRQ", 5 }, + { "TASKLET_SOFTIRQ", 6 }, + { "SCHED_SOFTIRQ", 7 }, + { "HRTIMER_SOFTIRQ", 8 }, + { "RCU_SOFTIRQ", 9 }, + + { "HRTIMER_NORESTART", 0 }, + { "HRTIMER_RESTART", 1 }, +}; + +unsigned long long eval_flag(const char *flag) +{ + int i; + + /* + * Some flags in the format files do not get converted. + * If the flag is not numeric, see if it is something that + * we already know about. + */ + if (isdigit(flag[0])) + return strtoull(flag, NULL, 0); + + for (i = 0; i < (int)(sizeof(flags)/sizeof(flags[0])); i++) + if (strcmp(flags[i].name, flag) == 0) + return flags[i].value; + + return 0; +} + +static void print_str_arg(void *data, int size, + struct event *event, struct print_arg *arg) +{ + struct print_flag_sym *flag; + unsigned long long val, fval; + char *str; + int print; + + switch (arg->type) { + case PRINT_NULL: + /* ?? */ + return; + case PRINT_ATOM: + printf("%s", arg->atom.atom); + return; + case PRINT_FIELD: + if (!arg->field.field) { + arg->field.field = find_any_field(event, arg->field.name); + if (!arg->field.field) + die("field %s not found", arg->field.name); + } + str = malloc_or_die(arg->field.field->size + 1); + memcpy(str, data + arg->field.field->offset, + arg->field.field->size); + str[arg->field.field->size] = 0; + printf("%s", str); + free(str); + break; + case PRINT_FLAGS: + val = eval_num_arg(data, size, event, arg->flags.field); + print = 0; + for (flag = arg->flags.flags; flag; flag = flag->next) { + fval = eval_flag(flag->value); + if (!val && !fval) { + printf("%s", flag->str); + break; + } + if (fval && (val & fval) == fval) { + if (print && arg->flags.delim) + printf("%s", arg->flags.delim); + printf("%s", flag->str); + print = 1; + val &= ~fval; + } + } + break; + case PRINT_SYMBOL: + val = eval_num_arg(data, size, event, arg->symbol.field); + for (flag = arg->symbol.symbols; flag; flag = flag->next) { + fval = eval_flag(flag->value); + if (val == fval) { + printf("%s", flag->str); + break; + } + } + break; + + case PRINT_TYPE: + break; + case PRINT_STRING: { + int str_offset; + + if (arg->string.offset == -1) { + struct format_field *f; + + f = find_any_field(event, arg->string.string); + arg->string.offset = f->offset; + } + str_offset = *(int *)(data + arg->string.offset); + str_offset &= 0xffff; + printf("%s", ((char *)data) + str_offset); + break; + } + case PRINT_OP: + /* + * The only op for string should be ? : + */ + if (arg->op.op[0] != '?') + return; + val = eval_num_arg(data, size, event, arg->op.left); + if (val) + print_str_arg(data, size, event, arg->op.right->op.left); + else + print_str_arg(data, size, event, arg->op.right->op.right); + break; + default: + /* well... */ + break; + } +} + +static struct print_arg *make_bprint_args(char *fmt, void *data, int size, struct event *event) +{ + static struct format_field *field, *ip_field; + struct print_arg *args, *arg, **next; + unsigned long long ip, val; + char *ptr; + void *bptr; + + if (!field) { + field = find_field(event, "buf"); + if (!field) + die("can't find buffer field for binary printk"); + ip_field = find_field(event, "ip"); + if (!ip_field) + die("can't find ip field for binary printk"); + } + + ip = read_size(data + ip_field->offset, ip_field->size); + + /* + * The first arg is the IP pointer. + */ + args = malloc_or_die(sizeof(*args)); + arg = args; + arg->next = NULL; + next = &arg->next; + + arg->type = PRINT_ATOM; + arg->atom.atom = malloc_or_die(32); + sprintf(arg->atom.atom, "%lld", ip); + + /* skip the first "%pf : " */ + for (ptr = fmt + 6, bptr = data + field->offset; + bptr < data + size && *ptr; ptr++) { + int ls = 0; + + if (*ptr == '%') { + process_again: + ptr++; + switch (*ptr) { + case '%': + break; + case 'l': + ls++; + goto process_again; + case 'L': + ls = 2; + goto process_again; + case '0' ... '9': + goto process_again; + case 'p': + ls = 1; + /* fall through */ + case 'd': + case 'u': + case 'x': + case 'i': + /* the pointers are always 4 bytes aligned */ + bptr = (void *)(((unsigned long)bptr + 3) & + ~3); + switch (ls) { + case 0: + case 1: + ls = long_size; + break; + case 2: + ls = 8; + default: + break; + } + val = read_size(bptr, ls); + bptr += ls; + arg = malloc_or_die(sizeof(*arg)); + arg->next = NULL; + arg->type = PRINT_ATOM; + arg->atom.atom = malloc_or_die(32); + sprintf(arg->atom.atom, "%lld", val); + *next = arg; + next = &arg->next; + break; + case 's': + arg = malloc_or_die(sizeof(*arg)); + arg->next = NULL; + arg->type = PRINT_STRING; + arg->string.string = strdup(bptr); + bptr += strlen(bptr) + 1; + *next = arg; + next = &arg->next; + default: + break; + } + } + } + + return args; +} + +static void free_args(struct print_arg *args) +{ + struct print_arg *next; + + while (args) { + next = args->next; + + if (args->type == PRINT_ATOM) + free(args->atom.atom); + else + free(args->string.string); + free(args); + args = next; + } +} + +static char *get_bprint_format(void *data, int size __unused, struct event *event) +{ + unsigned long long addr; + static struct format_field *field; + struct printk_map *printk; + char *format; + char *p; + + if (!field) { + field = find_field(event, "fmt"); + if (!field) + die("can't find format field for binary printk"); + printf("field->offset = %d size=%d\n", field->offset, field->size); + } + + addr = read_size(data + field->offset, field->size); + + printk = find_printk(addr); + if (!printk) { + format = malloc_or_die(45); + sprintf(format, "%%pf : (NO FORMAT FOUND at %llx)\n", + addr); + return format; + } + + p = printk->printk; + /* Remove any quotes. */ + if (*p == '"') + p++; + format = malloc_or_die(strlen(p) + 10); + sprintf(format, "%s : %s", "%pf", p); + /* remove ending quotes and new line since we will add one too */ + p = format + strlen(format) - 1; + if (*p == '"') + *p = 0; + + p -= 2; + if (strcmp(p, "\\n") == 0) + *p = 0; + + return format; +} + +static void pretty_print(void *data, int size, struct event *event) +{ + struct print_fmt *print_fmt = &event->print_fmt; + struct print_arg *arg = print_fmt->args; + struct print_arg *args = NULL; + const char *ptr = print_fmt->format; + unsigned long long val; + struct func_map *func; + const char *saveptr; + char *bprint_fmt = NULL; + char format[32]; + int show_func; + int len; + int ls; + + if (event->flags & EVENT_FL_ISFUNC) + ptr = " %pF <-- %pF"; + + if (event->flags & EVENT_FL_ISBPRINT) { + bprint_fmt = get_bprint_format(data, size, event); + args = make_bprint_args(bprint_fmt, data, size, event); + arg = args; + ptr = bprint_fmt; + } + + for (; *ptr; ptr++) { + ls = 0; + if (*ptr == '\\') { + ptr++; + switch (*ptr) { + case 'n': + printf("\n"); + break; + case 't': + printf("\t"); + break; + case 'r': + printf("\r"); + break; + case '\\': + printf("\\"); + break; + default: + printf("%c", *ptr); + break; + } + + } else if (*ptr == '%') { + saveptr = ptr; + show_func = 0; + cont_process: + ptr++; + switch (*ptr) { + case '%': + printf("%%"); + break; + case 'l': + ls++; + goto cont_process; + case 'L': + ls = 2; + goto cont_process; + case 'z': + case 'Z': + case '0' ... '9': + goto cont_process; + case 'p': + if (long_size == 4) + ls = 1; + else + ls = 2; + + if (*(ptr+1) == 'F' || + *(ptr+1) == 'f') { + ptr++; + show_func = *ptr; + } + + /* fall through */ + case 'd': + case 'i': + case 'x': + case 'X': + case 'u': + if (!arg) + die("no argument match"); + + len = ((unsigned long)ptr + 1) - + (unsigned long)saveptr; + + /* should never happen */ + if (len > 32) + die("bad format!"); + + memcpy(format, saveptr, len); + format[len] = 0; + + val = eval_num_arg(data, size, event, arg); + arg = arg->next; + + if (show_func) { + func = find_func(val); + if (func) { + printf("%s", func->func); + if (show_func == 'F') + printf("+0x%llx", + val - func->addr); + break; + } + } + switch (ls) { + case 0: + printf(format, (int)val); + break; + case 1: + printf(format, (long)val); + break; + case 2: + printf(format, (long long)val); + break; + default: + die("bad count (%d)", ls); + } + break; + case 's': + if (!arg) + die("no matching argument"); + + print_str_arg(data, size, event, arg); + arg = arg->next; + break; + default: + printf(">%c<", *ptr); + + } + } else + printf("%c", *ptr); + } + + if (args) { + free_args(args); + free(bprint_fmt); + } +} + +static inline int log10_cpu(int nb) +{ + if (nb / 100) + return 3; + if (nb / 10) + return 2; + return 1; +} + +static void print_lat_fmt(void *data, int size __unused) +{ + unsigned int lat_flags; + unsigned int pc; + int lock_depth; + int hardirq; + int softirq; + + lat_flags = parse_common_flags(data); + pc = parse_common_pc(data); + lock_depth = parse_common_lock_depth(data); + + hardirq = lat_flags & TRACE_FLAG_HARDIRQ; + softirq = lat_flags & TRACE_FLAG_SOFTIRQ; + + printf("%c%c%c", + (lat_flags & TRACE_FLAG_IRQS_OFF) ? 'd' : + (lat_flags & TRACE_FLAG_IRQS_NOSUPPORT) ? + 'X' : '.', + (lat_flags & TRACE_FLAG_NEED_RESCHED) ? + 'N' : '.', + (hardirq && softirq) ? 'H' : + hardirq ? 'h' : softirq ? 's' : '.'); + + if (pc) + printf("%x", pc); + else + printf("."); + + if (lock_depth < 0) + printf(". "); + else + printf("%d ", lock_depth); +} + +#define TRACE_GRAPH_INDENT 2 + +static struct record * +get_return_for_leaf(int cpu, int cur_pid, unsigned long long cur_func, + struct record *next) +{ + struct format_field *field; + struct event *event; + unsigned long val; + int type; + int pid; + + type = trace_parse_common_type(next->data); + event = trace_find_event(type); + if (!event) + return NULL; + + if (!(event->flags & EVENT_FL_ISFUNCRET)) + return NULL; + + pid = trace_parse_common_pid(next->data); + field = find_field(event, "func"); + if (!field) + die("function return does not have field func"); + + val = read_size(next->data + field->offset, field->size); + + if (cur_pid != pid || cur_func != val) + return NULL; + + /* this is a leaf, now advance the iterator */ + return trace_read_data(cpu); +} + +/* Signal a overhead of time execution to the output */ +static void print_graph_overhead(unsigned long long duration) +{ + /* Non nested entry or return */ + if (duration == ~0ULL) + return (void)printf(" "); + + /* Duration exceeded 100 msecs */ + if (duration > 100000ULL) + return (void)printf("! "); + + /* Duration exceeded 10 msecs */ + if (duration > 10000ULL) + return (void)printf("+ "); + + printf(" "); +} + +static void print_graph_duration(unsigned long long duration) +{ + unsigned long usecs = duration / 1000; + unsigned long nsecs_rem = duration % 1000; + /* log10(ULONG_MAX) + '\0' */ + char msecs_str[21]; + char nsecs_str[5]; + int len; + int i; + + sprintf(msecs_str, "%lu", usecs); + + /* Print msecs */ + len = printf("%lu", usecs); + + /* Print nsecs (we don't want to exceed 7 numbers) */ + if (len < 7) { + snprintf(nsecs_str, 8 - len, "%03lu", nsecs_rem); + len += printf(".%s", nsecs_str); + } + + printf(" us "); + + /* Print remaining spaces to fit the row's width */ + for (i = len; i < 7; i++) + printf(" "); + + printf("| "); +} + +static void +print_graph_entry_leaf(struct event *event, void *data, struct record *ret_rec) +{ + unsigned long long rettime, calltime; + unsigned long long duration, depth; + unsigned long long val; + struct format_field *field; + struct func_map *func; + struct event *ret_event; + int type; + int i; + + type = trace_parse_common_type(ret_rec->data); + ret_event = trace_find_event(type); + + field = find_field(ret_event, "rettime"); + if (!field) + die("can't find rettime in return graph"); + rettime = read_size(ret_rec->data + field->offset, field->size); + + field = find_field(ret_event, "calltime"); + if (!field) + die("can't find rettime in return graph"); + calltime = read_size(ret_rec->data + field->offset, field->size); + + duration = rettime - calltime; + + /* Overhead */ + print_graph_overhead(duration); + + /* Duration */ + print_graph_duration(duration); + + field = find_field(event, "depth"); + if (!field) + die("can't find depth in entry graph"); + depth = read_size(data + field->offset, field->size); + + /* Function */ + for (i = 0; i < (int)(depth * TRACE_GRAPH_INDENT); i++) + printf(" "); + + field = find_field(event, "func"); + if (!field) + die("can't find func in entry graph"); + val = read_size(data + field->offset, field->size); + func = find_func(val); + + if (func) + printf("%s();", func->func); + else + printf("%llx();", val); +} + +static void print_graph_nested(struct event *event, void *data) +{ + struct format_field *field; + unsigned long long depth; + unsigned long long val; + struct func_map *func; + int i; + + /* No overhead */ + print_graph_overhead(-1); + + /* No time */ + printf(" | "); + + field = find_field(event, "depth"); + if (!field) + die("can't find depth in entry graph"); + depth = read_size(data + field->offset, field->size); + + /* Function */ + for (i = 0; i < (int)(depth * TRACE_GRAPH_INDENT); i++) + printf(" "); + + field = find_field(event, "func"); + if (!field) + die("can't find func in entry graph"); + val = read_size(data + field->offset, field->size); + func = find_func(val); + + if (func) + printf("%s() {", func->func); + else + printf("%llx() {", val); +} + +static void +pretty_print_func_ent(void *data, int size, struct event *event, + int cpu, int pid) +{ + struct format_field *field; + struct record *rec; + void *copy_data; + unsigned long val; + + if (latency_format) { + print_lat_fmt(data, size); + printf(" | "); + } + + field = find_field(event, "func"); + if (!field) + die("function entry does not have func field"); + + val = read_size(data + field->offset, field->size); + + /* + * peek_data may unmap the data pointer. Copy it first. + */ + copy_data = malloc_or_die(size); + memcpy(copy_data, data, size); + data = copy_data; + + rec = trace_peek_data(cpu); + if (rec) { + rec = get_return_for_leaf(cpu, pid, val, rec); + if (rec) { + print_graph_entry_leaf(event, data, rec); + goto out_free; + } + } + print_graph_nested(event, data); +out_free: + free(data); +} + +static void +pretty_print_func_ret(void *data, int size __unused, struct event *event) +{ + unsigned long long rettime, calltime; + unsigned long long duration, depth; + struct format_field *field; + int i; + + if (latency_format) { + print_lat_fmt(data, size); + printf(" | "); + } + + field = find_field(event, "rettime"); + if (!field) + die("can't find rettime in return graph"); + rettime = read_size(data + field->offset, field->size); + + field = find_field(event, "calltime"); + if (!field) + die("can't find calltime in return graph"); + calltime = read_size(data + field->offset, field->size); + + duration = rettime - calltime; + + /* Overhead */ + print_graph_overhead(duration); + + /* Duration */ + print_graph_duration(duration); + + field = find_field(event, "depth"); + if (!field) + die("can't find depth in entry graph"); + depth = read_size(data + field->offset, field->size); + + /* Function */ + for (i = 0; i < (int)(depth * TRACE_GRAPH_INDENT); i++) + printf(" "); + + printf("}"); +} + +static void +pretty_print_func_graph(void *data, int size, struct event *event, + int cpu, int pid) +{ + if (event->flags & EVENT_FL_ISFUNCENT) + pretty_print_func_ent(data, size, event, cpu, pid); + else if (event->flags & EVENT_FL_ISFUNCRET) + pretty_print_func_ret(data, size, event); + printf("\n"); +} + +void print_trace_event(int cpu, void *data, int size) +{ + struct event *event; + int type; + int pid; + + type = trace_parse_common_type(data); + + event = trace_find_event(type); + if (!event) { + warning("ug! no event found for type %d", type); + return; + } + + pid = trace_parse_common_pid(data); + + if (event->flags & (EVENT_FL_ISFUNCENT | EVENT_FL_ISFUNCRET)) + return pretty_print_func_graph(data, size, event, cpu, pid); + + if (latency_format) + print_lat_fmt(data, size); + + if (event->flags & EVENT_FL_FAILED) { + printf("EVENT '%s' FAILED TO PARSE\n", + event->name); + return; + } + + pretty_print(data, size, event); +} + +static void print_fields(struct print_flag_sym *field) +{ + printf("{ %s, %s }", field->value, field->str); + if (field->next) { + printf(", "); + print_fields(field->next); + } +} + +static void print_args(struct print_arg *args) +{ + int print_paren = 1; + + switch (args->type) { + case PRINT_NULL: + printf("null"); + break; + case PRINT_ATOM: + printf("%s", args->atom.atom); + break; + case PRINT_FIELD: + printf("REC->%s", args->field.name); + break; + case PRINT_FLAGS: + printf("__print_flags("); + print_args(args->flags.field); + printf(", %s, ", args->flags.delim); + print_fields(args->flags.flags); + printf(")"); + break; + case PRINT_SYMBOL: + printf("__print_symbolic("); + print_args(args->symbol.field); + printf(", "); + print_fields(args->symbol.symbols); + printf(")"); + break; + case PRINT_STRING: + printf("__get_str(%s)", args->string.string); + break; + case PRINT_TYPE: + printf("(%s)", args->typecast.type); + print_args(args->typecast.item); + break; + case PRINT_OP: + if (strcmp(args->op.op, ":") == 0) + print_paren = 0; + if (print_paren) + printf("("); + print_args(args->op.left); + printf(" %s ", args->op.op); + print_args(args->op.right); + if (print_paren) + printf(")"); + break; + default: + /* we should warn... */ + return; + } + if (args->next) { + printf("\n"); + print_args(args->next); + } +} + +int parse_ftrace_file(char *buf, unsigned long size) +{ + struct format_field *field; + struct print_arg *arg, **list; + struct event *event; + int ret; + + init_input_buf(buf, size); + + event = alloc_event(); + if (!event) + return -ENOMEM; + + event->flags |= EVENT_FL_ISFTRACE; + + event->name = event_read_name(); + if (!event->name) + die("failed to read ftrace event name"); + + if (strcmp(event->name, "function") == 0) + event->flags |= EVENT_FL_ISFUNC; + + else if (strcmp(event->name, "funcgraph_entry") == 0) + event->flags |= EVENT_FL_ISFUNCENT; + + else if (strcmp(event->name, "funcgraph_exit") == 0) + event->flags |= EVENT_FL_ISFUNCRET; + + else if (strcmp(event->name, "bprint") == 0) + event->flags |= EVENT_FL_ISBPRINT; + + event->id = event_read_id(); + if (event->id < 0) + die("failed to read ftrace event id"); + + add_event(event); + + ret = event_read_format(event); + if (ret < 0) + die("failed to read ftrace event format"); + + ret = event_read_print(event); + if (ret < 0) + die("failed to read ftrace event print fmt"); + + /* New ftrace handles args */ + if (ret > 0) + return 0; + /* + * The arguments for ftrace files are parsed by the fields. + * Set up the fields as their arguments. + */ + list = &event->print_fmt.args; + for (field = event->format.fields; field; field = field->next) { + arg = malloc_or_die(sizeof(*arg)); + memset(arg, 0, sizeof(*arg)); + *list = arg; + list = &arg->next; + arg->type = PRINT_FIELD; + arg->field.name = field->name; + arg->field.field = field; + } + return 0; +} + +int parse_event_file(char *buf, unsigned long size, char *sys) +{ + struct event *event; + int ret; + + init_input_buf(buf, size); + + event = alloc_event(); + if (!event) + return -ENOMEM; + + event->name = event_read_name(); + if (!event->name) + die("failed to read event name"); + + event->id = event_read_id(); + if (event->id < 0) + die("failed to read event id"); + + ret = event_read_format(event); + if (ret < 0) { + warning("failed to read event format for %s", event->name); + goto event_failed; + } + + ret = event_read_print(event); + if (ret < 0) { + warning("failed to read event print fmt for %s", event->name); + goto event_failed; + } + + event->system = strdup(sys); + +#define PRINT_ARGS 0 + if (PRINT_ARGS && event->print_fmt.args) + print_args(event->print_fmt.args); + + add_event(event); + return 0; + + event_failed: + event->flags |= EVENT_FL_FAILED; + /* still add it even if it failed */ + add_event(event); + return -1; +} + +void parse_set_info(int nr_cpus, int long_sz) +{ + cpus = nr_cpus; + long_size = long_sz; +} + +int common_pc(struct scripting_context *context) +{ + return parse_common_pc(context->event_data); +} + +int common_flags(struct scripting_context *context) +{ + return parse_common_flags(context->event_data); +} + +int common_lock_depth(struct scripting_context *context) +{ + return parse_common_lock_depth(context->event_data); +} diff --git a/tools/perf/util/trace-event-read.c b/tools/perf/util/trace-event-read.c new file mode 100644 index 00000000..f55cc3a7 --- /dev/null +++ b/tools/perf/util/trace-event-read.c @@ -0,0 +1,539 @@ +/* + * Copyright (C) 2009, Steven Rostedt <srostedt@redhat.com> + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; version 2 of the License (not later!) + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + */ +#define _FILE_OFFSET_BITS 64 + +#include <dirent.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <getopt.h> +#include <stdarg.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <sys/wait.h> +#include <sys/mman.h> +#include <pthread.h> +#include <fcntl.h> +#include <unistd.h> +#include <ctype.h> +#include <errno.h> + +#include "../perf.h" +#include "util.h" +#include "trace-event.h" + +static int input_fd; + +static int read_page; + +int file_bigendian; +int host_bigendian; +static int long_size; + +static unsigned long page_size; + +static ssize_t calc_data_size; +static bool repipe; + +static int do_read(int fd, void *buf, int size) +{ + int rsize = size; + + while (size) { + int ret = read(fd, buf, size); + + if (ret <= 0) + return -1; + + if (repipe) { + int retw = write(STDOUT_FILENO, buf, ret); + + if (retw <= 0 || retw != ret) + die("repiping input file"); + } + + size -= ret; + buf += ret; + } + + return rsize; +} + +static int read_or_die(void *data, int size) +{ + int r; + + r = do_read(input_fd, data, size); + if (r <= 0) + die("reading input file (size expected=%d received=%d)", + size, r); + + if (calc_data_size) + calc_data_size += r; + + return r; +} + +/* If it fails, the next read will report it */ +static void skip(int size) +{ + char buf[BUFSIZ]; + int r; + + while (size) { + r = size > BUFSIZ ? BUFSIZ : size; + read_or_die(buf, r); + size -= r; + }; +} + +static unsigned int read4(void) +{ + unsigned int data; + + read_or_die(&data, 4); + return __data2host4(data); +} + +static unsigned long long read8(void) +{ + unsigned long long data; + + read_or_die(&data, 8); + return __data2host8(data); +} + +static char *read_string(void) +{ + char buf[BUFSIZ]; + char *str = NULL; + int size = 0; + off_t r; + char c; + + for (;;) { + r = read(input_fd, &c, 1); + if (r < 0) + die("reading input file"); + + if (!r) + die("no data"); + + if (repipe) { + int retw = write(STDOUT_FILENO, &c, 1); + + if (retw <= 0 || retw != r) + die("repiping input file string"); + } + + buf[size++] = c; + + if (!c) + break; + } + + if (calc_data_size) + calc_data_size += size; + + str = malloc_or_die(size); + memcpy(str, buf, size); + + return str; +} + +static void read_proc_kallsyms(void) +{ + unsigned int size; + char *buf; + + size = read4(); + if (!size) + return; + + buf = malloc_or_die(size + 1); + read_or_die(buf, size); + buf[size] = '\0'; + + parse_proc_kallsyms(buf, size); + + free(buf); +} + +static void read_ftrace_printk(void) +{ + unsigned int size; + char *buf; + + size = read4(); + if (!size) + return; + + buf = malloc_or_die(size); + read_or_die(buf, size); + + parse_ftrace_printk(buf, size); + + free(buf); +} + +static void read_header_files(void) +{ + unsigned long long size; + char *header_event; + char buf[BUFSIZ]; + + read_or_die(buf, 12); + + if (memcmp(buf, "header_page", 12) != 0) + die("did not read header page"); + + size = read8(); + skip(size); + + /* + * The size field in the page is of type long, + * use that instead, since it represents the kernel. + */ + long_size = header_page_size_size; + + read_or_die(buf, 13); + if (memcmp(buf, "header_event", 13) != 0) + die("did not read header event"); + + size = read8(); + header_event = malloc_or_die(size); + read_or_die(header_event, size); + free(header_event); +} + +static void read_ftrace_file(unsigned long long size) +{ + char *buf; + + buf = malloc_or_die(size); + read_or_die(buf, size); + parse_ftrace_file(buf, size); + free(buf); +} + +static void read_event_file(char *sys, unsigned long long size) +{ + char *buf; + + buf = malloc_or_die(size); + read_or_die(buf, size); + parse_event_file(buf, size, sys); + free(buf); +} + +static void read_ftrace_files(void) +{ + unsigned long long size; + int count; + int i; + + count = read4(); + + for (i = 0; i < count; i++) { + size = read8(); + read_ftrace_file(size); + } +} + +static void read_event_files(void) +{ + unsigned long long size; + char *sys; + int systems; + int count; + int i,x; + + systems = read4(); + + for (i = 0; i < systems; i++) { + sys = read_string(); + + count = read4(); + for (x=0; x < count; x++) { + size = read8(); + read_event_file(sys, size); + } + } +} + +struct cpu_data { + unsigned long long offset; + unsigned long long size; + unsigned long long timestamp; + struct record *next; + char *page; + int cpu; + int index; + int page_size; +}; + +static struct cpu_data *cpu_data; + +static void update_cpu_data_index(int cpu) +{ + cpu_data[cpu].offset += page_size; + cpu_data[cpu].size -= page_size; + cpu_data[cpu].index = 0; +} + +static void get_next_page(int cpu) +{ + off_t save_seek; + off_t ret; + + if (!cpu_data[cpu].page) + return; + + if (read_page) { + if (cpu_data[cpu].size <= page_size) { + free(cpu_data[cpu].page); + cpu_data[cpu].page = NULL; + return; + } + + update_cpu_data_index(cpu); + + /* other parts of the code may expect the pointer to not move */ + save_seek = lseek(input_fd, 0, SEEK_CUR); + + ret = lseek(input_fd, cpu_data[cpu].offset, SEEK_SET); + if (ret == (off_t)-1) + die("failed to lseek"); + ret = read(input_fd, cpu_data[cpu].page, page_size); + if (ret < 0) + die("failed to read page"); + + /* reset the file pointer back */ + lseek(input_fd, save_seek, SEEK_SET); + + return; + } + + munmap(cpu_data[cpu].page, page_size); + cpu_data[cpu].page = NULL; + + if (cpu_data[cpu].size <= page_size) + return; + + update_cpu_data_index(cpu); + + cpu_data[cpu].page = mmap(NULL, page_size, PROT_READ, MAP_PRIVATE, + input_fd, cpu_data[cpu].offset); + if (cpu_data[cpu].page == MAP_FAILED) + die("failed to mmap cpu %d at offset 0x%llx", + cpu, cpu_data[cpu].offset); +} + +static unsigned int type_len4host(unsigned int type_len_ts) +{ + if (file_bigendian) + return (type_len_ts >> 27) & ((1 << 5) - 1); + else + return type_len_ts & ((1 << 5) - 1); +} + +static unsigned int ts4host(unsigned int type_len_ts) +{ + if (file_bigendian) + return type_len_ts & ((1 << 27) - 1); + else + return type_len_ts >> 5; +} + +static int calc_index(void *ptr, int cpu) +{ + return (unsigned long)ptr - (unsigned long)cpu_data[cpu].page; +} + +struct record *trace_peek_data(int cpu) +{ + struct record *data; + void *page = cpu_data[cpu].page; + int idx = cpu_data[cpu].index; + void *ptr = page + idx; + unsigned long long extend; + unsigned int type_len_ts; + unsigned int type_len; + unsigned int delta; + unsigned int length = 0; + + if (cpu_data[cpu].next) + return cpu_data[cpu].next; + + if (!page) + return NULL; + + if (!idx) { + /* FIXME: handle header page */ + if (header_page_ts_size != 8) + die("expected a long long type for timestamp"); + cpu_data[cpu].timestamp = data2host8(ptr); + ptr += 8; + switch (header_page_size_size) { + case 4: + cpu_data[cpu].page_size = data2host4(ptr); + ptr += 4; + break; + case 8: + cpu_data[cpu].page_size = data2host8(ptr); + ptr += 8; + break; + default: + die("bad long size"); + } + ptr = cpu_data[cpu].page + header_page_data_offset; + } + +read_again: + idx = calc_index(ptr, cpu); + + if (idx >= cpu_data[cpu].page_size) { + get_next_page(cpu); + return trace_peek_data(cpu); + } + + type_len_ts = data2host4(ptr); + ptr += 4; + + type_len = type_len4host(type_len_ts); + delta = ts4host(type_len_ts); + + switch (type_len) { + case RINGBUF_TYPE_PADDING: + if (!delta) + die("error, hit unexpected end of page"); + length = data2host4(ptr); + ptr += 4; + length *= 4; + ptr += length; + goto read_again; + + case RINGBUF_TYPE_TIME_EXTEND: + extend = data2host4(ptr); + ptr += 4; + extend <<= TS_SHIFT; + extend += delta; + cpu_data[cpu].timestamp += extend; + goto read_again; + + case RINGBUF_TYPE_TIME_STAMP: + ptr += 12; + break; + case 0: + length = data2host4(ptr); + ptr += 4; + die("here! length=%d", length); + break; + default: + length = type_len * 4; + break; + } + + cpu_data[cpu].timestamp += delta; + + data = malloc_or_die(sizeof(*data)); + memset(data, 0, sizeof(*data)); + + data->ts = cpu_data[cpu].timestamp; + data->size = length; + data->data = ptr; + ptr += length; + + cpu_data[cpu].index = calc_index(ptr, cpu); + cpu_data[cpu].next = data; + + return data; +} + +struct record *trace_read_data(int cpu) +{ + struct record *data; + + data = trace_peek_data(cpu); + cpu_data[cpu].next = NULL; + + return data; +} + +ssize_t trace_report(int fd, bool __repipe) +{ + char buf[BUFSIZ]; + char test[] = { 23, 8, 68 }; + char *version; + int show_version = 0; + int show_funcs = 0; + int show_printk = 0; + ssize_t size; + + calc_data_size = 1; + repipe = __repipe; + + input_fd = fd; + + read_or_die(buf, 3); + if (memcmp(buf, test, 3) != 0) + die("no trace data in the file"); + + read_or_die(buf, 7); + if (memcmp(buf, "tracing", 7) != 0) + die("not a trace file (missing 'tracing' tag)"); + + version = read_string(); + if (show_version) + printf("version = %s\n", version); + free(version); + + read_or_die(buf, 1); + file_bigendian = buf[0]; + host_bigendian = bigendian(); + + read_or_die(buf, 1); + long_size = buf[0]; + + page_size = read4(); + + read_header_files(); + + read_ftrace_files(); + read_event_files(); + read_proc_kallsyms(); + read_ftrace_printk(); + + size = calc_data_size - 1; + calc_data_size = 0; + repipe = false; + + if (show_funcs) { + print_funcs(); + return size; + } + if (show_printk) { + print_printk(); + return size; + } + + return size; +} diff --git a/tools/perf/util/trace-event-scripting.c b/tools/perf/util/trace-event-scripting.c new file mode 100644 index 00000000..c9dcbec7 --- /dev/null +++ b/tools/perf/util/trace-event-scripting.c @@ -0,0 +1,167 @@ +/* + * trace-event-scripting. Scripting engine common and initialization code. + * + * Copyright (C) 2009-2010 Tom Zanussi <tzanussi@gmail.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + */ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <ctype.h> +#include <errno.h> + +#include "../perf.h" +#include "util.h" +#include "trace-event.h" + +struct scripting_context *scripting_context; + +static int stop_script_unsupported(void) +{ + return 0; +} + +static void process_event_unsupported(union perf_event *event __unused, + struct perf_sample *sample __unused, + struct perf_evsel *evsel __unused, + struct perf_session *session __unused, + struct thread *thread __unused) +{ +} + +static void print_python_unsupported_msg(void) +{ + fprintf(stderr, "Python scripting not supported." + " Install libpython and rebuild perf to enable it.\n" + "For example:\n # apt-get install python-dev (ubuntu)" + "\n # yum install python-devel (Fedora)" + "\n etc.\n"); +} + +static int python_start_script_unsupported(const char *script __unused, + int argc __unused, + const char **argv __unused) +{ + print_python_unsupported_msg(); + + return -1; +} + +static int python_generate_script_unsupported(const char *outfile __unused) +{ + print_python_unsupported_msg(); + + return -1; +} + +struct scripting_ops python_scripting_unsupported_ops = { + .name = "Python", + .start_script = python_start_script_unsupported, + .stop_script = stop_script_unsupported, + .process_event = process_event_unsupported, + .generate_script = python_generate_script_unsupported, +}; + +static void register_python_scripting(struct scripting_ops *scripting_ops) +{ + int err; + err = script_spec_register("Python", scripting_ops); + if (err) + die("error registering Python script extension"); + + err = script_spec_register("py", scripting_ops); + if (err) + die("error registering py script extension"); + + scripting_context = malloc(sizeof(struct scripting_context)); +} + +#ifdef NO_LIBPYTHON +void setup_python_scripting(void) +{ + register_python_scripting(&python_scripting_unsupported_ops); +} +#else +extern struct scripting_ops python_scripting_ops; + +void setup_python_scripting(void) +{ + register_python_scripting(&python_scripting_ops); +} +#endif + +static void print_perl_unsupported_msg(void) +{ + fprintf(stderr, "Perl scripting not supported." + " Install libperl and rebuild perf to enable it.\n" + "For example:\n # apt-get install libperl-dev (ubuntu)" + "\n # yum install 'perl(ExtUtils::Embed)' (Fedora)" + "\n etc.\n"); +} + +static int perl_start_script_unsupported(const char *script __unused, + int argc __unused, + const char **argv __unused) +{ + print_perl_unsupported_msg(); + + return -1; +} + +static int perl_generate_script_unsupported(const char *outfile __unused) +{ + print_perl_unsupported_msg(); + + return -1; +} + +struct scripting_ops perl_scripting_unsupported_ops = { + .name = "Perl", + .start_script = perl_start_script_unsupported, + .stop_script = stop_script_unsupported, + .process_event = process_event_unsupported, + .generate_script = perl_generate_script_unsupported, +}; + +static void register_perl_scripting(struct scripting_ops *scripting_ops) +{ + int err; + err = script_spec_register("Perl", scripting_ops); + if (err) + die("error registering Perl script extension"); + + err = script_spec_register("pl", scripting_ops); + if (err) + die("error registering pl script extension"); + + scripting_context = malloc(sizeof(struct scripting_context)); +} + +#ifdef NO_LIBPERL +void setup_perl_scripting(void) +{ + register_perl_scripting(&perl_scripting_unsupported_ops); +} +#else +extern struct scripting_ops perl_scripting_ops; + +void setup_perl_scripting(void) +{ + register_perl_scripting(&perl_scripting_ops); +} +#endif diff --git a/tools/perf/util/trace-event.h b/tools/perf/util/trace-event.h new file mode 100644 index 00000000..f674dda3 --- /dev/null +++ b/tools/perf/util/trace-event.h @@ -0,0 +1,302 @@ +#ifndef __PERF_TRACE_EVENTS_H +#define __PERF_TRACE_EVENTS_H + +#include <stdbool.h> +#include "parse-events.h" +#include "session.h" + +#define __unused __attribute__((unused)) + + +#ifndef PAGE_MASK +#define PAGE_MASK (page_size - 1) +#endif + +enum { + RINGBUF_TYPE_PADDING = 29, + RINGBUF_TYPE_TIME_EXTEND = 30, + RINGBUF_TYPE_TIME_STAMP = 31, +}; + +#ifndef TS_SHIFT +#define TS_SHIFT 27 +#endif + +#define NSECS_PER_SEC 1000000000ULL +#define NSECS_PER_USEC 1000ULL + +enum format_flags { + FIELD_IS_ARRAY = 1, + FIELD_IS_POINTER = 2, + FIELD_IS_SIGNED = 4, + FIELD_IS_STRING = 8, + FIELD_IS_DYNAMIC = 16, + FIELD_IS_FLAG = 32, + FIELD_IS_SYMBOLIC = 64, +}; + +struct format_field { + struct format_field *next; + char *type; + char *name; + int offset; + int size; + unsigned long flags; +}; + +struct format { + int nr_common; + int nr_fields; + struct format_field *common_fields; + struct format_field *fields; +}; + +struct print_arg_atom { + char *atom; +}; + +struct print_arg_string { + char *string; + int offset; +}; + +struct print_arg_field { + char *name; + struct format_field *field; +}; + +struct print_flag_sym { + struct print_flag_sym *next; + char *value; + char *str; +}; + +struct print_arg_typecast { + char *type; + struct print_arg *item; +}; + +struct print_arg_flags { + struct print_arg *field; + char *delim; + struct print_flag_sym *flags; +}; + +struct print_arg_symbol { + struct print_arg *field; + struct print_flag_sym *symbols; +}; + +struct print_arg; + +struct print_arg_op { + char *op; + int prio; + struct print_arg *left; + struct print_arg *right; +}; + +struct print_arg_func { + char *name; + struct print_arg *args; +}; + +enum print_arg_type { + PRINT_NULL, + PRINT_ATOM, + PRINT_FIELD, + PRINT_FLAGS, + PRINT_SYMBOL, + PRINT_TYPE, + PRINT_STRING, + PRINT_OP, +}; + +struct print_arg { + struct print_arg *next; + enum print_arg_type type; + union { + struct print_arg_atom atom; + struct print_arg_field field; + struct print_arg_typecast typecast; + struct print_arg_flags flags; + struct print_arg_symbol symbol; + struct print_arg_func func; + struct print_arg_string string; + struct print_arg_op op; + }; +}; + +struct print_fmt { + char *format; + struct print_arg *args; +}; + +struct event { + struct event *next; + char *name; + int id; + int flags; + struct format format; + struct print_fmt print_fmt; + char *system; +}; + +enum { + EVENT_FL_ISFTRACE = 0x01, + EVENT_FL_ISPRINT = 0x02, + EVENT_FL_ISBPRINT = 0x04, + EVENT_FL_ISFUNC = 0x08, + EVENT_FL_ISFUNCENT = 0x10, + EVENT_FL_ISFUNCRET = 0x20, + + EVENT_FL_FAILED = 0x80000000 +}; + +struct record { + unsigned long long ts; + int size; + void *data; +}; + +struct record *trace_peek_data(int cpu); +struct record *trace_read_data(int cpu); + +void parse_set_info(int nr_cpus, int long_sz); + +ssize_t trace_report(int fd, bool repipe); + +void *malloc_or_die(unsigned int size); + +void parse_cmdlines(char *file, int size); +void parse_proc_kallsyms(char *file, unsigned int size); +void parse_ftrace_printk(char *file, unsigned int size); + +void print_funcs(void); +void print_printk(void); + +int parse_ftrace_file(char *buf, unsigned long size); +int parse_event_file(char *buf, unsigned long size, char *sys); +void print_trace_event(int cpu, void *data, int size); + +extern int file_bigendian; +extern int host_bigendian; + +int bigendian(void); + +static inline unsigned short __data2host2(unsigned short data) +{ + unsigned short swap; + + if (host_bigendian == file_bigendian) + return data; + + swap = ((data & 0xffULL) << 8) | + ((data & (0xffULL << 8)) >> 8); + + return swap; +} + +static inline unsigned int __data2host4(unsigned int data) +{ + unsigned int swap; + + if (host_bigendian == file_bigendian) + return data; + + swap = ((data & 0xffULL) << 24) | + ((data & (0xffULL << 8)) << 8) | + ((data & (0xffULL << 16)) >> 8) | + ((data & (0xffULL << 24)) >> 24); + + return swap; +} + +static inline unsigned long long __data2host8(unsigned long long data) +{ + unsigned long long swap; + + if (host_bigendian == file_bigendian) + return data; + + swap = ((data & 0xffULL) << 56) | + ((data & (0xffULL << 8)) << 40) | + ((data & (0xffULL << 16)) << 24) | + ((data & (0xffULL << 24)) << 8) | + ((data & (0xffULL << 32)) >> 8) | + ((data & (0xffULL << 40)) >> 24) | + ((data & (0xffULL << 48)) >> 40) | + ((data & (0xffULL << 56)) >> 56); + + return swap; +} + +#define data2host2(ptr) __data2host2(*(unsigned short *)ptr) +#define data2host4(ptr) __data2host4(*(unsigned int *)ptr) +#define data2host8(ptr) ({ \ + unsigned long long __val; \ + \ + memcpy(&__val, (ptr), sizeof(unsigned long long)); \ + __data2host8(__val); \ +}) + +extern int header_page_ts_offset; +extern int header_page_ts_size; +extern int header_page_size_offset; +extern int header_page_size_size; +extern int header_page_data_offset; +extern int header_page_data_size; + +extern bool latency_format; + +int trace_parse_common_type(void *data); +int trace_parse_common_pid(void *data); +int parse_common_pc(void *data); +int parse_common_flags(void *data); +int parse_common_lock_depth(void *data); +struct event *trace_find_event(int id); +struct event *trace_find_next_event(struct event *event); +unsigned long long read_size(void *ptr, int size); +unsigned long long +raw_field_value(struct event *event, const char *name, void *data); +void *raw_field_ptr(struct event *event, const char *name, void *data); +unsigned long long eval_flag(const char *flag); + +int read_tracing_data(int fd, struct list_head *pattrs); +ssize_t read_tracing_data_size(int fd, struct list_head *pattrs); + +/* taken from kernel/trace/trace.h */ +enum trace_flag_type { + TRACE_FLAG_IRQS_OFF = 0x01, + TRACE_FLAG_IRQS_NOSUPPORT = 0x02, + TRACE_FLAG_NEED_RESCHED = 0x04, + TRACE_FLAG_HARDIRQ = 0x08, + TRACE_FLAG_SOFTIRQ = 0x10, +}; + +struct scripting_ops { + const char *name; + int (*start_script) (const char *script, int argc, const char **argv); + int (*stop_script) (void); + void (*process_event) (union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel, + struct perf_session *session, + struct thread *thread); + int (*generate_script) (const char *outfile); +}; + +int script_spec_register(const char *spec, struct scripting_ops *ops); + +void setup_perl_scripting(void); +void setup_python_scripting(void); + +struct scripting_context { + void *event_data; +}; + +int common_pc(struct scripting_context *context); +int common_flags(struct scripting_context *context); +int common_lock_depth(struct scripting_context *context); + +#endif /* __PERF_TRACE_EVENTS_H */ diff --git a/tools/perf/util/types.h b/tools/perf/util/types.h new file mode 100644 index 00000000..5f3689a3 --- /dev/null +++ b/tools/perf/util/types.h @@ -0,0 +1,19 @@ +#ifndef __PERF_TYPES_H +#define __PERF_TYPES_H + +#include <stdint.h> + +/* + * We define u64 as uint64_t for every architecture + * so that we can print it with "%"PRIx64 without getting warnings. + */ +typedef uint64_t u64; +typedef int64_t s64; +typedef unsigned int u32; +typedef signed int s32; +typedef unsigned short u16; +typedef signed short s16; +typedef unsigned char u8; +typedef signed char s8; + +#endif /* __PERF_TYPES_H */ diff --git a/tools/perf/util/ui/browser.c b/tools/perf/util/ui/browser.c new file mode 100644 index 00000000..611219f8 --- /dev/null +++ b/tools/perf/util/ui/browser.c @@ -0,0 +1,356 @@ +#include "libslang.h" +#include "ui.h" +#include <linux/compiler.h> +#include <linux/list.h> +#include <linux/rbtree.h> +#include <stdlib.h> +#include <sys/ttydefaults.h> +#include "browser.h" +#include "helpline.h" +#include "../color.h" +#include "../util.h" +#include <stdio.h> + +static int ui_browser__percent_color(double percent, bool current) +{ + if (current) + return HE_COLORSET_SELECTED; + if (percent >= MIN_RED) + return HE_COLORSET_TOP; + if (percent >= MIN_GREEN) + return HE_COLORSET_MEDIUM; + return HE_COLORSET_NORMAL; +} + +void ui_browser__set_color(struct ui_browser *self __used, int color) +{ + SLsmg_set_color(color); +} + +void ui_browser__set_percent_color(struct ui_browser *self, + double percent, bool current) +{ + int color = ui_browser__percent_color(percent, current); + ui_browser__set_color(self, color); +} + +void ui_browser__gotorc(struct ui_browser *self, int y, int x) +{ + SLsmg_gotorc(self->y + y, self->x + x); +} + +void ui_browser__list_head_seek(struct ui_browser *self, off_t offset, int whence) +{ + struct list_head *head = self->entries; + struct list_head *pos; + + switch (whence) { + case SEEK_SET: + pos = head->next; + break; + case SEEK_CUR: + pos = self->top; + break; + case SEEK_END: + pos = head->prev; + break; + default: + return; + } + + if (offset > 0) { + while (offset-- != 0) + pos = pos->next; + } else { + while (offset++ != 0) + pos = pos->prev; + } + + self->top = pos; +} + +void ui_browser__rb_tree_seek(struct ui_browser *self, off_t offset, int whence) +{ + struct rb_root *root = self->entries; + struct rb_node *nd; + + switch (whence) { + case SEEK_SET: + nd = rb_first(root); + break; + case SEEK_CUR: + nd = self->top; + break; + case SEEK_END: + nd = rb_last(root); + break; + default: + return; + } + + if (offset > 0) { + while (offset-- != 0) + nd = rb_next(nd); + } else { + while (offset++ != 0) + nd = rb_prev(nd); + } + + self->top = nd; +} + +unsigned int ui_browser__rb_tree_refresh(struct ui_browser *self) +{ + struct rb_node *nd; + int row = 0; + + if (self->top == NULL) + self->top = rb_first(self->entries); + + nd = self->top; + + while (nd != NULL) { + ui_browser__gotorc(self, row, 0); + self->write(self, nd, row); + if (++row == self->height) + break; + nd = rb_next(nd); + } + + return row; +} + +bool ui_browser__is_current_entry(struct ui_browser *self, unsigned row) +{ + return self->top_idx + row == self->index; +} + +void ui_browser__refresh_dimensions(struct ui_browser *self) +{ + int cols, rows; + newtGetScreenSize(&cols, &rows); + + self->width = cols - 1; + self->height = rows - 2; + self->y = 1; + self->x = 0; +} + +void ui_browser__reset_index(struct ui_browser *self) +{ + self->index = self->top_idx = 0; + self->seek(self, 0, SEEK_SET); +} + +void ui_browser__add_exit_key(struct ui_browser *self, int key) +{ + newtFormAddHotKey(self->form, key); +} + +void ui_browser__add_exit_keys(struct ui_browser *self, int keys[]) +{ + int i = 0; + + while (keys[i] && i < 64) { + ui_browser__add_exit_key(self, keys[i]); + ++i; + } +} + +void __ui_browser__show_title(struct ui_browser *browser, const char *title) +{ + SLsmg_gotorc(0, 0); + ui_browser__set_color(browser, NEWT_COLORSET_ROOT); + slsmg_write_nstring(title, browser->width); +} + +void ui_browser__show_title(struct ui_browser *browser, const char *title) +{ + pthread_mutex_lock(&ui__lock); + __ui_browser__show_title(browser, title); + pthread_mutex_unlock(&ui__lock); +} + +int ui_browser__show(struct ui_browser *self, const char *title, + const char *helpline, ...) +{ + va_list ap; + int keys[] = { NEWT_KEY_UP, NEWT_KEY_DOWN, NEWT_KEY_PGUP, + NEWT_KEY_PGDN, NEWT_KEY_HOME, NEWT_KEY_END, ' ', + NEWT_KEY_LEFT, NEWT_KEY_ESCAPE, 'q', CTRL('c'), 0 }; + + if (self->form != NULL) + newtFormDestroy(self->form); + + ui_browser__refresh_dimensions(self); + self->form = newtForm(NULL, NULL, 0); + if (self->form == NULL) + return -1; + + self->sb = newtVerticalScrollbar(self->width, 1, self->height, + HE_COLORSET_NORMAL, + HE_COLORSET_SELECTED); + if (self->sb == NULL) + return -1; + + pthread_mutex_lock(&ui__lock); + __ui_browser__show_title(self, title); + + ui_browser__add_exit_keys(self, keys); + newtFormAddComponent(self->form, self->sb); + + va_start(ap, helpline); + ui_helpline__vpush(helpline, ap); + va_end(ap); + pthread_mutex_unlock(&ui__lock); + return 0; +} + +void ui_browser__hide(struct ui_browser *self) +{ + pthread_mutex_lock(&ui__lock); + newtFormDestroy(self->form); + self->form = NULL; + ui_helpline__pop(); + pthread_mutex_unlock(&ui__lock); +} + +int ui_browser__refresh(struct ui_browser *self) +{ + int row; + + pthread_mutex_lock(&ui__lock); + newtScrollbarSet(self->sb, self->index, self->nr_entries - 1); + row = self->refresh(self); + ui_browser__set_color(self, HE_COLORSET_NORMAL); + SLsmg_fill_region(self->y + row, self->x, + self->height - row, self->width, ' '); + pthread_mutex_unlock(&ui__lock); + + return 0; +} + +int ui_browser__run(struct ui_browser *self) +{ + struct newtExitStruct es; + + if (ui_browser__refresh(self) < 0) + return -1; + + while (1) { + off_t offset; + + newtFormRun(self->form, &es); + + if (es.reason != NEWT_EXIT_HOTKEY) + break; + switch (es.u.key) { + case NEWT_KEY_DOWN: + if (self->index == self->nr_entries - 1) + break; + ++self->index; + if (self->index == self->top_idx + self->height) { + ++self->top_idx; + self->seek(self, +1, SEEK_CUR); + } + break; + case NEWT_KEY_UP: + if (self->index == 0) + break; + --self->index; + if (self->index < self->top_idx) { + --self->top_idx; + self->seek(self, -1, SEEK_CUR); + } + break; + case NEWT_KEY_PGDN: + case ' ': + if (self->top_idx + self->height > self->nr_entries - 1) + break; + + offset = self->height; + if (self->index + offset > self->nr_entries - 1) + offset = self->nr_entries - 1 - self->index; + self->index += offset; + self->top_idx += offset; + self->seek(self, +offset, SEEK_CUR); + break; + case NEWT_KEY_PGUP: + if (self->top_idx == 0) + break; + + if (self->top_idx < self->height) + offset = self->top_idx; + else + offset = self->height; + + self->index -= offset; + self->top_idx -= offset; + self->seek(self, -offset, SEEK_CUR); + break; + case NEWT_KEY_HOME: + ui_browser__reset_index(self); + break; + case NEWT_KEY_END: + offset = self->height - 1; + if (offset >= self->nr_entries) + offset = self->nr_entries - 1; + + self->index = self->nr_entries - 1; + self->top_idx = self->index - offset; + self->seek(self, -offset, SEEK_END); + break; + default: + return es.u.key; + } + if (ui_browser__refresh(self) < 0) + return -1; + } + return -1; +} + +unsigned int ui_browser__list_head_refresh(struct ui_browser *self) +{ + struct list_head *pos; + struct list_head *head = self->entries; + int row = 0; + + if (self->top == NULL || self->top == self->entries) + self->top = head->next; + + pos = self->top; + + list_for_each_from(pos, head) { + ui_browser__gotorc(self, row, 0); + self->write(self, pos, row); + if (++row == self->height) + break; + } + + return row; +} + +static struct newtPercentTreeColors { + const char *topColorFg, *topColorBg; + const char *mediumColorFg, *mediumColorBg; + const char *normalColorFg, *normalColorBg; + const char *selColorFg, *selColorBg; + const char *codeColorFg, *codeColorBg; +} defaultPercentTreeColors = { + "red", "lightgray", + "green", "lightgray", + "black", "lightgray", + "lightgray", "magenta", + "blue", "lightgray", +}; + +void ui_browser__init(void) +{ + struct newtPercentTreeColors *c = &defaultPercentTreeColors; + + sltt_set_color(HE_COLORSET_TOP, NULL, c->topColorFg, c->topColorBg); + sltt_set_color(HE_COLORSET_MEDIUM, NULL, c->mediumColorFg, c->mediumColorBg); + sltt_set_color(HE_COLORSET_NORMAL, NULL, c->normalColorFg, c->normalColorBg); + sltt_set_color(HE_COLORSET_SELECTED, NULL, c->selColorFg, c->selColorBg); + sltt_set_color(HE_COLORSET_CODE, NULL, c->codeColorFg, c->codeColorBg); +} diff --git a/tools/perf/util/ui/browser.h b/tools/perf/util/ui/browser.h new file mode 100644 index 00000000..fc63dda1 --- /dev/null +++ b/tools/perf/util/ui/browser.h @@ -0,0 +1,52 @@ +#ifndef _PERF_UI_BROWSER_H_ +#define _PERF_UI_BROWSER_H_ 1 + +#include <stdbool.h> +#include <newt.h> +#include <sys/types.h> +#include "../types.h" + +#define HE_COLORSET_TOP 50 +#define HE_COLORSET_MEDIUM 51 +#define HE_COLORSET_NORMAL 52 +#define HE_COLORSET_SELECTED 53 +#define HE_COLORSET_CODE 54 + +struct ui_browser { + newtComponent form, sb; + u64 index, top_idx; + void *top, *entries; + u16 y, x, width, height; + void *priv; + unsigned int (*refresh)(struct ui_browser *self); + void (*write)(struct ui_browser *self, void *entry, int row); + void (*seek)(struct ui_browser *self, off_t offset, int whence); + u32 nr_entries; +}; + +void ui_browser__set_color(struct ui_browser *self, int color); +void ui_browser__set_percent_color(struct ui_browser *self, + double percent, bool current); +bool ui_browser__is_current_entry(struct ui_browser *self, unsigned row); +void ui_browser__refresh_dimensions(struct ui_browser *self); +void ui_browser__reset_index(struct ui_browser *self); + +void ui_browser__gotorc(struct ui_browser *self, int y, int x); +void ui_browser__add_exit_key(struct ui_browser *self, int key); +void ui_browser__add_exit_keys(struct ui_browser *self, int keys[]); +void __ui_browser__show_title(struct ui_browser *browser, const char *title); +void ui_browser__show_title(struct ui_browser *browser, const char *title); +int ui_browser__show(struct ui_browser *self, const char *title, + const char *helpline, ...); +void ui_browser__hide(struct ui_browser *self); +int ui_browser__refresh(struct ui_browser *self); +int ui_browser__run(struct ui_browser *self); + +void ui_browser__rb_tree_seek(struct ui_browser *self, off_t offset, int whence); +unsigned int ui_browser__rb_tree_refresh(struct ui_browser *self); + +void ui_browser__list_head_seek(struct ui_browser *self, off_t offset, int whence); +unsigned int ui_browser__list_head_refresh(struct ui_browser *self); + +void ui_browser__init(void); +#endif /* _PERF_UI_BROWSER_H_ */ diff --git a/tools/perf/util/ui/browsers/annotate.c b/tools/perf/util/ui/browsers/annotate.c new file mode 100644 index 00000000..0229723a --- /dev/null +++ b/tools/perf/util/ui/browsers/annotate.c @@ -0,0 +1,302 @@ +#include "../browser.h" +#include "../helpline.h" +#include "../libslang.h" +#include "../../annotate.h" +#include "../../hist.h" +#include "../../sort.h" +#include "../../symbol.h" +#include <pthread.h> + +static void ui__error_window(const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + newtWinMessagev((char *)"Error", (char *)"Ok", (char *)fmt, ap); + va_end(ap); +} + +struct annotate_browser { + struct ui_browser b; + struct rb_root entries; + struct rb_node *curr_hot; +}; + +struct objdump_line_rb_node { + struct rb_node rb_node; + double percent; + u32 idx; +}; + +static inline +struct objdump_line_rb_node *objdump_line__rb(struct objdump_line *self) +{ + return (struct objdump_line_rb_node *)(self + 1); +} + +static void annotate_browser__write(struct ui_browser *self, void *entry, int row) +{ + struct objdump_line *ol = rb_entry(entry, struct objdump_line, node); + bool current_entry = ui_browser__is_current_entry(self, row); + int width = self->width; + + if (ol->offset != -1) { + struct objdump_line_rb_node *olrb = objdump_line__rb(ol); + ui_browser__set_percent_color(self, olrb->percent, current_entry); + slsmg_printf(" %7.2f ", olrb->percent); + } else { + ui_browser__set_percent_color(self, 0, current_entry); + slsmg_write_nstring(" ", 9); + } + + SLsmg_write_char(':'); + slsmg_write_nstring(" ", 8); + if (!*ol->line) + slsmg_write_nstring(" ", width - 18); + else + slsmg_write_nstring(ol->line, width - 18); + + if (!current_entry) + ui_browser__set_color(self, HE_COLORSET_CODE); +} + +static double objdump_line__calc_percent(struct objdump_line *self, + struct symbol *sym, int evidx) +{ + double percent = 0.0; + + if (self->offset != -1) { + int len = sym->end - sym->start; + unsigned int hits = 0; + struct annotation *notes = symbol__annotation(sym); + struct source_line *src_line = notes->src->lines; + struct sym_hist *h = annotation__histogram(notes, evidx); + s64 offset = self->offset; + struct objdump_line *next; + + next = objdump__get_next_ip_line(¬es->src->source, self); + while (offset < (s64)len && + (next == NULL || offset < next->offset)) { + if (src_line) { + percent += src_line[offset].percent; + } else + hits += h->addr[offset]; + + ++offset; + } + /* + * If the percentage wasn't already calculated in + * symbol__get_source_line, do it now: + */ + if (src_line == NULL && h->sum) + percent = 100.0 * hits / h->sum; + } + + return percent; +} + +static void objdump__insert_line(struct rb_root *self, + struct objdump_line_rb_node *line) +{ + struct rb_node **p = &self->rb_node; + struct rb_node *parent = NULL; + struct objdump_line_rb_node *l; + + while (*p != NULL) { + parent = *p; + l = rb_entry(parent, struct objdump_line_rb_node, rb_node); + if (line->percent < l->percent) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + rb_link_node(&line->rb_node, parent, p); + rb_insert_color(&line->rb_node, self); +} + +static void annotate_browser__set_top(struct annotate_browser *self, + struct rb_node *nd) +{ + struct objdump_line_rb_node *rbpos; + struct objdump_line *pos; + unsigned back; + + ui_browser__refresh_dimensions(&self->b); + back = self->b.height / 2; + rbpos = rb_entry(nd, struct objdump_line_rb_node, rb_node); + pos = ((struct objdump_line *)rbpos) - 1; + self->b.top_idx = self->b.index = rbpos->idx; + + while (self->b.top_idx != 0 && back != 0) { + pos = list_entry(pos->node.prev, struct objdump_line, node); + + --self->b.top_idx; + --back; + } + + self->b.top = pos; + self->curr_hot = nd; +} + +static void annotate_browser__calc_percent(struct annotate_browser *browser, + int evidx) +{ + struct symbol *sym = browser->b.priv; + struct annotation *notes = symbol__annotation(sym); + struct objdump_line *pos; + + browser->entries = RB_ROOT; + + pthread_mutex_lock(¬es->lock); + + list_for_each_entry(pos, ¬es->src->source, node) { + struct objdump_line_rb_node *rbpos = objdump_line__rb(pos); + rbpos->percent = objdump_line__calc_percent(pos, sym, evidx); + if (rbpos->percent < 0.01) { + RB_CLEAR_NODE(&rbpos->rb_node); + continue; + } + objdump__insert_line(&browser->entries, rbpos); + } + pthread_mutex_unlock(¬es->lock); + + browser->curr_hot = rb_last(&browser->entries); +} + +static int annotate_browser__run(struct annotate_browser *self, int evidx, + int refresh) +{ + struct rb_node *nd = NULL; + struct symbol *sym = self->b.priv; + /* + * RIGHT To allow builtin-annotate to cycle thru multiple symbols by + * examining the exit key for this function. + */ + int exit_keys[] = { 'H', NEWT_KEY_TAB, NEWT_KEY_UNTAB, + NEWT_KEY_RIGHT, 0 }; + int key; + + if (ui_browser__show(&self->b, sym->name, + "<-, -> or ESC: exit, TAB/shift+TAB: " + "cycle hottest lines, H: Hottest") < 0) + return -1; + + ui_browser__add_exit_keys(&self->b, exit_keys); + annotate_browser__calc_percent(self, evidx); + + if (self->curr_hot) + annotate_browser__set_top(self, self->curr_hot); + + nd = self->curr_hot; + + if (refresh != 0) + newtFormSetTimer(self->b.form, refresh); + + while (1) { + key = ui_browser__run(&self->b); + + if (refresh != 0) { + annotate_browser__calc_percent(self, evidx); + /* + * Current line focus got out of the list of most active + * lines, NULL it so that if TAB|UNTAB is pressed, we + * move to curr_hot (current hottest line). + */ + if (nd != NULL && RB_EMPTY_NODE(nd)) + nd = NULL; + } + + switch (key) { + case -1: + /* + * FIXME we need to check if it was + * es.reason == NEWT_EXIT_TIMER + */ + if (refresh != 0) + symbol__annotate_decay_histogram(sym, evidx); + continue; + case NEWT_KEY_TAB: + if (nd != NULL) { + nd = rb_prev(nd); + if (nd == NULL) + nd = rb_last(&self->entries); + } else + nd = self->curr_hot; + break; + case NEWT_KEY_UNTAB: + if (nd != NULL) + nd = rb_next(nd); + if (nd == NULL) + nd = rb_first(&self->entries); + else + nd = self->curr_hot; + break; + case 'H': + nd = self->curr_hot; + break; + default: + goto out; + } + + if (nd != NULL) + annotate_browser__set_top(self, nd); + } +out: + ui_browser__hide(&self->b); + return key; +} + +int hist_entry__tui_annotate(struct hist_entry *he, int evidx) +{ + return symbol__tui_annotate(he->ms.sym, he->ms.map, evidx, 0); +} + +int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx, + int refresh) +{ + struct objdump_line *pos, *n; + struct annotation *notes; + struct annotate_browser browser = { + .b = { + .refresh = ui_browser__list_head_refresh, + .seek = ui_browser__list_head_seek, + .write = annotate_browser__write, + .priv = sym, + }, + }; + int ret; + + if (sym == NULL) + return -1; + + if (map->dso->annotate_warned) + return -1; + + if (symbol__annotate(sym, map, sizeof(struct objdump_line_rb_node)) < 0) { + ui__error_window(ui_helpline__last_msg); + return -1; + } + + ui_helpline__push("Press <- or ESC to exit"); + + notes = symbol__annotation(sym); + + list_for_each_entry(pos, ¬es->src->source, node) { + struct objdump_line_rb_node *rbpos; + size_t line_len = strlen(pos->line); + + if (browser.b.width < line_len) + browser.b.width = line_len; + rbpos = objdump_line__rb(pos); + rbpos->idx = browser.b.nr_entries++; + } + + browser.b.entries = ¬es->src->source, + browser.b.width += 18; /* Percentage */ + ret = annotate_browser__run(&browser, evidx, refresh); + list_for_each_entry_safe(pos, n, ¬es->src->source, node) { + list_del(&pos->node); + objdump_line__free(pos); + } + return ret; +} diff --git a/tools/perf/util/ui/browsers/hists.c b/tools/perf/util/ui/browsers/hists.c new file mode 100644 index 00000000..5d767c62 --- /dev/null +++ b/tools/perf/util/ui/browsers/hists.c @@ -0,0 +1,1138 @@ +#define _GNU_SOURCE +#include <stdio.h> +#undef _GNU_SOURCE +#include "../libslang.h" +#include <stdlib.h> +#include <string.h> +#include <newt.h> +#include <linux/rbtree.h> + +#include "../../evsel.h" +#include "../../evlist.h" +#include "../../hist.h" +#include "../../pstack.h" +#include "../../sort.h" +#include "../../util.h" + +#include "../browser.h" +#include "../helpline.h" +#include "../util.h" +#include "map.h" + +struct hist_browser { + struct ui_browser b; + struct hists *hists; + struct hist_entry *he_selection; + struct map_symbol *selection; +}; + +static void hist_browser__refresh_dimensions(struct hist_browser *self) +{ + /* 3 == +/- toggle symbol before actual hist_entry rendering */ + self->b.width = 3 + (hists__sort_list_width(self->hists) + + sizeof("[k]")); +} + +static void hist_browser__reset(struct hist_browser *self) +{ + self->b.nr_entries = self->hists->nr_entries; + hist_browser__refresh_dimensions(self); + ui_browser__reset_index(&self->b); +} + +static char tree__folded_sign(bool unfolded) +{ + return unfolded ? '-' : '+'; +} + +static char map_symbol__folded(const struct map_symbol *self) +{ + return self->has_children ? tree__folded_sign(self->unfolded) : ' '; +} + +static char hist_entry__folded(const struct hist_entry *self) +{ + return map_symbol__folded(&self->ms); +} + +static char callchain_list__folded(const struct callchain_list *self) +{ + return map_symbol__folded(&self->ms); +} + +static void map_symbol__set_folding(struct map_symbol *self, bool unfold) +{ + self->unfolded = unfold ? self->has_children : false; +} + +static int callchain_node__count_rows_rb_tree(struct callchain_node *self) +{ + int n = 0; + struct rb_node *nd; + + for (nd = rb_first(&self->rb_root); nd; nd = rb_next(nd)) { + struct callchain_node *child = rb_entry(nd, struct callchain_node, rb_node); + struct callchain_list *chain; + char folded_sign = ' '; /* No children */ + + list_for_each_entry(chain, &child->val, list) { + ++n; + /* We need this because we may not have children */ + folded_sign = callchain_list__folded(chain); + if (folded_sign == '+') + break; + } + + if (folded_sign == '-') /* Have children and they're unfolded */ + n += callchain_node__count_rows_rb_tree(child); + } + + return n; +} + +static int callchain_node__count_rows(struct callchain_node *node) +{ + struct callchain_list *chain; + bool unfolded = false; + int n = 0; + + list_for_each_entry(chain, &node->val, list) { + ++n; + unfolded = chain->ms.unfolded; + } + + if (unfolded) + n += callchain_node__count_rows_rb_tree(node); + + return n; +} + +static int callchain__count_rows(struct rb_root *chain) +{ + struct rb_node *nd; + int n = 0; + + for (nd = rb_first(chain); nd; nd = rb_next(nd)) { + struct callchain_node *node = rb_entry(nd, struct callchain_node, rb_node); + n += callchain_node__count_rows(node); + } + + return n; +} + +static bool map_symbol__toggle_fold(struct map_symbol *self) +{ + if (!self->has_children) + return false; + + self->unfolded = !self->unfolded; + return true; +} + +static void callchain_node__init_have_children_rb_tree(struct callchain_node *self) +{ + struct rb_node *nd = rb_first(&self->rb_root); + + for (nd = rb_first(&self->rb_root); nd; nd = rb_next(nd)) { + struct callchain_node *child = rb_entry(nd, struct callchain_node, rb_node); + struct callchain_list *chain; + bool first = true; + + list_for_each_entry(chain, &child->val, list) { + if (first) { + first = false; + chain->ms.has_children = chain->list.next != &child->val || + !RB_EMPTY_ROOT(&child->rb_root); + } else + chain->ms.has_children = chain->list.next == &child->val && + !RB_EMPTY_ROOT(&child->rb_root); + } + + callchain_node__init_have_children_rb_tree(child); + } +} + +static void callchain_node__init_have_children(struct callchain_node *self) +{ + struct callchain_list *chain; + + list_for_each_entry(chain, &self->val, list) + chain->ms.has_children = !RB_EMPTY_ROOT(&self->rb_root); + + callchain_node__init_have_children_rb_tree(self); +} + +static void callchain__init_have_children(struct rb_root *self) +{ + struct rb_node *nd; + + for (nd = rb_first(self); nd; nd = rb_next(nd)) { + struct callchain_node *node = rb_entry(nd, struct callchain_node, rb_node); + callchain_node__init_have_children(node); + } +} + +static void hist_entry__init_have_children(struct hist_entry *self) +{ + if (!self->init_have_children) { + self->ms.has_children = !RB_EMPTY_ROOT(&self->sorted_chain); + callchain__init_have_children(&self->sorted_chain); + self->init_have_children = true; + } +} + +static bool hist_browser__toggle_fold(struct hist_browser *self) +{ + if (map_symbol__toggle_fold(self->selection)) { + struct hist_entry *he = self->he_selection; + + hist_entry__init_have_children(he); + self->hists->nr_entries -= he->nr_rows; + + if (he->ms.unfolded) + he->nr_rows = callchain__count_rows(&he->sorted_chain); + else + he->nr_rows = 0; + self->hists->nr_entries += he->nr_rows; + self->b.nr_entries = self->hists->nr_entries; + + return true; + } + + /* If it doesn't have children, no toggling performed */ + return false; +} + +static int callchain_node__set_folding_rb_tree(struct callchain_node *self, bool unfold) +{ + int n = 0; + struct rb_node *nd; + + for (nd = rb_first(&self->rb_root); nd; nd = rb_next(nd)) { + struct callchain_node *child = rb_entry(nd, struct callchain_node, rb_node); + struct callchain_list *chain; + bool has_children = false; + + list_for_each_entry(chain, &child->val, list) { + ++n; + map_symbol__set_folding(&chain->ms, unfold); + has_children = chain->ms.has_children; + } + + if (has_children) + n += callchain_node__set_folding_rb_tree(child, unfold); + } + + return n; +} + +static int callchain_node__set_folding(struct callchain_node *node, bool unfold) +{ + struct callchain_list *chain; + bool has_children = false; + int n = 0; + + list_for_each_entry(chain, &node->val, list) { + ++n; + map_symbol__set_folding(&chain->ms, unfold); + has_children = chain->ms.has_children; + } + + if (has_children) + n += callchain_node__set_folding_rb_tree(node, unfold); + + return n; +} + +static int callchain__set_folding(struct rb_root *chain, bool unfold) +{ + struct rb_node *nd; + int n = 0; + + for (nd = rb_first(chain); nd; nd = rb_next(nd)) { + struct callchain_node *node = rb_entry(nd, struct callchain_node, rb_node); + n += callchain_node__set_folding(node, unfold); + } + + return n; +} + +static void hist_entry__set_folding(struct hist_entry *self, bool unfold) +{ + hist_entry__init_have_children(self); + map_symbol__set_folding(&self->ms, unfold); + + if (self->ms.has_children) { + int n = callchain__set_folding(&self->sorted_chain, unfold); + self->nr_rows = unfold ? n : 0; + } else + self->nr_rows = 0; +} + +static void hists__set_folding(struct hists *self, bool unfold) +{ + struct rb_node *nd; + + self->nr_entries = 0; + + for (nd = rb_first(&self->entries); nd; nd = rb_next(nd)) { + struct hist_entry *he = rb_entry(nd, struct hist_entry, rb_node); + hist_entry__set_folding(he, unfold); + self->nr_entries += 1 + he->nr_rows; + } +} + +static void hist_browser__set_folding(struct hist_browser *self, bool unfold) +{ + hists__set_folding(self->hists, unfold); + self->b.nr_entries = self->hists->nr_entries; + /* Go to the start, we may be way after valid entries after a collapse */ + ui_browser__reset_index(&self->b); +} + +static int hist_browser__run(struct hist_browser *self, const char *title) +{ + int key; + int exit_keys[] = { 'a', '?', 'h', 'C', 'd', 'D', 'E', 't', + NEWT_KEY_ENTER, NEWT_KEY_RIGHT, NEWT_KEY_LEFT, + NEWT_KEY_TAB, NEWT_KEY_UNTAB, 0, }; + + self->b.entries = &self->hists->entries; + self->b.nr_entries = self->hists->nr_entries; + + hist_browser__refresh_dimensions(self); + + if (ui_browser__show(&self->b, title, + "Press '?' for help on key bindings") < 0) + return -1; + + ui_browser__add_exit_keys(&self->b, exit_keys); + + while (1) { + key = ui_browser__run(&self->b); + + switch (key) { + case 'D': { /* Debug */ + static int seq; + struct hist_entry *h = rb_entry(self->b.top, + struct hist_entry, rb_node); + ui_helpline__pop(); + ui_helpline__fpush("%d: nr_ent=(%d,%d), height=%d, idx=%d, fve: idx=%d, row_off=%d, nrows=%d", + seq++, self->b.nr_entries, + self->hists->nr_entries, + self->b.height, + self->b.index, + self->b.top_idx, + h->row_offset, h->nr_rows); + } + break; + case 'C': + /* Collapse the whole world. */ + hist_browser__set_folding(self, false); + break; + case 'E': + /* Expand the whole world. */ + hist_browser__set_folding(self, true); + break; + case NEWT_KEY_ENTER: + if (hist_browser__toggle_fold(self)) + break; + /* fall thru */ + default: + goto out; + } + } +out: + ui_browser__hide(&self->b); + return key; +} + +static char *callchain_list__sym_name(struct callchain_list *self, + char *bf, size_t bfsize) +{ + if (self->ms.sym) + return self->ms.sym->name; + + snprintf(bf, bfsize, "%#" PRIx64, self->ip); + return bf; +} + +#define LEVEL_OFFSET_STEP 3 + +static int hist_browser__show_callchain_node_rb_tree(struct hist_browser *self, + struct callchain_node *chain_node, + u64 total, int level, + unsigned short row, + off_t *row_offset, + bool *is_current_entry) +{ + struct rb_node *node; + int first_row = row, width, offset = level * LEVEL_OFFSET_STEP; + u64 new_total, remaining; + + if (callchain_param.mode == CHAIN_GRAPH_REL) + new_total = chain_node->children_hit; + else + new_total = total; + + remaining = new_total; + node = rb_first(&chain_node->rb_root); + while (node) { + struct callchain_node *child = rb_entry(node, struct callchain_node, rb_node); + struct rb_node *next = rb_next(node); + u64 cumul = callchain_cumul_hits(child); + struct callchain_list *chain; + char folded_sign = ' '; + int first = true; + int extra_offset = 0; + + remaining -= cumul; + + list_for_each_entry(chain, &child->val, list) { + char ipstr[BITS_PER_LONG / 4 + 1], *alloc_str; + const char *str; + int color; + bool was_first = first; + + if (first) + first = false; + else + extra_offset = LEVEL_OFFSET_STEP; + + folded_sign = callchain_list__folded(chain); + if (*row_offset != 0) { + --*row_offset; + goto do_next; + } + + alloc_str = NULL; + str = callchain_list__sym_name(chain, ipstr, sizeof(ipstr)); + if (was_first) { + double percent = cumul * 100.0 / new_total; + + if (asprintf(&alloc_str, "%2.2f%% %s", percent, str) < 0) + str = "Not enough memory!"; + else + str = alloc_str; + } + + color = HE_COLORSET_NORMAL; + width = self->b.width - (offset + extra_offset + 2); + if (ui_browser__is_current_entry(&self->b, row)) { + self->selection = &chain->ms; + color = HE_COLORSET_SELECTED; + *is_current_entry = true; + } + + ui_browser__set_color(&self->b, color); + ui_browser__gotorc(&self->b, row, 0); + slsmg_write_nstring(" ", offset + extra_offset); + slsmg_printf("%c ", folded_sign); + slsmg_write_nstring(str, width); + free(alloc_str); + + if (++row == self->b.height) + goto out; +do_next: + if (folded_sign == '+') + break; + } + + if (folded_sign == '-') { + const int new_level = level + (extra_offset ? 2 : 1); + row += hist_browser__show_callchain_node_rb_tree(self, child, new_total, + new_level, row, row_offset, + is_current_entry); + } + if (row == self->b.height) + goto out; + node = next; + } +out: + return row - first_row; +} + +static int hist_browser__show_callchain_node(struct hist_browser *self, + struct callchain_node *node, + int level, unsigned short row, + off_t *row_offset, + bool *is_current_entry) +{ + struct callchain_list *chain; + int first_row = row, + offset = level * LEVEL_OFFSET_STEP, + width = self->b.width - offset; + char folded_sign = ' '; + + list_for_each_entry(chain, &node->val, list) { + char ipstr[BITS_PER_LONG / 4 + 1], *s; + int color; + + folded_sign = callchain_list__folded(chain); + + if (*row_offset != 0) { + --*row_offset; + continue; + } + + color = HE_COLORSET_NORMAL; + if (ui_browser__is_current_entry(&self->b, row)) { + self->selection = &chain->ms; + color = HE_COLORSET_SELECTED; + *is_current_entry = true; + } + + s = callchain_list__sym_name(chain, ipstr, sizeof(ipstr)); + ui_browser__gotorc(&self->b, row, 0); + ui_browser__set_color(&self->b, color); + slsmg_write_nstring(" ", offset); + slsmg_printf("%c ", folded_sign); + slsmg_write_nstring(s, width - 2); + + if (++row == self->b.height) + goto out; + } + + if (folded_sign == '-') + row += hist_browser__show_callchain_node_rb_tree(self, node, + self->hists->stats.total_period, + level + 1, row, + row_offset, + is_current_entry); +out: + return row - first_row; +} + +static int hist_browser__show_callchain(struct hist_browser *self, + struct rb_root *chain, + int level, unsigned short row, + off_t *row_offset, + bool *is_current_entry) +{ + struct rb_node *nd; + int first_row = row; + + for (nd = rb_first(chain); nd; nd = rb_next(nd)) { + struct callchain_node *node = rb_entry(nd, struct callchain_node, rb_node); + + row += hist_browser__show_callchain_node(self, node, level, + row, row_offset, + is_current_entry); + if (row == self->b.height) + break; + } + + return row - first_row; +} + +static int hist_browser__show_entry(struct hist_browser *self, + struct hist_entry *entry, + unsigned short row) +{ + char s[256]; + double percent; + int printed = 0; + int color, width = self->b.width; + char folded_sign = ' '; + bool current_entry = ui_browser__is_current_entry(&self->b, row); + off_t row_offset = entry->row_offset; + + if (current_entry) { + self->he_selection = entry; + self->selection = &entry->ms; + } + + if (symbol_conf.use_callchain) { + hist_entry__init_have_children(entry); + folded_sign = hist_entry__folded(entry); + } + + if (row_offset == 0) { + hist_entry__snprintf(entry, s, sizeof(s), self->hists, NULL, false, + 0, false, self->hists->stats.total_period); + percent = (entry->period * 100.0) / self->hists->stats.total_period; + + color = HE_COLORSET_SELECTED; + if (!current_entry) { + if (percent >= MIN_RED) + color = HE_COLORSET_TOP; + else if (percent >= MIN_GREEN) + color = HE_COLORSET_MEDIUM; + else + color = HE_COLORSET_NORMAL; + } + + ui_browser__set_color(&self->b, color); + ui_browser__gotorc(&self->b, row, 0); + if (symbol_conf.use_callchain) { + slsmg_printf("%c ", folded_sign); + width -= 2; + } + slsmg_write_nstring(s, width); + ++row; + ++printed; + } else + --row_offset; + + if (folded_sign == '-' && row != self->b.height) { + printed += hist_browser__show_callchain(self, &entry->sorted_chain, + 1, row, &row_offset, + ¤t_entry); + if (current_entry) + self->he_selection = entry; + } + + return printed; +} + +static unsigned int hist_browser__refresh(struct ui_browser *self) +{ + unsigned row = 0; + struct rb_node *nd; + struct hist_browser *hb = container_of(self, struct hist_browser, b); + + if (self->top == NULL) + self->top = rb_first(&hb->hists->entries); + + for (nd = self->top; nd; nd = rb_next(nd)) { + struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node); + + if (h->filtered) + continue; + + row += hist_browser__show_entry(hb, h, row); + if (row == self->height) + break; + } + + return row; +} + +static struct rb_node *hists__filter_entries(struct rb_node *nd) +{ + while (nd != NULL) { + struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node); + if (!h->filtered) + return nd; + + nd = rb_next(nd); + } + + return NULL; +} + +static struct rb_node *hists__filter_prev_entries(struct rb_node *nd) +{ + while (nd != NULL) { + struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node); + if (!h->filtered) + return nd; + + nd = rb_prev(nd); + } + + return NULL; +} + +static void ui_browser__hists_seek(struct ui_browser *self, + off_t offset, int whence) +{ + struct hist_entry *h; + struct rb_node *nd; + bool first = true; + + if (self->nr_entries == 0) + return; + + switch (whence) { + case SEEK_SET: + nd = hists__filter_entries(rb_first(self->entries)); + break; + case SEEK_CUR: + nd = self->top; + goto do_offset; + case SEEK_END: + nd = hists__filter_prev_entries(rb_last(self->entries)); + first = false; + break; + default: + return; + } + + /* + * Moves not relative to the first visible entry invalidates its + * row_offset: + */ + h = rb_entry(self->top, struct hist_entry, rb_node); + h->row_offset = 0; + + /* + * Here we have to check if nd is expanded (+), if it is we can't go + * the next top level hist_entry, instead we must compute an offset of + * what _not_ to show and not change the first visible entry. + * + * This offset increments when we are going from top to bottom and + * decreases when we're going from bottom to top. + * + * As we don't have backpointers to the top level in the callchains + * structure, we need to always print the whole hist_entry callchain, + * skipping the first ones that are before the first visible entry + * and stop when we printed enough lines to fill the screen. + */ +do_offset: + if (offset > 0) { + do { + h = rb_entry(nd, struct hist_entry, rb_node); + if (h->ms.unfolded) { + u16 remaining = h->nr_rows - h->row_offset; + if (offset > remaining) { + offset -= remaining; + h->row_offset = 0; + } else { + h->row_offset += offset; + offset = 0; + self->top = nd; + break; + } + } + nd = hists__filter_entries(rb_next(nd)); + if (nd == NULL) + break; + --offset; + self->top = nd; + } while (offset != 0); + } else if (offset < 0) { + while (1) { + h = rb_entry(nd, struct hist_entry, rb_node); + if (h->ms.unfolded) { + if (first) { + if (-offset > h->row_offset) { + offset += h->row_offset; + h->row_offset = 0; + } else { + h->row_offset += offset; + offset = 0; + self->top = nd; + break; + } + } else { + if (-offset > h->nr_rows) { + offset += h->nr_rows; + h->row_offset = 0; + } else { + h->row_offset = h->nr_rows + offset; + offset = 0; + self->top = nd; + break; + } + } + } + + nd = hists__filter_prev_entries(rb_prev(nd)); + if (nd == NULL) + break; + ++offset; + self->top = nd; + if (offset == 0) { + /* + * Last unfiltered hist_entry, check if it is + * unfolded, if it is then we should have + * row_offset at its last entry. + */ + h = rb_entry(nd, struct hist_entry, rb_node); + if (h->ms.unfolded) + h->row_offset = h->nr_rows; + break; + } + first = false; + } + } else { + self->top = nd; + h = rb_entry(nd, struct hist_entry, rb_node); + h->row_offset = 0; + } +} + +static struct hist_browser *hist_browser__new(struct hists *hists) +{ + struct hist_browser *self = zalloc(sizeof(*self)); + + if (self) { + self->hists = hists; + self->b.refresh = hist_browser__refresh; + self->b.seek = ui_browser__hists_seek; + } + + return self; +} + +static void hist_browser__delete(struct hist_browser *self) +{ + free(self); +} + +static struct hist_entry *hist_browser__selected_entry(struct hist_browser *self) +{ + return self->he_selection; +} + +static struct thread *hist_browser__selected_thread(struct hist_browser *self) +{ + return self->he_selection->thread; +} + +static int hists__browser_title(struct hists *self, char *bf, size_t size, + const char *ev_name, const struct dso *dso, + const struct thread *thread) +{ + char unit; + int printed; + unsigned long nr_events = self->stats.nr_events[PERF_RECORD_SAMPLE]; + + nr_events = convert_unit(nr_events, &unit); + printed = snprintf(bf, size, "Events: %lu%c %s", nr_events, unit, ev_name); + + if (thread) + printed += snprintf(bf + printed, size - printed, + ", Thread: %s(%d)", + (thread->comm_set ? thread->comm : ""), + thread->pid); + if (dso) + printed += snprintf(bf + printed, size - printed, + ", DSO: %s", dso->short_name); + return printed; +} + +static int perf_evsel__hists_browse(struct perf_evsel *evsel, + const char *helpline, const char *ev_name, + bool left_exits) +{ + struct hists *self = &evsel->hists; + struct hist_browser *browser = hist_browser__new(self); + struct pstack *fstack; + const struct thread *thread_filter = NULL; + const struct dso *dso_filter = NULL; + char msg[160]; + int key = -1; + + if (browser == NULL) + return -1; + + fstack = pstack__new(2); + if (fstack == NULL) + goto out; + + ui_helpline__push(helpline); + + hists__browser_title(self, msg, sizeof(msg), ev_name, + dso_filter, thread_filter); + while (1) { + const struct thread *thread = NULL; + const struct dso *dso = NULL; + char *options[16]; + int nr_options = 0, choice = 0, i, + annotate = -2, zoom_dso = -2, zoom_thread = -2, + browse_map = -2; + + key = hist_browser__run(browser, msg); + + if (browser->he_selection != NULL) { + thread = hist_browser__selected_thread(browser); + dso = browser->selection->map ? browser->selection->map->dso : NULL; + } + + switch (key) { + case NEWT_KEY_TAB: + case NEWT_KEY_UNTAB: + /* + * Exit the browser, let hists__browser_tree + * go to the next or previous + */ + goto out_free_stack; + case 'a': + if (browser->selection == NULL || + browser->selection->sym == NULL || + browser->selection->map->dso->annotate_warned) + continue; + goto do_annotate; + case 'd': + goto zoom_dso; + case 't': + goto zoom_thread; + case NEWT_KEY_F1: + case 'h': + case '?': + ui__help_window("-> Zoom into DSO/Threads & Annotate current symbol\n" + "<- Zoom out\n" + "a Annotate current symbol\n" + "h/?/F1 Show this window\n" + "C Collapse all callchains\n" + "E Expand all callchains\n" + "d Zoom into current DSO\n" + "t Zoom into current Thread\n" + "TAB/UNTAB Switch events\n" + "q/CTRL+C Exit browser"); + continue; + case NEWT_KEY_ENTER: + case NEWT_KEY_RIGHT: + /* menu */ + break; + case NEWT_KEY_LEFT: { + const void *top; + + if (pstack__empty(fstack)) { + /* + * Go back to the perf_evsel_menu__run or other user + */ + if (left_exits) + goto out_free_stack; + continue; + } + top = pstack__pop(fstack); + if (top == &dso_filter) + goto zoom_out_dso; + if (top == &thread_filter) + goto zoom_out_thread; + continue; + } + case NEWT_KEY_ESCAPE: + if (!left_exits && + !ui__dialog_yesno("Do you really want to exit?")) + continue; + /* Fall thru */ + default: + goto out_free_stack; + } + + if (browser->selection != NULL && + browser->selection->sym != NULL && + !browser->selection->map->dso->annotate_warned && + asprintf(&options[nr_options], "Annotate %s", + browser->selection->sym->name) > 0) + annotate = nr_options++; + + if (thread != NULL && + asprintf(&options[nr_options], "Zoom %s %s(%d) thread", + (thread_filter ? "out of" : "into"), + (thread->comm_set ? thread->comm : ""), + thread->pid) > 0) + zoom_thread = nr_options++; + + if (dso != NULL && + asprintf(&options[nr_options], "Zoom %s %s DSO", + (dso_filter ? "out of" : "into"), + (dso->kernel ? "the Kernel" : dso->short_name)) > 0) + zoom_dso = nr_options++; + + if (browser->selection != NULL && + browser->selection->map != NULL && + asprintf(&options[nr_options], "Browse map details") > 0) + browse_map = nr_options++; + + options[nr_options++] = (char *)"Exit"; + + choice = ui__popup_menu(nr_options, options); + + for (i = 0; i < nr_options - 1; ++i) + free(options[i]); + + if (choice == nr_options - 1) + break; + + if (choice == -1) + continue; + + if (choice == annotate) { + struct hist_entry *he; +do_annotate: + he = hist_browser__selected_entry(browser); + if (he == NULL) + continue; + + hist_entry__tui_annotate(he, evsel->idx); + } else if (choice == browse_map) + map__browse(browser->selection->map); + else if (choice == zoom_dso) { +zoom_dso: + if (dso_filter) { + pstack__remove(fstack, &dso_filter); +zoom_out_dso: + ui_helpline__pop(); + dso_filter = NULL; + } else { + if (dso == NULL) + continue; + ui_helpline__fpush("To zoom out press <- or -> + \"Zoom out of %s DSO\"", + dso->kernel ? "the Kernel" : dso->short_name); + dso_filter = dso; + pstack__push(fstack, &dso_filter); + } + hists__filter_by_dso(self, dso_filter); + hists__browser_title(self, msg, sizeof(msg), ev_name, + dso_filter, thread_filter); + hist_browser__reset(browser); + } else if (choice == zoom_thread) { +zoom_thread: + if (thread_filter) { + pstack__remove(fstack, &thread_filter); +zoom_out_thread: + ui_helpline__pop(); + thread_filter = NULL; + } else { + ui_helpline__fpush("To zoom out press <- or -> + \"Zoom out of %s(%d) thread\"", + thread->comm_set ? thread->comm : "", + thread->pid); + thread_filter = thread; + pstack__push(fstack, &thread_filter); + } + hists__filter_by_thread(self, thread_filter); + hists__browser_title(self, msg, sizeof(msg), ev_name, + dso_filter, thread_filter); + hist_browser__reset(browser); + } + } +out_free_stack: + pstack__delete(fstack); +out: + hist_browser__delete(browser); + return key; +} + +struct perf_evsel_menu { + struct ui_browser b; + struct perf_evsel *selection; +}; + +static void perf_evsel_menu__write(struct ui_browser *browser, + void *entry, int row) +{ + struct perf_evsel_menu *menu = container_of(browser, + struct perf_evsel_menu, b); + struct perf_evsel *evsel = list_entry(entry, struct perf_evsel, node); + bool current_entry = ui_browser__is_current_entry(browser, row); + unsigned long nr_events = evsel->hists.stats.nr_events[PERF_RECORD_SAMPLE]; + const char *ev_name = event_name(evsel); + char bf[256], unit; + + ui_browser__set_color(browser, current_entry ? HE_COLORSET_SELECTED : + HE_COLORSET_NORMAL); + + nr_events = convert_unit(nr_events, &unit); + snprintf(bf, sizeof(bf), "%lu%c%s%s", nr_events, + unit, unit == ' ' ? "" : " ", ev_name); + slsmg_write_nstring(bf, browser->width); + + if (current_entry) + menu->selection = evsel; +} + +static int perf_evsel_menu__run(struct perf_evsel_menu *menu, const char *help) +{ + int exit_keys[] = { NEWT_KEY_ENTER, NEWT_KEY_RIGHT, 0, }; + struct perf_evlist *evlist = menu->b.priv; + struct perf_evsel *pos; + const char *ev_name, *title = "Available samples"; + int key; + + if (ui_browser__show(&menu->b, title, + "ESC: exit, ENTER|->: Browse histograms") < 0) + return -1; + + ui_browser__add_exit_keys(&menu->b, exit_keys); + + while (1) { + key = ui_browser__run(&menu->b); + + switch (key) { + case NEWT_KEY_RIGHT: + case NEWT_KEY_ENTER: + if (!menu->selection) + continue; + pos = menu->selection; +browse_hists: + ev_name = event_name(pos); + key = perf_evsel__hists_browse(pos, help, ev_name, true); + ui_browser__show_title(&menu->b, title); + break; + case NEWT_KEY_LEFT: + continue; + case NEWT_KEY_ESCAPE: + if (!ui__dialog_yesno("Do you really want to exit?")) + continue; + /* Fall thru */ + default: + goto out; + } + + switch (key) { + case NEWT_KEY_TAB: + if (pos->node.next == &evlist->entries) + pos = list_entry(evlist->entries.next, struct perf_evsel, node); + else + pos = list_entry(pos->node.next, struct perf_evsel, node); + goto browse_hists; + case NEWT_KEY_UNTAB: + if (pos->node.prev == &evlist->entries) + pos = list_entry(evlist->entries.prev, struct perf_evsel, node); + else + pos = list_entry(pos->node.prev, struct perf_evsel, node); + goto browse_hists; + case 'q': + case CTRL('c'): + goto out; + default: + break; + } + } + +out: + ui_browser__hide(&menu->b); + return key; +} + +static int __perf_evlist__tui_browse_hists(struct perf_evlist *evlist, + const char *help) +{ + struct perf_evsel *pos; + struct perf_evsel_menu menu = { + .b = { + .entries = &evlist->entries, + .refresh = ui_browser__list_head_refresh, + .seek = ui_browser__list_head_seek, + .write = perf_evsel_menu__write, + .nr_entries = evlist->nr_entries, + .priv = evlist, + }, + }; + + ui_helpline__push("Press ESC to exit"); + + list_for_each_entry(pos, &evlist->entries, node) { + const char *ev_name = event_name(pos); + size_t line_len = strlen(ev_name) + 7; + + if (menu.b.width < line_len) + menu.b.width = line_len; + /* + * Cache the evsel name, tracepoints have a _high_ cost per + * event_name() call. + */ + if (pos->name == NULL) + pos->name = strdup(ev_name); + } + + return perf_evsel_menu__run(&menu, help); +} + +int perf_evlist__tui_browse_hists(struct perf_evlist *evlist, const char *help) +{ + + if (evlist->nr_entries == 1) { + struct perf_evsel *first = list_entry(evlist->entries.next, + struct perf_evsel, node); + const char *ev_name = event_name(first); + return perf_evsel__hists_browse(first, help, ev_name, false); + } + + return __perf_evlist__tui_browse_hists(evlist, help); +} diff --git a/tools/perf/util/ui/browsers/map.c b/tools/perf/util/ui/browsers/map.c new file mode 100644 index 00000000..8462bffe --- /dev/null +++ b/tools/perf/util/ui/browsers/map.c @@ -0,0 +1,156 @@ +#include "../libslang.h" +#include <elf.h> +#include <inttypes.h> +#include <sys/ttydefaults.h> +#include <ctype.h> +#include <string.h> +#include <linux/bitops.h> +#include "../../debug.h" +#include "../../symbol.h" +#include "../browser.h" +#include "../helpline.h" +#include "map.h" + +static int ui_entry__read(const char *title, char *bf, size_t size, int width) +{ + struct newtExitStruct es; + newtComponent form, entry; + const char *result; + int err = -1; + + newtCenteredWindow(width, 1, title); + form = newtForm(NULL, NULL, 0); + if (form == NULL) + return -1; + + entry = newtEntry(0, 0, "0x", width, &result, NEWT_FLAG_SCROLL); + if (entry == NULL) + goto out_free_form; + + newtFormAddComponent(form, entry); + newtFormAddHotKey(form, NEWT_KEY_ENTER); + newtFormAddHotKey(form, NEWT_KEY_ESCAPE); + newtFormAddHotKey(form, NEWT_KEY_LEFT); + newtFormAddHotKey(form, CTRL('c')); + newtFormRun(form, &es); + + if (result != NULL) { + strncpy(bf, result, size); + err = 0; + } +out_free_form: + newtPopWindow(); + newtFormDestroy(form); + return err; +} + +struct map_browser { + struct ui_browser b; + struct map *map; + u8 addrlen; +}; + +static void map_browser__write(struct ui_browser *self, void *nd, int row) +{ + struct symbol *sym = rb_entry(nd, struct symbol, rb_node); + struct map_browser *mb = container_of(self, struct map_browser, b); + bool current_entry = ui_browser__is_current_entry(self, row); + int width; + + ui_browser__set_percent_color(self, 0, current_entry); + slsmg_printf("%*" PRIx64 " %*" PRIx64 " %c ", + mb->addrlen, sym->start, mb->addrlen, sym->end, + sym->binding == STB_GLOBAL ? 'g' : + sym->binding == STB_LOCAL ? 'l' : 'w'); + width = self->width - ((mb->addrlen * 2) + 4); + if (width > 0) + slsmg_write_nstring(sym->name, width); +} + +/* FIXME uber-kludgy, see comment on cmd_report... */ +static u32 *symbol__browser_index(struct symbol *self) +{ + return ((void *)self) - sizeof(struct rb_node) - sizeof(u32); +} + +static int map_browser__search(struct map_browser *self) +{ + char target[512]; + struct symbol *sym; + int err = ui_entry__read("Search by name/addr", target, sizeof(target), 40); + + if (err) + return err; + + if (target[0] == '0' && tolower(target[1]) == 'x') { + u64 addr = strtoull(target, NULL, 16); + sym = map__find_symbol(self->map, addr, NULL); + } else + sym = map__find_symbol_by_name(self->map, target, NULL); + + if (sym != NULL) { + u32 *idx = symbol__browser_index(sym); + + self->b.top = &sym->rb_node; + self->b.index = self->b.top_idx = *idx; + } else + ui_helpline__fpush("%s not found!", target); + + return 0; +} + +static int map_browser__run(struct map_browser *self) +{ + int key; + + if (ui_browser__show(&self->b, self->map->dso->long_name, + "Press <- or ESC to exit, %s / to search", + verbose ? "" : "restart with -v to use") < 0) + return -1; + + if (verbose) + ui_browser__add_exit_key(&self->b, '/'); + + while (1) { + key = ui_browser__run(&self->b); + + if (verbose && key == '/') + map_browser__search(self); + else + break; + } + + ui_browser__hide(&self->b); + return key; +} + +int map__browse(struct map *self) +{ + struct map_browser mb = { + .b = { + .entries = &self->dso->symbols[self->type], + .refresh = ui_browser__rb_tree_refresh, + .seek = ui_browser__rb_tree_seek, + .write = map_browser__write, + }, + .map = self, + }; + struct rb_node *nd; + char tmp[BITS_PER_LONG / 4]; + u64 maxaddr = 0; + + for (nd = rb_first(mb.b.entries); nd; nd = rb_next(nd)) { + struct symbol *pos = rb_entry(nd, struct symbol, rb_node); + + if (maxaddr < pos->end) + maxaddr = pos->end; + if (verbose) { + u32 *idx = symbol__browser_index(pos); + *idx = mb.b.nr_entries; + } + ++mb.b.nr_entries; + } + + mb.addrlen = snprintf(tmp, sizeof(tmp), "%" PRIx64, maxaddr); + return map_browser__run(&mb); +} diff --git a/tools/perf/util/ui/browsers/map.h b/tools/perf/util/ui/browsers/map.h new file mode 100644 index 00000000..df8581a4 --- /dev/null +++ b/tools/perf/util/ui/browsers/map.h @@ -0,0 +1,6 @@ +#ifndef _PERF_UI_MAP_BROWSER_H_ +#define _PERF_UI_MAP_BROWSER_H_ 1 +struct map; + +int map__browse(struct map *self); +#endif /* _PERF_UI_MAP_BROWSER_H_ */ diff --git a/tools/perf/util/ui/browsers/top.c b/tools/perf/util/ui/browsers/top.c new file mode 100644 index 00000000..5a065385 --- /dev/null +++ b/tools/perf/util/ui/browsers/top.c @@ -0,0 +1,213 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Parts came from builtin-{top,stat,record}.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ +#include "../browser.h" +#include "../../annotate.h" +#include "../helpline.h" +#include "../libslang.h" +#include "../util.h" +#include "../../evlist.h" +#include "../../hist.h" +#include "../../sort.h" +#include "../../symbol.h" +#include "../../top.h" + +struct perf_top_browser { + struct ui_browser b; + struct rb_root root; + struct sym_entry *selection; + float sum_ksamples; + int dso_width; + int dso_short_width; + int sym_width; +}; + +static void perf_top_browser__write(struct ui_browser *browser, void *entry, int row) +{ + struct perf_top_browser *top_browser = container_of(browser, struct perf_top_browser, b); + struct sym_entry *syme = rb_entry(entry, struct sym_entry, rb_node); + bool current_entry = ui_browser__is_current_entry(browser, row); + struct symbol *symbol = sym_entry__symbol(syme); + struct perf_top *top = browser->priv; + int width = browser->width; + double pcnt; + + pcnt = 100.0 - (100.0 * ((top_browser->sum_ksamples - syme->snap_count) / + top_browser->sum_ksamples)); + ui_browser__set_percent_color(browser, pcnt, current_entry); + + if (top->evlist->nr_entries == 1 || !top->display_weighted) { + slsmg_printf("%20.2f ", syme->weight); + width -= 24; + } else { + slsmg_printf("%9.1f %10ld ", syme->weight, syme->snap_count); + width -= 23; + } + + slsmg_printf("%4.1f%%", pcnt); + width -= 7; + + if (verbose) { + slsmg_printf(" %016" PRIx64, symbol->start); + width -= 17; + } + + slsmg_printf(" %-*.*s ", top_browser->sym_width, top_browser->sym_width, + symbol->name); + width -= top_browser->sym_width; + slsmg_write_nstring(width >= syme->map->dso->long_name_len ? + syme->map->dso->long_name : + syme->map->dso->short_name, width); + + if (current_entry) + top_browser->selection = syme; +} + +static void perf_top_browser__update_rb_tree(struct perf_top_browser *browser) +{ + struct perf_top *top = browser->b.priv; + u64 top_idx = browser->b.top_idx; + + browser->root = RB_ROOT; + browser->b.top = NULL; + browser->sum_ksamples = perf_top__decay_samples(top, &browser->root); + /* + * No active symbols + */ + if (top->rb_entries == 0) + return; + + perf_top__find_widths(top, &browser->root, &browser->dso_width, + &browser->dso_short_width, + &browser->sym_width); + if (browser->sym_width + browser->dso_width > browser->b.width - 29) { + browser->dso_width = browser->dso_short_width; + if (browser->sym_width + browser->dso_width > browser->b.width - 29) + browser->sym_width = browser->b.width - browser->dso_width - 29; + } + + /* + * Adjust the ui_browser indexes since the entries in the browser->root + * rb_tree may have changed, then seek it from start, so that we get a + * possible new top of the screen. + */ + browser->b.nr_entries = top->rb_entries; + + if (top_idx >= browser->b.nr_entries) { + if (browser->b.height >= browser->b.nr_entries) + top_idx = browser->b.nr_entries - browser->b.height; + else + top_idx = 0; + } + + if (browser->b.index >= top_idx + browser->b.height) + browser->b.index = top_idx + browser->b.index - browser->b.top_idx; + + if (browser->b.index >= browser->b.nr_entries) + browser->b.index = browser->b.nr_entries - 1; + + browser->b.top_idx = top_idx; + browser->b.seek(&browser->b, top_idx, SEEK_SET); +} + +static void perf_top_browser__annotate(struct perf_top_browser *browser) +{ + struct sym_entry *syme = browser->selection; + struct symbol *sym = sym_entry__symbol(syme); + struct annotation *notes = symbol__annotation(sym); + struct perf_top *top = browser->b.priv; + + if (notes->src != NULL) + goto do_annotation; + + pthread_mutex_lock(¬es->lock); + + top->sym_filter_entry = NULL; + + if (symbol__alloc_hist(sym, top->evlist->nr_entries) < 0) { + pr_err("Not enough memory for annotating '%s' symbol!\n", + sym->name); + pthread_mutex_unlock(¬es->lock); + return; + } + + top->sym_filter_entry = syme; + + pthread_mutex_unlock(¬es->lock); +do_annotation: + symbol__tui_annotate(sym, syme->map, 0, top->delay_secs * 1000); +} + +static int perf_top_browser__run(struct perf_top_browser *browser) +{ + int key; + char title[160]; + struct perf_top *top = browser->b.priv; + int delay_msecs = top->delay_secs * 1000; + int exit_keys[] = { 'a', NEWT_KEY_ENTER, NEWT_KEY_RIGHT, 0, }; + + perf_top_browser__update_rb_tree(browser); + perf_top__header_snprintf(top, title, sizeof(title)); + perf_top__reset_sample_counters(top); + + if (ui_browser__show(&browser->b, title, + "ESC: exit, ENTER|->|a: Live Annotate") < 0) + return -1; + + newtFormSetTimer(browser->b.form, delay_msecs); + ui_browser__add_exit_keys(&browser->b, exit_keys); + + while (1) { + key = ui_browser__run(&browser->b); + + switch (key) { + case -1: + /* FIXME we need to check if it was es.reason == NEWT_EXIT_TIMER */ + perf_top_browser__update_rb_tree(browser); + perf_top__header_snprintf(top, title, sizeof(title)); + perf_top__reset_sample_counters(top); + ui_browser__set_color(&browser->b, NEWT_COLORSET_ROOT); + SLsmg_gotorc(0, 0); + slsmg_write_nstring(title, browser->b.width); + break; + case 'a': + case NEWT_KEY_RIGHT: + case NEWT_KEY_ENTER: + if (browser->selection) + perf_top_browser__annotate(browser); + break; + case NEWT_KEY_LEFT: + continue; + case NEWT_KEY_ESCAPE: + if (!ui__dialog_yesno("Do you really want to exit?")) + continue; + /* Fall thru */ + default: + goto out; + } + } +out: + ui_browser__hide(&browser->b); + return key; +} + +int perf_top__tui_browser(struct perf_top *top) +{ + struct perf_top_browser browser = { + .b = { + .entries = &browser.root, + .refresh = ui_browser__rb_tree_refresh, + .seek = ui_browser__rb_tree_seek, + .write = perf_top_browser__write, + .priv = top, + }, + }; + + ui_helpline__push("Press <- or ESC to exit"); + return perf_top_browser__run(&browser); +} diff --git a/tools/perf/util/ui/helpline.c b/tools/perf/util/ui/helpline.c new file mode 100644 index 00000000..f36d2ff5 --- /dev/null +++ b/tools/perf/util/ui/helpline.c @@ -0,0 +1,72 @@ +#define _GNU_SOURCE +#include <stdio.h> +#include <stdlib.h> +#include <newt.h> + +#include "../debug.h" +#include "helpline.h" +#include "ui.h" + +void ui_helpline__pop(void) +{ + newtPopHelpLine(); +} + +void ui_helpline__push(const char *msg) +{ + newtPushHelpLine(msg); +} + +void ui_helpline__vpush(const char *fmt, va_list ap) +{ + char *s; + + if (vasprintf(&s, fmt, ap) < 0) + vfprintf(stderr, fmt, ap); + else { + ui_helpline__push(s); + free(s); + } +} + +void ui_helpline__fpush(const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + ui_helpline__vpush(fmt, ap); + va_end(ap); +} + +void ui_helpline__puts(const char *msg) +{ + ui_helpline__pop(); + ui_helpline__push(msg); +} + +void ui_helpline__init(void) +{ + ui_helpline__puts(" "); +} + +char ui_helpline__last_msg[1024]; + +int ui_helpline__show_help(const char *format, va_list ap) +{ + int ret; + static int backlog; + + pthread_mutex_lock(&ui__lock); + ret = vsnprintf(ui_helpline__last_msg + backlog, + sizeof(ui_helpline__last_msg) - backlog, format, ap); + backlog += ret; + + if (ui_helpline__last_msg[backlog - 1] == '\n') { + ui_helpline__puts(ui_helpline__last_msg); + newtRefresh(); + backlog = 0; + } + pthread_mutex_unlock(&ui__lock); + + return ret; +} diff --git a/tools/perf/util/ui/helpline.h b/tools/perf/util/ui/helpline.h new file mode 100644 index 00000000..ab6028d0 --- /dev/null +++ b/tools/perf/util/ui/helpline.h @@ -0,0 +1,11 @@ +#ifndef _PERF_UI_HELPLINE_H_ +#define _PERF_UI_HELPLINE_H_ 1 + +void ui_helpline__init(void); +void ui_helpline__pop(void); +void ui_helpline__push(const char *msg); +void ui_helpline__vpush(const char *fmt, va_list ap); +void ui_helpline__fpush(const char *fmt, ...); +void ui_helpline__puts(const char *msg); + +#endif /* _PERF_UI_HELPLINE_H_ */ diff --git a/tools/perf/util/ui/libslang.h b/tools/perf/util/ui/libslang.h new file mode 100644 index 00000000..2b63e1c9 --- /dev/null +++ b/tools/perf/util/ui/libslang.h @@ -0,0 +1,27 @@ +#ifndef _PERF_UI_SLANG_H_ +#define _PERF_UI_SLANG_H_ 1 +/* + * slang versions <= 2.0.6 have a "#if HAVE_LONG_LONG" that breaks + * the build if it isn't defined. Use the equivalent one that glibc + * has on features.h. + */ +#include <features.h> +#ifndef HAVE_LONG_LONG +#define HAVE_LONG_LONG __GLIBC_HAVE_LONG_LONG +#endif +#include <slang.h> + +#if SLANG_VERSION < 20104 +#define slsmg_printf(msg, args...) \ + SLsmg_printf((char *)(msg), ##args) +#define slsmg_write_nstring(msg, len) \ + SLsmg_write_nstring((char *)(msg), len) +#define sltt_set_color(obj, name, fg, bg) \ + SLtt_set_color(obj,(char *)(name), (char *)(fg), (char *)(bg)) +#else +#define slsmg_printf SLsmg_printf +#define slsmg_write_nstring SLsmg_write_nstring +#define sltt_set_color SLtt_set_color +#endif + +#endif /* _PERF_UI_SLANG_H_ */ diff --git a/tools/perf/util/ui/progress.c b/tools/perf/util/ui/progress.c new file mode 100644 index 00000000..d7fc399d --- /dev/null +++ b/tools/perf/util/ui/progress.c @@ -0,0 +1,60 @@ +#include <stdlib.h> +#include <newt.h> +#include "../cache.h" +#include "progress.h" + +struct ui_progress { + newtComponent form, scale; +}; + +struct ui_progress *ui_progress__new(const char *title, u64 total) +{ + struct ui_progress *self = malloc(sizeof(*self)); + + if (self != NULL) { + int cols; + + if (use_browser <= 0) + return self; + newtGetScreenSize(&cols, NULL); + cols -= 4; + newtCenteredWindow(cols, 1, title); + self->form = newtForm(NULL, NULL, 0); + if (self->form == NULL) + goto out_free_self; + self->scale = newtScale(0, 0, cols, total); + if (self->scale == NULL) + goto out_free_form; + newtFormAddComponent(self->form, self->scale); + newtRefresh(); + } + + return self; + +out_free_form: + newtFormDestroy(self->form); +out_free_self: + free(self); + return NULL; +} + +void ui_progress__update(struct ui_progress *self, u64 curr) +{ + /* + * FIXME: We should have a per UI backend way of showing progress, + * stdio will just show a percentage as NN%, etc. + */ + if (use_browser <= 0) + return; + newtScaleSet(self->scale, curr); + newtRefresh(); +} + +void ui_progress__delete(struct ui_progress *self) +{ + if (use_browser > 0) { + newtFormDestroy(self->form); + newtPopWindow(); + } + free(self); +} diff --git a/tools/perf/util/ui/progress.h b/tools/perf/util/ui/progress.h new file mode 100644 index 00000000..a3820a0b --- /dev/null +++ b/tools/perf/util/ui/progress.h @@ -0,0 +1,11 @@ +#ifndef _PERF_UI_PROGRESS_H_ +#define _PERF_UI_PROGRESS_H_ 1 + +struct ui_progress; + +struct ui_progress *ui_progress__new(const char *title, u64 total); +void ui_progress__delete(struct ui_progress *self); + +void ui_progress__update(struct ui_progress *self, u64 curr); + +#endif diff --git a/tools/perf/util/ui/setup.c b/tools/perf/util/ui/setup.c new file mode 100644 index 00000000..ee46d671 --- /dev/null +++ b/tools/perf/util/ui/setup.c @@ -0,0 +1,46 @@ +#include <newt.h> +#include <signal.h> +#include <stdbool.h> + +#include "../cache.h" +#include "../debug.h" +#include "browser.h" +#include "helpline.h" +#include "ui.h" + +pthread_mutex_t ui__lock = PTHREAD_MUTEX_INITIALIZER; + +static void newt_suspend(void *d __used) +{ + newtSuspend(); + raise(SIGTSTP); + newtResume(); +} + +void setup_browser(bool fallback_to_pager) +{ + if (!isatty(1) || !use_browser || dump_trace) { + use_browser = 0; + if (fallback_to_pager) + setup_pager(); + return; + } + + use_browser = 1; + newtInit(); + newtCls(); + newtSetSuspendCallback(newt_suspend, NULL); + ui_helpline__init(); + ui_browser__init(); +} + +void exit_browser(bool wait_for_ok) +{ + if (use_browser > 0) { + if (wait_for_ok) { + char title[] = "Fatal Error", ok[] = "Ok"; + newtWinMessage(title, ok, ui_helpline__last_msg); + } + newtFinished(); + } +} diff --git a/tools/perf/util/ui/ui.h b/tools/perf/util/ui/ui.h new file mode 100644 index 00000000..d264e059 --- /dev/null +++ b/tools/perf/util/ui/ui.h @@ -0,0 +1,8 @@ +#ifndef _PERF_UI_H_ +#define _PERF_UI_H_ 1 + +#include <pthread.h> + +extern pthread_mutex_t ui__lock; + +#endif /* _PERF_UI_H_ */ diff --git a/tools/perf/util/ui/util.c b/tools/perf/util/ui/util.c new file mode 100644 index 00000000..fdf1fc8f --- /dev/null +++ b/tools/perf/util/ui/util.c @@ -0,0 +1,130 @@ +#include <newt.h> +#include <signal.h> +#include <stdio.h> +#include <stdbool.h> +#include <string.h> +#include <sys/ttydefaults.h> + +#include "../cache.h" +#include "../debug.h" +#include "browser.h" +#include "helpline.h" +#include "ui.h" +#include "util.h" + +static void newt_form__set_exit_keys(newtComponent self) +{ + newtFormAddHotKey(self, NEWT_KEY_LEFT); + newtFormAddHotKey(self, NEWT_KEY_ESCAPE); + newtFormAddHotKey(self, 'Q'); + newtFormAddHotKey(self, 'q'); + newtFormAddHotKey(self, CTRL('c')); +} + +static newtComponent newt_form__new(void) +{ + newtComponent self = newtForm(NULL, NULL, 0); + if (self) + newt_form__set_exit_keys(self); + return self; +} + +int ui__popup_menu(int argc, char * const argv[]) +{ + struct newtExitStruct es; + int i, rc = -1, max_len = 5; + newtComponent listbox, form = newt_form__new(); + + if (form == NULL) + return -1; + + listbox = newtListbox(0, 0, argc, NEWT_FLAG_RETURNEXIT); + if (listbox == NULL) + goto out_destroy_form; + + newtFormAddComponent(form, listbox); + + for (i = 0; i < argc; ++i) { + int len = strlen(argv[i]); + if (len > max_len) + max_len = len; + if (newtListboxAddEntry(listbox, argv[i], (void *)(long)i)) + goto out_destroy_form; + } + + newtCenteredWindow(max_len, argc, NULL); + newtFormRun(form, &es); + rc = newtListboxGetCurrent(listbox) - NULL; + if (es.reason == NEWT_EXIT_HOTKEY) + rc = -1; + newtPopWindow(); +out_destroy_form: + newtFormDestroy(form); + return rc; +} + +int ui__help_window(const char *text) +{ + struct newtExitStruct es; + newtComponent tb, form = newt_form__new(); + int rc = -1; + int max_len = 0, nr_lines = 0; + const char *t; + + if (form == NULL) + return -1; + + t = text; + while (1) { + const char *sep = strchr(t, '\n'); + int len; + + if (sep == NULL) + sep = strchr(t, '\0'); + len = sep - t; + if (max_len < len) + max_len = len; + ++nr_lines; + if (*sep == '\0') + break; + t = sep + 1; + } + + tb = newtTextbox(0, 0, max_len, nr_lines, 0); + if (tb == NULL) + goto out_destroy_form; + + newtTextboxSetText(tb, text); + newtFormAddComponent(form, tb); + newtCenteredWindow(max_len, nr_lines, NULL); + newtFormRun(form, &es); + newtPopWindow(); + rc = 0; +out_destroy_form: + newtFormDestroy(form); + return rc; +} + +static const char yes[] = "Yes", no[] = "No", + warning_str[] = "Warning!", ok[] = "Ok"; + +bool ui__dialog_yesno(const char *msg) +{ + /* newtWinChoice should really be accepting const char pointers... */ + return newtWinChoice(NULL, (char *)yes, (char *)no, (char *)msg) == 1; +} + +void ui__warning(const char *format, ...) +{ + va_list args; + + va_start(args, format); + if (use_browser > 0) { + pthread_mutex_lock(&ui__lock); + newtWinMessagev((char *)warning_str, (char *)ok, + (char *)format, args); + pthread_mutex_unlock(&ui__lock); + } else + vfprintf(stderr, format, args); + va_end(args); +} diff --git a/tools/perf/util/ui/util.h b/tools/perf/util/ui/util.h new file mode 100644 index 00000000..afcbc1d9 --- /dev/null +++ b/tools/perf/util/ui/util.h @@ -0,0 +1,10 @@ +#ifndef _PERF_UI_UTIL_H_ +#define _PERF_UI_UTIL_H_ 1 + +#include <stdbool.h> + +int ui__popup_menu(int argc, char * const argv[]); +int ui__help_window(const char *text); +bool ui__dialog_yesno(const char *msg); + +#endif /* _PERF_UI_UTIL_H_ */ diff --git a/tools/perf/util/usage.c b/tools/perf/util/usage.c new file mode 100644 index 00000000..e16bf9a7 --- /dev/null +++ b/tools/perf/util/usage.c @@ -0,0 +1,80 @@ +/* + * GIT - The information manager from hell + * + * Copyright (C) Linus Torvalds, 2005 + */ +#include "util.h" + +static void report(const char *prefix, const char *err, va_list params) +{ + char msg[1024]; + vsnprintf(msg, sizeof(msg), err, params); + fprintf(stderr, " %s%s\n", prefix, msg); +} + +static NORETURN void usage_builtin(const char *err) +{ + fprintf(stderr, "\n Usage: %s\n", err); + exit(129); +} + +static NORETURN void die_builtin(const char *err, va_list params) +{ + report(" Fatal: ", err, params); + exit(128); +} + +static void error_builtin(const char *err, va_list params) +{ + report(" Error: ", err, params); +} + +static void warn_builtin(const char *warn, va_list params) +{ + report(" Warning: ", warn, params); +} + +/* If we are in a dlopen()ed .so write to a global variable would segfault + * (ugh), so keep things static. */ +static void (*usage_routine)(const char *err) NORETURN = usage_builtin; +static void (*die_routine)(const char *err, va_list params) NORETURN = die_builtin; +static void (*error_routine)(const char *err, va_list params) = error_builtin; +static void (*warn_routine)(const char *err, va_list params) = warn_builtin; + +void set_die_routine(void (*routine)(const char *err, va_list params) NORETURN) +{ + die_routine = routine; +} + +void usage(const char *err) +{ + usage_routine(err); +} + +void die(const char *err, ...) +{ + va_list params; + + va_start(params, err); + die_routine(err, params); + va_end(params); +} + +int error(const char *err, ...) +{ + va_list params; + + va_start(params, err); + error_routine(err, params); + va_end(params); + return -1; +} + +void warning(const char *warn, ...) +{ + va_list params; + + va_start(params, warn); + warn_routine(warn, params); + va_end(params); +} diff --git a/tools/perf/util/util.c b/tools/perf/util/util.c new file mode 100644 index 00000000..5b3ea49a --- /dev/null +++ b/tools/perf/util/util.c @@ -0,0 +1,133 @@ +#include "util.h" +#include <sys/mman.h> + +int mkdir_p(char *path, mode_t mode) +{ + struct stat st; + int err; + char *d = path; + + if (*d != '/') + return -1; + + if (stat(path, &st) == 0) + return 0; + + while (*++d == '/'); + + while ((d = strchr(d, '/'))) { + *d = '\0'; + err = stat(path, &st) && mkdir(path, mode); + *d++ = '/'; + if (err) + return -1; + while (*d == '/') + ++d; + } + return (stat(path, &st) && mkdir(path, mode)) ? -1 : 0; +} + +static int slow_copyfile(const char *from, const char *to) +{ + int err = 0; + char *line = NULL; + size_t n; + FILE *from_fp = fopen(from, "r"), *to_fp; + + if (from_fp == NULL) + goto out; + + to_fp = fopen(to, "w"); + if (to_fp == NULL) + goto out_fclose_from; + + while (getline(&line, &n, from_fp) > 0) + if (fputs(line, to_fp) == EOF) + goto out_fclose_to; + err = 0; +out_fclose_to: + fclose(to_fp); + free(line); +out_fclose_from: + fclose(from_fp); +out: + return err; +} + +int copyfile(const char *from, const char *to) +{ + int fromfd, tofd; + struct stat st; + void *addr; + int err = -1; + + if (stat(from, &st)) + goto out; + + if (st.st_size == 0) /* /proc? do it slowly... */ + return slow_copyfile(from, to); + + fromfd = open(from, O_RDONLY); + if (fromfd < 0) + goto out; + + tofd = creat(to, 0755); + if (tofd < 0) + goto out_close_from; + + addr = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fromfd, 0); + if (addr == MAP_FAILED) + goto out_close_to; + + if (write(tofd, addr, st.st_size) == st.st_size) + err = 0; + + munmap(addr, st.st_size); +out_close_to: + close(tofd); + if (err) + unlink(to); +out_close_from: + close(fromfd); +out: + return err; +} + +unsigned long convert_unit(unsigned long value, char *unit) +{ + *unit = ' '; + + if (value > 1000) { + value /= 1000; + *unit = 'K'; + } + + if (value > 1000) { + value /= 1000; + *unit = 'M'; + } + + if (value > 1000) { + value /= 1000; + *unit = 'G'; + } + + return value; +} + +int readn(int fd, void *buf, size_t n) +{ + void *buf_start = buf; + + while (n) { + int ret = read(fd, buf, n); + + if (ret <= 0) + return ret; + + n -= ret; + buf += ret; + } + + return buf - buf_start; +} diff --git a/tools/perf/util/util.h b/tools/perf/util/util.h new file mode 100644 index 00000000..fc784284 --- /dev/null +++ b/tools/perf/util/util.h @@ -0,0 +1,247 @@ +#ifndef GIT_COMPAT_UTIL_H +#define GIT_COMPAT_UTIL_H + +#define _FILE_OFFSET_BITS 64 + +#ifndef FLEX_ARRAY +/* + * See if our compiler is known to support flexible array members. + */ +#if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) +# define FLEX_ARRAY /* empty */ +#elif defined(__GNUC__) +# if (__GNUC__ >= 3) +# define FLEX_ARRAY /* empty */ +# else +# define FLEX_ARRAY 0 /* older GNU extension */ +# endif +#endif + +/* + * Otherwise, default to safer but a bit wasteful traditional style + */ +#ifndef FLEX_ARRAY +# define FLEX_ARRAY 1 +#endif +#endif + +#define ARRAY_SIZE(x) (sizeof(x)/sizeof(x[0])) + +#ifdef __GNUC__ +#define TYPEOF(x) (__typeof__(x)) +#else +#define TYPEOF(x) +#endif + +#define MSB(x, bits) ((x) & TYPEOF(x)(~0ULL << (sizeof(x) * 8 - (bits)))) +#define HAS_MULTI_BITS(i) ((i) & ((i) - 1)) /* checks if an integer has more than 1 bit set */ + +/* Approximation of the length of the decimal representation of this type. */ +#define decimal_length(x) ((int)(sizeof(x) * 2.56 + 0.5) + 1) + +#define _ALL_SOURCE 1 +#define _GNU_SOURCE 1 +#define _BSD_SOURCE 1 +#define HAS_BOOL + +#include <unistd.h> +#include <stdio.h> +#include <sys/stat.h> +#include <sys/statfs.h> +#include <fcntl.h> +#include <stdbool.h> +#include <stddef.h> +#include <stdlib.h> +#include <stdarg.h> +#include <string.h> +#include <errno.h> +#include <limits.h> +#include <sys/param.h> +#include <sys/types.h> +#include <dirent.h> +#include <sys/time.h> +#include <time.h> +#include <signal.h> +#include <fnmatch.h> +#include <assert.h> +#include <regex.h> +#include <utime.h> +#include <sys/wait.h> +#include <sys/poll.h> +#include <sys/socket.h> +#include <sys/ioctl.h> +#include <sys/select.h> +#include <netinet/in.h> +#include <netinet/tcp.h> +#include <arpa/inet.h> +#include <netdb.h> +#include <pwd.h> +#include <inttypes.h> +#include "../../../include/linux/magic.h" +#include "types.h" +#include <sys/ttydefaults.h> + +extern const char *graph_line; +extern const char *graph_dotted_line; +extern char buildid_dir[]; + +/* On most systems <limits.h> would have given us this, but + * not on some systems (e.g. GNU/Hurd). + */ +#ifndef PATH_MAX +#define PATH_MAX 4096 +#endif + +#ifndef PRIuMAX +#define PRIuMAX "llu" +#endif + +#ifndef PRIu32 +#define PRIu32 "u" +#endif + +#ifndef PRIx32 +#define PRIx32 "x" +#endif + +#ifndef PATH_SEP +#define PATH_SEP ':' +#endif + +#ifndef STRIP_EXTENSION +#define STRIP_EXTENSION "" +#endif + +#ifndef has_dos_drive_prefix +#define has_dos_drive_prefix(path) 0 +#endif + +#ifndef is_dir_sep +#define is_dir_sep(c) ((c) == '/') +#endif + +#ifdef __GNUC__ +#define NORETURN __attribute__((__noreturn__)) +#else +#define NORETURN +#ifndef __attribute__ +#define __attribute__(x) +#endif +#endif + +/* General helper functions */ +extern void usage(const char *err) NORETURN; +extern void die(const char *err, ...) NORETURN __attribute__((format (printf, 1, 2))); +extern int error(const char *err, ...) __attribute__((format (printf, 1, 2))); +extern void warning(const char *err, ...) __attribute__((format (printf, 1, 2))); + +#include "../../../include/linux/stringify.h" + +#define DIE_IF(cnd) \ + do { if (cnd) \ + die(" at (" __FILE__ ":" __stringify(__LINE__) "): " \ + __stringify(cnd) "\n"); \ + } while (0) + + +extern void set_die_routine(void (*routine)(const char *err, va_list params) NORETURN); + +extern int prefixcmp(const char *str, const char *prefix); +extern void set_buildid_dir(void); +extern void disable_buildid_cache(void); + +static inline const char *skip_prefix(const char *str, const char *prefix) +{ + size_t len = strlen(prefix); + return strncmp(str, prefix, len) ? NULL : str + len; +} + +#ifdef __GLIBC_PREREQ +#if __GLIBC_PREREQ(2, 1) +#define HAVE_STRCHRNUL +#endif +#endif + +#ifndef HAVE_STRCHRNUL +#define strchrnul gitstrchrnul +static inline char *gitstrchrnul(const char *s, int c) +{ + while (*s && *s != c) + s++; + return (char *)s; +} +#endif + +/* + * Wrappers: + */ +extern char *xstrdup(const char *str); +extern void *xrealloc(void *ptr, size_t size) __attribute__((weak)); + + +static inline void *zalloc(size_t size) +{ + return calloc(1, size); +} + +static inline int has_extension(const char *filename, const char *ext) +{ + size_t len = strlen(filename); + size_t extlen = strlen(ext); + + return len > extlen && !memcmp(filename + len - extlen, ext, extlen); +} + +/* Sane ctype - no locale, and works with signed chars */ +#undef isascii +#undef isspace +#undef isdigit +#undef isxdigit +#undef isalpha +#undef isprint +#undef isalnum +#undef tolower +#undef toupper + +extern unsigned char sane_ctype[256]; +#define GIT_SPACE 0x01 +#define GIT_DIGIT 0x02 +#define GIT_ALPHA 0x04 +#define GIT_GLOB_SPECIAL 0x08 +#define GIT_REGEX_SPECIAL 0x10 +#define GIT_PRINT_EXTRA 0x20 +#define GIT_PRINT 0x3E +#define sane_istest(x,mask) ((sane_ctype[(unsigned char)(x)] & (mask)) != 0) +#define isascii(x) (((x) & ~0x7f) == 0) +#define isspace(x) sane_istest(x,GIT_SPACE) +#define isdigit(x) sane_istest(x,GIT_DIGIT) +#define isxdigit(x) \ + (sane_istest(toupper(x), GIT_ALPHA | GIT_DIGIT) && toupper(x) < 'G') +#define isalpha(x) sane_istest(x,GIT_ALPHA) +#define isalnum(x) sane_istest(x,GIT_ALPHA | GIT_DIGIT) +#define isprint(x) sane_istest(x,GIT_PRINT) +#define tolower(x) sane_case((unsigned char)(x), 0x20) +#define toupper(x) sane_case((unsigned char)(x), 0) + +static inline int sane_case(int x, int high) +{ + if (sane_istest(x, GIT_ALPHA)) + x = (x & ~0x20) | high; + return x; +} + +int mkdir_p(char *path, mode_t mode); +int copyfile(const char *from, const char *to); + +s64 perf_atoll(const char *str); +char **argv_split(const char *str, int *argcp); +void argv_free(char **argv); +bool strglobmatch(const char *str, const char *pat); +bool strlazymatch(const char *str, const char *pat); +unsigned long convert_unit(unsigned long value, char *unit); +int readn(int fd, void *buf, size_t size); + +#define _STR(x) #x +#define STR(x) _STR(x) + +#endif diff --git a/tools/perf/util/values.c b/tools/perf/util/values.c new file mode 100644 index 00000000..bdd33470 --- /dev/null +++ b/tools/perf/util/values.c @@ -0,0 +1,231 @@ +#include <stdlib.h> + +#include "util.h" +#include "values.h" + +void perf_read_values_init(struct perf_read_values *values) +{ + values->threads_max = 16; + values->pid = malloc(values->threads_max * sizeof(*values->pid)); + values->tid = malloc(values->threads_max * sizeof(*values->tid)); + values->value = malloc(values->threads_max * sizeof(*values->value)); + if (!values->pid || !values->tid || !values->value) + die("failed to allocate read_values threads arrays"); + values->threads = 0; + + values->counters_max = 16; + values->counterrawid = malloc(values->counters_max + * sizeof(*values->counterrawid)); + values->countername = malloc(values->counters_max + * sizeof(*values->countername)); + if (!values->counterrawid || !values->countername) + die("failed to allocate read_values counters arrays"); + values->counters = 0; +} + +void perf_read_values_destroy(struct perf_read_values *values) +{ + int i; + + if (!values->threads_max || !values->counters_max) + return; + + for (i = 0; i < values->threads; i++) + free(values->value[i]); + free(values->pid); + free(values->tid); + free(values->counterrawid); + for (i = 0; i < values->counters; i++) + free(values->countername[i]); + free(values->countername); +} + +static void perf_read_values__enlarge_threads(struct perf_read_values *values) +{ + values->threads_max *= 2; + values->pid = realloc(values->pid, + values->threads_max * sizeof(*values->pid)); + values->tid = realloc(values->tid, + values->threads_max * sizeof(*values->tid)); + values->value = realloc(values->value, + values->threads_max * sizeof(*values->value)); + if (!values->pid || !values->tid || !values->value) + die("failed to enlarge read_values threads arrays"); +} + +static int perf_read_values__findnew_thread(struct perf_read_values *values, + u32 pid, u32 tid) +{ + int i; + + for (i = 0; i < values->threads; i++) + if (values->pid[i] == pid && values->tid[i] == tid) + return i; + + if (values->threads == values->threads_max) + perf_read_values__enlarge_threads(values); + + i = values->threads++; + values->pid[i] = pid; + values->tid[i] = tid; + values->value[i] = malloc(values->counters_max * sizeof(**values->value)); + if (!values->value[i]) + die("failed to allocate read_values counters array"); + + return i; +} + +static void perf_read_values__enlarge_counters(struct perf_read_values *values) +{ + int i; + + values->counters_max *= 2; + values->counterrawid = realloc(values->counterrawid, + values->counters_max * sizeof(*values->counterrawid)); + values->countername = realloc(values->countername, + values->counters_max * sizeof(*values->countername)); + if (!values->counterrawid || !values->countername) + die("failed to enlarge read_values counters arrays"); + + for (i = 0; i < values->threads; i++) { + values->value[i] = realloc(values->value[i], + values->counters_max * sizeof(**values->value)); + if (!values->value[i]) + die("failed to enlarge read_values counters arrays"); + } +} + +static int perf_read_values__findnew_counter(struct perf_read_values *values, + u64 rawid, const char *name) +{ + int i; + + for (i = 0; i < values->counters; i++) + if (values->counterrawid[i] == rawid) + return i; + + if (values->counters == values->counters_max) + perf_read_values__enlarge_counters(values); + + i = values->counters++; + values->counterrawid[i] = rawid; + values->countername[i] = strdup(name); + + return i; +} + +void perf_read_values_add_value(struct perf_read_values *values, + u32 pid, u32 tid, + u64 rawid, const char *name, u64 value) +{ + int tindex, cindex; + + tindex = perf_read_values__findnew_thread(values, pid, tid); + cindex = perf_read_values__findnew_counter(values, rawid, name); + + values->value[tindex][cindex] = value; +} + +static void perf_read_values__display_pretty(FILE *fp, + struct perf_read_values *values) +{ + int i, j; + int pidwidth, tidwidth; + int *counterwidth; + + counterwidth = malloc(values->counters * sizeof(*counterwidth)); + if (!counterwidth) + die("failed to allocate counterwidth array"); + tidwidth = 3; + pidwidth = 3; + for (j = 0; j < values->counters; j++) + counterwidth[j] = strlen(values->countername[j]); + for (i = 0; i < values->threads; i++) { + int width; + + width = snprintf(NULL, 0, "%d", values->pid[i]); + if (width > pidwidth) + pidwidth = width; + width = snprintf(NULL, 0, "%d", values->tid[i]); + if (width > tidwidth) + tidwidth = width; + for (j = 0; j < values->counters; j++) { + width = snprintf(NULL, 0, "%" PRIu64, values->value[i][j]); + if (width > counterwidth[j]) + counterwidth[j] = width; + } + } + + fprintf(fp, "# %*s %*s", pidwidth, "PID", tidwidth, "TID"); + for (j = 0; j < values->counters; j++) + fprintf(fp, " %*s", counterwidth[j], values->countername[j]); + fprintf(fp, "\n"); + + for (i = 0; i < values->threads; i++) { + fprintf(fp, " %*d %*d", pidwidth, values->pid[i], + tidwidth, values->tid[i]); + for (j = 0; j < values->counters; j++) + fprintf(fp, " %*" PRIu64, + counterwidth[j], values->value[i][j]); + fprintf(fp, "\n"); + } + free(counterwidth); +} + +static void perf_read_values__display_raw(FILE *fp, + struct perf_read_values *values) +{ + int width, pidwidth, tidwidth, namewidth, rawwidth, countwidth; + int i, j; + + tidwidth = 3; /* TID */ + pidwidth = 3; /* PID */ + namewidth = 4; /* "Name" */ + rawwidth = 3; /* "Raw" */ + countwidth = 5; /* "Count" */ + + for (i = 0; i < values->threads; i++) { + width = snprintf(NULL, 0, "%d", values->pid[i]); + if (width > pidwidth) + pidwidth = width; + width = snprintf(NULL, 0, "%d", values->tid[i]); + if (width > tidwidth) + tidwidth = width; + } + for (j = 0; j < values->counters; j++) { + width = strlen(values->countername[j]); + if (width > namewidth) + namewidth = width; + width = snprintf(NULL, 0, "%" PRIx64, values->counterrawid[j]); + if (width > rawwidth) + rawwidth = width; + } + for (i = 0; i < values->threads; i++) { + for (j = 0; j < values->counters; j++) { + width = snprintf(NULL, 0, "%" PRIu64, values->value[i][j]); + if (width > countwidth) + countwidth = width; + } + } + + fprintf(fp, "# %*s %*s %*s %*s %*s\n", + pidwidth, "PID", tidwidth, "TID", + namewidth, "Name", rawwidth, "Raw", + countwidth, "Count"); + for (i = 0; i < values->threads; i++) + for (j = 0; j < values->counters; j++) + fprintf(fp, " %*d %*d %*s %*" PRIx64 " %*" PRIu64, + pidwidth, values->pid[i], + tidwidth, values->tid[i], + namewidth, values->countername[j], + rawwidth, values->counterrawid[j], + countwidth, values->value[i][j]); +} + +void perf_read_values_display(FILE *fp, struct perf_read_values *values, int raw) +{ + if (raw) + perf_read_values__display_raw(fp, values); + else + perf_read_values__display_pretty(fp, values); +} diff --git a/tools/perf/util/values.h b/tools/perf/util/values.h new file mode 100644 index 00000000..2fa967e1 --- /dev/null +++ b/tools/perf/util/values.h @@ -0,0 +1,27 @@ +#ifndef __PERF_VALUES_H +#define __PERF_VALUES_H + +#include "types.h" + +struct perf_read_values { + int threads; + int threads_max; + u32 *pid, *tid; + int counters; + int counters_max; + u64 *counterrawid; + char **countername; + u64 **value; +}; + +void perf_read_values_init(struct perf_read_values *values); +void perf_read_values_destroy(struct perf_read_values *values); + +void perf_read_values_add_value(struct perf_read_values *values, + u32 pid, u32 tid, + u64 rawid, const char *name, u64 value); + +void perf_read_values_display(FILE *fp, struct perf_read_values *values, + int raw); + +#endif /* __PERF_VALUES_H */ diff --git a/tools/perf/util/wrapper.c b/tools/perf/util/wrapper.c new file mode 100644 index 00000000..73e900ed --- /dev/null +++ b/tools/perf/util/wrapper.c @@ -0,0 +1,40 @@ +/* + * Various trivial helper wrappers around standard functions + */ +#include "cache.h" + +/* + * There's no pack memory to release - but stay close to the Git + * version so wrap this away: + */ +static inline void release_pack_memory(size_t size __used, int flag __used) +{ +} + +char *xstrdup(const char *str) +{ + char *ret = strdup(str); + if (!ret) { + release_pack_memory(strlen(str) + 1, -1); + ret = strdup(str); + if (!ret) + die("Out of memory, strdup failed"); + } + return ret; +} + +void *xrealloc(void *ptr, size_t size) +{ + void *ret = realloc(ptr, size); + if (!ret && !size) + ret = realloc(ptr, 1); + if (!ret) { + release_pack_memory(size, -1); + ret = realloc(ptr, size); + if (!ret && !size) + ret = realloc(ptr, 1); + if (!ret) + die("Out of memory, realloc failed"); + } + return ret; +} diff --git a/tools/perf/util/xyarray.c b/tools/perf/util/xyarray.c new file mode 100644 index 00000000..22afbf6c --- /dev/null +++ b/tools/perf/util/xyarray.c @@ -0,0 +1,20 @@ +#include "xyarray.h" +#include "util.h" + +struct xyarray *xyarray__new(int xlen, int ylen, size_t entry_size) +{ + size_t row_size = ylen * entry_size; + struct xyarray *xy = zalloc(sizeof(*xy) + xlen * row_size); + + if (xy != NULL) { + xy->entry_size = entry_size; + xy->row_size = row_size; + } + + return xy; +} + +void xyarray__delete(struct xyarray *xy) +{ + free(xy); +} diff --git a/tools/perf/util/xyarray.h b/tools/perf/util/xyarray.h new file mode 100644 index 00000000..c488a072 --- /dev/null +++ b/tools/perf/util/xyarray.h @@ -0,0 +1,20 @@ +#ifndef _PERF_XYARRAY_H_ +#define _PERF_XYARRAY_H_ 1 + +#include <sys/types.h> + +struct xyarray { + size_t row_size; + size_t entry_size; + char contents[]; +}; + +struct xyarray *xyarray__new(int xlen, int ylen, size_t entry_size); +void xyarray__delete(struct xyarray *xy); + +static inline void *xyarray__entry(struct xyarray *xy, int x, int y) +{ + return &xy->contents[x * xy->row_size + y * xy->entry_size]; +} + +#endif /* _PERF_XYARRAY_H_ */ diff --git a/tools/power/x86/turbostat/Makefile b/tools/power/x86/turbostat/Makefile new file mode 100644 index 00000000..fd8e1f12 --- /dev/null +++ b/tools/power/x86/turbostat/Makefile @@ -0,0 +1,8 @@ +turbostat : turbostat.c + +clean : + rm -f turbostat + +install : + install turbostat /usr/bin/turbostat + install turbostat.8 /usr/share/man/man8 diff --git a/tools/power/x86/turbostat/turbostat.8 b/tools/power/x86/turbostat/turbostat.8 new file mode 100644 index 00000000..ff75125d --- /dev/null +++ b/tools/power/x86/turbostat/turbostat.8 @@ -0,0 +1,172 @@ +.TH TURBOSTAT 8 +.SH NAME +turbostat \- Report processor frequency and idle statistics +.SH SYNOPSIS +.ft B +.B turbostat +.RB [ "\-v" ] +.RB [ "\-M MSR#" ] +.RB command +.br +.B turbostat +.RB [ "\-v" ] +.RB [ "\-M MSR#" ] +.RB [ "\-i interval_sec" ] +.SH DESCRIPTION +\fBturbostat \fP reports processor topology, frequency +and idle power state statistics on modern X86 processors. +Either \fBcommand\fP is forked and statistics are printed +upon its completion, or statistics are printed periodically. + +\fBturbostat \fP +requires that the processor +supports an "invariant" TSC, plus the APERF and MPERF MSRs. +\fBturbostat \fP will report idle cpu power state residency +on processors that additionally support C-state residency counters. + +.SS Options +The \fB-v\fP option increases verbosity. +.PP +The \fB-M MSR#\fP option dumps the specified MSR, +in addition to the usual frequency and idle statistics. +.PP +The \fB-i interval_sec\fP option prints statistics every \fiinterval_sec\fP seconds. +The default is 5 seconds. +.PP +The \fBcommand\fP parameter forks \fBcommand\fP and upon its exit, +displays the statistics gathered since it was forked. +.PP +.SH FIELD DESCRIPTIONS +.nf +\fBpkg\fP processor package number. +\fBcore\fP processor core number. +\fBCPU\fP Linux CPU (logical processor) number. +\fB%c0\fP percent of the interval that the CPU retired instructions. +\fBGHz\fP average clock rate while the CPU was in c0 state. +\fBTSC\fP average GHz that the TSC ran during the entire interval. +\fB%c1, %c3, %c6\fP show the percentage residency in hardware core idle states. +\fB%pc3, %pc6\fP percentage residency in hardware package idle states. +.fi +.PP +.SH EXAMPLE +Without any parameters, turbostat prints out counters ever 5 seconds. +(override interval with "-i sec" option, or specify a command +for turbostat to fork). + +The first row of statistics reflect the average for the entire system. +Subsequent rows show per-CPU statistics. + +.nf +[root@x980]# ./turbostat +core CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6 + 0.04 1.62 3.38 0.11 0.00 99.85 0.00 95.07 + 0 0 0.04 1.62 3.38 0.06 0.00 99.90 0.00 95.07 + 0 6 0.02 1.62 3.38 0.08 0.00 99.90 0.00 95.07 + 1 2 0.10 1.62 3.38 0.29 0.00 99.61 0.00 95.07 + 1 8 0.11 1.62 3.38 0.28 0.00 99.61 0.00 95.07 + 2 4 0.01 1.62 3.38 0.01 0.00 99.98 0.00 95.07 + 2 10 0.01 1.61 3.38 0.02 0.00 99.98 0.00 95.07 + 8 1 0.07 1.62 3.38 0.15 0.00 99.78 0.00 95.07 + 8 7 0.03 1.62 3.38 0.19 0.00 99.78 0.00 95.07 + 9 3 0.01 1.62 3.38 0.02 0.00 99.98 0.00 95.07 + 9 9 0.01 1.62 3.38 0.02 0.00 99.98 0.00 95.07 + 10 5 0.01 1.62 3.38 0.13 0.00 99.86 0.00 95.07 + 10 11 0.08 1.62 3.38 0.05 0.00 99.86 0.00 95.07 +.fi +.SH VERBOSE EXAMPLE +The "-v" option adds verbosity to the output: + +.nf +GenuineIntel 11 CPUID levels; family:model:stepping 0x6:2c:2 (6:44:2) +12 * 133 = 1600 MHz max efficiency +25 * 133 = 3333 MHz TSC frequency +26 * 133 = 3467 MHz max turbo 4 active cores +26 * 133 = 3467 MHz max turbo 3 active cores +27 * 133 = 3600 MHz max turbo 2 active cores +27 * 133 = 3600 MHz max turbo 1 active cores + +.fi +The \fBmax efficiency\fP frequency, a.k.a. Low Frequency Mode, is the frequency +available at the minimum package voltage. The \fBTSC frequency\fP is the nominal +maximum frequency of the processor if turbo-mode were not available. This frequency +should be sustainable on all CPUs indefinitely, given nominal power and cooling. +The remaining rows show what maximum turbo frequency is possible +depending on the number of idle cores. Note that this information is +not available on all processors. +.SH FORK EXAMPLE +If turbostat is invoked with a command, it will fork that command +and output the statistics gathered when the command exits. +eg. Here a cycle soaker is run on 1 CPU (see %c0) for a few seconds +until ^C while the other CPUs are mostly idle: + +.nf +[root@x980 lenb]# ./turbostat cat /dev/zero > /dev/null + +^Ccore CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6 + 8.49 3.63 3.38 16.23 0.66 74.63 0.00 0.00 + 0 0 1.22 3.62 3.38 32.18 0.00 66.60 0.00 0.00 + 0 6 0.40 3.61 3.38 33.00 0.00 66.60 0.00 0.00 + 1 2 0.11 3.14 3.38 0.19 3.95 95.75 0.00 0.00 + 1 8 0.05 2.88 3.38 0.25 3.95 95.75 0.00 0.00 + 2 4 0.00 3.13 3.38 0.02 0.00 99.98 0.00 0.00 + 2 10 0.00 3.09 3.38 0.02 0.00 99.98 0.00 0.00 + 8 1 0.04 3.50 3.38 14.43 0.00 85.54 0.00 0.00 + 8 7 0.03 2.98 3.38 14.43 0.00 85.54 0.00 0.00 + 9 3 0.00 3.16 3.38 100.00 0.00 0.00 0.00 0.00 + 9 9 99.93 3.63 3.38 0.06 0.00 0.00 0.00 0.00 + 10 5 0.01 2.82 3.38 0.08 0.00 99.91 0.00 0.00 + 10 11 0.02 3.36 3.38 0.06 0.00 99.91 0.00 0.00 +6.950866 sec + +.fi +Above the cycle soaker drives cpu9 up 3.6 Ghz turbo limit +while the other processors are generally in various states of idle. + +Note that cpu3 is an HT sibling sharing core9 +with cpu9, and thus it is unable to get to an idle state +deeper than c1 while cpu9 is busy. + +Note that turbostat reports average GHz of 3.61, while +the arithmetic average of the GHz column above is 3.24. +This is a weighted average, where the weight is %c0. ie. it is the total number of +un-halted cycles elapsed per time divided by the number of CPUs. +.SH NOTES + +.B "turbostat " +must be run as root. + +.B "turbostat " +reads hardware counters, but doesn't write them. +So it will not interfere with the OS or other programs, including +multiple invocations of itself. + +\fBturbostat \fP +may work poorly on Linux-2.6.20 through 2.6.29, +as \fBacpi-cpufreq \fPperiodically cleared the APERF and MPERF +in those kernels. + +The APERF, MPERF MSRs are defined to count non-halted cycles. +Although it is not guaranteed by the architecture, turbostat assumes +that they count at TSC rate, which is true on all processors tested to date. + +.SH REFERENCES +"Intel® Turbo Boost Technology +in Intel® Coreâ„¢ Microarchitecture (Nehalem) Based Processors" +http://download.intel.com/design/processor/applnots/320354.pdf + +"Intel® 64 and IA-32 Architectures Software Developer's Manual +Volume 3B: System Programming Guide" +http://www.intel.com/products/processor/manuals/ + +.SH FILES +.ta +.nf +/dev/cpu/*/msr +.fi + +.SH "SEE ALSO" +msr(4), vmstat(8) +.PP +.SH AUTHORS +.nf +Written by Len Brown <len.brown@intel.com> diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c new file mode 100644 index 00000000..6d8ef4a3 --- /dev/null +++ b/tools/power/x86/turbostat/turbostat.c @@ -0,0 +1,1044 @@ +/* + * turbostat -- show CPU frequency and C-state residency + * on modern Intel turbo-capable processors. + * + * Copyright (c) 2010, Intel Corporation. + * Len Brown <len.brown@intel.com> + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + */ + +#include <stdio.h> +#include <unistd.h> +#include <sys/types.h> +#include <sys/wait.h> +#include <sys/stat.h> +#include <sys/resource.h> +#include <fcntl.h> +#include <signal.h> +#include <sys/time.h> +#include <stdlib.h> +#include <dirent.h> +#include <string.h> +#include <ctype.h> + +#define MSR_TSC 0x10 +#define MSR_NEHALEM_PLATFORM_INFO 0xCE +#define MSR_NEHALEM_TURBO_RATIO_LIMIT 0x1AD +#define MSR_APERF 0xE8 +#define MSR_MPERF 0xE7 +#define MSR_PKG_C2_RESIDENCY 0x60D /* SNB only */ +#define MSR_PKG_C3_RESIDENCY 0x3F8 +#define MSR_PKG_C6_RESIDENCY 0x3F9 +#define MSR_PKG_C7_RESIDENCY 0x3FA /* SNB only */ +#define MSR_CORE_C3_RESIDENCY 0x3FC +#define MSR_CORE_C6_RESIDENCY 0x3FD +#define MSR_CORE_C7_RESIDENCY 0x3FE /* SNB only */ + +char *proc_stat = "/proc/stat"; +unsigned int interval_sec = 5; /* set with -i interval_sec */ +unsigned int verbose; /* set with -v */ +unsigned int skip_c0; +unsigned int skip_c1; +unsigned int do_nhm_cstates; +unsigned int do_snb_cstates; +unsigned int has_aperf; +unsigned int units = 1000000000; /* Ghz etc */ +unsigned int genuine_intel; +unsigned int has_invariant_tsc; +unsigned int do_nehalem_platform_info; +unsigned int do_nehalem_turbo_ratio_limit; +unsigned int extra_msr_offset; +double bclk; +unsigned int show_pkg; +unsigned int show_core; +unsigned int show_cpu; + +int aperf_mperf_unstable; +int backwards_count; +char *progname; +int need_reinitialize; + +int num_cpus; + +struct counters { + unsigned long long tsc; /* per thread */ + unsigned long long aperf; /* per thread */ + unsigned long long mperf; /* per thread */ + unsigned long long c1; /* per thread (calculated) */ + unsigned long long c3; /* per core */ + unsigned long long c6; /* per core */ + unsigned long long c7; /* per core */ + unsigned long long pc2; /* per package */ + unsigned long long pc3; /* per package */ + unsigned long long pc6; /* per package */ + unsigned long long pc7; /* per package */ + unsigned long long extra_msr; /* per thread */ + int pkg; + int core; + int cpu; + struct counters *next; +}; + +struct counters *cnt_even; +struct counters *cnt_odd; +struct counters *cnt_delta; +struct counters *cnt_average; +struct timeval tv_even; +struct timeval tv_odd; +struct timeval tv_delta; + +unsigned long long get_msr(int cpu, off_t offset) +{ + ssize_t retval; + unsigned long long msr; + char pathname[32]; + int fd; + + sprintf(pathname, "/dev/cpu/%d/msr", cpu); + fd = open(pathname, O_RDONLY); + if (fd < 0) { + perror(pathname); + need_reinitialize = 1; + return 0; + } + + retval = pread(fd, &msr, sizeof msr, offset); + if (retval != sizeof msr) { + fprintf(stderr, "cpu%d pread(..., 0x%zx) = %jd\n", + cpu, offset, retval); + exit(-2); + } + + close(fd); + return msr; +} + +void print_header(void) +{ + if (show_pkg) + fprintf(stderr, "pkg "); + if (show_core) + fprintf(stderr, "core"); + if (show_cpu) + fprintf(stderr, " CPU"); + if (do_nhm_cstates) + fprintf(stderr, " %%c0 "); + if (has_aperf) + fprintf(stderr, " GHz"); + fprintf(stderr, " TSC"); + if (do_nhm_cstates) + fprintf(stderr, " %%c1 "); + if (do_nhm_cstates) + fprintf(stderr, " %%c3 "); + if (do_nhm_cstates) + fprintf(stderr, " %%c6 "); + if (do_snb_cstates) + fprintf(stderr, " %%c7 "); + if (do_snb_cstates) + fprintf(stderr, " %%pc2 "); + if (do_nhm_cstates) + fprintf(stderr, " %%pc3 "); + if (do_nhm_cstates) + fprintf(stderr, " %%pc6 "); + if (do_snb_cstates) + fprintf(stderr, " %%pc7 "); + if (extra_msr_offset) + fprintf(stderr, " MSR 0x%x ", extra_msr_offset); + + putc('\n', stderr); +} + +void dump_cnt(struct counters *cnt) +{ + fprintf(stderr, "package: %d ", cnt->pkg); + fprintf(stderr, "core:: %d ", cnt->core); + fprintf(stderr, "CPU: %d ", cnt->cpu); + fprintf(stderr, "TSC: %016llX\n", cnt->tsc); + fprintf(stderr, "c3: %016llX\n", cnt->c3); + fprintf(stderr, "c6: %016llX\n", cnt->c6); + fprintf(stderr, "c7: %016llX\n", cnt->c7); + fprintf(stderr, "aperf: %016llX\n", cnt->aperf); + fprintf(stderr, "pc2: %016llX\n", cnt->pc2); + fprintf(stderr, "pc3: %016llX\n", cnt->pc3); + fprintf(stderr, "pc6: %016llX\n", cnt->pc6); + fprintf(stderr, "pc7: %016llX\n", cnt->pc7); + fprintf(stderr, "msr0x%x: %016llX\n", extra_msr_offset, cnt->extra_msr); +} + +void dump_list(struct counters *cnt) +{ + printf("dump_list 0x%p\n", cnt); + + for (; cnt; cnt = cnt->next) + dump_cnt(cnt); +} + +void print_cnt(struct counters *p) +{ + double interval_float; + + interval_float = tv_delta.tv_sec + tv_delta.tv_usec/1000000.0; + + /* topology columns, print blanks on 1st (average) line */ + if (p == cnt_average) { + if (show_pkg) + fprintf(stderr, " "); + if (show_core) + fprintf(stderr, " "); + if (show_cpu) + fprintf(stderr, " "); + } else { + if (show_pkg) + fprintf(stderr, "%4d", p->pkg); + if (show_core) + fprintf(stderr, "%4d", p->core); + if (show_cpu) + fprintf(stderr, "%4d", p->cpu); + } + + /* %c0 */ + if (do_nhm_cstates) { + if (!skip_c0) + fprintf(stderr, "%7.2f", 100.0 * p->mperf/p->tsc); + else + fprintf(stderr, " ****"); + } + + /* GHz */ + if (has_aperf) { + if (!aperf_mperf_unstable) { + fprintf(stderr, "%5.2f", + 1.0 * p->tsc / units * p->aperf / + p->mperf / interval_float); + } else { + if (p->aperf > p->tsc || p->mperf > p->tsc) { + fprintf(stderr, " ****"); + } else { + fprintf(stderr, "%4.1f*", + 1.0 * p->tsc / + units * p->aperf / + p->mperf / interval_float); + } + } + } + + /* TSC */ + fprintf(stderr, "%5.2f", 1.0 * p->tsc/units/interval_float); + + if (do_nhm_cstates) { + if (!skip_c1) + fprintf(stderr, "%7.2f", 100.0 * p->c1/p->tsc); + else + fprintf(stderr, " ****"); + } + if (do_nhm_cstates) + fprintf(stderr, "%7.2f", 100.0 * p->c3/p->tsc); + if (do_nhm_cstates) + fprintf(stderr, "%7.2f", 100.0 * p->c6/p->tsc); + if (do_snb_cstates) + fprintf(stderr, "%7.2f", 100.0 * p->c7/p->tsc); + if (do_snb_cstates) + fprintf(stderr, "%7.2f", 100.0 * p->pc2/p->tsc); + if (do_nhm_cstates) + fprintf(stderr, "%7.2f", 100.0 * p->pc3/p->tsc); + if (do_nhm_cstates) + fprintf(stderr, "%7.2f", 100.0 * p->pc6/p->tsc); + if (do_snb_cstates) + fprintf(stderr, "%7.2f", 100.0 * p->pc7/p->tsc); + if (extra_msr_offset) + fprintf(stderr, " 0x%016llx", p->extra_msr); + putc('\n', stderr); +} + +void print_counters(struct counters *counters) +{ + struct counters *cnt; + + print_header(); + + if (num_cpus > 1) + print_cnt(cnt_average); + + for (cnt = counters; cnt != NULL; cnt = cnt->next) + print_cnt(cnt); + +} + +#define SUBTRACT_COUNTER(after, before, delta) (delta = (after - before), (before > after)) + +int compute_delta(struct counters *after, + struct counters *before, struct counters *delta) +{ + int errors = 0; + int perf_err = 0; + + skip_c0 = skip_c1 = 0; + + for ( ; after && before && delta; + after = after->next, before = before->next, delta = delta->next) { + if (before->cpu != after->cpu) { + printf("cpu configuration changed: %d != %d\n", + before->cpu, after->cpu); + return -1; + } + + if (SUBTRACT_COUNTER(after->tsc, before->tsc, delta->tsc)) { + fprintf(stderr, "cpu%d TSC went backwards %llX to %llX\n", + before->cpu, before->tsc, after->tsc); + errors++; + } + /* check for TSC < 1 Mcycles over interval */ + if (delta->tsc < (1000 * 1000)) { + fprintf(stderr, "Insanely slow TSC rate," + " TSC stops in idle?\n"); + fprintf(stderr, "You can disable all c-states" + " by booting with \"idle=poll\"\n"); + fprintf(stderr, "or just the deep ones with" + " \"processor.max_cstate=1\"\n"); + exit(-3); + } + if (SUBTRACT_COUNTER(after->c3, before->c3, delta->c3)) { + fprintf(stderr, "cpu%d c3 counter went backwards %llX to %llX\n", + before->cpu, before->c3, after->c3); + errors++; + } + if (SUBTRACT_COUNTER(after->c6, before->c6, delta->c6)) { + fprintf(stderr, "cpu%d c6 counter went backwards %llX to %llX\n", + before->cpu, before->c6, after->c6); + errors++; + } + if (SUBTRACT_COUNTER(after->c7, before->c7, delta->c7)) { + fprintf(stderr, "cpu%d c7 counter went backwards %llX to %llX\n", + before->cpu, before->c7, after->c7); + errors++; + } + if (SUBTRACT_COUNTER(after->pc2, before->pc2, delta->pc2)) { + fprintf(stderr, "cpu%d pc2 counter went backwards %llX to %llX\n", + before->cpu, before->pc2, after->pc2); + errors++; + } + if (SUBTRACT_COUNTER(after->pc3, before->pc3, delta->pc3)) { + fprintf(stderr, "cpu%d pc3 counter went backwards %llX to %llX\n", + before->cpu, before->pc3, after->pc3); + errors++; + } + if (SUBTRACT_COUNTER(after->pc6, before->pc6, delta->pc6)) { + fprintf(stderr, "cpu%d pc6 counter went backwards %llX to %llX\n", + before->cpu, before->pc6, after->pc6); + errors++; + } + if (SUBTRACT_COUNTER(after->pc7, before->pc7, delta->pc7)) { + fprintf(stderr, "cpu%d pc7 counter went backwards %llX to %llX\n", + before->cpu, before->pc7, after->pc7); + errors++; + } + + perf_err = SUBTRACT_COUNTER(after->aperf, before->aperf, delta->aperf); + if (perf_err) { + fprintf(stderr, "cpu%d aperf counter went backwards %llX to %llX\n", + before->cpu, before->aperf, after->aperf); + } + perf_err |= SUBTRACT_COUNTER(after->mperf, before->mperf, delta->mperf); + if (perf_err) { + fprintf(stderr, "cpu%d mperf counter went backwards %llX to %llX\n", + before->cpu, before->mperf, after->mperf); + } + if (perf_err) { + if (!aperf_mperf_unstable) { + fprintf(stderr, "%s: APERF or MPERF went backwards *\n", progname); + fprintf(stderr, "* Frequency results do not cover entire interval *\n"); + fprintf(stderr, "* fix this by running Linux-2.6.30 or later *\n"); + + aperf_mperf_unstable = 1; + } + /* + * mperf delta is likely a huge "positive" number + * can not use it for calculating c0 time + */ + skip_c0 = 1; + skip_c1 = 1; + } + + /* + * As mperf and tsc collection are not atomic, + * it is possible for mperf's non-halted cycles + * to exceed TSC's all cycles: show c1 = 0% in that case. + */ + if (delta->mperf > delta->tsc) + delta->c1 = 0; + else /* normal case, derive c1 */ + delta->c1 = delta->tsc - delta->mperf + - delta->c3 - delta->c6 - delta->c7; + + if (delta->mperf == 0) + delta->mperf = 1; /* divide by 0 protection */ + + /* + * for "extra msr", just copy the latest w/o subtracting + */ + delta->extra_msr = after->extra_msr; + if (errors) { + fprintf(stderr, "ERROR cpu%d before:\n", before->cpu); + dump_cnt(before); + fprintf(stderr, "ERROR cpu%d after:\n", before->cpu); + dump_cnt(after); + errors = 0; + } + } + return 0; +} + +void compute_average(struct counters *delta, struct counters *avg) +{ + struct counters *sum; + + sum = calloc(1, sizeof(struct counters)); + if (sum == NULL) { + perror("calloc sum"); + exit(1); + } + + for (; delta; delta = delta->next) { + sum->tsc += delta->tsc; + sum->c1 += delta->c1; + sum->c3 += delta->c3; + sum->c6 += delta->c6; + sum->c7 += delta->c7; + sum->aperf += delta->aperf; + sum->mperf += delta->mperf; + sum->pc2 += delta->pc2; + sum->pc3 += delta->pc3; + sum->pc6 += delta->pc6; + sum->pc7 += delta->pc7; + } + avg->tsc = sum->tsc/num_cpus; + avg->c1 = sum->c1/num_cpus; + avg->c3 = sum->c3/num_cpus; + avg->c6 = sum->c6/num_cpus; + avg->c7 = sum->c7/num_cpus; + avg->aperf = sum->aperf/num_cpus; + avg->mperf = sum->mperf/num_cpus; + avg->pc2 = sum->pc2/num_cpus; + avg->pc3 = sum->pc3/num_cpus; + avg->pc6 = sum->pc6/num_cpus; + avg->pc7 = sum->pc7/num_cpus; + + free(sum); +} + +void get_counters(struct counters *cnt) +{ + for ( ; cnt; cnt = cnt->next) { + cnt->tsc = get_msr(cnt->cpu, MSR_TSC); + if (do_nhm_cstates) + cnt->c3 = get_msr(cnt->cpu, MSR_CORE_C3_RESIDENCY); + if (do_nhm_cstates) + cnt->c6 = get_msr(cnt->cpu, MSR_CORE_C6_RESIDENCY); + if (do_snb_cstates) + cnt->c7 = get_msr(cnt->cpu, MSR_CORE_C7_RESIDENCY); + if (has_aperf) + cnt->aperf = get_msr(cnt->cpu, MSR_APERF); + if (has_aperf) + cnt->mperf = get_msr(cnt->cpu, MSR_MPERF); + if (do_snb_cstates) + cnt->pc2 = get_msr(cnt->cpu, MSR_PKG_C2_RESIDENCY); + if (do_nhm_cstates) + cnt->pc3 = get_msr(cnt->cpu, MSR_PKG_C3_RESIDENCY); + if (do_nhm_cstates) + cnt->pc6 = get_msr(cnt->cpu, MSR_PKG_C6_RESIDENCY); + if (do_snb_cstates) + cnt->pc7 = get_msr(cnt->cpu, MSR_PKG_C7_RESIDENCY); + if (extra_msr_offset) + cnt->extra_msr = get_msr(cnt->cpu, extra_msr_offset); + } +} + +void print_nehalem_info(void) +{ + unsigned long long msr; + unsigned int ratio; + + if (!do_nehalem_platform_info) + return; + + msr = get_msr(0, MSR_NEHALEM_PLATFORM_INFO); + + ratio = (msr >> 40) & 0xFF; + fprintf(stderr, "%d * %.0f = %.0f MHz max efficiency\n", + ratio, bclk, ratio * bclk); + + ratio = (msr >> 8) & 0xFF; + fprintf(stderr, "%d * %.0f = %.0f MHz TSC frequency\n", + ratio, bclk, ratio * bclk); + + if (verbose > 1) + fprintf(stderr, "MSR_NEHALEM_PLATFORM_INFO: 0x%llx\n", msr); + + if (!do_nehalem_turbo_ratio_limit) + return; + + msr = get_msr(0, MSR_NEHALEM_TURBO_RATIO_LIMIT); + + ratio = (msr >> 24) & 0xFF; + if (ratio) + fprintf(stderr, "%d * %.0f = %.0f MHz max turbo 4 active cores\n", + ratio, bclk, ratio * bclk); + + ratio = (msr >> 16) & 0xFF; + if (ratio) + fprintf(stderr, "%d * %.0f = %.0f MHz max turbo 3 active cores\n", + ratio, bclk, ratio * bclk); + + ratio = (msr >> 8) & 0xFF; + if (ratio) + fprintf(stderr, "%d * %.0f = %.0f MHz max turbo 2 active cores\n", + ratio, bclk, ratio * bclk); + + ratio = (msr >> 0) & 0xFF; + if (ratio) + fprintf(stderr, "%d * %.0f = %.0f MHz max turbo 1 active cores\n", + ratio, bclk, ratio * bclk); + +} + +void free_counter_list(struct counters *list) +{ + struct counters *p; + + for (p = list; p; ) { + struct counters *free_me; + + free_me = p; + p = p->next; + free(free_me); + } +} + +void free_all_counters(void) +{ + free_counter_list(cnt_even); + cnt_even = NULL; + + free_counter_list(cnt_odd); + cnt_odd = NULL; + + free_counter_list(cnt_delta); + cnt_delta = NULL; + + free_counter_list(cnt_average); + cnt_average = NULL; +} + +void insert_counters(struct counters **list, + struct counters *new) +{ + struct counters *prev; + + /* + * list was empty + */ + if (*list == NULL) { + new->next = *list; + *list = new; + return; + } + + show_cpu = 1; /* there is more than one CPU */ + + /* + * insert on front of list. + * It is sorted by ascending package#, core#, cpu# + */ + if (((*list)->pkg > new->pkg) || + (((*list)->pkg == new->pkg) && ((*list)->core > new->core)) || + (((*list)->pkg == new->pkg) && ((*list)->core == new->core) && ((*list)->cpu > new->cpu))) { + new->next = *list; + *list = new; + return; + } + + prev = *list; + + while (prev->next && (prev->next->pkg < new->pkg)) { + prev = prev->next; + show_pkg = 1; /* there is more than 1 package */ + } + + while (prev->next && (prev->next->pkg == new->pkg) + && (prev->next->core < new->core)) { + prev = prev->next; + show_core = 1; /* there is more than 1 core */ + } + + while (prev->next && (prev->next->pkg == new->pkg) + && (prev->next->core == new->core) + && (prev->next->cpu < new->cpu)) { + prev = prev->next; + } + + /* + * insert after "prev" + */ + new->next = prev->next; + prev->next = new; +} + +void alloc_new_counters(int pkg, int core, int cpu) +{ + struct counters *new; + + if (verbose > 1) + printf("pkg%d core%d, cpu%d\n", pkg, core, cpu); + + new = (struct counters *)calloc(1, sizeof(struct counters)); + if (new == NULL) { + perror("calloc"); + exit(1); + } + new->pkg = pkg; + new->core = core; + new->cpu = cpu; + insert_counters(&cnt_odd, new); + + new = (struct counters *)calloc(1, + sizeof(struct counters)); + if (new == NULL) { + perror("calloc"); + exit(1); + } + new->pkg = pkg; + new->core = core; + new->cpu = cpu; + insert_counters(&cnt_even, new); + + new = (struct counters *)calloc(1, sizeof(struct counters)); + if (new == NULL) { + perror("calloc"); + exit(1); + } + new->pkg = pkg; + new->core = core; + new->cpu = cpu; + insert_counters(&cnt_delta, new); + + new = (struct counters *)calloc(1, sizeof(struct counters)); + if (new == NULL) { + perror("calloc"); + exit(1); + } + new->pkg = pkg; + new->core = core; + new->cpu = cpu; + cnt_average = new; +} + +int get_physical_package_id(int cpu) +{ + char path[64]; + FILE *filep; + int pkg; + + sprintf(path, "/sys/devices/system/cpu/cpu%d/topology/physical_package_id", cpu); + filep = fopen(path, "r"); + if (filep == NULL) { + perror(path); + exit(1); + } + fscanf(filep, "%d", &pkg); + fclose(filep); + return pkg; +} + +int get_core_id(int cpu) +{ + char path[64]; + FILE *filep; + int core; + + sprintf(path, "/sys/devices/system/cpu/cpu%d/topology/core_id", cpu); + filep = fopen(path, "r"); + if (filep == NULL) { + perror(path); + exit(1); + } + fscanf(filep, "%d", &core); + fclose(filep); + return core; +} + +/* + * run func(index, cpu) on every cpu in /proc/stat + */ + +int for_all_cpus(void (func)(int, int, int)) +{ + FILE *fp; + int cpu_count; + int retval; + + fp = fopen(proc_stat, "r"); + if (fp == NULL) { + perror(proc_stat); + exit(1); + } + + retval = fscanf(fp, "cpu %*d %*d %*d %*d %*d %*d %*d %*d %*d %*d\n"); + if (retval != 0) { + perror("/proc/stat format"); + exit(1); + } + + for (cpu_count = 0; ; cpu_count++) { + int cpu; + + retval = fscanf(fp, "cpu%u %*d %*d %*d %*d %*d %*d %*d %*d %*d %*d\n", &cpu); + if (retval != 1) + break; + + func(get_physical_package_id(cpu), get_core_id(cpu), cpu); + } + fclose(fp); + return cpu_count; +} + +void re_initialize(void) +{ + printf("turbostat: topology changed, re-initializing.\n"); + free_all_counters(); + num_cpus = for_all_cpus(alloc_new_counters); + need_reinitialize = 0; + printf("num_cpus is now %d\n", num_cpus); +} + +void dummy(int pkg, int core, int cpu) { return; } +/* + * check to see if a cpu came on-line + */ +void verify_num_cpus(void) +{ + int new_num_cpus; + + new_num_cpus = for_all_cpus(dummy); + + if (new_num_cpus != num_cpus) { + if (verbose) + printf("num_cpus was %d, is now %d\n", + num_cpus, new_num_cpus); + need_reinitialize = 1; + } +} + +void turbostat_loop() +{ +restart: + get_counters(cnt_even); + gettimeofday(&tv_even, (struct timezone *)NULL); + + while (1) { + verify_num_cpus(); + if (need_reinitialize) { + re_initialize(); + goto restart; + } + sleep(interval_sec); + get_counters(cnt_odd); + gettimeofday(&tv_odd, (struct timezone *)NULL); + + compute_delta(cnt_odd, cnt_even, cnt_delta); + timersub(&tv_odd, &tv_even, &tv_delta); + compute_average(cnt_delta, cnt_average); + print_counters(cnt_delta); + if (need_reinitialize) { + re_initialize(); + goto restart; + } + sleep(interval_sec); + get_counters(cnt_even); + gettimeofday(&tv_even, (struct timezone *)NULL); + compute_delta(cnt_even, cnt_odd, cnt_delta); + timersub(&tv_even, &tv_odd, &tv_delta); + compute_average(cnt_delta, cnt_average); + print_counters(cnt_delta); + } +} + +void check_dev_msr() +{ + struct stat sb; + + if (stat("/dev/cpu/0/msr", &sb)) { + fprintf(stderr, "no /dev/cpu/0/msr\n"); + fprintf(stderr, "Try \"# modprobe msr\"\n"); + exit(-5); + } +} + +void check_super_user() +{ + if (getuid() != 0) { + fprintf(stderr, "must be root\n"); + exit(-6); + } +} + +int has_nehalem_turbo_ratio_limit(unsigned int family, unsigned int model) +{ + if (!genuine_intel) + return 0; + + if (family != 6) + return 0; + + switch (model) { + case 0x1A: /* Core i7, Xeon 5500 series - Bloomfield, Gainstown NHM-EP */ + case 0x1E: /* Core i7 and i5 Processor - Clarksfield, Lynnfield, Jasper Forest */ + case 0x1F: /* Core i7 and i5 Processor - Nehalem */ + case 0x25: /* Westmere Client - Clarkdale, Arrandale */ + case 0x2C: /* Westmere EP - Gulftown */ + case 0x2A: /* SNB */ + case 0x2D: /* SNB Xeon */ + return 1; + case 0x2E: /* Nehalem-EX Xeon - Beckton */ + case 0x2F: /* Westmere-EX Xeon - Eagleton */ + default: + return 0; + } +} + +int is_snb(unsigned int family, unsigned int model) +{ + if (!genuine_intel) + return 0; + + switch (model) { + case 0x2A: + case 0x2D: + return 1; + } + return 0; +} + +double discover_bclk(unsigned int family, unsigned int model) +{ + if (is_snb(family, model)) + return 100.00; + else + return 133.33; +} + +void check_cpuid() +{ + unsigned int eax, ebx, ecx, edx, max_level; + unsigned int fms, family, model, stepping; + + eax = ebx = ecx = edx = 0; + + asm("cpuid" : "=a" (max_level), "=b" (ebx), "=c" (ecx), "=d" (edx) : "a" (0)); + + if (ebx == 0x756e6547 && edx == 0x49656e69 && ecx == 0x6c65746e) + genuine_intel = 1; + + if (verbose) + fprintf(stderr, "%.4s%.4s%.4s ", + (char *)&ebx, (char *)&edx, (char *)&ecx); + + asm("cpuid" : "=a" (fms), "=c" (ecx), "=d" (edx) : "a" (1) : "ebx"); + family = (fms >> 8) & 0xf; + model = (fms >> 4) & 0xf; + stepping = fms & 0xf; + if (family == 6 || family == 0xf) + model += ((fms >> 16) & 0xf) << 4; + + if (verbose) + fprintf(stderr, "%d CPUID levels; family:model:stepping 0x%x:%x:%x (%d:%d:%d)\n", + max_level, family, model, stepping, family, model, stepping); + + if (!(edx & (1 << 5))) { + fprintf(stderr, "CPUID: no MSR\n"); + exit(1); + } + + /* + * check max extended function levels of CPUID. + * This is needed to check for invariant TSC. + * This check is valid for both Intel and AMD. + */ + ebx = ecx = edx = 0; + asm("cpuid" : "=a" (max_level), "=b" (ebx), "=c" (ecx), "=d" (edx) : "a" (0x80000000)); + + if (max_level < 0x80000007) { + fprintf(stderr, "CPUID: no invariant TSC (max_level 0x%x)\n", max_level); + exit(1); + } + + /* + * Non-Stop TSC is advertised by CPUID.EAX=0x80000007: EDX.bit8 + * this check is valid for both Intel and AMD + */ + asm("cpuid" : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx) : "a" (0x80000007)); + has_invariant_tsc = edx & (1 << 8); + + if (!has_invariant_tsc) { + fprintf(stderr, "No invariant TSC\n"); + exit(1); + } + + /* + * APERF/MPERF is advertised by CPUID.EAX=0x6: ECX.bit0 + * this check is valid for both Intel and AMD + */ + + asm("cpuid" : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx) : "a" (0x6)); + has_aperf = ecx & (1 << 0); + if (!has_aperf) { + fprintf(stderr, "No APERF MSR\n"); + exit(1); + } + + do_nehalem_platform_info = genuine_intel && has_invariant_tsc; + do_nhm_cstates = genuine_intel; /* all Intel w/ non-stop TSC have NHM counters */ + do_snb_cstates = is_snb(family, model); + bclk = discover_bclk(family, model); + + do_nehalem_turbo_ratio_limit = has_nehalem_turbo_ratio_limit(family, model); +} + + +void usage() +{ + fprintf(stderr, "%s: [-v] [-M MSR#] [-i interval_sec | command ...]\n", + progname); + exit(1); +} + + +/* + * in /dev/cpu/ return success for names that are numbers + * ie. filter out ".", "..", "microcode". + */ +int dir_filter(const struct dirent *dirp) +{ + if (isdigit(dirp->d_name[0])) + return 1; + else + return 0; +} + +int open_dev_cpu_msr(int dummy1) +{ + return 0; +} + +void turbostat_init() +{ + check_cpuid(); + + check_dev_msr(); + check_super_user(); + + num_cpus = for_all_cpus(alloc_new_counters); + + if (verbose) + print_nehalem_info(); +} + +int fork_it(char **argv) +{ + int retval; + pid_t child_pid; + get_counters(cnt_even); + gettimeofday(&tv_even, (struct timezone *)NULL); + + child_pid = fork(); + if (!child_pid) { + /* child */ + execvp(argv[0], argv); + } else { + int status; + + /* parent */ + if (child_pid == -1) { + perror("fork"); + exit(1); + } + + signal(SIGINT, SIG_IGN); + signal(SIGQUIT, SIG_IGN); + if (waitpid(child_pid, &status, 0) == -1) { + perror("wait"); + exit(1); + } + } + get_counters(cnt_odd); + gettimeofday(&tv_odd, (struct timezone *)NULL); + retval = compute_delta(cnt_odd, cnt_even, cnt_delta); + + timersub(&tv_odd, &tv_even, &tv_delta); + compute_average(cnt_delta, cnt_average); + if (!retval) + print_counters(cnt_delta); + + fprintf(stderr, "%.6f sec\n", tv_delta.tv_sec + tv_delta.tv_usec/1000000.0); + + return 0; +} + +void cmdline(int argc, char **argv) +{ + int opt; + + progname = argv[0]; + + while ((opt = getopt(argc, argv, "+vi:M:")) != -1) { + switch (opt) { + case 'v': + verbose++; + break; + case 'i': + interval_sec = atoi(optarg); + break; + case 'M': + sscanf(optarg, "%x", &extra_msr_offset); + if (verbose > 1) + fprintf(stderr, "MSR 0x%X\n", extra_msr_offset); + break; + default: + usage(); + } + } +} + +int main(int argc, char **argv) +{ + cmdline(argc, argv); + + if (verbose > 1) + fprintf(stderr, "turbostat Dec 6, 2010" + " - Len Brown <lenb@kernel.org>\n"); + if (verbose > 1) + fprintf(stderr, "http://userweb.kernel.org/~lenb/acpi/utils/pmtools/turbostat/\n"); + + turbostat_init(); + + /* + * if any params left, it must be a command to fork + */ + if (argc - optind) + return fork_it(argv + optind); + else + turbostat_loop(); + + return 0; +} diff --git a/tools/power/x86/x86_energy_perf_policy/Makefile b/tools/power/x86/x86_energy_perf_policy/Makefile new file mode 100644 index 00000000..f458237f --- /dev/null +++ b/tools/power/x86/x86_energy_perf_policy/Makefile @@ -0,0 +1,8 @@ +x86_energy_perf_policy : x86_energy_perf_policy.c + +clean : + rm -f x86_energy_perf_policy + +install : + install x86_energy_perf_policy /usr/bin/ + install x86_energy_perf_policy.8 /usr/share/man/man8/ diff --git a/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.8 b/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.8 new file mode 100644 index 00000000..8eaaad64 --- /dev/null +++ b/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.8 @@ -0,0 +1,104 @@ +.\" This page Copyright (C) 2010 Len Brown <len.brown@intel.com> +.\" Distributed under the GPL, Copyleft 1994. +.TH X86_ENERGY_PERF_POLICY 8 +.SH NAME +x86_energy_perf_policy \- read or write MSR_IA32_ENERGY_PERF_BIAS +.SH SYNOPSIS +.ft B +.B x86_energy_perf_policy +.RB [ "\-c cpu" ] +.RB [ "\-v" ] +.RB "\-r" +.br +.B x86_energy_perf_policy +.RB [ "\-c cpu" ] +.RB [ "\-v" ] +.RB 'performance' +.br +.B x86_energy_perf_policy +.RB [ "\-c cpu" ] +.RB [ "\-v" ] +.RB 'normal' +.br +.B x86_energy_perf_policy +.RB [ "\-c cpu" ] +.RB [ "\-v" ] +.RB 'powersave' +.br +.B x86_energy_perf_policy +.RB [ "\-c cpu" ] +.RB [ "\-v" ] +.RB n +.br +.SH DESCRIPTION +\fBx86_energy_perf_policy\fP +allows software to convey +its policy for the relative importance of performance +versus energy savings to the processor. + +The processor uses this information in model-specific ways +when it must select trade-offs between performance and +energy efficiency. + +This policy hint does not supersede Processor Performance states +(P-states) or CPU Idle power states (C-states), but allows +software to have influence where it would otherwise be unable +to express a preference. + +For example, this setting may tell the hardware how +aggressively or conservatively to control frequency +in the "turbo range" above the explicitly OS-controlled +P-state frequency range. It may also tell the hardware +how aggressively is should enter the OS requested C-states. + +Support for this feature is indicated by CPUID.06H.ECX.bit3 +per the Intel Architectures Software Developer's Manual. + +.SS Options +\fB-c\fP limits operation to a single CPU. +The default is to operate on all CPUs. +Note that MSR_IA32_ENERGY_PERF_BIAS is defined per +logical processor, but that the initial implementations +of the MSR were shared among all processors in each package. +.PP +\fB-v\fP increases verbosity. By default +x86_energy_perf_policy is silent. +.PP +\fB-r\fP is for "read-only" mode - the unchanged state +is read and displayed. +.PP +.I performance +Set a policy where performance is paramount. +The processor will be unwilling to sacrifice any performance +for the sake of energy saving. This is the hardware default. +.PP +.I normal +Set a policy with a normal balance between performance and energy efficiency. +The processor will tolerate minor performance compromise +for potentially significant energy savings. +This reasonable default for most desktops and servers. +.PP +.I powersave +Set a policy where the processor can accept +a measurable performance hit to maximize energy efficiency. +.PP +.I n +Set MSR_IA32_ENERGY_PERF_BIAS to the specified number. +The range of valid numbers is 0-15, where 0 is maximum +performance and 15 is maximum energy efficiency. + +.SH NOTES +.B "x86_energy_perf_policy " +runs only as root. +.SH FILES +.ta +.nf +/dev/cpu/*/msr +.fi + +.SH "SEE ALSO" +msr(4) +.PP +.SH AUTHORS +.nf +Written by Len Brown <len.brown@intel.com> diff --git a/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.c b/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.c new file mode 100644 index 00000000..2618ef2b --- /dev/null +++ b/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.c @@ -0,0 +1,325 @@ +/* + * x86_energy_perf_policy -- set the energy versus performance + * policy preference bias on recent X86 processors. + */ +/* + * Copyright (c) 2010, Intel Corporation. + * Len Brown <len.brown@intel.com> + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + */ + +#include <stdio.h> +#include <unistd.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <sys/resource.h> +#include <fcntl.h> +#include <signal.h> +#include <sys/time.h> +#include <stdlib.h> +#include <string.h> + +unsigned int verbose; /* set with -v */ +unsigned int read_only; /* set with -r */ +char *progname; +unsigned long long new_bias; +int cpu = -1; + +/* + * Usage: + * + * -c cpu: limit action to a single CPU (default is all CPUs) + * -v: verbose output (can invoke more than once) + * -r: read-only, don't change any settings + * + * performance + * Performance is paramount. + * Unwilling to sacrifice any performance + * for the sake of energy saving. (hardware default) + * + * normal + * Can tolerate minor performance compromise + * for potentially significant energy savings. + * (reasonable default for most desktops and servers) + * + * powersave + * Can tolerate significant performance hit + * to maximize energy savings. + * + * n + * a numerical value to write to the underlying MSR. + */ +void usage(void) +{ + printf("%s: [-c cpu] [-v] " + "(-r | 'performance' | 'normal' | 'powersave' | n)\n", + progname); + exit(1); +} + +#define MSR_IA32_ENERGY_PERF_BIAS 0x000001b0 + +#define BIAS_PERFORMANCE 0 +#define BIAS_BALANCE 6 +#define BIAS_POWERSAVE 15 + +void cmdline(int argc, char **argv) +{ + int opt; + + progname = argv[0]; + + while ((opt = getopt(argc, argv, "+rvc:")) != -1) { + switch (opt) { + case 'c': + cpu = atoi(optarg); + break; + case 'r': + read_only = 1; + break; + case 'v': + verbose++; + break; + default: + usage(); + } + } + /* if -r, then should be no additional optind */ + if (read_only && (argc > optind)) + usage(); + + /* + * if no -r , then must be one additional optind + */ + if (!read_only) { + + if (argc != optind + 1) { + printf("must supply -r or policy param\n"); + usage(); + } + + if (!strcmp("performance", argv[optind])) { + new_bias = BIAS_PERFORMANCE; + } else if (!strcmp("normal", argv[optind])) { + new_bias = BIAS_BALANCE; + } else if (!strcmp("powersave", argv[optind])) { + new_bias = BIAS_POWERSAVE; + } else { + char *endptr; + + new_bias = strtoull(argv[optind], &endptr, 0); + if (endptr == argv[optind] || + new_bias > BIAS_POWERSAVE) { + fprintf(stderr, "invalid value: %s\n", + argv[optind]); + usage(); + } + } + } +} + +/* + * validate_cpuid() + * returns on success, quietly exits on failure (make verbose with -v) + */ +void validate_cpuid(void) +{ + unsigned int eax, ebx, ecx, edx, max_level; + char brand[16]; + unsigned int fms, family, model, stepping; + + eax = ebx = ecx = edx = 0; + + asm("cpuid" : "=a" (max_level), "=b" (ebx), "=c" (ecx), + "=d" (edx) : "a" (0)); + + if (ebx != 0x756e6547 || edx != 0x49656e69 || ecx != 0x6c65746e) { + if (verbose) + fprintf(stderr, "%.4s%.4s%.4s != GenuineIntel", + (char *)&ebx, (char *)&edx, (char *)&ecx); + exit(1); + } + + asm("cpuid" : "=a" (fms), "=c" (ecx), "=d" (edx) : "a" (1) : "ebx"); + family = (fms >> 8) & 0xf; + model = (fms >> 4) & 0xf; + stepping = fms & 0xf; + if (family == 6 || family == 0xf) + model += ((fms >> 16) & 0xf) << 4; + + if (verbose > 1) + printf("CPUID %s %d levels family:model:stepping " + "0x%x:%x:%x (%d:%d:%d)\n", brand, max_level, + family, model, stepping, family, model, stepping); + + if (!(edx & (1 << 5))) { + if (verbose) + printf("CPUID: no MSR\n"); + exit(1); + } + + /* + * Support for MSR_IA32_ENERGY_PERF_BIAS + * is indicated by CPUID.06H.ECX.bit3 + */ + asm("cpuid" : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx) : "a" (6)); + if (verbose) + printf("CPUID.06H.ECX: 0x%x\n", ecx); + if (!(ecx & (1 << 3))) { + if (verbose) + printf("CPUID: No MSR_IA32_ENERGY_PERF_BIAS\n"); + exit(1); + } + return; /* success */ +} + +unsigned long long get_msr(int cpu, int offset) +{ + unsigned long long msr; + char msr_path[32]; + int retval; + int fd; + + sprintf(msr_path, "/dev/cpu/%d/msr", cpu); + fd = open(msr_path, O_RDONLY); + if (fd < 0) { + printf("Try \"# modprobe msr\"\n"); + perror(msr_path); + exit(1); + } + + retval = pread(fd, &msr, sizeof msr, offset); + + if (retval != sizeof msr) { + printf("pread cpu%d 0x%x = %d\n", cpu, offset, retval); + exit(-2); + } + close(fd); + return msr; +} + +unsigned long long put_msr(int cpu, unsigned long long new_msr, int offset) +{ + unsigned long long old_msr; + char msr_path[32]; + int retval; + int fd; + + sprintf(msr_path, "/dev/cpu/%d/msr", cpu); + fd = open(msr_path, O_RDWR); + if (fd < 0) { + perror(msr_path); + exit(1); + } + + retval = pread(fd, &old_msr, sizeof old_msr, offset); + if (retval != sizeof old_msr) { + perror("pwrite"); + printf("pread cpu%d 0x%x = %d\n", cpu, offset, retval); + exit(-2); + } + + retval = pwrite(fd, &new_msr, sizeof new_msr, offset); + if (retval != sizeof new_msr) { + perror("pwrite"); + printf("pwrite cpu%d 0x%x = %d\n", cpu, offset, retval); + exit(-2); + } + + close(fd); + + return old_msr; +} + +void print_msr(int cpu) +{ + printf("cpu%d: 0x%016llx\n", + cpu, get_msr(cpu, MSR_IA32_ENERGY_PERF_BIAS)); +} + +void update_msr(int cpu) +{ + unsigned long long previous_msr; + + previous_msr = put_msr(cpu, new_bias, MSR_IA32_ENERGY_PERF_BIAS); + + if (verbose) + printf("cpu%d msr0x%x 0x%016llx -> 0x%016llx\n", + cpu, MSR_IA32_ENERGY_PERF_BIAS, previous_msr, new_bias); + + return; +} + +char *proc_stat = "/proc/stat"; +/* + * run func() on every cpu in /dev/cpu + */ +void for_every_cpu(void (func)(int)) +{ + FILE *fp; + int retval; + + fp = fopen(proc_stat, "r"); + if (fp == NULL) { + perror(proc_stat); + exit(1); + } + + retval = fscanf(fp, "cpu %*d %*d %*d %*d %*d %*d %*d %*d %*d %*d\n"); + if (retval != 0) { + perror("/proc/stat format"); + exit(1); + } + + while (1) { + int cpu; + + retval = fscanf(fp, + "cpu%u %*d %*d %*d %*d %*d %*d %*d %*d %*d %*d\n", + &cpu); + if (retval != 1) + return; + + func(cpu); + } + fclose(fp); +} + +int main(int argc, char **argv) +{ + cmdline(argc, argv); + + if (verbose > 1) + printf("x86_energy_perf_policy Nov 24, 2010" + " - Len Brown <lenb@kernel.org>\n"); + if (verbose > 1 && !read_only) + printf("new_bias %lld\n", new_bias); + + validate_cpuid(); + + if (cpu != -1) { + if (read_only) + print_msr(cpu); + else + update_msr(cpu); + } else { + if (read_only) + for_every_cpu(print_msr); + else + for_every_cpu(update_msr); + } + + return 0; +} diff --git a/tools/slub/slabinfo.c b/tools/slub/slabinfo.c new file mode 100644 index 00000000..516551c9 --- /dev/null +++ b/tools/slub/slabinfo.c @@ -0,0 +1,1364 @@ +/* + * Slabinfo: Tool to get reports about slabs + * + * (C) 2007 sgi, Christoph Lameter + * + * Compile by: + * + * gcc -o slabinfo slabinfo.c + */ +#include <stdio.h> +#include <stdlib.h> +#include <sys/types.h> +#include <dirent.h> +#include <strings.h> +#include <string.h> +#include <unistd.h> +#include <stdarg.h> +#include <getopt.h> +#include <regex.h> +#include <errno.h> + +#define MAX_SLABS 500 +#define MAX_ALIASES 500 +#define MAX_NODES 1024 + +struct slabinfo { + char *name; + int alias; + int refs; + int aliases, align, cache_dma, cpu_slabs, destroy_by_rcu; + int hwcache_align, object_size, objs_per_slab; + int sanity_checks, slab_size, store_user, trace; + int order, poison, reclaim_account, red_zone; + unsigned long partial, objects, slabs, objects_partial, objects_total; + unsigned long alloc_fastpath, alloc_slowpath; + unsigned long free_fastpath, free_slowpath; + unsigned long free_frozen, free_add_partial, free_remove_partial; + unsigned long alloc_from_partial, alloc_slab, free_slab, alloc_refill; + unsigned long cpuslab_flush, deactivate_full, deactivate_empty; + unsigned long deactivate_to_head, deactivate_to_tail; + unsigned long deactivate_remote_frees, order_fallback; + int numa[MAX_NODES]; + int numa_partial[MAX_NODES]; +} slabinfo[MAX_SLABS]; + +struct aliasinfo { + char *name; + char *ref; + struct slabinfo *slab; +} aliasinfo[MAX_ALIASES]; + +int slabs = 0; +int actual_slabs = 0; +int aliases = 0; +int alias_targets = 0; +int highest_node = 0; + +char buffer[4096]; + +int show_empty = 0; +int show_report = 0; +int show_alias = 0; +int show_slab = 0; +int skip_zero = 1; +int show_numa = 0; +int show_track = 0; +int show_first_alias = 0; +int validate = 0; +int shrink = 0; +int show_inverted = 0; +int show_single_ref = 0; +int show_totals = 0; +int sort_size = 0; +int sort_active = 0; +int set_debug = 0; +int show_ops = 0; +int show_activity = 0; + +/* Debug options */ +int sanity = 0; +int redzone = 0; +int poison = 0; +int tracking = 0; +int tracing = 0; + +int page_size; + +regex_t pattern; + +static void fatal(const char *x, ...) +{ + va_list ap; + + va_start(ap, x); + vfprintf(stderr, x, ap); + va_end(ap); + exit(EXIT_FAILURE); +} + +static void usage(void) +{ + printf("slabinfo 5/7/2007. (c) 2007 sgi.\n\n" + "slabinfo [-ahnpvtsz] [-d debugopts] [slab-regexp]\n" + "-a|--aliases Show aliases\n" + "-A|--activity Most active slabs first\n" + "-d<options>|--debug=<options> Set/Clear Debug options\n" + "-D|--display-active Switch line format to activity\n" + "-e|--empty Show empty slabs\n" + "-f|--first-alias Show first alias\n" + "-h|--help Show usage information\n" + "-i|--inverted Inverted list\n" + "-l|--slabs Show slabs\n" + "-n|--numa Show NUMA information\n" + "-o|--ops Show kmem_cache_ops\n" + "-s|--shrink Shrink slabs\n" + "-r|--report Detailed report on single slabs\n" + "-S|--Size Sort by size\n" + "-t|--tracking Show alloc/free information\n" + "-T|--Totals Show summary information\n" + "-v|--validate Validate slabs\n" + "-z|--zero Include empty slabs\n" + "-1|--1ref Single reference\n" + "\nValid debug options (FZPUT may be combined)\n" + "a / A Switch on all debug options (=FZUP)\n" + "- Switch off all debug options\n" + "f / F Sanity Checks (SLAB_DEBUG_FREE)\n" + "z / Z Redzoning\n" + "p / P Poisoning\n" + "u / U Tracking\n" + "t / T Tracing\n" + ); +} + +static unsigned long read_obj(const char *name) +{ + FILE *f = fopen(name, "r"); + + if (!f) + buffer[0] = 0; + else { + if (!fgets(buffer, sizeof(buffer), f)) + buffer[0] = 0; + fclose(f); + if (buffer[strlen(buffer)] == '\n') + buffer[strlen(buffer)] = 0; + } + return strlen(buffer); +} + + +/* + * Get the contents of an attribute + */ +static unsigned long get_obj(const char *name) +{ + if (!read_obj(name)) + return 0; + + return atol(buffer); +} + +static unsigned long get_obj_and_str(const char *name, char **x) +{ + unsigned long result = 0; + char *p; + + *x = NULL; + + if (!read_obj(name)) { + x = NULL; + return 0; + } + result = strtoul(buffer, &p, 10); + while (*p == ' ') + p++; + if (*p) + *x = strdup(p); + return result; +} + +static void set_obj(struct slabinfo *s, const char *name, int n) +{ + char x[100]; + FILE *f; + + snprintf(x, 100, "%s/%s", s->name, name); + f = fopen(x, "w"); + if (!f) + fatal("Cannot write to %s\n", x); + + fprintf(f, "%d\n", n); + fclose(f); +} + +static unsigned long read_slab_obj(struct slabinfo *s, const char *name) +{ + char x[100]; + FILE *f; + size_t l; + + snprintf(x, 100, "%s/%s", s->name, name); + f = fopen(x, "r"); + if (!f) { + buffer[0] = 0; + l = 0; + } else { + l = fread(buffer, 1, sizeof(buffer), f); + buffer[l] = 0; + fclose(f); + } + return l; +} + + +/* + * Put a size string together + */ +static int store_size(char *buffer, unsigned long value) +{ + unsigned long divisor = 1; + char trailer = 0; + int n; + + if (value > 1000000000UL) { + divisor = 100000000UL; + trailer = 'G'; + } else if (value > 1000000UL) { + divisor = 100000UL; + trailer = 'M'; + } else if (value > 1000UL) { + divisor = 100; + trailer = 'K'; + } + + value /= divisor; + n = sprintf(buffer, "%ld",value); + if (trailer) { + buffer[n] = trailer; + n++; + buffer[n] = 0; + } + if (divisor != 1) { + memmove(buffer + n - 2, buffer + n - 3, 4); + buffer[n-2] = '.'; + n++; + } + return n; +} + +static void decode_numa_list(int *numa, char *t) +{ + int node; + int nr; + + memset(numa, 0, MAX_NODES * sizeof(int)); + + if (!t) + return; + + while (*t == 'N') { + t++; + node = strtoul(t, &t, 10); + if (*t == '=') { + t++; + nr = strtoul(t, &t, 10); + numa[node] = nr; + if (node > highest_node) + highest_node = node; + } + while (*t == ' ') + t++; + } +} + +static void slab_validate(struct slabinfo *s) +{ + if (strcmp(s->name, "*") == 0) + return; + + set_obj(s, "validate", 1); +} + +static void slab_shrink(struct slabinfo *s) +{ + if (strcmp(s->name, "*") == 0) + return; + + set_obj(s, "shrink", 1); +} + +int line = 0; + +static void first_line(void) +{ + if (show_activity) + printf("Name Objects Alloc Free %%Fast Fallb O\n"); + else + printf("Name Objects Objsize Space " + "Slabs/Part/Cpu O/S O %%Fr %%Ef Flg\n"); +} + +/* + * Find the shortest alias of a slab + */ +static struct aliasinfo *find_one_alias(struct slabinfo *find) +{ + struct aliasinfo *a; + struct aliasinfo *best = NULL; + + for(a = aliasinfo;a < aliasinfo + aliases; a++) { + if (a->slab == find && + (!best || strlen(best->name) < strlen(a->name))) { + best = a; + if (strncmp(a->name,"kmall", 5) == 0) + return best; + } + } + return best; +} + +static unsigned long slab_size(struct slabinfo *s) +{ + return s->slabs * (page_size << s->order); +} + +static unsigned long slab_activity(struct slabinfo *s) +{ + return s->alloc_fastpath + s->free_fastpath + + s->alloc_slowpath + s->free_slowpath; +} + +static void slab_numa(struct slabinfo *s, int mode) +{ + int node; + + if (strcmp(s->name, "*") == 0) + return; + + if (!highest_node) { + printf("\n%s: No NUMA information available.\n", s->name); + return; + } + + if (skip_zero && !s->slabs) + return; + + if (!line) { + printf("\n%-21s:", mode ? "NUMA nodes" : "Slab"); + for(node = 0; node <= highest_node; node++) + printf(" %4d", node); + printf("\n----------------------"); + for(node = 0; node <= highest_node; node++) + printf("-----"); + printf("\n"); + } + printf("%-21s ", mode ? "All slabs" : s->name); + for(node = 0; node <= highest_node; node++) { + char b[20]; + + store_size(b, s->numa[node]); + printf(" %4s", b); + } + printf("\n"); + if (mode) { + printf("%-21s ", "Partial slabs"); + for(node = 0; node <= highest_node; node++) { + char b[20]; + + store_size(b, s->numa_partial[node]); + printf(" %4s", b); + } + printf("\n"); + } + line++; +} + +static void show_tracking(struct slabinfo *s) +{ + printf("\n%s: Kernel object allocation\n", s->name); + printf("-----------------------------------------------------------------------\n"); + if (read_slab_obj(s, "alloc_calls")) + printf(buffer); + else + printf("No Data\n"); + + printf("\n%s: Kernel object freeing\n", s->name); + printf("------------------------------------------------------------------------\n"); + if (read_slab_obj(s, "free_calls")) + printf(buffer); + else + printf("No Data\n"); + +} + +static void ops(struct slabinfo *s) +{ + if (strcmp(s->name, "*") == 0) + return; + + if (read_slab_obj(s, "ops")) { + printf("\n%s: kmem_cache operations\n", s->name); + printf("--------------------------------------------\n"); + printf(buffer); + } else + printf("\n%s has no kmem_cache operations\n", s->name); +} + +static const char *onoff(int x) +{ + if (x) + return "On "; + return "Off"; +} + +static void slab_stats(struct slabinfo *s) +{ + unsigned long total_alloc; + unsigned long total_free; + unsigned long total; + + if (!s->alloc_slab) + return; + + total_alloc = s->alloc_fastpath + s->alloc_slowpath; + total_free = s->free_fastpath + s->free_slowpath; + + if (!total_alloc) + return; + + printf("\n"); + printf("Slab Perf Counter Alloc Free %%Al %%Fr\n"); + printf("--------------------------------------------------\n"); + printf("Fastpath %8lu %8lu %3lu %3lu\n", + s->alloc_fastpath, s->free_fastpath, + s->alloc_fastpath * 100 / total_alloc, + s->free_fastpath * 100 / total_free); + printf("Slowpath %8lu %8lu %3lu %3lu\n", + total_alloc - s->alloc_fastpath, s->free_slowpath, + (total_alloc - s->alloc_fastpath) * 100 / total_alloc, + s->free_slowpath * 100 / total_free); + printf("Page Alloc %8lu %8lu %3lu %3lu\n", + s->alloc_slab, s->free_slab, + s->alloc_slab * 100 / total_alloc, + s->free_slab * 100 / total_free); + printf("Add partial %8lu %8lu %3lu %3lu\n", + s->deactivate_to_head + s->deactivate_to_tail, + s->free_add_partial, + (s->deactivate_to_head + s->deactivate_to_tail) * 100 / total_alloc, + s->free_add_partial * 100 / total_free); + printf("Remove partial %8lu %8lu %3lu %3lu\n", + s->alloc_from_partial, s->free_remove_partial, + s->alloc_from_partial * 100 / total_alloc, + s->free_remove_partial * 100 / total_free); + + printf("RemoteObj/SlabFrozen %8lu %8lu %3lu %3lu\n", + s->deactivate_remote_frees, s->free_frozen, + s->deactivate_remote_frees * 100 / total_alloc, + s->free_frozen * 100 / total_free); + + printf("Total %8lu %8lu\n\n", total_alloc, total_free); + + if (s->cpuslab_flush) + printf("Flushes %8lu\n", s->cpuslab_flush); + + if (s->alloc_refill) + printf("Refill %8lu\n", s->alloc_refill); + + total = s->deactivate_full + s->deactivate_empty + + s->deactivate_to_head + s->deactivate_to_tail; + + if (total) + printf("Deactivate Full=%lu(%lu%%) Empty=%lu(%lu%%) " + "ToHead=%lu(%lu%%) ToTail=%lu(%lu%%)\n", + s->deactivate_full, (s->deactivate_full * 100) / total, + s->deactivate_empty, (s->deactivate_empty * 100) / total, + s->deactivate_to_head, (s->deactivate_to_head * 100) / total, + s->deactivate_to_tail, (s->deactivate_to_tail * 100) / total); +} + +static void report(struct slabinfo *s) +{ + if (strcmp(s->name, "*") == 0) + return; + + printf("\nSlabcache: %-20s Aliases: %2d Order : %2d Objects: %lu\n", + s->name, s->aliases, s->order, s->objects); + if (s->hwcache_align) + printf("** Hardware cacheline aligned\n"); + if (s->cache_dma) + printf("** Memory is allocated in a special DMA zone\n"); + if (s->destroy_by_rcu) + printf("** Slabs are destroyed via RCU\n"); + if (s->reclaim_account) + printf("** Reclaim accounting active\n"); + + printf("\nSizes (bytes) Slabs Debug Memory\n"); + printf("------------------------------------------------------------------------\n"); + printf("Object : %7d Total : %7ld Sanity Checks : %s Total: %7ld\n", + s->object_size, s->slabs, onoff(s->sanity_checks), + s->slabs * (page_size << s->order)); + printf("SlabObj: %7d Full : %7ld Redzoning : %s Used : %7ld\n", + s->slab_size, s->slabs - s->partial - s->cpu_slabs, + onoff(s->red_zone), s->objects * s->object_size); + printf("SlabSiz: %7d Partial: %7ld Poisoning : %s Loss : %7ld\n", + page_size << s->order, s->partial, onoff(s->poison), + s->slabs * (page_size << s->order) - s->objects * s->object_size); + printf("Loss : %7d CpuSlab: %7d Tracking : %s Lalig: %7ld\n", + s->slab_size - s->object_size, s->cpu_slabs, onoff(s->store_user), + (s->slab_size - s->object_size) * s->objects); + printf("Align : %7d Objects: %7d Tracing : %s Lpadd: %7ld\n", + s->align, s->objs_per_slab, onoff(s->trace), + ((page_size << s->order) - s->objs_per_slab * s->slab_size) * + s->slabs); + + ops(s); + show_tracking(s); + slab_numa(s, 1); + slab_stats(s); +} + +static void slabcache(struct slabinfo *s) +{ + char size_str[20]; + char dist_str[40]; + char flags[20]; + char *p = flags; + + if (strcmp(s->name, "*") == 0) + return; + + if (actual_slabs == 1) { + report(s); + return; + } + + if (skip_zero && !show_empty && !s->slabs) + return; + + if (show_empty && s->slabs) + return; + + store_size(size_str, slab_size(s)); + snprintf(dist_str, 40, "%lu/%lu/%d", s->slabs - s->cpu_slabs, + s->partial, s->cpu_slabs); + + if (!line++) + first_line(); + + if (s->aliases) + *p++ = '*'; + if (s->cache_dma) + *p++ = 'd'; + if (s->hwcache_align) + *p++ = 'A'; + if (s->poison) + *p++ = 'P'; + if (s->reclaim_account) + *p++ = 'a'; + if (s->red_zone) + *p++ = 'Z'; + if (s->sanity_checks) + *p++ = 'F'; + if (s->store_user) + *p++ = 'U'; + if (s->trace) + *p++ = 'T'; + + *p = 0; + if (show_activity) { + unsigned long total_alloc; + unsigned long total_free; + + total_alloc = s->alloc_fastpath + s->alloc_slowpath; + total_free = s->free_fastpath + s->free_slowpath; + + printf("%-21s %8ld %10ld %10ld %3ld %3ld %5ld %1d\n", + s->name, s->objects, + total_alloc, total_free, + total_alloc ? (s->alloc_fastpath * 100 / total_alloc) : 0, + total_free ? (s->free_fastpath * 100 / total_free) : 0, + s->order_fallback, s->order); + } + else + printf("%-21s %8ld %7d %8s %14s %4d %1d %3ld %3ld %s\n", + s->name, s->objects, s->object_size, size_str, dist_str, + s->objs_per_slab, s->order, + s->slabs ? (s->partial * 100) / s->slabs : 100, + s->slabs ? (s->objects * s->object_size * 100) / + (s->slabs * (page_size << s->order)) : 100, + flags); +} + +/* + * Analyze debug options. Return false if something is amiss. + */ +static int debug_opt_scan(char *opt) +{ + if (!opt || !opt[0] || strcmp(opt, "-") == 0) + return 1; + + if (strcasecmp(opt, "a") == 0) { + sanity = 1; + poison = 1; + redzone = 1; + tracking = 1; + return 1; + } + + for ( ; *opt; opt++) + switch (*opt) { + case 'F' : case 'f': + if (sanity) + return 0; + sanity = 1; + break; + case 'P' : case 'p': + if (poison) + return 0; + poison = 1; + break; + + case 'Z' : case 'z': + if (redzone) + return 0; + redzone = 1; + break; + + case 'U' : case 'u': + if (tracking) + return 0; + tracking = 1; + break; + + case 'T' : case 't': + if (tracing) + return 0; + tracing = 1; + break; + default: + return 0; + } + return 1; +} + +static int slab_empty(struct slabinfo *s) +{ + if (s->objects > 0) + return 0; + + /* + * We may still have slabs even if there are no objects. Shrinking will + * remove them. + */ + if (s->slabs != 0) + set_obj(s, "shrink", 1); + + return 1; +} + +static void slab_debug(struct slabinfo *s) +{ + if (strcmp(s->name, "*") == 0) + return; + + if (sanity && !s->sanity_checks) { + set_obj(s, "sanity", 1); + } + if (!sanity && s->sanity_checks) { + if (slab_empty(s)) + set_obj(s, "sanity", 0); + else + fprintf(stderr, "%s not empty cannot disable sanity checks\n", s->name); + } + if (redzone && !s->red_zone) { + if (slab_empty(s)) + set_obj(s, "red_zone", 1); + else + fprintf(stderr, "%s not empty cannot enable redzoning\n", s->name); + } + if (!redzone && s->red_zone) { + if (slab_empty(s)) + set_obj(s, "red_zone", 0); + else + fprintf(stderr, "%s not empty cannot disable redzoning\n", s->name); + } + if (poison && !s->poison) { + if (slab_empty(s)) + set_obj(s, "poison", 1); + else + fprintf(stderr, "%s not empty cannot enable poisoning\n", s->name); + } + if (!poison && s->poison) { + if (slab_empty(s)) + set_obj(s, "poison", 0); + else + fprintf(stderr, "%s not empty cannot disable poisoning\n", s->name); + } + if (tracking && !s->store_user) { + if (slab_empty(s)) + set_obj(s, "store_user", 1); + else + fprintf(stderr, "%s not empty cannot enable tracking\n", s->name); + } + if (!tracking && s->store_user) { + if (slab_empty(s)) + set_obj(s, "store_user", 0); + else + fprintf(stderr, "%s not empty cannot disable tracking\n", s->name); + } + if (tracing && !s->trace) { + if (slabs == 1) + set_obj(s, "trace", 1); + else + fprintf(stderr, "%s can only enable trace for one slab at a time\n", s->name); + } + if (!tracing && s->trace) + set_obj(s, "trace", 1); +} + +static void totals(void) +{ + struct slabinfo *s; + + int used_slabs = 0; + char b1[20], b2[20], b3[20], b4[20]; + unsigned long long max = 1ULL << 63; + + /* Object size */ + unsigned long long min_objsize = max, max_objsize = 0, avg_objsize; + + /* Number of partial slabs in a slabcache */ + unsigned long long min_partial = max, max_partial = 0, + avg_partial, total_partial = 0; + + /* Number of slabs in a slab cache */ + unsigned long long min_slabs = max, max_slabs = 0, + avg_slabs, total_slabs = 0; + + /* Size of the whole slab */ + unsigned long long min_size = max, max_size = 0, + avg_size, total_size = 0; + + /* Bytes used for object storage in a slab */ + unsigned long long min_used = max, max_used = 0, + avg_used, total_used = 0; + + /* Waste: Bytes used for alignment and padding */ + unsigned long long min_waste = max, max_waste = 0, + avg_waste, total_waste = 0; + /* Number of objects in a slab */ + unsigned long long min_objects = max, max_objects = 0, + avg_objects, total_objects = 0; + /* Waste per object */ + unsigned long long min_objwaste = max, + max_objwaste = 0, avg_objwaste, + total_objwaste = 0; + + /* Memory per object */ + unsigned long long min_memobj = max, + max_memobj = 0, avg_memobj, + total_objsize = 0; + + /* Percentage of partial slabs per slab */ + unsigned long min_ppart = 100, max_ppart = 0, + avg_ppart, total_ppart = 0; + + /* Number of objects in partial slabs */ + unsigned long min_partobj = max, max_partobj = 0, + avg_partobj, total_partobj = 0; + + /* Percentage of partial objects of all objects in a slab */ + unsigned long min_ppartobj = 100, max_ppartobj = 0, + avg_ppartobj, total_ppartobj = 0; + + + for (s = slabinfo; s < slabinfo + slabs; s++) { + unsigned long long size; + unsigned long used; + unsigned long long wasted; + unsigned long long objwaste; + unsigned long percentage_partial_slabs; + unsigned long percentage_partial_objs; + + if (!s->slabs || !s->objects) + continue; + + used_slabs++; + + size = slab_size(s); + used = s->objects * s->object_size; + wasted = size - used; + objwaste = s->slab_size - s->object_size; + + percentage_partial_slabs = s->partial * 100 / s->slabs; + if (percentage_partial_slabs > 100) + percentage_partial_slabs = 100; + + percentage_partial_objs = s->objects_partial * 100 + / s->objects; + + if (percentage_partial_objs > 100) + percentage_partial_objs = 100; + + if (s->object_size < min_objsize) + min_objsize = s->object_size; + if (s->partial < min_partial) + min_partial = s->partial; + if (s->slabs < min_slabs) + min_slabs = s->slabs; + if (size < min_size) + min_size = size; + if (wasted < min_waste) + min_waste = wasted; + if (objwaste < min_objwaste) + min_objwaste = objwaste; + if (s->objects < min_objects) + min_objects = s->objects; + if (used < min_used) + min_used = used; + if (s->objects_partial < min_partobj) + min_partobj = s->objects_partial; + if (percentage_partial_slabs < min_ppart) + min_ppart = percentage_partial_slabs; + if (percentage_partial_objs < min_ppartobj) + min_ppartobj = percentage_partial_objs; + if (s->slab_size < min_memobj) + min_memobj = s->slab_size; + + if (s->object_size > max_objsize) + max_objsize = s->object_size; + if (s->partial > max_partial) + max_partial = s->partial; + if (s->slabs > max_slabs) + max_slabs = s->slabs; + if (size > max_size) + max_size = size; + if (wasted > max_waste) + max_waste = wasted; + if (objwaste > max_objwaste) + max_objwaste = objwaste; + if (s->objects > max_objects) + max_objects = s->objects; + if (used > max_used) + max_used = used; + if (s->objects_partial > max_partobj) + max_partobj = s->objects_partial; + if (percentage_partial_slabs > max_ppart) + max_ppart = percentage_partial_slabs; + if (percentage_partial_objs > max_ppartobj) + max_ppartobj = percentage_partial_objs; + if (s->slab_size > max_memobj) + max_memobj = s->slab_size; + + total_partial += s->partial; + total_slabs += s->slabs; + total_size += size; + total_waste += wasted; + + total_objects += s->objects; + total_used += used; + total_partobj += s->objects_partial; + total_ppart += percentage_partial_slabs; + total_ppartobj += percentage_partial_objs; + + total_objwaste += s->objects * objwaste; + total_objsize += s->objects * s->slab_size; + } + + if (!total_objects) { + printf("No objects\n"); + return; + } + if (!used_slabs) { + printf("No slabs\n"); + return; + } + + /* Per slab averages */ + avg_partial = total_partial / used_slabs; + avg_slabs = total_slabs / used_slabs; + avg_size = total_size / used_slabs; + avg_waste = total_waste / used_slabs; + + avg_objects = total_objects / used_slabs; + avg_used = total_used / used_slabs; + avg_partobj = total_partobj / used_slabs; + avg_ppart = total_ppart / used_slabs; + avg_ppartobj = total_ppartobj / used_slabs; + + /* Per object object sizes */ + avg_objsize = total_used / total_objects; + avg_objwaste = total_objwaste / total_objects; + avg_partobj = total_partobj * 100 / total_objects; + avg_memobj = total_objsize / total_objects; + + printf("Slabcache Totals\n"); + printf("----------------\n"); + printf("Slabcaches : %3d Aliases : %3d->%-3d Active: %3d\n", + slabs, aliases, alias_targets, used_slabs); + + store_size(b1, total_size);store_size(b2, total_waste); + store_size(b3, total_waste * 100 / total_used); + printf("Memory used: %6s # Loss : %6s MRatio:%6s%%\n", b1, b2, b3); + + store_size(b1, total_objects);store_size(b2, total_partobj); + store_size(b3, total_partobj * 100 / total_objects); + printf("# Objects : %6s # PartObj: %6s ORatio:%6s%%\n", b1, b2, b3); + + printf("\n"); + printf("Per Cache Average Min Max Total\n"); + printf("---------------------------------------------------------\n"); + + store_size(b1, avg_objects);store_size(b2, min_objects); + store_size(b3, max_objects);store_size(b4, total_objects); + printf("#Objects %10s %10s %10s %10s\n", + b1, b2, b3, b4); + + store_size(b1, avg_slabs);store_size(b2, min_slabs); + store_size(b3, max_slabs);store_size(b4, total_slabs); + printf("#Slabs %10s %10s %10s %10s\n", + b1, b2, b3, b4); + + store_size(b1, avg_partial);store_size(b2, min_partial); + store_size(b3, max_partial);store_size(b4, total_partial); + printf("#PartSlab %10s %10s %10s %10s\n", + b1, b2, b3, b4); + store_size(b1, avg_ppart);store_size(b2, min_ppart); + store_size(b3, max_ppart); + store_size(b4, total_partial * 100 / total_slabs); + printf("%%PartSlab%10s%% %10s%% %10s%% %10s%%\n", + b1, b2, b3, b4); + + store_size(b1, avg_partobj);store_size(b2, min_partobj); + store_size(b3, max_partobj); + store_size(b4, total_partobj); + printf("PartObjs %10s %10s %10s %10s\n", + b1, b2, b3, b4); + + store_size(b1, avg_ppartobj);store_size(b2, min_ppartobj); + store_size(b3, max_ppartobj); + store_size(b4, total_partobj * 100 / total_objects); + printf("%% PartObj%10s%% %10s%% %10s%% %10s%%\n", + b1, b2, b3, b4); + + store_size(b1, avg_size);store_size(b2, min_size); + store_size(b3, max_size);store_size(b4, total_size); + printf("Memory %10s %10s %10s %10s\n", + b1, b2, b3, b4); + + store_size(b1, avg_used);store_size(b2, min_used); + store_size(b3, max_used);store_size(b4, total_used); + printf("Used %10s %10s %10s %10s\n", + b1, b2, b3, b4); + + store_size(b1, avg_waste);store_size(b2, min_waste); + store_size(b3, max_waste);store_size(b4, total_waste); + printf("Loss %10s %10s %10s %10s\n", + b1, b2, b3, b4); + + printf("\n"); + printf("Per Object Average Min Max\n"); + printf("---------------------------------------------\n"); + + store_size(b1, avg_memobj);store_size(b2, min_memobj); + store_size(b3, max_memobj); + printf("Memory %10s %10s %10s\n", + b1, b2, b3); + store_size(b1, avg_objsize);store_size(b2, min_objsize); + store_size(b3, max_objsize); + printf("User %10s %10s %10s\n", + b1, b2, b3); + + store_size(b1, avg_objwaste);store_size(b2, min_objwaste); + store_size(b3, max_objwaste); + printf("Loss %10s %10s %10s\n", + b1, b2, b3); +} + +static void sort_slabs(void) +{ + struct slabinfo *s1,*s2; + + for (s1 = slabinfo; s1 < slabinfo + slabs; s1++) { + for (s2 = s1 + 1; s2 < slabinfo + slabs; s2++) { + int result; + + if (sort_size) + result = slab_size(s1) < slab_size(s2); + else if (sort_active) + result = slab_activity(s1) < slab_activity(s2); + else + result = strcasecmp(s1->name, s2->name); + + if (show_inverted) + result = -result; + + if (result > 0) { + struct slabinfo t; + + memcpy(&t, s1, sizeof(struct slabinfo)); + memcpy(s1, s2, sizeof(struct slabinfo)); + memcpy(s2, &t, sizeof(struct slabinfo)); + } + } + } +} + +static void sort_aliases(void) +{ + struct aliasinfo *a1,*a2; + + for (a1 = aliasinfo; a1 < aliasinfo + aliases; a1++) { + for (a2 = a1 + 1; a2 < aliasinfo + aliases; a2++) { + char *n1, *n2; + + n1 = a1->name; + n2 = a2->name; + if (show_alias && !show_inverted) { + n1 = a1->ref; + n2 = a2->ref; + } + if (strcasecmp(n1, n2) > 0) { + struct aliasinfo t; + + memcpy(&t, a1, sizeof(struct aliasinfo)); + memcpy(a1, a2, sizeof(struct aliasinfo)); + memcpy(a2, &t, sizeof(struct aliasinfo)); + } + } + } +} + +static void link_slabs(void) +{ + struct aliasinfo *a; + struct slabinfo *s; + + for (a = aliasinfo; a < aliasinfo + aliases; a++) { + + for (s = slabinfo; s < slabinfo + slabs; s++) + if (strcmp(a->ref, s->name) == 0) { + a->slab = s; + s->refs++; + break; + } + if (s == slabinfo + slabs) + fatal("Unresolved alias %s\n", a->ref); + } +} + +static void alias(void) +{ + struct aliasinfo *a; + char *active = NULL; + + sort_aliases(); + link_slabs(); + + for(a = aliasinfo; a < aliasinfo + aliases; a++) { + + if (!show_single_ref && a->slab->refs == 1) + continue; + + if (!show_inverted) { + if (active) { + if (strcmp(a->slab->name, active) == 0) { + printf(" %s", a->name); + continue; + } + } + printf("\n%-12s <- %s", a->slab->name, a->name); + active = a->slab->name; + } + else + printf("%-20s -> %s\n", a->name, a->slab->name); + } + if (active) + printf("\n"); +} + + +static void rename_slabs(void) +{ + struct slabinfo *s; + struct aliasinfo *a; + + for (s = slabinfo; s < slabinfo + slabs; s++) { + if (*s->name != ':') + continue; + + if (s->refs > 1 && !show_first_alias) + continue; + + a = find_one_alias(s); + + if (a) + s->name = a->name; + else { + s->name = "*"; + actual_slabs--; + } + } +} + +static int slab_mismatch(char *slab) +{ + return regexec(&pattern, slab, 0, NULL, 0); +} + +static void read_slab_dir(void) +{ + DIR *dir; + struct dirent *de; + struct slabinfo *slab = slabinfo; + struct aliasinfo *alias = aliasinfo; + char *p; + char *t; + int count; + + if (chdir("/sys/kernel/slab") && chdir("/sys/slab")) + fatal("SYSFS support for SLUB not active\n"); + + dir = opendir("."); + while ((de = readdir(dir))) { + if (de->d_name[0] == '.' || + (de->d_name[0] != ':' && slab_mismatch(de->d_name))) + continue; + switch (de->d_type) { + case DT_LNK: + alias->name = strdup(de->d_name); + count = readlink(de->d_name, buffer, sizeof(buffer)); + + if (count < 0) + fatal("Cannot read symlink %s\n", de->d_name); + + buffer[count] = 0; + p = buffer + count; + while (p > buffer && p[-1] != '/') + p--; + alias->ref = strdup(p); + alias++; + break; + case DT_DIR: + if (chdir(de->d_name)) + fatal("Unable to access slab %s\n", slab->name); + slab->name = strdup(de->d_name); + slab->alias = 0; + slab->refs = 0; + slab->aliases = get_obj("aliases"); + slab->align = get_obj("align"); + slab->cache_dma = get_obj("cache_dma"); + slab->cpu_slabs = get_obj("cpu_slabs"); + slab->destroy_by_rcu = get_obj("destroy_by_rcu"); + slab->hwcache_align = get_obj("hwcache_align"); + slab->object_size = get_obj("object_size"); + slab->objects = get_obj("objects"); + slab->objects_partial = get_obj("objects_partial"); + slab->objects_total = get_obj("objects_total"); + slab->objs_per_slab = get_obj("objs_per_slab"); + slab->order = get_obj("order"); + slab->partial = get_obj("partial"); + slab->partial = get_obj_and_str("partial", &t); + decode_numa_list(slab->numa_partial, t); + free(t); + slab->poison = get_obj("poison"); + slab->reclaim_account = get_obj("reclaim_account"); + slab->red_zone = get_obj("red_zone"); + slab->sanity_checks = get_obj("sanity_checks"); + slab->slab_size = get_obj("slab_size"); + slab->slabs = get_obj_and_str("slabs", &t); + decode_numa_list(slab->numa, t); + free(t); + slab->store_user = get_obj("store_user"); + slab->trace = get_obj("trace"); + slab->alloc_fastpath = get_obj("alloc_fastpath"); + slab->alloc_slowpath = get_obj("alloc_slowpath"); + slab->free_fastpath = get_obj("free_fastpath"); + slab->free_slowpath = get_obj("free_slowpath"); + slab->free_frozen= get_obj("free_frozen"); + slab->free_add_partial = get_obj("free_add_partial"); + slab->free_remove_partial = get_obj("free_remove_partial"); + slab->alloc_from_partial = get_obj("alloc_from_partial"); + slab->alloc_slab = get_obj("alloc_slab"); + slab->alloc_refill = get_obj("alloc_refill"); + slab->free_slab = get_obj("free_slab"); + slab->cpuslab_flush = get_obj("cpuslab_flush"); + slab->deactivate_full = get_obj("deactivate_full"); + slab->deactivate_empty = get_obj("deactivate_empty"); + slab->deactivate_to_head = get_obj("deactivate_to_head"); + slab->deactivate_to_tail = get_obj("deactivate_to_tail"); + slab->deactivate_remote_frees = get_obj("deactivate_remote_frees"); + slab->order_fallback = get_obj("order_fallback"); + chdir(".."); + if (slab->name[0] == ':') + alias_targets++; + slab++; + break; + default : + fatal("Unknown file type %lx\n", de->d_type); + } + } + closedir(dir); + slabs = slab - slabinfo; + actual_slabs = slabs; + aliases = alias - aliasinfo; + if (slabs > MAX_SLABS) + fatal("Too many slabs\n"); + if (aliases > MAX_ALIASES) + fatal("Too many aliases\n"); +} + +static void output_slabs(void) +{ + struct slabinfo *slab; + + for (slab = slabinfo; slab < slabinfo + slabs; slab++) { + + if (slab->alias) + continue; + + + if (show_numa) + slab_numa(slab, 0); + else if (show_track) + show_tracking(slab); + else if (validate) + slab_validate(slab); + else if (shrink) + slab_shrink(slab); + else if (set_debug) + slab_debug(slab); + else if (show_ops) + ops(slab); + else if (show_slab) + slabcache(slab); + else if (show_report) + report(slab); + } +} + +struct option opts[] = { + { "aliases", 0, NULL, 'a' }, + { "activity", 0, NULL, 'A' }, + { "debug", 2, NULL, 'd' }, + { "display-activity", 0, NULL, 'D' }, + { "empty", 0, NULL, 'e' }, + { "first-alias", 0, NULL, 'f' }, + { "help", 0, NULL, 'h' }, + { "inverted", 0, NULL, 'i'}, + { "numa", 0, NULL, 'n' }, + { "ops", 0, NULL, 'o' }, + { "report", 0, NULL, 'r' }, + { "shrink", 0, NULL, 's' }, + { "slabs", 0, NULL, 'l' }, + { "track", 0, NULL, 't'}, + { "validate", 0, NULL, 'v' }, + { "zero", 0, NULL, 'z' }, + { "1ref", 0, NULL, '1'}, + { NULL, 0, NULL, 0 } +}; + +int main(int argc, char *argv[]) +{ + int c; + int err; + char *pattern_source; + + page_size = getpagesize(); + + while ((c = getopt_long(argc, argv, "aAd::Defhil1noprstvzTS", + opts, NULL)) != -1) + switch (c) { + case '1': + show_single_ref = 1; + break; + case 'a': + show_alias = 1; + break; + case 'A': + sort_active = 1; + break; + case 'd': + set_debug = 1; + if (!debug_opt_scan(optarg)) + fatal("Invalid debug option '%s'\n", optarg); + break; + case 'D': + show_activity = 1; + break; + case 'e': + show_empty = 1; + break; + case 'f': + show_first_alias = 1; + break; + case 'h': + usage(); + return 0; + case 'i': + show_inverted = 1; + break; + case 'n': + show_numa = 1; + break; + case 'o': + show_ops = 1; + break; + case 'r': + show_report = 1; + break; + case 's': + shrink = 1; + break; + case 'l': + show_slab = 1; + break; + case 't': + show_track = 1; + break; + case 'v': + validate = 1; + break; + case 'z': + skip_zero = 0; + break; + case 'T': + show_totals = 1; + break; + case 'S': + sort_size = 1; + break; + + default: + fatal("%s: Invalid option '%c'\n", argv[0], optopt); + + } + + if (!show_slab && !show_alias && !show_track && !show_report + && !validate && !shrink && !set_debug && !show_ops) + show_slab = 1; + + if (argc > optind) + pattern_source = argv[optind]; + else + pattern_source = ".*"; + + err = regcomp(&pattern, pattern_source, REG_ICASE|REG_NOSUB); + if (err) + fatal("%s: Invalid pattern '%s' code %d\n", + argv[0], pattern_source, err); + read_slab_dir(); + if (show_alias) + alias(); + else + if (show_totals) + totals(); + else { + link_slabs(); + rename_slabs(); + sort_slabs(); + output_slabs(); + } + return 0; +} diff --git a/tools/testing/ktest/compare-ktest-sample.pl b/tools/testing/ktest/compare-ktest-sample.pl new file mode 100755 index 00000000..9a571e71 --- /dev/null +++ b/tools/testing/ktest/compare-ktest-sample.pl @@ -0,0 +1,30 @@ +#!/usr/bin/perl + +open (IN,"ktest.pl"); +while (<IN>) { + if (/\$opt\{"?([A-Z].*?)(\[.*\])?"?\}/ || + /set_test_option\("(.*?)"/) { + $opt{$1} = 1; + } +} +close IN; + +open (IN, "sample.conf"); +while (<IN>) { + if (/^\s*#?\s*(\S+)\s*=/) { + $samp{$1} = 1; + } +} +close IN; + +foreach $opt (keys %opt) { + if (!defined($samp{$opt})) { + print "opt = $opt\n"; + } +} + +foreach $samp (keys %samp) { + if (!defined($opt{$samp})) { + print "samp = $samp\n"; + } +} diff --git a/tools/testing/ktest/ktest.pl b/tools/testing/ktest/ktest.pl new file mode 100755 index 00000000..cef28e66 --- /dev/null +++ b/tools/testing/ktest/ktest.pl @@ -0,0 +1,2264 @@ +#!/usr/bin/perl -w +# +# Copyright 2010 - Steven Rostedt <srostedt@redhat.com>, Red Hat Inc. +# Licensed under the terms of the GNU GPL License version 2 +# + +use strict; +use IPC::Open2; +use Fcntl qw(F_GETFL F_SETFL O_NONBLOCK); +use File::Path qw(mkpath); +use File::Copy qw(cp); +use FileHandle; + +my $VERSION = "0.2"; + +$| = 1; + +my %opt; +my %repeat_tests; +my %repeats; +my %default; + +#default opts +$default{"NUM_TESTS"} = 1; +$default{"REBOOT_TYPE"} = "grub"; +$default{"TEST_TYPE"} = "test"; +$default{"BUILD_TYPE"} = "randconfig"; +$default{"MAKE_CMD"} = "make"; +$default{"TIMEOUT"} = 120; +$default{"TMP_DIR"} = "/tmp/ktest"; +$default{"SLEEP_TIME"} = 60; # sleep time between tests +$default{"BUILD_NOCLEAN"} = 0; +$default{"REBOOT_ON_ERROR"} = 0; +$default{"POWEROFF_ON_ERROR"} = 0; +$default{"REBOOT_ON_SUCCESS"} = 1; +$default{"POWEROFF_ON_SUCCESS"} = 0; +$default{"BUILD_OPTIONS"} = ""; +$default{"BISECT_SLEEP_TIME"} = 60; # sleep time between bisects +$default{"PATCHCHECK_SLEEP_TIME"} = 60; # sleep time between patch checks +$default{"CLEAR_LOG"} = 0; +$default{"BISECT_MANUAL"} = 0; +$default{"BISECT_SKIP"} = 1; +$default{"SUCCESS_LINE"} = "login:"; +$default{"BOOTED_TIMEOUT"} = 1; +$default{"DIE_ON_FAILURE"} = 1; +$default{"SSH_EXEC"} = "ssh \$SSH_USER\@\$MACHINE \$SSH_COMMAND"; +$default{"SCP_TO_TARGET"} = "scp \$SRC_FILE \$SSH_USER\@\$MACHINE:\$DST_FILE"; +$default{"REBOOT"} = "ssh \$SSH_USER\@\$MACHINE reboot"; +$default{"STOP_AFTER_SUCCESS"} = 10; +$default{"STOP_AFTER_FAILURE"} = 60; +$default{"STOP_TEST_AFTER"} = 600; +$default{"LOCALVERSION"} = "-test"; + +my $ktest_config; +my $version; +my $machine; +my $ssh_user; +my $tmpdir; +my $builddir; +my $outputdir; +my $output_config; +my $test_type; +my $build_type; +my $build_options; +my $reboot_type; +my $reboot_script; +my $power_cycle; +my $reboot; +my $reboot_on_error; +my $poweroff_on_error; +my $die_on_failure; +my $powercycle_after_reboot; +my $poweroff_after_halt; +my $ssh_exec; +my $scp_to_target; +my $power_off; +my $grub_menu; +my $grub_number; +my $target; +my $make; +my $post_install; +my $noclean; +my $minconfig; +my $addconfig; +my $in_bisect = 0; +my $bisect_bad = ""; +my $reverse_bisect; +my $bisect_manual; +my $bisect_skip; +my $in_patchcheck = 0; +my $run_test; +my $redirect; +my $buildlog; +my $dmesg; +my $monitor_fp; +my $monitor_pid; +my $monitor_cnt = 0; +my $sleep_time; +my $bisect_sleep_time; +my $patchcheck_sleep_time; +my $store_failures; +my $timeout; +my $booted_timeout; +my $console; +my $success_line; +my $stop_after_success; +my $stop_after_failure; +my $stop_test_after; +my $build_target; +my $target_image; +my $localversion; +my $iteration = 0; +my $successes = 0; + +my %entered_configs; +my %config_help; +my %variable; + +$config_help{"MACHINE"} = << "EOF" + The machine hostname that you will test. +EOF + ; +$config_help{"SSH_USER"} = << "EOF" + The box is expected to have ssh on normal bootup, provide the user + (most likely root, since you need privileged operations) +EOF + ; +$config_help{"BUILD_DIR"} = << "EOF" + The directory that contains the Linux source code (full path). +EOF + ; +$config_help{"OUTPUT_DIR"} = << "EOF" + The directory that the objects will be built (full path). + (can not be same as BUILD_DIR) +EOF + ; +$config_help{"BUILD_TARGET"} = << "EOF" + The location of the compiled file to copy to the target. + (relative to OUTPUT_DIR) +EOF + ; +$config_help{"TARGET_IMAGE"} = << "EOF" + The place to put your image on the test machine. +EOF + ; +$config_help{"POWER_CYCLE"} = << "EOF" + A script or command to reboot the box. + + Here is a digital loggers power switch example + POWER_CYCLE = wget --no-proxy -O /dev/null -q --auth-no-challenge 'http://admin:admin\@power/outlet?5=CCL' + + Here is an example to reboot a virtual box on the current host + with the name "Guest". + POWER_CYCLE = virsh destroy Guest; sleep 5; virsh start Guest +EOF + ; +$config_help{"CONSOLE"} = << "EOF" + The script or command that reads the console + + If you use ttywatch server, something like the following would work. +CONSOLE = nc -d localhost 3001 + + For a virtual machine with guest name "Guest". +CONSOLE = virsh console Guest +EOF + ; +$config_help{"LOCALVERSION"} = << "EOF" + Required version ending to differentiate the test + from other linux builds on the system. +EOF + ; +$config_help{"REBOOT_TYPE"} = << "EOF" + Way to reboot the box to the test kernel. + Only valid options so far are "grub" and "script". + + If you specify grub, it will assume grub version 1 + and will search in /boot/grub/menu.lst for the title \$GRUB_MENU + and select that target to reboot to the kernel. If this is not + your setup, then specify "script" and have a command or script + specified in REBOOT_SCRIPT to boot to the target. + + The entry in /boot/grub/menu.lst must be entered in manually. + The test will not modify that file. +EOF + ; +$config_help{"GRUB_MENU"} = << "EOF" + The grub title name for the test kernel to boot + (Only mandatory if REBOOT_TYPE = grub) + + Note, ktest.pl will not update the grub menu.lst, you need to + manually add an option for the test. ktest.pl will search + the grub menu.lst for this option to find what kernel to + reboot into. + + For example, if in the /boot/grub/menu.lst the test kernel title has: + title Test Kernel + kernel vmlinuz-test + GRUB_MENU = Test Kernel +EOF + ; +$config_help{"REBOOT_SCRIPT"} = << "EOF" + A script to reboot the target into the test kernel + (Only mandatory if REBOOT_TYPE = script) +EOF + ; + + +sub get_ktest_config { + my ($config) = @_; + + return if (defined($opt{$config})); + + if (defined($config_help{$config})) { + print "\n"; + print $config_help{$config}; + } + + for (;;) { + print "$config = "; + if (defined($default{$config})) { + print "\[$default{$config}\] "; + } + $entered_configs{$config} = <STDIN>; + $entered_configs{$config} =~ s/^\s*(.*\S)\s*$/$1/; + if ($entered_configs{$config} =~ /^\s*$/) { + if ($default{$config}) { + $entered_configs{$config} = $default{$config}; + } else { + print "Your answer can not be blank\n"; + next; + } + } + last; + } +} + +sub get_ktest_configs { + get_ktest_config("MACHINE"); + get_ktest_config("SSH_USER"); + get_ktest_config("BUILD_DIR"); + get_ktest_config("OUTPUT_DIR"); + get_ktest_config("BUILD_TARGET"); + get_ktest_config("TARGET_IMAGE"); + get_ktest_config("POWER_CYCLE"); + get_ktest_config("CONSOLE"); + get_ktest_config("LOCALVERSION"); + + my $rtype = $opt{"REBOOT_TYPE"}; + + if (!defined($rtype)) { + if (!defined($opt{"GRUB_MENU"})) { + get_ktest_config("REBOOT_TYPE"); + $rtype = $entered_configs{"REBOOT_TYPE"}; + } else { + $rtype = "grub"; + } + } + + if ($rtype eq "grub") { + get_ktest_config("GRUB_MENU"); + } else { + get_ktest_config("REBOOT_SCRIPT"); + } +} + +sub process_variables { + my ($value) = @_; + my $retval = ""; + + # We want to check for '\', and it is just easier + # to check the previous characet of '$' and not need + # to worry if '$' is the first character. By adding + # a space to $value, we can just check [^\\]\$ and + # it will still work. + $value = " $value"; + + while ($value =~ /(.*?[^\\])\$\{(.*?)\}(.*)/) { + my $begin = $1; + my $var = $2; + my $end = $3; + # append beginning of value to retval + $retval = "$retval$begin"; + if (defined($variable{$var})) { + $retval = "$retval$variable{$var}"; + } else { + # put back the origin piece. + $retval = "$retval\$\{$var\}"; + } + $value = $end; + } + $retval = "$retval$value"; + + # remove the space added in the beginning + $retval =~ s/ //; + + return "$retval" +} + +sub set_value { + my ($lvalue, $rvalue) = @_; + + if (defined($opt{$lvalue})) { + die "Error: Option $lvalue defined more than once!\n"; + } + if ($rvalue =~ /^\s*$/) { + delete $opt{$lvalue}; + } else { + $rvalue = process_variables($rvalue); + $opt{$lvalue} = $rvalue; + } +} + +sub set_variable { + my ($lvalue, $rvalue) = @_; + + if ($rvalue =~ /^\s*$/) { + delete $variable{$lvalue}; + } else { + $rvalue = process_variables($rvalue); + $variable{$lvalue} = $rvalue; + } +} + +sub read_config { + my ($config) = @_; + + open(IN, $config) || die "can't read file $config"; + + my $name = $config; + $name =~ s,.*/(.*),$1,; + + my $test_num = 0; + my $default = 1; + my $repeat = 1; + my $num_tests_set = 0; + my $skip = 0; + my $rest; + + while (<IN>) { + + # ignore blank lines and comments + next if (/^\s*$/ || /\s*\#/); + + if (/^\s*TEST_START(.*)/) { + + $rest = $1; + + if ($num_tests_set) { + die "$name: $.: Can not specify both NUM_TESTS and TEST_START\n"; + } + + my $old_test_num = $test_num; + my $old_repeat = $repeat; + + $test_num += $repeat; + $default = 0; + $repeat = 1; + + if ($rest =~ /\s+SKIP(.*)/) { + $rest = $1; + $skip = 1; + } else { + $skip = 0; + } + + if ($rest =~ /\s+ITERATE\s+(\d+)(.*)$/) { + $repeat = $1; + $rest = $2; + $repeat_tests{"$test_num"} = $repeat; + } + + if ($rest =~ /\s+SKIP(.*)/) { + $rest = $1; + $skip = 1; + } + + if ($rest !~ /^\s*$/) { + die "$name: $.: Gargbage found after TEST_START\n$_"; + } + + if ($skip) { + $test_num = $old_test_num; + $repeat = $old_repeat; + } + + } elsif (/^\s*DEFAULTS(.*)$/) { + $default = 1; + + $rest = $1; + + if ($rest =~ /\s+SKIP(.*)/) { + $rest = $1; + $skip = 1; + } else { + $skip = 0; + } + + if ($rest !~ /^\s*$/) { + die "$name: $.: Gargbage found after DEFAULTS\n$_"; + } + + } elsif (/^\s*([A-Z_\[\]\d]+)\s*=\s*(.*?)\s*$/) { + + next if ($skip); + + my $lvalue = $1; + my $rvalue = $2; + + if (!$default && + ($lvalue eq "NUM_TESTS" || + $lvalue eq "LOG_FILE" || + $lvalue eq "CLEAR_LOG")) { + die "$name: $.: $lvalue must be set in DEFAULTS section\n"; + } + + if ($lvalue eq "NUM_TESTS") { + if ($test_num) { + die "$name: $.: Can not specify both NUM_TESTS and TEST_START\n"; + } + if (!$default) { + die "$name: $.: NUM_TESTS must be set in default section\n"; + } + $num_tests_set = 1; + } + + if ($default || $lvalue =~ /\[\d+\]$/) { + set_value($lvalue, $rvalue); + } else { + my $val = "$lvalue\[$test_num\]"; + set_value($val, $rvalue); + + if ($repeat > 1) { + $repeats{$val} = $repeat; + } + } + } elsif (/^\s*([A-Z_\[\]\d]+)\s*:=\s*(.*?)\s*$/) { + next if ($skip); + + my $lvalue = $1; + my $rvalue = $2; + + # process config variables. + # Config variables are only active while reading the + # config and can be defined anywhere. They also ignore + # TEST_START and DEFAULTS, but are skipped if they are in + # on of these sections that have SKIP defined. + # The save variable can be + # defined multiple times and the new one simply overrides + # the prevous one. + set_variable($lvalue, $rvalue); + + } else { + die "$name: $.: Garbage found in config\n$_"; + } + } + + close(IN); + + if ($test_num) { + $test_num += $repeat - 1; + $opt{"NUM_TESTS"} = $test_num; + } + + # make sure we have all mandatory configs + get_ktest_configs; + + # set any defaults + + foreach my $default (keys %default) { + if (!defined($opt{$default})) { + $opt{$default} = $default{$default}; + } + } +} + +sub _logit { + if (defined($opt{"LOG_FILE"})) { + open(OUT, ">> $opt{LOG_FILE}") or die "Can't write to $opt{LOG_FILE}"; + print OUT @_; + close(OUT); + } +} + +sub logit { + if (defined($opt{"LOG_FILE"})) { + _logit @_; + } else { + print @_; + } +} + +sub doprint { + print @_; + _logit @_; +} + +sub run_command; + +sub reboot { + # try to reboot normally + if (run_command $reboot) { + if (defined($powercycle_after_reboot)) { + sleep $powercycle_after_reboot; + run_command "$power_cycle"; + } + } else { + # nope? power cycle it. + run_command "$power_cycle"; + } +} + +sub do_not_reboot { + my $i = $iteration; + + return $test_type eq "build" || + ($test_type eq "patchcheck" && $opt{"PATCHCHECK_TYPE[$i]"} eq "build") || + ($test_type eq "bisect" && $opt{"BISECT_TYPE[$i]"} eq "build"); +} + +sub dodie { + doprint "CRITICAL FAILURE... ", @_, "\n"; + + my $i = $iteration; + + if ($reboot_on_error && !do_not_reboot) { + + doprint "REBOOTING\n"; + reboot; + + } elsif ($poweroff_on_error && defined($power_off)) { + doprint "POWERING OFF\n"; + `$power_off`; + } + + if (defined($opt{"LOG_FILE"})) { + print " See $opt{LOG_FILE} for more info.\n"; + } + + die @_, "\n"; +} + +sub open_console { + my ($fp) = @_; + + my $flags; + + my $pid = open($fp, "$console|") or + dodie "Can't open console $console"; + + $flags = fcntl($fp, F_GETFL, 0) or + dodie "Can't get flags for the socket: $!"; + $flags = fcntl($fp, F_SETFL, $flags | O_NONBLOCK) or + dodie "Can't set flags for the socket: $!"; + + return $pid; +} + +sub close_console { + my ($fp, $pid) = @_; + + doprint "kill child process $pid\n"; + kill 2, $pid; + + print "closing!\n"; + close($fp); +} + +sub start_monitor { + if ($monitor_cnt++) { + return; + } + $monitor_fp = \*MONFD; + $monitor_pid = open_console $monitor_fp; + + return; + + open(MONFD, "Stop perl from warning about single use of MONFD"); +} + +sub end_monitor { + if (--$monitor_cnt) { + return; + } + close_console($monitor_fp, $monitor_pid); +} + +sub wait_for_monitor { + my ($time) = @_; + my $line; + + doprint "** Wait for monitor to settle down **\n"; + + # read the monitor and wait for the system to calm down + do { + $line = wait_for_input($monitor_fp, $time); + print "$line" if (defined($line)); + } while (defined($line)); + print "** Monitor flushed **\n"; +} + +sub fail { + + if ($die_on_failure) { + dodie @_; + } + + doprint "FAILED\n"; + + my $i = $iteration; + + # no need to reboot for just building. + if (!do_not_reboot) { + doprint "REBOOTING\n"; + reboot; + start_monitor; + wait_for_monitor $sleep_time; + end_monitor; + } + + doprint "%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%\n"; + doprint "%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%\n"; + doprint "KTEST RESULT: TEST $i Failed: ", @_, "\n"; + doprint "%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%\n"; + doprint "%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%\n"; + + return 1 if (!defined($store_failures)); + + my @t = localtime; + my $date = sprintf "%04d%02d%02d%02d%02d%02d", + 1900+$t[5],$t[4],$t[3],$t[2],$t[1],$t[0]; + + my $type = $build_type; + if ($type =~ /useconfig/) { + $type = "useconfig"; + } + + my $dir = "$machine-$test_type-$type-fail-$date"; + my $faildir = "$store_failures/$dir"; + + if (!-d $faildir) { + mkpath($faildir) or + die "can't create $faildir"; + } + if (-f "$output_config") { + cp "$output_config", "$faildir/config" or + die "failed to copy .config"; + } + if (-f $buildlog) { + cp $buildlog, "$faildir/buildlog" or + die "failed to move $buildlog"; + } + if (-f $dmesg) { + cp $dmesg, "$faildir/dmesg" or + die "failed to move $dmesg"; + } + + doprint "*** Saved info to $faildir ***\n"; + + return 1; +} + +sub run_command { + my ($command) = @_; + my $dolog = 0; + my $dord = 0; + my $pid; + + $command =~ s/\$SSH_USER/$ssh_user/g; + $command =~ s/\$MACHINE/$machine/g; + + doprint("$command ... "); + + $pid = open(CMD, "$command 2>&1 |") or + (fail "unable to exec $command" and return 0); + + if (defined($opt{"LOG_FILE"})) { + open(LOG, ">>$opt{LOG_FILE}") or + dodie "failed to write to log"; + $dolog = 1; + } + + if (defined($redirect)) { + open (RD, ">$redirect") or + dodie "failed to write to redirect $redirect"; + $dord = 1; + } + + while (<CMD>) { + print LOG if ($dolog); + print RD if ($dord); + } + + waitpid($pid, 0); + my $failed = $?; + + close(CMD); + close(LOG) if ($dolog); + close(RD) if ($dord); + + if ($failed) { + doprint "FAILED!\n"; + } else { + doprint "SUCCESS\n"; + } + + return !$failed; +} + +sub run_ssh { + my ($cmd) = @_; + my $cp_exec = $ssh_exec; + + $cp_exec =~ s/\$SSH_COMMAND/$cmd/g; + return run_command "$cp_exec"; +} + +sub run_scp { + my ($src, $dst) = @_; + my $cp_scp = $scp_to_target; + + $cp_scp =~ s/\$SRC_FILE/$src/g; + $cp_scp =~ s/\$DST_FILE/$dst/g; + + return run_command "$cp_scp"; +} + +sub get_grub_index { + + if ($reboot_type ne "grub") { + return; + } + return if (defined($grub_number)); + + doprint "Find grub menu ... "; + $grub_number = -1; + + my $ssh_grub = $ssh_exec; + $ssh_grub =~ s,\$SSH_COMMAND,cat /boot/grub/menu.lst,g; + + open(IN, "$ssh_grub |") + or die "unable to get menu.lst"; + + while (<IN>) { + if (/^\s*title\s+$grub_menu\s*$/) { + $grub_number++; + last; + } elsif (/^\s*title\s/) { + $grub_number++; + } + } + close(IN); + + die "Could not find '$grub_menu' in /boot/grub/menu on $machine" + if ($grub_number < 0); + doprint "$grub_number\n"; +} + +sub wait_for_input +{ + my ($fp, $time) = @_; + my $rin; + my $ready; + my $line; + my $ch; + + if (!defined($time)) { + $time = $timeout; + } + + $rin = ''; + vec($rin, fileno($fp), 1) = 1; + $ready = select($rin, undef, undef, $time); + + $line = ""; + + # try to read one char at a time + while (sysread $fp, $ch, 1) { + $line .= $ch; + last if ($ch eq "\n"); + } + + if (!length($line)) { + return undef; + } + + return $line; +} + +sub reboot_to { + if ($reboot_type eq "grub") { + run_ssh "'(echo \"savedefault --default=$grub_number --once\" | grub --batch && reboot)'"; + return; + } + + run_command "$reboot_script"; +} + +sub get_sha1 { + my ($commit) = @_; + + doprint "git rev-list --max-count=1 $commit ... "; + my $sha1 = `git rev-list --max-count=1 $commit`; + my $ret = $?; + + logit $sha1; + + if ($ret) { + doprint "FAILED\n"; + dodie "Failed to get git $commit"; + } + + print "SUCCESS\n"; + + chomp $sha1; + + return $sha1; +} + +sub monitor { + my $booted = 0; + my $bug = 0; + my $skip_call_trace = 0; + my $loops; + + wait_for_monitor 5; + + my $line; + my $full_line = ""; + + open(DMESG, "> $dmesg") or + die "unable to write to $dmesg"; + + reboot_to; + + my $success_start; + my $failure_start; + my $monitor_start = time; + my $done = 0; + + while (!$done) { + + if ($booted) { + $line = wait_for_input($monitor_fp, $booted_timeout); + } else { + $line = wait_for_input($monitor_fp); + } + + last if (!defined($line)); + + doprint $line; + print DMESG $line; + + # we are not guaranteed to get a full line + $full_line .= $line; + + if ($full_line =~ /$success_line/) { + $booted = 1; + $success_start = time; + } + + if ($booted && defined($stop_after_success) && + $stop_after_success >= 0) { + my $now = time; + if ($now - $success_start >= $stop_after_success) { + doprint "Test forced to stop after $stop_after_success seconds after success\n"; + last; + } + } + + if ($full_line =~ /\[ backtrace testing \]/) { + $skip_call_trace = 1; + } + + if ($full_line =~ /call trace:/i) { + if (!$bug && !$skip_call_trace) { + $bug = 1; + $failure_start = time; + } + } + + if ($bug && defined($stop_after_failure) && + $stop_after_failure >= 0) { + my $now = time; + if ($now - $failure_start >= $stop_after_failure) { + doprint "Test forced to stop after $stop_after_failure seconds after failure\n"; + last; + } + } + + if ($full_line =~ /\[ end of backtrace testing \]/) { + $skip_call_trace = 0; + } + + if ($full_line =~ /Kernel panic -/) { + $failure_start = time; + $bug = 1; + } + + if ($line =~ /\n/) { + $full_line = ""; + } + + if ($stop_test_after > 0 && !$booted && !$bug) { + if (time - $monitor_start > $stop_test_after) { + doprint "STOP_TEST_AFTER ($stop_test_after seconds) timed out\n"; + $done = 1; + } + } + } + + close(DMESG); + + if ($bug) { + return 0 if ($in_bisect); + fail "failed - got a bug report" and return 0; + } + + if (!$booted) { + return 0 if ($in_bisect); + fail "failed - never got a boot prompt." and return 0; + } + + return 1; +} + +sub install { + + run_scp "$outputdir/$build_target", "$target_image" or + dodie "failed to copy image"; + + my $install_mods = 0; + + # should we process modules? + $install_mods = 0; + open(IN, "$output_config") or dodie("Can't read config file"); + while (<IN>) { + if (/CONFIG_MODULES(=y)?/) { + $install_mods = 1 if (defined($1)); + last; + } + } + close(IN); + + if (!$install_mods) { + doprint "No modules needed\n"; + return; + } + + run_command "$make INSTALL_MOD_PATH=$tmpdir modules_install" or + dodie "Failed to install modules"; + + my $modlib = "/lib/modules/$version"; + my $modtar = "ktest-mods.tar.bz2"; + + run_ssh "rm -rf $modlib" or + dodie "failed to remove old mods: $modlib"; + + # would be nice if scp -r did not follow symbolic links + run_command "cd $tmpdir && tar -cjf $modtar lib/modules/$version" or + dodie "making tarball"; + + run_scp "$tmpdir/$modtar", "/tmp" or + dodie "failed to copy modules"; + + unlink "$tmpdir/$modtar"; + + run_ssh "'(cd / && tar xf /tmp/$modtar)'" or + dodie "failed to tar modules"; + + run_ssh "rm -f /tmp/$modtar"; + + return if (!defined($post_install)); + + my $cp_post_install = $post_install; + $cp_post_install =~ s/\$KERNEL_VERSION/$version/g; + run_command "$cp_post_install" or + dodie "Failed to run post install"; +} + +sub check_buildlog { + my ($patch) = @_; + + my @files = `git show $patch | diffstat -l`; + + open(IN, "git show $patch |") or + dodie "failed to show $patch"; + while (<IN>) { + if (m,^--- a/(.*),) { + chomp $1; + $files[$#files] = $1; + } + } + close(IN); + + open(IN, $buildlog) or dodie "Can't open $buildlog"; + while (<IN>) { + if (/^\s*(.*?):.*(warning|error)/) { + my $err = $1; + foreach my $file (@files) { + my $fullpath = "$builddir/$file"; + if ($file eq $err || $fullpath eq $err) { + fail "$file built with warnings" and return 0; + } + } + } + } + close(IN); + + return 1; +} + +sub make_oldconfig { + my ($defconfig) = @_; + + if (!run_command "$defconfig $make oldnoconfig") { + # Perhaps oldnoconfig doesn't exist in this version of the kernel + # try a yes '' | oldconfig + doprint "oldnoconfig failed, trying yes '' | make oldconfig\n"; + run_command "yes '' | $defconfig $make oldconfig" or + dodie "failed make config oldconfig"; + } +} + +sub build { + my ($type) = @_; + my $defconfig = ""; + + unlink $buildlog; + + if ($type =~ /^useconfig:(.*)/) { + run_command "cp $1 $output_config" or + dodie "could not copy $1 to .config"; + + $type = "oldconfig"; + } + + # old config can ask questions + if ($type eq "oldconfig") { + $type = "oldnoconfig"; + + # allow for empty configs + run_command "touch $output_config"; + + run_command "mv $output_config $outputdir/config_temp" or + dodie "moving .config"; + + if (!$noclean && !run_command "$make mrproper") { + dodie "make mrproper"; + } + + run_command "mv $outputdir/config_temp $output_config" or + dodie "moving config_temp"; + + } elsif (!$noclean) { + unlink "$output_config"; + run_command "$make mrproper" or + dodie "make mrproper"; + } + + # add something to distinguish this build + open(OUT, "> $outputdir/localversion") or dodie("Can't make localversion file"); + print OUT "$localversion\n"; + close(OUT); + + if (defined($minconfig)) { + $defconfig = "KCONFIG_ALLCONFIG=$minconfig"; + } + + if ($type eq "oldnoconfig") { + make_oldconfig $defconfig; + } else { + run_command "$defconfig $make $type" or + dodie "failed make config"; + } + + $redirect = "$buildlog"; + if (!run_command "$make $build_options") { + undef $redirect; + # bisect may need this to pass + return 0 if ($in_bisect); + fail "failed build" and return 0; + } + undef $redirect; + + return 1; +} + +sub halt { + if (!run_ssh "halt" or defined($power_off)) { + if (defined($poweroff_after_halt)) { + sleep $poweroff_after_halt; + run_command "$power_off"; + } + } else { + # nope? the zap it! + run_command "$power_off"; + } +} + +sub success { + my ($i) = @_; + + $successes++; + + doprint "\n\n*******************************************\n"; + doprint "*******************************************\n"; + doprint "KTEST RESULT: TEST $i SUCCESS!!!! **\n"; + doprint "*******************************************\n"; + doprint "*******************************************\n"; + + if ($i != $opt{"NUM_TESTS"} && !do_not_reboot) { + doprint "Reboot and wait $sleep_time seconds\n"; + reboot; + start_monitor; + wait_for_monitor $sleep_time; + end_monitor; + } +} + +sub get_version { + # get the release name + doprint "$make kernelrelease ... "; + $version = `$make kernelrelease | tail -1`; + chomp($version); + doprint "$version\n"; +} + +sub answer_bisect { + for (;;) { + doprint "Pass or fail? [p/f]"; + my $ans = <STDIN>; + chomp $ans; + if ($ans eq "p" || $ans eq "P") { + return 1; + } elsif ($ans eq "f" || $ans eq "F") { + return 0; + } else { + print "Please answer 'P' or 'F'\n"; + } + } +} + +sub child_run_test { + my $failed = 0; + + # child should have no power + $reboot_on_error = 0; + $poweroff_on_error = 0; + $die_on_failure = 1; + + run_command $run_test or $failed = 1; + exit $failed; +} + +my $child_done; + +sub child_finished { + $child_done = 1; +} + +sub do_run_test { + my $child_pid; + my $child_exit; + my $line; + my $full_line; + my $bug = 0; + + wait_for_monitor 1; + + doprint "run test $run_test\n"; + + $child_done = 0; + + $SIG{CHLD} = qw(child_finished); + + $child_pid = fork; + + child_run_test if (!$child_pid); + + $full_line = ""; + + do { + $line = wait_for_input($monitor_fp, 1); + if (defined($line)) { + + # we are not guaranteed to get a full line + $full_line .= $line; + doprint $line; + + if ($full_line =~ /call trace:/i) { + $bug = 1; + } + + if ($full_line =~ /Kernel panic -/) { + $bug = 1; + } + + if ($line =~ /\n/) { + $full_line = ""; + } + } + } while (!$child_done && !$bug); + + if ($bug) { + my $failure_start = time; + my $now; + do { + $line = wait_for_input($monitor_fp, 1); + if (defined($line)) { + doprint $line; + } + $now = time; + if ($now - $failure_start >= $stop_after_failure) { + last; + } + } while (defined($line)); + + doprint "Detected kernel crash!\n"; + # kill the child with extreme prejudice + kill 9, $child_pid; + } + + waitpid $child_pid, 0; + $child_exit = $?; + + if ($bug || $child_exit) { + return 0 if $in_bisect; + fail "test failed" and return 0; + } + return 1; +} + +sub run_git_bisect { + my ($command) = @_; + + doprint "$command ... "; + + my $output = `$command 2>&1`; + my $ret = $?; + + logit $output; + + if ($ret) { + doprint "FAILED\n"; + dodie "Failed to git bisect"; + } + + doprint "SUCCESS\n"; + if ($output =~ m/^(Bisecting: .*\(roughly \d+ steps?\))\s+\[([[:xdigit:]]+)\]/) { + doprint "$1 [$2]\n"; + } elsif ($output =~ m/^([[:xdigit:]]+) is the first bad commit/) { + $bisect_bad = $1; + doprint "Found bad commit... $1\n"; + return 0; + } else { + # we already logged it, just print it now. + print $output; + } + + return 1; +} + +sub bisect_reboot { + doprint "Reboot and sleep $bisect_sleep_time seconds\n"; + reboot; + start_monitor; + wait_for_monitor $bisect_sleep_time; + end_monitor; +} + +# returns 1 on success, 0 on failure, -1 on skip +sub run_bisect_test { + my ($type, $buildtype) = @_; + + my $failed = 0; + my $result; + my $output; + my $ret; + + $in_bisect = 1; + + build $buildtype or $failed = 1; + + if ($type ne "build") { + if ($failed && $bisect_skip) { + $in_bisect = 0; + return -1; + } + dodie "Failed on build" if $failed; + + # Now boot the box + get_grub_index; + get_version; + install; + + start_monitor; + monitor or $failed = 1; + + if ($type ne "boot") { + if ($failed && $bisect_skip) { + end_monitor; + bisect_reboot; + $in_bisect = 0; + return -1; + } + dodie "Failed on boot" if $failed; + + do_run_test or $failed = 1; + } + end_monitor; + } + + if ($failed) { + $result = 0; + } else { + $result = 1; + } + + # reboot the box to a kernel we can ssh to + if ($type ne "build") { + bisect_reboot; + } + $in_bisect = 0; + + return $result; +} + +sub run_bisect { + my ($type) = @_; + my $buildtype = "oldconfig"; + + # We should have a minconfig to use? + if (defined($minconfig)) { + $buildtype = "useconfig:$minconfig"; + } + + my $ret = run_bisect_test $type, $buildtype; + + if ($bisect_manual) { + $ret = answer_bisect; + } + + # Are we looking for where it worked, not failed? + if ($reverse_bisect) { + $ret = !$ret; + } + + if ($ret > 0) { + return "good"; + } elsif ($ret == 0) { + return "bad"; + } elsif ($bisect_skip) { + doprint "HIT A BAD COMMIT ... SKIPPING\n"; + return "skip"; + } +} + +sub bisect { + my ($i) = @_; + + my $result; + + die "BISECT_GOOD[$i] not defined\n" if (!defined($opt{"BISECT_GOOD[$i]"})); + die "BISECT_BAD[$i] not defined\n" if (!defined($opt{"BISECT_BAD[$i]"})); + die "BISECT_TYPE[$i] not defined\n" if (!defined($opt{"BISECT_TYPE[$i]"})); + + my $good = $opt{"BISECT_GOOD[$i]"}; + my $bad = $opt{"BISECT_BAD[$i]"}; + my $type = $opt{"BISECT_TYPE[$i]"}; + my $start = $opt{"BISECT_START[$i]"}; + my $replay = $opt{"BISECT_REPLAY[$i]"}; + my $start_files = $opt{"BISECT_FILES[$i]"}; + + if (defined($start_files)) { + $start_files = " -- " . $start_files; + } else { + $start_files = ""; + } + + # convert to true sha1's + $good = get_sha1($good); + $bad = get_sha1($bad); + + if (defined($opt{"BISECT_REVERSE[$i]"}) && + $opt{"BISECT_REVERSE[$i]"} == 1) { + doprint "Performing a reverse bisect (bad is good, good is bad!)\n"; + $reverse_bisect = 1; + } else { + $reverse_bisect = 0; + } + + # Can't have a test without having a test to run + if ($type eq "test" && !defined($run_test)) { + $type = "boot"; + } + + my $check = $opt{"BISECT_CHECK[$i]"}; + if (defined($check) && $check ne "0") { + + # get current HEAD + my $head = get_sha1("HEAD"); + + if ($check ne "good") { + doprint "TESTING BISECT BAD [$bad]\n"; + run_command "git checkout $bad" or + die "Failed to checkout $bad"; + + $result = run_bisect $type; + + if ($result ne "bad") { + fail "Tested BISECT_BAD [$bad] and it succeeded" and return 0; + } + } + + if ($check ne "bad") { + doprint "TESTING BISECT GOOD [$good]\n"; + run_command "git checkout $good" or + die "Failed to checkout $good"; + + $result = run_bisect $type; + + if ($result ne "good") { + fail "Tested BISECT_GOOD [$good] and it failed" and return 0; + } + } + + # checkout where we started + run_command "git checkout $head" or + die "Failed to checkout $head"; + } + + run_command "git bisect start$start_files" or + dodie "could not start bisect"; + + run_command "git bisect good $good" or + dodie "could not set bisect good to $good"; + + run_git_bisect "git bisect bad $bad" or + dodie "could not set bisect bad to $bad"; + + if (defined($replay)) { + run_command "git bisect replay $replay" or + dodie "failed to run replay"; + } + + if (defined($start)) { + run_command "git checkout $start" or + dodie "failed to checkout $start"; + } + + my $test; + do { + $result = run_bisect $type; + $test = run_git_bisect "git bisect $result"; + } while ($test); + + run_command "git bisect log" or + dodie "could not capture git bisect log"; + + run_command "git bisect reset" or + dodie "could not reset git bisect"; + + doprint "Bad commit was [$bisect_bad]\n"; + + success $i; +} + +my %config_ignore; +my %config_set; + +my %config_list; +my %null_config; + +my %dependency; + +sub process_config_ignore { + my ($config) = @_; + + open (IN, $config) + or dodie "Failed to read $config"; + + while (<IN>) { + if (/^((CONFIG\S*)=.*)/) { + $config_ignore{$2} = $1; + } + } + + close(IN); +} + +sub read_current_config { + my ($config_ref) = @_; + + %{$config_ref} = (); + undef %{$config_ref}; + + my @key = keys %{$config_ref}; + if ($#key >= 0) { + print "did not delete!\n"; + exit; + } + open (IN, "$output_config"); + + while (<IN>) { + if (/^(CONFIG\S+)=(.*)/) { + ${$config_ref}{$1} = $2; + } + } + close(IN); +} + +sub get_dependencies { + my ($config) = @_; + + my $arr = $dependency{$config}; + if (!defined($arr)) { + return (); + } + + my @deps = @{$arr}; + + foreach my $dep (@{$arr}) { + print "ADD DEP $dep\n"; + @deps = (@deps, get_dependencies $dep); + } + + return @deps; +} + +sub create_config { + my @configs = @_; + + open(OUT, ">$output_config") or dodie "Can not write to $output_config"; + + foreach my $config (@configs) { + print OUT "$config_set{$config}\n"; + my @deps = get_dependencies $config; + foreach my $dep (@deps) { + print OUT "$config_set{$dep}\n"; + } + } + + foreach my $config (keys %config_ignore) { + print OUT "$config_ignore{$config}\n"; + } + close(OUT); + +# exit; + make_oldconfig ""; +} + +sub compare_configs { + my (%a, %b) = @_; + + foreach my $item (keys %a) { + if (!defined($b{$item})) { + print "diff $item\n"; + return 1; + } + delete $b{$item}; + } + + my @keys = keys %b; + if ($#keys) { + print "diff2 $keys[0]\n"; + } + return -1 if ($#keys >= 0); + + return 0; +} + +sub run_config_bisect_test { + my ($type) = @_; + + return run_bisect_test $type, "oldconfig"; +} + +sub process_passed { + my (%configs) = @_; + + doprint "These configs had no failure: (Enabling them for further compiles)\n"; + # Passed! All these configs are part of a good compile. + # Add them to the min options. + foreach my $config (keys %configs) { + if (defined($config_list{$config})) { + doprint " removing $config\n"; + $config_ignore{$config} = $config_list{$config}; + delete $config_list{$config}; + } + } + doprint "config copied to $outputdir/config_good\n"; + run_command "cp -f $output_config $outputdir/config_good"; +} + +sub process_failed { + my ($config) = @_; + + doprint "\n\n***************************************\n"; + doprint "Found bad config: $config\n"; + doprint "***************************************\n\n"; +} + +sub run_config_bisect { + + my @start_list = keys %config_list; + + if ($#start_list < 0) { + doprint "No more configs to test!!!\n"; + return -1; + } + + doprint "***** RUN TEST ***\n"; + my $type = $opt{"CONFIG_BISECT_TYPE[$iteration]"}; + my $ret; + my %current_config; + + my $count = $#start_list + 1; + doprint " $count configs to test\n"; + + my $half = int($#start_list / 2); + + do { + my @tophalf = @start_list[0 .. $half]; + + create_config @tophalf; + read_current_config \%current_config; + + $count = $#tophalf + 1; + doprint "Testing $count configs\n"; + my $found = 0; + # make sure we test something + foreach my $config (@tophalf) { + if (defined($current_config{$config})) { + logit " $config\n"; + $found = 1; + } + } + if (!$found) { + # try the other half + doprint "Top half produced no set configs, trying bottom half\n"; + @tophalf = @start_list[$half + 1 .. $#start_list]; + create_config @tophalf; + read_current_config \%current_config; + foreach my $config (@tophalf) { + if (defined($current_config{$config})) { + logit " $config\n"; + $found = 1; + } + } + if (!$found) { + doprint "Failed: Can't make new config with current configs\n"; + foreach my $config (@start_list) { + doprint " CONFIG: $config\n"; + } + return -1; + } + $count = $#tophalf + 1; + doprint "Testing $count configs\n"; + } + + $ret = run_config_bisect_test $type; + if ($bisect_manual) { + $ret = answer_bisect; + } + if ($ret) { + process_passed %current_config; + return 0; + } + + doprint "This config had a failure.\n"; + doprint "Removing these configs that were not set in this config:\n"; + doprint "config copied to $outputdir/config_bad\n"; + run_command "cp -f $output_config $outputdir/config_bad"; + + # A config exists in this group that was bad. + foreach my $config (keys %config_list) { + if (!defined($current_config{$config})) { + doprint " removing $config\n"; + delete $config_list{$config}; + } + } + + @start_list = @tophalf; + + if ($#start_list == 0) { + process_failed $start_list[0]; + return 1; + } + + # remove half the configs we are looking at and see if + # they are good. + $half = int($#start_list / 2); + } while ($#start_list > 0); + + # we found a single config, try it again unless we are running manually + + if ($bisect_manual) { + process_failed $start_list[0]; + return 1; + } + + my @tophalf = @start_list[0 .. 0]; + + $ret = run_config_bisect_test $type; + if ($ret) { + process_passed %current_config; + return 0; + } + + process_failed $start_list[0]; + return 1; +} + +sub config_bisect { + my ($i) = @_; + + my $start_config = $opt{"CONFIG_BISECT[$i]"}; + + my $tmpconfig = "$tmpdir/use_config"; + + # Make the file with the bad config and the min config + if (defined($minconfig)) { + # read the min config for things to ignore + run_command "cp $minconfig $tmpconfig" or + dodie "failed to copy $minconfig to $tmpconfig"; + } else { + unlink $tmpconfig; + } + + # Add other configs + if (defined($addconfig)) { + run_command "cat $addconfig >> $tmpconfig" or + dodie "failed to append $addconfig"; + } + + my $defconfig = ""; + if (-f $tmpconfig) { + $defconfig = "KCONFIG_ALLCONFIG=$tmpconfig"; + process_config_ignore $tmpconfig; + } + + # now process the start config + run_command "cp $start_config $output_config" or + dodie "failed to copy $start_config to $output_config"; + + # read directly what we want to check + my %config_check; + open (IN, $output_config) + or dodie "faied to open $output_config"; + + while (<IN>) { + if (/^((CONFIG\S*)=.*)/) { + $config_check{$2} = $1; + } + } + close(IN); + + # Now run oldconfig with the minconfig (and addconfigs) + make_oldconfig $defconfig; + + # check to see what we lost (or gained) + open (IN, $output_config) + or dodie "Failed to read $start_config"; + + my %removed_configs; + my %added_configs; + + while (<IN>) { + if (/^((CONFIG\S*)=.*)/) { + # save off all options + $config_set{$2} = $1; + if (defined($config_check{$2})) { + if (defined($config_ignore{$2})) { + $removed_configs{$2} = $1; + } else { + $config_list{$2} = $1; + } + } elsif (!defined($config_ignore{$2})) { + $added_configs{$2} = $1; + $config_list{$2} = $1; + } + } + } + close(IN); + + my @confs = keys %removed_configs; + if ($#confs >= 0) { + doprint "Configs overridden by default configs and removed from check:\n"; + foreach my $config (@confs) { + doprint " $config\n"; + } + } + @confs = keys %added_configs; + if ($#confs >= 0) { + doprint "Configs appearing in make oldconfig and added:\n"; + foreach my $config (@confs) { + doprint " $config\n"; + } + } + + my %config_test; + my $once = 0; + + # Sometimes kconfig does weird things. We must make sure + # that the config we autocreate has everything we need + # to test, otherwise we may miss testing configs, or + # may not be able to create a new config. + # Here we create a config with everything set. + create_config (keys %config_list); + read_current_config \%config_test; + foreach my $config (keys %config_list) { + if (!defined($config_test{$config})) { + if (!$once) { + $once = 1; + doprint "Configs not produced by kconfig (will not be checked):\n"; + } + doprint " $config\n"; + delete $config_list{$config}; + } + } + my $ret; + do { + $ret = run_config_bisect; + } while (!$ret); + + return $ret if ($ret < 0); + + success $i; +} + +sub patchcheck_reboot { + doprint "Reboot and sleep $patchcheck_sleep_time seconds\n"; + reboot; + start_monitor; + wait_for_monitor $patchcheck_sleep_time; + end_monitor; +} + +sub patchcheck { + my ($i) = @_; + + die "PATCHCHECK_START[$i] not defined\n" + if (!defined($opt{"PATCHCHECK_START[$i]"})); + die "PATCHCHECK_TYPE[$i] not defined\n" + if (!defined($opt{"PATCHCHECK_TYPE[$i]"})); + + my $start = $opt{"PATCHCHECK_START[$i]"}; + + my $end = "HEAD"; + if (defined($opt{"PATCHCHECK_END[$i]"})) { + $end = $opt{"PATCHCHECK_END[$i]"}; + } + + # Get the true sha1's since we can use things like HEAD~3 + $start = get_sha1($start); + $end = get_sha1($end); + + my $type = $opt{"PATCHCHECK_TYPE[$i]"}; + + # Can't have a test without having a test to run + if ($type eq "test" && !defined($run_test)) { + $type = "boot"; + } + + open (IN, "git log --pretty=oneline $end|") or + dodie "could not get git list"; + + my @list; + + while (<IN>) { + chomp; + $list[$#list+1] = $_; + last if (/^$start/); + } + close(IN); + + if ($list[$#list] !~ /^$start/) { + fail "SHA1 $start not found"; + } + + # go backwards in the list + @list = reverse @list; + + my $save_clean = $noclean; + + $in_patchcheck = 1; + foreach my $item (@list) { + my $sha1 = $item; + $sha1 =~ s/^([[:xdigit:]]+).*/$1/; + + doprint "\nProcessing commit $item\n\n"; + + run_command "git checkout $sha1" or + die "Failed to checkout $sha1"; + + # only clean on the first and last patch + if ($item eq $list[0] || + $item eq $list[$#list]) { + $noclean = $save_clean; + } else { + $noclean = 1; + } + + if (defined($minconfig)) { + build "useconfig:$minconfig" or return 0; + } else { + # ?? no config to use? + build "oldconfig" or return 0; + } + + check_buildlog $sha1 or return 0; + + next if ($type eq "build"); + + get_grub_index; + get_version; + install; + + my $failed = 0; + + start_monitor; + monitor or $failed = 1; + + if (!$failed && $type ne "boot"){ + do_run_test or $failed = 1; + } + end_monitor; + return 0 if ($failed); + + patchcheck_reboot; + + } + $in_patchcheck = 0; + success $i; + + return 1; +} + +$#ARGV < 1 or die "ktest.pl version: $VERSION\n usage: ktest.pl config-file\n"; + +if ($#ARGV == 0) { + $ktest_config = $ARGV[0]; + if (! -f $ktest_config) { + print "$ktest_config does not exist.\n"; + my $ans; + for (;;) { + print "Create it? [Y/n] "; + $ans = <STDIN>; + chomp $ans; + if ($ans =~ /^\s*$/) { + $ans = "y"; + } + last if ($ans =~ /^y$/i || $ans =~ /^n$/i); + print "Please answer either 'y' or 'n'.\n"; + } + if ($ans !~ /^y$/i) { + exit 0; + } + } +} else { + $ktest_config = "ktest.conf"; +} + +if (! -f $ktest_config) { + open(OUT, ">$ktest_config") or die "Can not create $ktest_config"; + print OUT << "EOF" +# Generated by ktest.pl +# +# Define each test with TEST_START +# The config options below it will override the defaults +TEST_START + +DEFAULTS +EOF +; + close(OUT); +} +read_config $ktest_config; + +# Append any configs entered in manually to the config file. +my @new_configs = keys %entered_configs; +if ($#new_configs >= 0) { + print "\nAppending entered in configs to $ktest_config\n"; + open(OUT, ">>$ktest_config") or die "Can not append to $ktest_config"; + foreach my $config (@new_configs) { + print OUT "$config = $entered_configs{$config}\n"; + $opt{$config} = $entered_configs{$config}; + } +} + +if ($opt{"CLEAR_LOG"} && defined($opt{"LOG_FILE"})) { + unlink $opt{"LOG_FILE"}; +} + +doprint "\n\nSTARTING AUTOMATED TESTS\n\n"; + +for (my $i = 0, my $repeat = 1; $i <= $opt{"NUM_TESTS"}; $i += $repeat) { + + if (!$i) { + doprint "DEFAULT OPTIONS:\n"; + } else { + doprint "\nTEST $i OPTIONS"; + if (defined($repeat_tests{$i})) { + $repeat = $repeat_tests{$i}; + doprint " ITERATE $repeat"; + } + doprint "\n"; + } + + foreach my $option (sort keys %opt) { + + if ($option =~ /\[(\d+)\]$/) { + next if ($i != $1); + } else { + next if ($i); + } + + doprint "$option = $opt{$option}\n"; + } +} + +sub __set_test_option { + my ($name, $i) = @_; + + my $option = "$name\[$i\]"; + + if (defined($opt{$option})) { + return $opt{$option}; + } + + foreach my $test (keys %repeat_tests) { + if ($i >= $test && + $i < $test + $repeat_tests{$test}) { + $option = "$name\[$test\]"; + if (defined($opt{$option})) { + return $opt{$option}; + } + } + } + + if (defined($opt{$name})) { + return $opt{$name}; + } + + return undef; +} + +sub eval_option { + my ($option, $i) = @_; + + # Add space to evaluate the character before $ + $option = " $option"; + my $retval = ""; + + while ($option =~ /(.*?[^\\])\$\{(.*?)\}(.*)/) { + my $start = $1; + my $var = $2; + my $end = $3; + + # Append beginning of line + $retval = "$retval$start"; + + # If the iteration option OPT[$i] exists, then use that. + # otherwise see if the default OPT (without [$i]) exists. + + my $o = "$var\[$i\]"; + + if (defined($opt{$o})) { + $o = $opt{$o}; + $retval = "$retval$o"; + } elsif (defined($opt{$var})) { + $o = $opt{$var}; + $retval = "$retval$o"; + } else { + $retval = "$retval\$\{$var\}"; + } + + $option = $end; + } + + $retval = "$retval$option"; + + $retval =~ s/^ //; + + return $retval; +} + +sub set_test_option { + my ($name, $i) = @_; + + my $option = __set_test_option($name, $i); + return $option if (!defined($option)); + + my $prev = ""; + + # Since an option can evaluate to another option, + # keep iterating until we do not evaluate any more + # options. + my $r = 0; + while ($prev ne $option) { + # Check for recursive evaluations. + # 100 deep should be more than enough. + if ($r++ > 100) { + die "Over 100 evaluations accurred with $name\n" . + "Check for recursive variables\n"; + } + $prev = $option; + $option = eval_option($option, $i); + } + + return $option; +} + +# First we need to do is the builds +for (my $i = 1; $i <= $opt{"NUM_TESTS"}; $i++) { + + $iteration = $i; + + my $makecmd = set_test_option("MAKE_CMD", $i); + + $machine = set_test_option("MACHINE", $i); + $ssh_user = set_test_option("SSH_USER", $i); + $tmpdir = set_test_option("TMP_DIR", $i); + $outputdir = set_test_option("OUTPUT_DIR", $i); + $builddir = set_test_option("BUILD_DIR", $i); + $test_type = set_test_option("TEST_TYPE", $i); + $build_type = set_test_option("BUILD_TYPE", $i); + $build_options = set_test_option("BUILD_OPTIONS", $i); + $power_cycle = set_test_option("POWER_CYCLE", $i); + $reboot = set_test_option("REBOOT", $i); + $noclean = set_test_option("BUILD_NOCLEAN", $i); + $minconfig = set_test_option("MIN_CONFIG", $i); + $run_test = set_test_option("TEST", $i); + $addconfig = set_test_option("ADD_CONFIG", $i); + $reboot_type = set_test_option("REBOOT_TYPE", $i); + $grub_menu = set_test_option("GRUB_MENU", $i); + $post_install = set_test_option("POST_INSTALL", $i); + $reboot_script = set_test_option("REBOOT_SCRIPT", $i); + $reboot_on_error = set_test_option("REBOOT_ON_ERROR", $i); + $poweroff_on_error = set_test_option("POWEROFF_ON_ERROR", $i); + $die_on_failure = set_test_option("DIE_ON_FAILURE", $i); + $power_off = set_test_option("POWER_OFF", $i); + $powercycle_after_reboot = set_test_option("POWERCYCLE_AFTER_REBOOT", $i); + $poweroff_after_halt = set_test_option("POWEROFF_AFTER_HALT", $i); + $sleep_time = set_test_option("SLEEP_TIME", $i); + $bisect_sleep_time = set_test_option("BISECT_SLEEP_TIME", $i); + $patchcheck_sleep_time = set_test_option("PATCHCHECK_SLEEP_TIME", $i); + $bisect_manual = set_test_option("BISECT_MANUAL", $i); + $bisect_skip = set_test_option("BISECT_SKIP", $i); + $store_failures = set_test_option("STORE_FAILURES", $i); + $timeout = set_test_option("TIMEOUT", $i); + $booted_timeout = set_test_option("BOOTED_TIMEOUT", $i); + $console = set_test_option("CONSOLE", $i); + $success_line = set_test_option("SUCCESS_LINE", $i); + $stop_after_success = set_test_option("STOP_AFTER_SUCCESS", $i); + $stop_after_failure = set_test_option("STOP_AFTER_FAILURE", $i); + $stop_test_after = set_test_option("STOP_TEST_AFTER", $i); + $build_target = set_test_option("BUILD_TARGET", $i); + $ssh_exec = set_test_option("SSH_EXEC", $i); + $scp_to_target = set_test_option("SCP_TO_TARGET", $i); + $target_image = set_test_option("TARGET_IMAGE", $i); + $localversion = set_test_option("LOCALVERSION", $i); + + chdir $builddir || die "can't change directory to $builddir"; + + if (!-d $tmpdir) { + mkpath($tmpdir) or + die "can't create $tmpdir"; + } + + $ENV{"SSH_USER"} = $ssh_user; + $ENV{"MACHINE"} = $machine; + + $target = "$ssh_user\@$machine"; + + $buildlog = "$tmpdir/buildlog-$machine"; + $dmesg = "$tmpdir/dmesg-$machine"; + $make = "$makecmd O=$outputdir"; + $output_config = "$outputdir/.config"; + + if ($reboot_type eq "grub") { + dodie "GRUB_MENU not defined" if (!defined($grub_menu)); + } elsif (!defined($reboot_script)) { + dodie "REBOOT_SCRIPT not defined" + } + + my $run_type = $build_type; + if ($test_type eq "patchcheck") { + $run_type = $opt{"PATCHCHECK_TYPE[$i]"}; + } elsif ($test_type eq "bisect") { + $run_type = $opt{"BISECT_TYPE[$i]"}; + } elsif ($test_type eq "config_bisect") { + $run_type = $opt{"CONFIG_BISECT_TYPE[$i]"}; + } + + # mistake in config file? + if (!defined($run_type)) { + $run_type = "ERROR"; + } + + doprint "\n\n"; + doprint "RUNNING TEST $i of $opt{NUM_TESTS} with option $test_type $run_type\n\n"; + + unlink $dmesg; + unlink $buildlog; + + if (!defined($minconfig)) { + $minconfig = $addconfig; + + } elsif (defined($addconfig)) { + run_command "cat $addconfig $minconfig > $tmpdir/add_config" or + dodie "Failed to create temp config"; + $minconfig = "$tmpdir/add_config"; + } + + my $checkout = $opt{"CHECKOUT[$i]"}; + if (defined($checkout)) { + run_command "git checkout $checkout" or + die "failed to checkout $checkout"; + } + + if ($test_type eq "bisect") { + bisect $i; + next; + } elsif ($test_type eq "config_bisect") { + config_bisect $i; + next; + } elsif ($test_type eq "patchcheck") { + patchcheck $i; + next; + } + + if ($build_type ne "nobuild") { + build $build_type or next; + } + + if ($test_type ne "build") { + get_grub_index; + get_version; + install; + + my $failed = 0; + start_monitor; + monitor or $failed = 1;; + + if (!$failed && $test_type ne "boot" && defined($run_test)) { + do_run_test or $failed = 1; + } + end_monitor; + next if ($failed); + } + + success $i; +} + +if ($opt{"POWEROFF_ON_SUCCESS"}) { + halt; +} elsif ($opt{"REBOOT_ON_SUCCESS"} && !do_not_reboot) { + reboot; +} + +doprint "\n $successes of $opt{NUM_TESTS} tests were successful\n\n"; + +exit 0; diff --git a/tools/testing/ktest/sample.conf b/tools/testing/ktest/sample.conf new file mode 100644 index 00000000..48cbcc80 --- /dev/null +++ b/tools/testing/ktest/sample.conf @@ -0,0 +1,757 @@ +# +# Config file for ktest.pl +# +# Note, all paths must be absolute +# + +# Options set in the beginning of the file are considered to be +# default options. These options can be overriden by test specific +# options, with the following exceptions: +# +# LOG_FILE +# CLEAR_LOG +# POWEROFF_ON_SUCCESS +# REBOOT_ON_SUCCESS +# +# Test specific options are set after the label: +# +# TEST_START +# +# The options after a TEST_START label are specific to that test. +# Each TEST_START label will set up a new test. If you want to +# perform a test more than once, you can add the ITERATE label +# to it followed by the number of times you want that test +# to iterate. If the ITERATE is left off, the test will only +# be performed once. +# +# TEST_START ITERATE 10 +# +# You can skip a test by adding SKIP (before or after the ITERATE +# and number) +# +# TEST_START SKIP +# +# TEST_START SKIP ITERATE 10 +# +# TEST_START ITERATE 10 SKIP +# +# The SKIP label causes the options and the test itself to be ignored. +# This is useful to set up several different tests in one config file, and +# only enabling the ones you want to use for a current test run. +# +# You can add default options anywhere in the file as well +# with the DEFAULTS tag. This allows you to have default options +# after the test options to keep the test options at the top +# of the file. You can even place the DEFAULTS tag between +# test cases (but not in the middle of a single test case) +# +# TEST_START +# MIN_CONFIG = /home/test/config-test1 +# +# DEFAULTS +# MIN_CONFIG = /home/test/config-default +# +# TEST_START ITERATE 10 +# +# The above will run the first test with MIN_CONFIG set to +# /home/test/config-test-1. Then 10 tests will be executed +# with MIN_CONFIG with /home/test/config-default. +# +# You can also disable defaults with the SKIP option +# +# DEFAULTS SKIP +# MIN_CONFIG = /home/test/config-use-sometimes +# +# DEFAULTS +# MIN_CONFIG = /home/test/config-most-times +# +# The above will ignore the first MIN_CONFIG. If you want to +# use the first MIN_CONFIG, remove the SKIP from the first +# DEFAULTS tag and add it to the second. Be careful, options +# may only be declared once per test or default. If you have +# the same option name under the same test or as default +# ktest will fail to execute, and no tests will run. +# + +#### Config variables #### +# +# This config file can also contain "config variables". +# These are assigned with ":=" instead of the ktest option +# assigment "=". +# +# The difference between ktest options and config variables +# is that config variables can be used multiple times, +# where each instance will override the previous instance. +# And that they only live at time of processing this config. +# +# The advantage to config variables are that they can be used +# by any option or any other config variables to define thing +# that you may use over and over again in the options. +# +# For example: +# +# USER := root +# TARGET := mybox +# TEST_CASE := ssh ${USER}@${TARGET} /path/to/my/test +# +# TEST_START +# MIN_CONFIG = config1 +# TEST = ${TEST_CASE} +# +# TEST_START +# MIN_CONFIG = config2 +# TEST = ${TEST_CASE} +# +# TEST_CASE := ssh ${USER}@${TARGET} /path/to/my/test2 +# +# TEST_START +# MIN_CONFIG = config1 +# TEST = ${TEST_CASE} +# +# TEST_START +# MIN_CONFIG = config2 +# TEST = ${TEST_CASE} +# +# TEST_DIR := /home/me/test +# +# BUILD_DIR = ${TEST_DIR}/linux.git +# OUTPUT_DIR = ${TEST_DIR}/test +# +# Note, the config variables are evaluated immediately, thus +# updating TARGET after TEST_CASE has been assigned does nothing +# to TEST_CASE. +# +# As shown in the example, to evaluate a config variable, you +# use the ${X} convention. Simple $X will not work. +# +# If the config variable does not exist, the ${X} will not +# be evaluated. Thus: +# +# MAKE_CMD = PATH=/mypath:${PATH} make +# +# If PATH is not a config variable, then the ${PATH} in +# the MAKE_CMD option will be evaluated by the shell when +# the MAKE_CMD option is passed into shell processing. + +#### Using options in other options #### +# +# Options that are defined in the config file may also be used +# by other options. All options are evaulated at time of +# use (except that config variables are evaluated at config +# processing time). +# +# If an ktest option is used within another option, instead of +# typing it again in that option you can simply use the option +# just like you can config variables. +# +# MACHINE = mybox +# +# TEST = ssh root@${MACHINE} /path/to/test +# +# The option will be used per test case. Thus: +# +# TEST_TYPE = test +# TEST = ssh root@{MACHINE} +# +# TEST_START +# MACHINE = box1 +# +# TEST_START +# MACHINE = box2 +# +# For both test cases, MACHINE will be evaluated at the time +# of the test case. The first test will run ssh root@box1 +# and the second will run ssh root@box2. + +#### Mandatory Default Options #### + +# These options must be in the default section, although most +# may be overridden by test options. + +# The machine hostname that you will test +#MACHINE = target + +# The box is expected to have ssh on normal bootup, provide the user +# (most likely root, since you need privileged operations) +#SSH_USER = root + +# The directory that contains the Linux source code +#BUILD_DIR = /home/test/linux.git + +# The directory that the objects will be built +# (can not be same as BUILD_DIR) +#OUTPUT_DIR = /home/test/build/target + +# The location of the compiled file to copy to the target +# (relative to OUTPUT_DIR) +#BUILD_TARGET = arch/x86/boot/bzImage + +# The place to put your image on the test machine +#TARGET_IMAGE = /boot/vmlinuz-test + +# A script or command to reboot the box +# +# Here is a digital loggers power switch example +#POWER_CYCLE = wget --no-proxy -O /dev/null -q --auth-no-challenge 'http://admin:admin@power/outlet?5=CCL' +# +# Here is an example to reboot a virtual box on the current host +# with the name "Guest". +#POWER_CYCLE = virsh destroy Guest; sleep 5; virsh start Guest + +# The script or command that reads the console +# +# If you use ttywatch server, something like the following would work. +#CONSOLE = nc -d localhost 3001 +# +# For a virtual machine with guest name "Guest". +#CONSOLE = virsh console Guest + +# Required version ending to differentiate the test +# from other linux builds on the system. +#LOCALVERSION = -test + +# The grub title name for the test kernel to boot +# (Only mandatory if REBOOT_TYPE = grub) +# +# Note, ktest.pl will not update the grub menu.lst, you need to +# manually add an option for the test. ktest.pl will search +# the grub menu.lst for this option to find what kernel to +# reboot into. +# +# For example, if in the /boot/grub/menu.lst the test kernel title has: +# title Test Kernel +# kernel vmlinuz-test +#GRUB_MENU = Test Kernel + +# A script to reboot the target into the test kernel +# (Only mandatory if REBOOT_TYPE = script) +#REBOOT_SCRIPT = + +#### Optional Config Options (all have defaults) #### + +# Start a test setup. If you leave this off, all options +# will be default and the test will run once. +# This is a label and not really an option (it takes no value). +# You can append ITERATE and a number after it to iterate the +# test a number of times, or SKIP to ignore this test. +# +#TEST_START +#TEST_START ITERATE 5 +#TEST_START SKIP + +# Have the following options as default again. Used after tests +# have already been defined by TEST_START. Optionally, you can +# just define all default options before the first TEST_START +# and you do not need this option. +# +# This is a label and not really an option (it takes no value). +# You can append SKIP to this label and the options within this +# section will be ignored. +# +# DEFAULTS +# DEFAULTS SKIP + +# The default test type (default test) +# The test types may be: +# build - only build the kernel, do nothing else +# boot - build and boot the kernel +# test - build, boot and if TEST is set, run the test script +# (If TEST is not set, it defaults back to boot) +# bisect - Perform a bisect on the kernel (see BISECT_TYPE below) +# patchcheck - Do a test on a series of commits in git (see PATCHCHECK below) +#TEST_TYPE = test + +# Test to run if there is a successful boot and TEST_TYPE is test. +# Must exit with 0 on success and non zero on error +# default (undefined) +#TEST = ssh user@machine /root/run_test + +# The build type is any make config type or special command +# (default randconfig) +# nobuild - skip the clean and build step +# useconfig:/path/to/config - use the given config and run +# oldconfig on it. +# This option is ignored if TEST_TYPE is patchcheck or bisect +#BUILD_TYPE = randconfig + +# The make command (default make) +# If you are building a 32bit x86 on a 64 bit host +#MAKE_CMD = CC=i386-gcc AS=i386-as make ARCH=i386 + +# Any build options for the make of the kernel (not for other makes, like configs) +# (default "") +#BUILD_OPTIONS = -j20 + +# If you need an initrd, you can add a script or code here to install +# it. The environment variable KERNEL_VERSION will be set to the +# kernel version that is used. Remember to add the initrd line +# to your grub menu.lst file. +# +# Here's a couple of examples to use: +#POST_INSTALL = ssh user@target /sbin/mkinitrd --allow-missing -f /boot/initramfs-test.img $KERNEL_VERSION +# +# or on some systems: +#POST_INSTALL = ssh user@target /sbin/dracut -f /boot/initramfs-test.img $KERNEL_VERSION + +# Way to reboot the box to the test kernel. +# Only valid options so far are "grub" and "script" +# (default grub) +# If you specify grub, it will assume grub version 1 +# and will search in /boot/grub/menu.lst for the title $GRUB_MENU +# and select that target to reboot to the kernel. If this is not +# your setup, then specify "script" and have a command or script +# specified in REBOOT_SCRIPT to boot to the target. +# +# The entry in /boot/grub/menu.lst must be entered in manually. +# The test will not modify that file. +#REBOOT_TYPE = grub + +# The min config that is needed to build for the machine +# A nice way to create this is with the following: +# +# $ ssh target +# $ lsmod > mymods +# $ scp mymods host:/tmp +# $ exit +# $ cd linux.git +# $ rm .config +# $ make LSMOD=mymods localyesconfig +# $ grep '^CONFIG' .config > /home/test/config-min +# +# If you want even less configs: +# +# log in directly to target (do not ssh) +# +# $ su +# # lsmod | cut -d' ' -f1 | xargs rmmod +# +# repeat the above several times +# +# # lsmod > mymods +# # reboot +# +# May need to reboot to get your network back to copy the mymods +# to the host, and then remove the previous .config and run the +# localyesconfig again. The CONFIG_MIN generated like this will +# not guarantee network activity to the box so the TEST_TYPE of +# test may fail. +# +# You might also want to set: +# CONFIG_CMDLINE="<your options here>" +# randconfig may set the above and override your real command +# line options. +# (default undefined) +#MIN_CONFIG = /home/test/config-min + +# Sometimes there's options that just break the boot and +# you do not care about. Here are a few: +# # CONFIG_STAGING is not set +# Staging drivers are horrible, and can break the build. +# # CONFIG_SCSI_DEBUG is not set +# SCSI_DEBUG may change your root partition +# # CONFIG_KGDB_SERIAL_CONSOLE is not set +# KGDB may cause oops waiting for a connection that's not there. +# This option points to the file containing config options that will be prepended +# to the MIN_CONFIG (or be the MIN_CONFIG if it is not set) +# +# Note, config options in MIN_CONFIG will override these options. +# +# (default undefined) +#ADD_CONFIG = /home/test/config-broken + +# The location on the host where to write temp files +# (default /tmp/ktest) +#TMP_DIR = /tmp/ktest + +# Optional log file to write the status (recommended) +# Note, this is a DEFAULT section only option. +# (default undefined) +#LOG_FILE = /home/test/logfiles/target.log + +# Remove old logfile if it exists before starting all tests. +# Note, this is a DEFAULT section only option. +# (default 0) +#CLEAR_LOG = 0 + +# Line to define a successful boot up in console output. +# This is what the line contains, not the entire line. If you need +# the entire line to match, then use regural expression syntax like: +# (do not add any quotes around it) +# +# SUCCESS_LINE = ^MyBox Login:$ +# +# (default "login:") +#SUCCESS_LINE = login: + +# In case the console constantly fills the screen, having +# a specified time to stop the test after success is recommended. +# (in seconds) +# (default 10) +#STOP_AFTER_SUCCESS = 10 + +# In case the console constantly fills the screen, having +# a specified time to stop the test after failure is recommended. +# (in seconds) +# (default 60) +#STOP_AFTER_FAILURE = 60 + +# In case the console constantly fills the screen, having +# a specified time to stop the test if it never succeeds nor fails +# is recommended. +# Note: this is ignored if a success or failure is detected. +# (in seconds) +# (default 600, -1 is to never stop) +#STOP_TEST_AFTER = 600 + +# Stop testing if a build fails. If set, the script will end if +# a failure is detected, otherwise it will save off the .config, +# dmesg and bootlog in a directory called +# MACHINE-TEST_TYPE_BUILD_TYPE-fail-yyyymmddhhmmss +# if the STORE_FAILURES directory is set. +# (default 1) +# Note, even if this is set to zero, there are some errors that still +# stop the tests. +#DIE_ON_FAILURE = 1 + +# Directory to store failure directories on failure. If this is not +# set, DIE_ON_FAILURE=0 will not save off the .config, dmesg and +# bootlog. This option is ignored if DIE_ON_FAILURE is not set. +# (default undefined) +#STORE_FAILURES = /home/test/failures + +# Build without doing a make mrproper, or removing .config +# (default 0) +#BUILD_NOCLEAN = 0 + +# As the test reads the console, after it hits the SUCCESS_LINE +# the time it waits for the monitor to settle down between reads +# can usually be lowered. +# (in seconds) (default 1) +#BOOTED_TIMEOUT = 1 + +# The timeout in seconds when we consider the box hung after +# the console stop producing output. Be sure to leave enough +# time here to get pass a reboot. Some machines may not produce +# any console output for a long time during a reboot. You do +# not want the test to fail just because the system was in +# the process of rebooting to the test kernel. +# (default 120) +#TIMEOUT = 120 + +# In between tests, a reboot of the box may occur, and this +# is the time to wait for the console after it stops producing +# output. Some machines may not produce a large lag on reboot +# so this should accommodate it. +# The difference between this and TIMEOUT, is that TIMEOUT happens +# when rebooting to the test kernel. This sleep time happens +# after a test has completed and we are about to start running +# another test. If a reboot to the reliable kernel happens, +# we wait SLEEP_TIME for the console to stop producing output +# before starting the next test. +# (default 60) +#SLEEP_TIME = 60 + +# The time in between bisects to sleep (in seconds) +# (default 60) +#BISECT_SLEEP_TIME = 60 + +# The time in between patch checks to sleep (in seconds) +# (default 60) +#PATCHCHECK_SLEEP_TIME = 60 + +# Reboot the target box on error (default 0) +#REBOOT_ON_ERROR = 0 + +# Power off the target on error (ignored if REBOOT_ON_ERROR is set) +# Note, this is a DEFAULT section only option. +# (default 0) +#POWEROFF_ON_ERROR = 0 + +# Power off the target after all tests have completed successfully +# Note, this is a DEFAULT section only option. +# (default 0) +#POWEROFF_ON_SUCCESS = 0 + +# Reboot the target after all test completed successfully (default 1) +# (ignored if POWEROFF_ON_SUCCESS is set) +#REBOOT_ON_SUCCESS = 1 + +# In case there are isses with rebooting, you can specify this +# to always powercycle after this amount of time after calling +# reboot. +# Note, POWERCYCLE_AFTER_REBOOT = 0 does NOT disable it. It just +# makes it powercycle immediately after rebooting. Do not define +# it if you do not want it. +# (default undefined) +#POWERCYCLE_AFTER_REBOOT = 5 + +# In case there's isses with halting, you can specify this +# to always poweroff after this amount of time after calling +# halt. +# Note, POWEROFF_AFTER_HALT = 0 does NOT disable it. It just +# makes it poweroff immediately after halting. Do not define +# it if you do not want it. +# (default undefined) +#POWEROFF_AFTER_HALT = 20 + +# A script or command to power off the box (default undefined) +# Needed for POWEROFF_ON_ERROR and SUCCESS +# +# Example for digital loggers power switch: +#POWER_OFF = wget --no-proxy -O /dev/null -q --auth-no-challenge 'http://admin:admin@power/outlet?5=OFF' +# +# Example for a virtual guest call "Guest". +#POWER_OFF = virsh destroy Guest + +# The way to execute a command on the target +# (default ssh $SSH_USER@$MACHINE $SSH_COMMAND";) +# The variables SSH_USER, MACHINE and SSH_COMMAND are defined +#SSH_EXEC = ssh $SSH_USER@$MACHINE $SSH_COMMAND"; + +# The way to copy a file to the target +# (default scp $SRC_FILE $SSH_USER@$MACHINE:$DST_FILE) +# The variables SSH_USER, MACHINE, SRC_FILE and DST_FILE are defined. +#SCP_TO_TARGET = scp $SRC_FILE $SSH_USER@$MACHINE:$DST_FILE + +# The nice way to reboot the target +# (default ssh $SSH_USER@$MACHINE reboot) +# The variables SSH_USER and MACHINE are defined. +#REBOOT = ssh $SSH_USER@$MACHINE reboot + +#### Per test run options #### +# The following options are only allowed in TEST_START sections. +# They are ignored in the DEFAULTS sections. +# +# All of these are optional and undefined by default, although +# some of these options are required for TEST_TYPE of patchcheck +# and bisect. +# +# +# CHECKOUT = branch +# +# If the BUILD_DIR is a git repository, then you can set this option +# to checkout the given branch before running the TEST. If you +# specify this for the first run, that branch will be used for +# all preceding tests until a new CHECKOUT is set. +# +# +# +# For TEST_TYPE = patchcheck +# +# This expects the BUILD_DIR to be a git repository, and +# will checkout the PATCHCHECK_START commit. +# +# The option BUILD_TYPE will be ignored. +# +# The MIN_CONFIG will be used for all builds of the patchcheck. The build type +# used for patchcheck is oldconfig. +# +# PATCHCHECK_START is required and is the first patch to +# test (the SHA1 of the commit). You may also specify anything +# that git checkout allows (branch name, tage, HEAD~3). +# +# PATCHCHECK_END is the last patch to check (default HEAD) +# +# PATCHCHECK_TYPE is required and is the type of test to run: +# build, boot, test. +# +# Note, the build test will look for warnings, if a warning occurred +# in a file that a commit touches, the build will fail. +# +# If BUILD_NOCLEAN is set, then make mrproper will not be run on +# any of the builds, just like all other TEST_TYPE tests. But +# what makes patchcheck different from the other tests, is if +# BUILD_NOCLEAN is not set, only the first and last patch run +# make mrproper. This helps speed up the test. +# +# Example: +# TEST_START +# TEST_TYPE = patchcheck +# CHECKOUT = mybranch +# PATCHCHECK_TYPE = boot +# PATCHCHECK_START = 747e94ae3d1b4c9bf5380e569f614eb9040b79e7 +# PATCHCHECK_END = HEAD~2 +# +# +# +# For TEST_TYPE = bisect +# +# You can specify a git bisect if the BUILD_DIR is a git repository. +# The MIN_CONFIG will be used for all builds of the bisect. The build type +# used for bisecting is oldconfig. +# +# The option BUILD_TYPE will be ignored. +# +# BISECT_TYPE is the type of test to perform: +# build - bad fails to build +# boot - bad builds but fails to boot +# test - bad boots but fails a test +# +# BISECT_GOOD is the commit (SHA1) to label as good (accepts all git good commit types) +# BISECT_BAD is the commit to label as bad (accepts all git bad commit types) +# +# The above three options are required for a bisect operation. +# +# BISECT_REPLAY = /path/to/replay/file (optional, default undefined) +# +# If an operation failed in the bisect that was not expected to +# fail. Then the test ends. The state of the BUILD_DIR will be +# left off at where the failure occurred. You can examine the +# reason for the failure, and perhaps even find a git commit +# that would work to continue with. You can run: +# +# git bisect log > /path/to/replay/file +# +# The adding: +# +# BISECT_REPLAY= /path/to/replay/file +# +# And running the test again. The test will perform the initial +# git bisect start, git bisect good, and git bisect bad, and +# then it will run git bisect replay on this file, before +# continuing with the bisect. +# +# BISECT_START = commit (optional, default undefined) +# +# As with BISECT_REPLAY, if the test failed on a commit that +# just happen to have a bad commit in the middle of the bisect, +# and you need to skip it. If BISECT_START is defined, it +# will checkout that commit after doing the initial git bisect start, +# git bisect good, git bisect bad, and running the git bisect replay +# if the BISECT_REPLAY is set. +# +# BISECT_SKIP = 1 (optional, default 0) +# +# If BISECT_TYPE is set to test but the build fails, ktest will +# simply fail the test and end their. You could use BISECT_REPLAY +# and BISECT_START to resume after you found a new starting point, +# or you could set BISECT_SKIP to 1. If BISECT_SKIP is set to 1, +# when something other than the BISECT_TYPE fails, ktest.pl will +# run "git bisect skip" and try again. +# +# BISECT_FILES = <path> (optional, default undefined) +# +# To just run the git bisect on a specific path, set BISECT_FILES. +# For example: +# +# BISECT_FILES = arch/x86 kernel/time +# +# Will run the bisect with "git bisect start -- arch/x86 kernel/time" +# +# BISECT_REVERSE = 1 (optional, default 0) +# +# In those strange instances where it was broken forever +# and you are trying to find where it started to work! +# Set BISECT_GOOD to the commit that was last known to fail +# Set BISECT_BAD to the commit that is known to start working. +# With BISECT_REVERSE = 1, The test will consider failures as +# good, and success as bad. +# +# BISECT_MANUAL = 1 (optional, default 0) +# +# In case there's a problem with automating the bisect for +# whatever reason. (Can't reboot, want to inspect each iteration) +# Doing a BISECT_MANUAL will have the test wait for you to +# tell it if the test passed or failed after each iteration. +# This is basicall the same as running git bisect yourself +# but ktest will rebuild and install the kernel for you. +# +# BISECT_CHECK = 1 (optional, default 0) +# +# Just to be sure the good is good and bad is bad, setting +# BISECT_CHECK to 1 will start the bisect by first checking +# out BISECT_BAD and makes sure it fails, then it will check +# out BISECT_GOOD and makes sure it succeeds before starting +# the bisect (it works for BISECT_REVERSE too). +# +# You can limit the test to just check BISECT_GOOD or +# BISECT_BAD with BISECT_CHECK = good or +# BISECT_CHECK = bad, respectively. +# +# Example: +# TEST_START +# TEST_TYPE = bisect +# BISECT_GOOD = v2.6.36 +# BISECT_BAD = b5153163ed580e00c67bdfecb02b2e3843817b3e +# BISECT_TYPE = build +# MIN_CONFIG = /home/test/config-bisect +# +# +# +# For TEST_TYPE = config_bisect +# +# In those cases that you have two different configs. One of them +# work, the other does not, and you do not know what config causes +# the problem. +# The TEST_TYPE config_bisect will bisect the bad config looking for +# what config causes the failure. +# +# The way it works is this: +# +# First it finds a config to work with. Since a different version, or +# MIN_CONFIG may cause different dependecies, it must run through this +# preparation. +# +# Overwrites any config set in the bad config with a config set in +# either the MIN_CONFIG or ADD_CONFIG. Thus, make sure these configs +# are minimal and do not disable configs you want to test: +# (ie. # CONFIG_FOO is not set). +# +# An oldconfig is run on the bad config and any new config that +# appears will be added to the configs to test. +# +# Finally, it generates a config with the above result and runs it +# again through make oldconfig to produce a config that should be +# satisfied by kconfig. +# +# Then it starts the bisect. +# +# The configs to test are cut in half. If all the configs in this +# half depend on a config in the other half, then the other half +# is tested instead. If no configs are enabled by either half, then +# this means a circular dependency exists and the test fails. +# +# A config is created with the test half, and the bisect test is run. +# +# If the bisect succeeds, then all configs in the generated config +# are removed from the configs to test and added to the configs that +# will be enabled for all builds (they will be enabled, but not be part +# of the configs to examine). +# +# If the bisect fails, then all test configs that were not enabled by +# the config file are removed from the test. These configs will not +# be enabled in future tests. Since current config failed, we consider +# this to be a subset of the config that we started with. +# +# When we are down to one config, it is considered the bad config. +# +# Note, the config chosen may not be the true bad config. Due to +# dependencies and selections of the kbuild system, mulitple +# configs may be needed to cause a failure. If you disable the +# config that was found and restart the test, if the test fails +# again, it is recommended to rerun the config_bisect with a new +# bad config without the found config enabled. +# +# The option BUILD_TYPE will be ignored. +# +# CONFIG_BISECT_TYPE is the type of test to perform: +# build - bad fails to build +# boot - bad builds but fails to boot +# test - bad boots but fails a test +# +# CONFIG_BISECT is the config that failed to boot +# +# If BISECT_MANUAL is set, it will pause between iterations. +# This is useful to use just ktest.pl just for the config bisect. +# If you set it to build, it will run the bisect and you can +# control what happens in between iterations. It will ask you if +# the test succeeded or not and continue the config bisect. +# +# Example: +# TEST_START +# TEST_TYPE = config_bisect +# CONFIG_BISECT_TYPE = build +# CONFIG_BISECT = /home/test/¢onfig-bad +# MIN_CONFIG = /home/test/config-min +# BISECT_MANUAL = 1 +# diff --git a/tools/usb/Makefile b/tools/usb/Makefile new file mode 100644 index 00000000..8b704af1 --- /dev/null +++ b/tools/usb/Makefile @@ -0,0 +1,13 @@ +# Makefile for USB tools + +CC = $(CROSS_COMPILE)gcc +PTHREAD_LIBS = -lpthread +WARNINGS = -Wall -Wextra +CFLAGS = $(WARNINGS) -g $(PTHREAD_LIBS) + +all: testusb ffs-test +%: %.c + $(CC) $(CFLAGS) -o $@ $^ + +clean: + $(RM) testusb ffs-test diff --git a/tools/usb/ffs-test.c b/tools/usb/ffs-test.c new file mode 100644 index 00000000..f17dfee1 --- /dev/null +++ b/tools/usb/ffs-test.c @@ -0,0 +1,554 @@ +/* + * ffs-test.c.c -- user mode filesystem api for usb composite function + * + * Copyright (C) 2010 Samsung Electronics + * Author: Michal Nazarewicz <m.nazarewicz@samsung.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +/* $(CROSS_COMPILE)cc -Wall -Wextra -g -o ffs-test ffs-test.c -lpthread */ + + +#define _BSD_SOURCE /* for endian.h */ + +#include <endian.h> +#include <errno.h> +#include <fcntl.h> +#include <pthread.h> +#include <stdarg.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <sys/ioctl.h> +#include <sys/stat.h> +#include <sys/types.h> +#include <unistd.h> + +#include "../../include/linux/usb/functionfs.h" + + +/******************** Little Endian Handling ********************************/ + +#define cpu_to_le16(x) htole16(x) +#define cpu_to_le32(x) htole32(x) +#define le32_to_cpu(x) le32toh(x) +#define le16_to_cpu(x) le16toh(x) + +static inline __u16 get_unaligned_le16(const void *_ptr) +{ + const __u8 *ptr = _ptr; + return ptr[0] | (ptr[1] << 8); +} + +static inline __u32 get_unaligned_le32(const void *_ptr) +{ + const __u8 *ptr = _ptr; + return ptr[0] | (ptr[1] << 8) | (ptr[2] << 16) | (ptr[3] << 24); +} + +static inline void put_unaligned_le16(__u16 val, void *_ptr) +{ + __u8 *ptr = _ptr; + *ptr++ = val; + *ptr++ = val >> 8; +} + +static inline void put_unaligned_le32(__u32 val, void *_ptr) +{ + __u8 *ptr = _ptr; + *ptr++ = val; + *ptr++ = val >> 8; + *ptr++ = val >> 16; + *ptr++ = val >> 24; +} + + +/******************** Messages and Errors ***********************************/ + +static const char argv0[] = "ffs-test"; + +static unsigned verbosity = 7; + +static void _msg(unsigned level, const char *fmt, ...) +{ + if (level < 2) + level = 2; + else if (level > 7) + level = 7; + + if (level <= verbosity) { + static const char levels[8][6] = { + [2] = "crit:", + [3] = "err: ", + [4] = "warn:", + [5] = "note:", + [6] = "info:", + [7] = "dbg: " + }; + + int _errno = errno; + va_list ap; + + fprintf(stderr, "%s: %s ", argv0, levels[level]); + va_start(ap, fmt); + vfprintf(stderr, fmt, ap); + va_end(ap); + + if (fmt[strlen(fmt) - 1] != '\n') { + char buffer[128]; + strerror_r(_errno, buffer, sizeof buffer); + fprintf(stderr, ": (-%d) %s\n", _errno, buffer); + } + + fflush(stderr); + } +} + +#define die(...) (_msg(2, __VA_ARGS__), exit(1)) +#define err(...) _msg(3, __VA_ARGS__) +#define warn(...) _msg(4, __VA_ARGS__) +#define note(...) _msg(5, __VA_ARGS__) +#define info(...) _msg(6, __VA_ARGS__) +#define debug(...) _msg(7, __VA_ARGS__) + +#define die_on(cond, ...) do { \ + if (cond) \ + die(__VA_ARGS__); \ + } while (0) + + +/******************** Descriptors and Strings *******************************/ + +static const struct { + struct usb_functionfs_descs_head header; + struct { + struct usb_interface_descriptor intf; + struct usb_endpoint_descriptor_no_audio sink; + struct usb_endpoint_descriptor_no_audio source; + } __attribute__((packed)) fs_descs, hs_descs; +} __attribute__((packed)) descriptors = { + .header = { + .magic = cpu_to_le32(FUNCTIONFS_DESCRIPTORS_MAGIC), + .length = cpu_to_le32(sizeof descriptors), + .fs_count = 3, + .hs_count = 3, + }, + .fs_descs = { + .intf = { + .bLength = sizeof descriptors.fs_descs.intf, + .bDescriptorType = USB_DT_INTERFACE, + .bNumEndpoints = 2, + .bInterfaceClass = USB_CLASS_VENDOR_SPEC, + .iInterface = 1, + }, + .sink = { + .bLength = sizeof descriptors.fs_descs.sink, + .bDescriptorType = USB_DT_ENDPOINT, + .bEndpointAddress = 1 | USB_DIR_IN, + .bmAttributes = USB_ENDPOINT_XFER_BULK, + /* .wMaxPacketSize = autoconfiguration (kernel) */ + }, + .source = { + .bLength = sizeof descriptors.fs_descs.source, + .bDescriptorType = USB_DT_ENDPOINT, + .bEndpointAddress = 2 | USB_DIR_OUT, + .bmAttributes = USB_ENDPOINT_XFER_BULK, + /* .wMaxPacketSize = autoconfiguration (kernel) */ + }, + }, + .hs_descs = { + .intf = { + .bLength = sizeof descriptors.fs_descs.intf, + .bDescriptorType = USB_DT_INTERFACE, + .bNumEndpoints = 2, + .bInterfaceClass = USB_CLASS_VENDOR_SPEC, + .iInterface = 1, + }, + .sink = { + .bLength = sizeof descriptors.hs_descs.sink, + .bDescriptorType = USB_DT_ENDPOINT, + .bEndpointAddress = 1 | USB_DIR_IN, + .bmAttributes = USB_ENDPOINT_XFER_BULK, + .wMaxPacketSize = cpu_to_le16(512), + }, + .source = { + .bLength = sizeof descriptors.hs_descs.source, + .bDescriptorType = USB_DT_ENDPOINT, + .bEndpointAddress = 2 | USB_DIR_OUT, + .bmAttributes = USB_ENDPOINT_XFER_BULK, + .wMaxPacketSize = cpu_to_le16(512), + .bInterval = 1, /* NAK every 1 uframe */ + }, + }, +}; + + +#define STR_INTERFACE_ "Source/Sink" + +static const struct { + struct usb_functionfs_strings_head header; + struct { + __le16 code; + const char str1[sizeof STR_INTERFACE_]; + } __attribute__((packed)) lang0; +} __attribute__((packed)) strings = { + .header = { + .magic = cpu_to_le32(FUNCTIONFS_STRINGS_MAGIC), + .length = cpu_to_le32(sizeof strings), + .str_count = cpu_to_le32(1), + .lang_count = cpu_to_le32(1), + }, + .lang0 = { + cpu_to_le16(0x0409), /* en-us */ + STR_INTERFACE_, + }, +}; + +#define STR_INTERFACE strings.lang0.str1 + + +/******************** Files and Threads Handling ****************************/ + +struct thread; + +static ssize_t read_wrap(struct thread *t, void *buf, size_t nbytes); +static ssize_t write_wrap(struct thread *t, const void *buf, size_t nbytes); +static ssize_t ep0_consume(struct thread *t, const void *buf, size_t nbytes); +static ssize_t fill_in_buf(struct thread *t, void *buf, size_t nbytes); +static ssize_t empty_out_buf(struct thread *t, const void *buf, size_t nbytes); + + +static struct thread { + const char *const filename; + size_t buf_size; + + ssize_t (*in)(struct thread *, void *, size_t); + const char *const in_name; + + ssize_t (*out)(struct thread *, const void *, size_t); + const char *const out_name; + + int fd; + pthread_t id; + void *buf; + ssize_t status; +} threads[] = { + { + "ep0", 4 * sizeof(struct usb_functionfs_event), + read_wrap, NULL, + ep0_consume, "<consume>", + 0, 0, NULL, 0 + }, + { + "ep1", 8 * 1024, + fill_in_buf, "<in>", + write_wrap, NULL, + 0, 0, NULL, 0 + }, + { + "ep2", 8 * 1024, + read_wrap, NULL, + empty_out_buf, "<out>", + 0, 0, NULL, 0 + }, +}; + + +static void init_thread(struct thread *t) +{ + t->buf = malloc(t->buf_size); + die_on(!t->buf, "malloc"); + + t->fd = open(t->filename, O_RDWR); + die_on(t->fd < 0, "%s", t->filename); +} + +static void cleanup_thread(void *arg) +{ + struct thread *t = arg; + int ret, fd; + + fd = t->fd; + if (t->fd < 0) + return; + t->fd = -1; + + /* test the FIFO ioctls (non-ep0 code paths) */ + if (t != threads) { + ret = ioctl(fd, FUNCTIONFS_FIFO_STATUS); + if (ret < 0) { + /* ENODEV reported after disconnect */ + if (errno != ENODEV) + err("%s: get fifo status", t->filename); + } else if (ret) { + warn("%s: unclaimed = %d\n", t->filename, ret); + if (ioctl(fd, FUNCTIONFS_FIFO_FLUSH) < 0) + err("%s: fifo flush", t->filename); + } + } + + if (close(fd) < 0) + err("%s: close", t->filename); + + free(t->buf); + t->buf = NULL; +} + +static void *start_thread_helper(void *arg) +{ + const char *name, *op, *in_name, *out_name; + struct thread *t = arg; + ssize_t ret; + + info("%s: starts\n", t->filename); + in_name = t->in_name ? t->in_name : t->filename; + out_name = t->out_name ? t->out_name : t->filename; + + pthread_cleanup_push(cleanup_thread, arg); + + for (;;) { + pthread_testcancel(); + + ret = t->in(t, t->buf, t->buf_size); + if (ret > 0) { + ret = t->out(t, t->buf, ret); + name = out_name; + op = "write"; + } else { + name = in_name; + op = "read"; + } + + if (ret > 0) { + /* nop */ + } else if (!ret) { + debug("%s: %s: EOF", name, op); + break; + } else if (errno == EINTR || errno == EAGAIN) { + debug("%s: %s", name, op); + } else { + warn("%s: %s", name, op); + break; + } + } + + pthread_cleanup_pop(1); + + t->status = ret; + info("%s: ends\n", t->filename); + return NULL; +} + +static void start_thread(struct thread *t) +{ + debug("%s: starting\n", t->filename); + + die_on(pthread_create(&t->id, NULL, start_thread_helper, t) < 0, + "pthread_create(%s)", t->filename); +} + +static void join_thread(struct thread *t) +{ + int ret = pthread_join(t->id, NULL); + + if (ret < 0) + err("%s: joining thread", t->filename); + else + debug("%s: joined\n", t->filename); +} + + +static ssize_t read_wrap(struct thread *t, void *buf, size_t nbytes) +{ + return read(t->fd, buf, nbytes); +} + +static ssize_t write_wrap(struct thread *t, const void *buf, size_t nbytes) +{ + return write(t->fd, buf, nbytes); +} + + +/******************** Empty/Fill buffer routines ****************************/ + +/* 0 -- stream of zeros, 1 -- i % 63, 2 -- pipe */ +enum pattern { PAT_ZERO, PAT_SEQ, PAT_PIPE }; +static enum pattern pattern; + +static ssize_t +fill_in_buf(struct thread *ignore, void *buf, size_t nbytes) +{ + size_t i; + __u8 *p; + + (void)ignore; + + switch (pattern) { + case PAT_ZERO: + memset(buf, 0, nbytes); + break; + + case PAT_SEQ: + for (p = buf, i = 0; i < nbytes; ++i, ++p) + *p = i % 63; + break; + + case PAT_PIPE: + return fread(buf, 1, nbytes, stdin); + } + + return nbytes; +} + +static ssize_t +empty_out_buf(struct thread *ignore, const void *buf, size_t nbytes) +{ + const __u8 *p; + __u8 expected; + ssize_t ret; + size_t len; + + (void)ignore; + + switch (pattern) { + case PAT_ZERO: + expected = 0; + for (p = buf, len = 0; len < nbytes; ++p, ++len) + if (*p) + goto invalid; + break; + + case PAT_SEQ: + for (p = buf, len = 0; len < nbytes; ++p, ++len) + if (*p != len % 63) { + expected = len % 63; + goto invalid; + } + break; + + case PAT_PIPE: + ret = fwrite(buf, nbytes, 1, stdout); + if (ret > 0) + fflush(stdout); + break; + +invalid: + err("bad OUT byte %zd, expected %02x got %02x\n", + len, expected, *p); + for (p = buf, len = 0; len < nbytes; ++p, ++len) { + if (0 == (len % 32)) + fprintf(stderr, "%4zd:", len); + fprintf(stderr, " %02x", *p); + if (31 == (len % 32)) + fprintf(stderr, "\n"); + } + fflush(stderr); + errno = EILSEQ; + return -1; + } + + return len; +} + + +/******************** Endpoints routines ************************************/ + +static void handle_setup(const struct usb_ctrlrequest *setup) +{ + printf("bRequestType = %d\n", setup->bRequestType); + printf("bRequest = %d\n", setup->bRequest); + printf("wValue = %d\n", le16_to_cpu(setup->wValue)); + printf("wIndex = %d\n", le16_to_cpu(setup->wIndex)); + printf("wLength = %d\n", le16_to_cpu(setup->wLength)); +} + +static ssize_t +ep0_consume(struct thread *ignore, const void *buf, size_t nbytes) +{ + static const char *const names[] = { + [FUNCTIONFS_BIND] = "BIND", + [FUNCTIONFS_UNBIND] = "UNBIND", + [FUNCTIONFS_ENABLE] = "ENABLE", + [FUNCTIONFS_DISABLE] = "DISABLE", + [FUNCTIONFS_SETUP] = "SETUP", + [FUNCTIONFS_SUSPEND] = "SUSPEND", + [FUNCTIONFS_RESUME] = "RESUME", + }; + + const struct usb_functionfs_event *event = buf; + size_t n; + + (void)ignore; + + for (n = nbytes / sizeof *event; n; --n, ++event) + switch (event->type) { + case FUNCTIONFS_BIND: + case FUNCTIONFS_UNBIND: + case FUNCTIONFS_ENABLE: + case FUNCTIONFS_DISABLE: + case FUNCTIONFS_SETUP: + case FUNCTIONFS_SUSPEND: + case FUNCTIONFS_RESUME: + printf("Event %s\n", names[event->type]); + if (event->type == FUNCTIONFS_SETUP) + handle_setup(&event->u.setup); + break; + + default: + printf("Event %03u (unknown)\n", event->type); + } + + return nbytes; +} + +static void ep0_init(struct thread *t) +{ + ssize_t ret; + + info("%s: writing descriptors\n", t->filename); + ret = write(t->fd, &descriptors, sizeof descriptors); + die_on(ret < 0, "%s: write: descriptors", t->filename); + + info("%s: writing strings\n", t->filename); + ret = write(t->fd, &strings, sizeof strings); + die_on(ret < 0, "%s: write: strings", t->filename); +} + + +/******************** Main **************************************************/ + +int main(void) +{ + unsigned i; + + /* XXX TODO: Argument parsing missing */ + + init_thread(threads); + ep0_init(threads); + + for (i = 1; i < sizeof threads / sizeof *threads; ++i) + init_thread(threads + i); + + for (i = 1; i < sizeof threads / sizeof *threads; ++i) + start_thread(threads + i); + + start_thread_helper(threads); + + for (i = 1; i < sizeof threads / sizeof *threads; ++i) + join_thread(threads + i); + + return 0; +} diff --git a/tools/usb/hcd-tests.sh b/tools/usb/hcd-tests.sh new file mode 100644 index 00000000..b30b3dc4 --- /dev/null +++ b/tools/usb/hcd-tests.sh @@ -0,0 +1,275 @@ +#!/bin/sh +# +# test types can be passed on the command line: +# +# - control: any device can do this +# - out, in: out needs 'bulk sink' firmware, in needs 'bulk src' +# - iso-out, iso-in: out needs 'iso sink' firmware, in needs 'iso src' +# - halt: needs bulk sink+src, tests halt set/clear from host +# - unlink: needs bulk sink and/or src, test HCD unlink processing +# - loop: needs firmware that will buffer N transfers +# +# run it for hours, days, weeks. +# + +# +# this default provides a steady test load for a bulk device +# +TYPES='control out in' +#TYPES='control out in halt' + +# +# to test HCD code +# +# - include unlink tests +# - add some ${RANDOM}ness +# - connect several devices concurrently (same HC) +# - keep HC's IRQ lines busy with unrelated traffic (IDE, net, ...) +# - add other concurrent system loads +# + +declare -i COUNT BUFLEN + +COUNT=50000 +BUFLEN=2048 + +# NOTE: the 'in' and 'out' cases are usually bulk, but can be +# set up to use interrupt transfers by 'usbtest' module options + + +if [ "$DEVICE" = "" ]; then + echo "testing ALL recognized usbtest devices" + echo "" + TEST_ARGS="-a" +else + TEST_ARGS="" +fi + +do_test () +{ + if ! ./testusb $TEST_ARGS -s $BUFLEN -c $COUNT $* 2>/dev/null + then + echo "FAIL" + exit 1 + fi +} + +ARGS="$*" + +if [ "$ARGS" = "" ]; +then + ARGS="$TYPES" +fi + +# FIXME use /sys/bus/usb/device/$THIS/bConfigurationValue to +# check and change configs + +CONFIG='' + +check_config () +{ + if [ "$CONFIG" = "" ]; then + CONFIG=$1 + echo "assuming $CONFIG configuration" + return + fi + if [ "$CONFIG" = $1 ]; then + return + fi + + echo "** device must be in $1 config, but it's $CONFIG instead" + exit 1 +} + + +echo "TESTING: $ARGS" + +while : true +do + echo $(date) + + for TYPE in $ARGS + do + # restore defaults + COUNT=5000 + BUFLEN=2048 + + # FIXME automatically multiply COUNT by 10 when + # /sys/bus/usb/device/$THIS/speed == "480" + +# COUNT=50000 + + case $TYPE in + control) + # any device, in any configuration, can use this. + echo '** Control test cases:' + + echo "test 9: ch9 postconfig" + do_test -t 9 -c 5000 + echo "test 10: control queueing" + do_test -t 10 -c 5000 + + # this relies on some vendor-specific commands + echo "test 14: control writes" + do_test -t 14 -c 15000 -s 256 -v 1 + + echo "test 21: control writes, unaligned" + do_test -t 21 -c 100 -s 256 -v 1 + + ;; + + out) + check_config sink-src + echo '** Host Write (OUT) test cases:' + + echo "test 1: $COUNT transfers, same size" + do_test -t 1 + echo "test 3: $COUNT transfers, variable/short size" + do_test -t 3 -v 421 + + COUNT=100 + echo "test 17: $COUNT transfers, unaligned DMA map by core" + do_test -t 17 + + echo "test 19: $COUNT transfers, unaligned DMA map by usb_alloc_coherent" + do_test -t 19 + + COUNT=2000 + echo "test 5: $COUNT scatterlists, same size entries" + do_test -t 5 + + # try to trigger short OUT processing bugs + echo "test 7a: $COUNT scatterlists, variable size/short entries" + do_test -t 7 -v 579 + BUFLEN=4096 + echo "test 7b: $COUNT scatterlists, variable size/bigger entries" + do_test -t 7 -v 41 + BUFLEN=64 + echo "test 7c: $COUNT scatterlists, variable size/micro entries" + do_test -t 7 -v 63 + ;; + + iso-out) + check_config sink-src + echo '** Host ISOCHRONOUS Write (OUT) test cases:' + + # at peak iso transfer rates: + # - usb 2.0 high bandwidth, this is one frame. + # - usb 1.1, it's twenty-four frames. + BUFLEN=24500 + + COUNT=1000 + +# COUNT=10000 + + echo "test 15: $COUNT transfers, same size" + # do_test -t 15 -g 3 -v 0 + BUFLEN=32768 + do_test -t 15 -g 8 -v 0 + + # FIXME it'd make sense to have an iso OUT test issuing + # short writes on more packets than the last one + + COUNT=100 + echo "test 22: $COUNT transfers, non aligned" + do_test -t 22 -g 8 -v 0 + + ;; + + in) + check_config sink-src + echo '** Host Read (IN) test cases:' + + # NOTE: these "variable size" reads are just multiples + # of 512 bytes, no EOVERFLOW testing is done yet + + echo "test 2: $COUNT transfers, same size" + do_test -t 2 + echo "test 4: $COUNT transfers, variable size" + do_test -t 4 + + COUNT=100 + echo "test 18: $COUNT transfers, unaligned DMA map by core" + do_test -t 18 + + echo "test 20: $COUNT transfers, unaligned DMA map by usb_alloc_coherent" + do_test -t 20 + + COUNT=2000 + echo "test 6: $COUNT scatterlists, same size entries" + do_test -t 6 + echo "test 8: $COUNT scatterlists, variable size entries" + do_test -t 8 + ;; + + iso-in) + check_config sink-src + echo '** Host ISOCHRONOUS Read (IN) test cases:' + + # at peak iso transfer rates: + # - usb 2.0 high bandwidth, this is one frame. + # - usb 1.1, it's twenty-four frames. + BUFLEN=24500 + + COUNT=1000 + +# COUNT=10000 + + echo "test 16: $COUNT transfers, same size" + # do_test -t 16 -g 3 -v 0 + BUFLEN=32768 + do_test -t 16 -g 8 -v 0 + + # FIXME since iso expects faults, it'd make sense + # to have an iso IN test issuing short reads ... + + COUNT=100 + echo "test 23: $COUNT transfers, unaligned" + do_test -t 23 -g 8 -v 0 + + ;; + + halt) + # NOTE: sometimes hardware doesn't cooperate well with halting + # endpoints from the host side. so long as mass-storage class + # firmware can halt them from the device, don't worry much if + # you can't make this test work on your device. + COUNT=2000 + echo "test 13: $COUNT halt set/clear" + do_test -t 13 + ;; + + unlink) + COUNT=2000 + echo "test 11: $COUNT read unlinks" + do_test -t 11 + + echo "test 12: $COUNT write unlinks" + do_test -t 12 + ;; + + loop) + # defaults need too much buffering for ez-usb devices + BUFLEN=2048 + COUNT=32 + + # modprobe g_zero qlen=$COUNT buflen=$BUFLEN loopdefault + check_config loopback + + # FIXME someone needs to write and merge a version of this + + echo "write $COUNT buffers of $BUFLEN bytes, read them back" + + echo "write $COUNT variable size buffers, read them back" + + ;; + + *) + echo "Don't understand test type $TYPE" + exit 1; + esac + echo '' + done +done + +# vim: sw=4 diff --git a/tools/usb/testusb.c b/tools/usb/testusb.c new file mode 100644 index 00000000..f08e8946 --- /dev/null +++ b/tools/usb/testusb.c @@ -0,0 +1,547 @@ +/* $(CROSS_COMPILE)cc -Wall -Wextra -g -lpthread -o testusb testusb.c */ + +/* + * Copyright (c) 2002 by David Brownell + * Copyright (c) 2010 by Samsung Electronics + * Author: Michal Nazarewicz <m.nazarewicz@samsung.com> + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2 of the License, or (at your + * option) any later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY + * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License + * for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +/* + * This program issues ioctls to perform the tests implemented by the + * kernel driver. It can generate a variety of transfer patterns; you + * should make sure to test both regular streaming and mixes of + * transfer sizes (including short transfers). + * + * For more information on how this can be used and on USB testing + * refer to <URL:http://www.linux-usb.org/usbtest/>. + */ + +#include <stdio.h> +#include <string.h> +#include <ftw.h> +#include <stdlib.h> +#include <pthread.h> +#include <unistd.h> +#include <errno.h> +#include <limits.h> + +#include <sys/types.h> +#include <sys/stat.h> +#include <fcntl.h> + +#include <sys/ioctl.h> +#include <linux/usbdevice_fs.h> + +/*-------------------------------------------------------------------------*/ + +#define TEST_CASES 30 + +// FIXME make these public somewhere; usbdevfs.h? + +struct usbtest_param { + // inputs + unsigned test_num; /* 0..(TEST_CASES-1) */ + unsigned iterations; + unsigned length; + unsigned vary; + unsigned sglen; + + // outputs + struct timeval duration; +}; +#define USBTEST_REQUEST _IOWR('U', 100, struct usbtest_param) + +/*-------------------------------------------------------------------------*/ + +/* #include <linux/usb_ch9.h> */ + +#define USB_DT_DEVICE 0x01 +#define USB_DT_INTERFACE 0x04 + +#define USB_CLASS_PER_INTERFACE 0 /* for DeviceClass */ +#define USB_CLASS_VENDOR_SPEC 0xff + + +struct usb_device_descriptor { + __u8 bLength; + __u8 bDescriptorType; + __u16 bcdUSB; + __u8 bDeviceClass; + __u8 bDeviceSubClass; + __u8 bDeviceProtocol; + __u8 bMaxPacketSize0; + __u16 idVendor; + __u16 idProduct; + __u16 bcdDevice; + __u8 iManufacturer; + __u8 iProduct; + __u8 iSerialNumber; + __u8 bNumConfigurations; +} __attribute__ ((packed)); + +struct usb_interface_descriptor { + __u8 bLength; + __u8 bDescriptorType; + + __u8 bInterfaceNumber; + __u8 bAlternateSetting; + __u8 bNumEndpoints; + __u8 bInterfaceClass; + __u8 bInterfaceSubClass; + __u8 bInterfaceProtocol; + __u8 iInterface; +} __attribute__ ((packed)); + +enum usb_device_speed { + USB_SPEED_UNKNOWN = 0, /* enumerating */ + USB_SPEED_LOW, USB_SPEED_FULL, /* usb 1.1 */ + USB_SPEED_HIGH /* usb 2.0 */ +}; + +/*-------------------------------------------------------------------------*/ + +static char *speed (enum usb_device_speed s) +{ + switch (s) { + case USB_SPEED_UNKNOWN: return "unknown"; + case USB_SPEED_LOW: return "low"; + case USB_SPEED_FULL: return "full"; + case USB_SPEED_HIGH: return "high"; + default: return "??"; + } +} + +struct testdev { + struct testdev *next; + char *name; + pthread_t thread; + enum usb_device_speed speed; + unsigned ifnum : 8; + unsigned forever : 1; + int test; + + struct usbtest_param param; +}; +static struct testdev *testdevs; + +static int testdev_ffs_ifnum(FILE *fd) +{ + union { + char buf[255]; + struct usb_interface_descriptor intf; + } u; + + for (;;) { + if (fread(u.buf, 1, 1, fd) != 1) + return -1; + if (fread(u.buf + 1, (unsigned char)u.buf[0] - 1, 1, fd) != 1) + return -1; + + if (u.intf.bLength == sizeof u.intf + && u.intf.bDescriptorType == USB_DT_INTERFACE + && u.intf.bNumEndpoints == 2 + && u.intf.bInterfaceClass == USB_CLASS_VENDOR_SPEC + && u.intf.bInterfaceSubClass == 0 + && u.intf.bInterfaceProtocol == 0) + return (unsigned char)u.intf.bInterfaceNumber; + } +} + +static int testdev_ifnum(FILE *fd) +{ + struct usb_device_descriptor dev; + + if (fread(&dev, sizeof dev, 1, fd) != 1) + return -1; + + if (dev.bLength != sizeof dev || dev.bDescriptorType != USB_DT_DEVICE) + return -1; + + /* FX2 with (tweaked) bulksrc firmware */ + if (dev.idVendor == 0x0547 && dev.idProduct == 0x1002) + return 0; + + /*----------------------------------------------------*/ + + /* devices that start up using the EZ-USB default device and + * which we can use after loading simple firmware. hotplug + * can fxload it, and then run this test driver. + * + * we return false positives in two cases: + * - the device has a "real" driver (maybe usb-serial) that + * renumerates. the device should vanish quickly. + * - the device doesn't have the test firmware installed. + */ + + /* generic EZ-USB FX controller */ + if (dev.idVendor == 0x0547 && dev.idProduct == 0x2235) + return 0; + + /* generic EZ-USB FX2 controller */ + if (dev.idVendor == 0x04b4 && dev.idProduct == 0x8613) + return 0; + + /* CY3671 development board with EZ-USB FX */ + if (dev.idVendor == 0x0547 && dev.idProduct == 0x0080) + return 0; + + /* Keyspan 19Qi uses an21xx (original EZ-USB) */ + if (dev.idVendor == 0x06cd && dev.idProduct == 0x010b) + return 0; + + /*----------------------------------------------------*/ + + /* "gadget zero", Linux-USB test software */ + if (dev.idVendor == 0x0525 && dev.idProduct == 0xa4a0) + return 0; + + /* user mode subset of that */ + if (dev.idVendor == 0x0525 && dev.idProduct == 0xa4a4) + return testdev_ffs_ifnum(fd); + /* return 0; */ + + /* iso version of usermode code */ + if (dev.idVendor == 0x0525 && dev.idProduct == 0xa4a3) + return 0; + + /* some GPL'd test firmware uses these IDs */ + + if (dev.idVendor == 0xfff0 && dev.idProduct == 0xfff0) + return 0; + + /*----------------------------------------------------*/ + + /* iBOT2 high speed webcam */ + if (dev.idVendor == 0x0b62 && dev.idProduct == 0x0059) + return 0; + + /*----------------------------------------------------*/ + + /* the FunctionFS gadget can have the source/sink interface + * anywhere. We look for an interface descriptor that match + * what we expect. We ignore configuratiens thou. */ + + if (dev.idVendor == 0x0525 && dev.idProduct == 0xa4ac + && (dev.bDeviceClass == USB_CLASS_PER_INTERFACE + || dev.bDeviceClass == USB_CLASS_VENDOR_SPEC)) + return testdev_ffs_ifnum(fd); + + return -1; +} + +static int find_testdev(const char *name, const struct stat *sb, int flag) +{ + FILE *fd; + int ifnum; + struct testdev *entry; + + (void)sb; /* unused */ + + if (flag != FTW_F) + return 0; + /* ignore /proc/bus/usb/{devices,drivers} */ + if (strrchr(name, '/')[1] == 'd') + return 0; + + fd = fopen(name, "rb"); + if (!fd) { + perror(name); + return 0; + } + + ifnum = testdev_ifnum(fd); + fclose(fd); + if (ifnum < 0) + return 0; + + entry = calloc(1, sizeof *entry); + if (!entry) + goto nomem; + + entry->name = strdup(name); + if (!entry->name) { + free(entry); +nomem: + perror("malloc"); + return 0; + } + + entry->ifnum = ifnum; + + /* FIXME ask usbfs what speed; update USBDEVFS_CONNECTINFO so + * it tells about high speed etc */ + + fprintf(stderr, "%s speed\t%s\t%u\n", + speed(entry->speed), entry->name, entry->ifnum); + + entry->next = testdevs; + testdevs = entry; + return 0; +} + +static int +usbdev_ioctl (int fd, int ifno, unsigned request, void *param) +{ + struct usbdevfs_ioctl wrapper; + + wrapper.ifno = ifno; + wrapper.ioctl_code = request; + wrapper.data = param; + + return ioctl (fd, USBDEVFS_IOCTL, &wrapper); +} + +static void *handle_testdev (void *arg) +{ + struct testdev *dev = arg; + int fd, i; + int status; + + if ((fd = open (dev->name, O_RDWR)) < 0) { + perror ("can't open dev file r/w"); + return 0; + } + +restart: + for (i = 0; i < TEST_CASES; i++) { + if (dev->test != -1 && dev->test != i) + continue; + dev->param.test_num = i; + + status = usbdev_ioctl (fd, dev->ifnum, + USBTEST_REQUEST, &dev->param); + if (status < 0 && errno == EOPNOTSUPP) + continue; + + /* FIXME need a "syslog it" option for background testing */ + + /* NOTE: each thread emits complete lines; no fragments! */ + if (status < 0) { + char buf [80]; + int err = errno; + + if (strerror_r (errno, buf, sizeof buf)) { + snprintf (buf, sizeof buf, "error %d", err); + errno = err; + } + printf ("%s test %d --> %d (%s)\n", + dev->name, i, errno, buf); + } else + printf ("%s test %d, %4d.%.06d secs\n", dev->name, i, + (int) dev->param.duration.tv_sec, + (int) dev->param.duration.tv_usec); + + fflush (stdout); + } + if (dev->forever) + goto restart; + + close (fd); + return arg; +} + +static const char *usbfs_dir_find(void) +{ + static char usbfs_path_0[] = "/dev/usb/devices"; + static char usbfs_path_1[] = "/proc/bus/usb/devices"; + + static char *const usbfs_paths[] = { + usbfs_path_0, usbfs_path_1 + }; + + static char *const * + end = usbfs_paths + sizeof usbfs_paths / sizeof *usbfs_paths; + + char *const *it = usbfs_paths; + do { + int fd = open(*it, O_RDONLY); + close(fd); + if (fd >= 0) { + strrchr(*it, '/')[0] = '\0'; + return *it; + } + } while (++it != end); + + return NULL; +} + +static int parse_num(unsigned *num, const char *str) +{ + unsigned long val; + char *end; + + errno = 0; + val = strtoul(str, &end, 0); + if (errno || *end || val > UINT_MAX) + return -1; + *num = val; + return 0; +} + +int main (int argc, char **argv) +{ + + int c; + struct testdev *entry; + char *device; + const char *usbfs_dir = NULL; + int all = 0, forever = 0, not = 0; + int test = -1 /* all */; + struct usbtest_param param; + + /* pick defaults that works with all speeds, without short packets. + * + * Best per-frame data rates: + * high speed, bulk 512 * 13 * 8 = 53248 + * interrupt 1024 * 3 * 8 = 24576 + * full speed, bulk/intr 64 * 19 = 1216 + * interrupt 64 * 1 = 64 + * low speed, interrupt 8 * 1 = 8 + */ + param.iterations = 1000; + param.length = 512; + param.vary = 512; + param.sglen = 32; + + /* for easy use when hotplugging */ + device = getenv ("DEVICE"); + + while ((c = getopt (argc, argv, "D:aA:c:g:hns:t:v:")) != EOF) + switch (c) { + case 'D': /* device, if only one */ + device = optarg; + continue; + case 'A': /* use all devices with specified usbfs dir */ + usbfs_dir = optarg; + /* FALL THROUGH */ + case 'a': /* use all devices */ + device = NULL; + all = 1; + continue; + case 'c': /* count iterations */ + if (parse_num(¶m.iterations, optarg)) + goto usage; + continue; + case 'g': /* scatter/gather entries */ + if (parse_num(¶m.sglen, optarg)) + goto usage; + continue; + case 'l': /* loop forever */ + forever = 1; + continue; + case 'n': /* no test running! */ + not = 1; + continue; + case 's': /* size of packet */ + if (parse_num(¶m.length, optarg)) + goto usage; + continue; + case 't': /* run just one test */ + test = atoi (optarg); + if (test < 0) + goto usage; + continue; + case 'v': /* vary packet size by ... */ + if (parse_num(¶m.vary, optarg)) + goto usage; + continue; + case '?': + case 'h': + default: +usage: + fprintf (stderr, "usage: %s [-n] [-D dev | -a | -A usbfs-dir]\n" + "\t[-c iterations] [-t testnum]\n" + "\t[-s packetsize] [-g sglen] [-v vary]\n", + argv [0]); + return 1; + } + if (optind != argc) + goto usage; + if (!all && !device) { + fprintf (stderr, "must specify '-a' or '-D dev', " + "or DEVICE=/proc/bus/usb/BBB/DDD in env\n"); + goto usage; + } + + /* Find usbfs mount point */ + if (!usbfs_dir) { + usbfs_dir = usbfs_dir_find(); + if (!usbfs_dir) { + fputs ("usbfs files are missing\n", stderr); + return -1; + } + } + + /* collect and list the test devices */ + if (ftw (usbfs_dir, find_testdev, 3) != 0) { + fputs ("ftw failed; is usbfs missing?\n", stderr); + return -1; + } + + /* quit, run single test, or create test threads */ + if (!testdevs && !device) { + fputs ("no test devices recognized\n", stderr); + return -1; + } + if (not) + return 0; + if (testdevs && testdevs->next == 0 && !device) + device = testdevs->name; + for (entry = testdevs; entry; entry = entry->next) { + int status; + + entry->param = param; + entry->forever = forever; + entry->test = test; + + if (device) { + if (strcmp (entry->name, device)) + continue; + return handle_testdev (entry) != entry; + } + status = pthread_create (&entry->thread, 0, handle_testdev, entry); + if (status) { + perror ("pthread_create"); + continue; + } + } + if (device) { + struct testdev dev; + + /* kernel can recognize test devices we don't */ + fprintf (stderr, "%s: %s may see only control tests\n", + argv [0], device); + + memset (&dev, 0, sizeof dev); + dev.name = device; + dev.param = param; + dev.forever = forever; + dev.test = test; + return handle_testdev (&dev) != &dev; + } + + /* wait for tests to complete */ + for (entry = testdevs; entry; entry = entry->next) { + void *retval; + + if (pthread_join (entry->thread, &retval)) + perror ("pthread_join"); + /* testing errors discarded! */ + } + + return 0; +} diff --git a/tools/virtio/Makefile b/tools/virtio/Makefile new file mode 100644 index 00000000..d1d442ed --- /dev/null +++ b/tools/virtio/Makefile @@ -0,0 +1,12 @@ +all: test mod +test: virtio_test +virtio_test: virtio_ring.o virtio_test.o +CFLAGS += -g -O2 -Wall -I. -I ../../usr/include/ -Wno-pointer-sign -fno-strict-overflow -MMD +vpath %.c ../../drivers/virtio +mod: + ${MAKE} -C `pwd`/../.. M=`pwd`/vhost_test +.PHONY: all test mod clean +clean: + ${RM} *.o vhost_test/*.o vhost_test/.*.cmd \ + vhost_test/Module.symvers vhost_test/modules.order *.d +-include *.d diff --git a/tools/virtio/linux/device.h b/tools/virtio/linux/device.h new file mode 100644 index 00000000..4ad7e1df --- /dev/null +++ b/tools/virtio/linux/device.h @@ -0,0 +1,2 @@ +#ifndef LINUX_DEVICE_H +#endif diff --git a/tools/virtio/linux/slab.h b/tools/virtio/linux/slab.h new file mode 100644 index 00000000..81baeac8 --- /dev/null +++ b/tools/virtio/linux/slab.h @@ -0,0 +1,2 @@ +#ifndef LINUX_SLAB_H +#endif diff --git a/tools/virtio/linux/virtio.h b/tools/virtio/linux/virtio.h new file mode 100644 index 00000000..669bcdd4 --- /dev/null +++ b/tools/virtio/linux/virtio.h @@ -0,0 +1,223 @@ +#ifndef LINUX_VIRTIO_H +#define LINUX_VIRTIO_H + +#include <stdbool.h> +#include <stdlib.h> +#include <stddef.h> +#include <stdio.h> +#include <string.h> +#include <assert.h> + +#include <linux/types.h> +#include <errno.h> + +typedef unsigned long long dma_addr_t; + +struct scatterlist { + unsigned long page_link; + unsigned int offset; + unsigned int length; + dma_addr_t dma_address; +}; + +struct page { + unsigned long long dummy; +}; + +#define BUG_ON(__BUG_ON_cond) assert(!(__BUG_ON_cond)) + +/* Physical == Virtual */ +#define virt_to_phys(p) ((unsigned long)p) +#define phys_to_virt(a) ((void *)(unsigned long)(a)) +/* Page address: Virtual / 4K */ +#define virt_to_page(p) ((struct page*)((virt_to_phys(p) / 4096) * \ + sizeof(struct page))) +#define offset_in_page(p) (((unsigned long)p) % 4096) +#define sg_phys(sg) ((sg->page_link & ~0x3) / sizeof(struct page) * 4096 + \ + sg->offset) +static inline void sg_mark_end(struct scatterlist *sg) +{ + /* + * Set termination bit, clear potential chain bit + */ + sg->page_link |= 0x02; + sg->page_link &= ~0x01; +} +static inline void sg_init_table(struct scatterlist *sgl, unsigned int nents) +{ + memset(sgl, 0, sizeof(*sgl) * nents); + sg_mark_end(&sgl[nents - 1]); +} +static inline void sg_assign_page(struct scatterlist *sg, struct page *page) +{ + unsigned long page_link = sg->page_link & 0x3; + + /* + * In order for the low bit stealing approach to work, pages + * must be aligned at a 32-bit boundary as a minimum. + */ + BUG_ON((unsigned long) page & 0x03); + sg->page_link = page_link | (unsigned long) page; +} + +static inline void sg_set_page(struct scatterlist *sg, struct page *page, + unsigned int len, unsigned int offset) +{ + sg_assign_page(sg, page); + sg->offset = offset; + sg->length = len; +} + +static inline void sg_set_buf(struct scatterlist *sg, const void *buf, + unsigned int buflen) +{ + sg_set_page(sg, virt_to_page(buf), buflen, offset_in_page(buf)); +} + +static inline void sg_init_one(struct scatterlist *sg, const void *buf, unsigned int buflen) +{ + sg_init_table(sg, 1); + sg_set_buf(sg, buf, buflen); +} + +typedef __u16 u16; + +typedef enum { + GFP_KERNEL, + GFP_ATOMIC, +} gfp_t; +typedef enum { + IRQ_NONE, + IRQ_HANDLED +} irqreturn_t; + +static inline void *kmalloc(size_t s, gfp_t gfp) +{ + return malloc(s); +} + +static inline void kfree(void *p) +{ + free(p); +} + +#define container_of(ptr, type, member) ({ \ + const typeof( ((type *)0)->member ) *__mptr = (ptr); \ + (type *)( (char *)__mptr - offsetof(type,member) );}) + +#define uninitialized_var(x) x = x + +# ifndef likely +# define likely(x) (__builtin_expect(!!(x), 1)) +# endif +# ifndef unlikely +# define unlikely(x) (__builtin_expect(!!(x), 0)) +# endif + +#define pr_err(format, ...) fprintf (stderr, format, ## __VA_ARGS__) +#ifdef DEBUG +#define pr_debug(format, ...) fprintf (stderr, format, ## __VA_ARGS__) +#else +#define pr_debug(format, ...) do {} while (0) +#endif +#define dev_err(dev, format, ...) fprintf (stderr, format, ## __VA_ARGS__) +#define dev_warn(dev, format, ...) fprintf (stderr, format, ## __VA_ARGS__) + +/* TODO: empty stubs for now. Broken but enough for virtio_ring.c */ +#define list_add_tail(a, b) do {} while (0) +#define list_del(a) do {} while (0) + +#define BIT_WORD(nr) ((nr) / BITS_PER_LONG) +#define BITS_PER_BYTE 8 +#define BITS_PER_LONG (sizeof(long) * BITS_PER_BYTE) +#define BIT_MASK(nr) (1UL << ((nr) % BITS_PER_LONG)) +/* TODO: Not atomic as it should be: + * we don't use this for anything important. */ +static inline void clear_bit(int nr, volatile unsigned long *addr) +{ + unsigned long mask = BIT_MASK(nr); + unsigned long *p = ((unsigned long *)addr) + BIT_WORD(nr); + + *p &= ~mask; +} + +static inline int test_bit(int nr, const volatile unsigned long *addr) +{ + return 1UL & (addr[BIT_WORD(nr)] >> (nr & (BITS_PER_LONG-1))); +} + +/* The only feature we care to support */ +#define virtio_has_feature(dev, feature) \ + test_bit((feature), (dev)->features) +/* end of stubs */ + +struct virtio_device { + void *dev; + unsigned long features[1]; +}; + +struct virtqueue { + /* TODO: commented as list macros are empty stubs for now. + * Broken but enough for virtio_ring.c + * struct list_head list; */ + void (*callback)(struct virtqueue *vq); + const char *name; + struct virtio_device *vdev; + void *priv; +}; + +#define EXPORT_SYMBOL_GPL(__EXPORT_SYMBOL_GPL_name) \ + void __EXPORT_SYMBOL_GPL##__EXPORT_SYMBOL_GPL_name() { \ +} +#define MODULE_LICENSE(__MODULE_LICENSE_value) \ + const char *__MODULE_LICENSE_name = __MODULE_LICENSE_value + +#define CONFIG_SMP + +#if defined(__i386__) || defined(__x86_64__) +#define barrier() asm volatile("" ::: "memory") +#define mb() __sync_synchronize() + +#define smp_mb() mb() +# define smp_rmb() barrier() +# define smp_wmb() barrier() +#else +#error Please fill in barrier macros +#endif + +/* Interfaces exported by virtio_ring. */ +int virtqueue_add_buf_gfp(struct virtqueue *vq, + struct scatterlist sg[], + unsigned int out_num, + unsigned int in_num, + void *data, + gfp_t gfp); + +static inline int virtqueue_add_buf(struct virtqueue *vq, + struct scatterlist sg[], + unsigned int out_num, + unsigned int in_num, + void *data) +{ + return virtqueue_add_buf_gfp(vq, sg, out_num, in_num, data, GFP_ATOMIC); +} + +void virtqueue_kick(struct virtqueue *vq); + +void *virtqueue_get_buf(struct virtqueue *vq, unsigned int *len); + +void virtqueue_disable_cb(struct virtqueue *vq); + +bool virtqueue_enable_cb(struct virtqueue *vq); + +void *virtqueue_detach_unused_buf(struct virtqueue *vq); +struct virtqueue *vring_new_virtqueue(unsigned int num, + unsigned int vring_align, + struct virtio_device *vdev, + void *pages, + void (*notify)(struct virtqueue *vq), + void (*callback)(struct virtqueue *vq), + const char *name); +void vring_del_virtqueue(struct virtqueue *vq); + +#endif diff --git a/tools/virtio/vhost_test/Makefile b/tools/virtio/vhost_test/Makefile new file mode 100644 index 00000000..a1d35b81 --- /dev/null +++ b/tools/virtio/vhost_test/Makefile @@ -0,0 +1,2 @@ +obj-m += vhost_test.o +EXTRA_CFLAGS += -Idrivers/vhost diff --git a/tools/virtio/vhost_test/vhost_test.c b/tools/virtio/vhost_test/vhost_test.c new file mode 100644 index 00000000..18735189 --- /dev/null +++ b/tools/virtio/vhost_test/vhost_test.c @@ -0,0 +1 @@ +#include "test.c" diff --git a/tools/virtio/virtio_test.c b/tools/virtio/virtio_test.c new file mode 100644 index 00000000..74d3331b --- /dev/null +++ b/tools/virtio/virtio_test.c @@ -0,0 +1,263 @@ +#define _GNU_SOURCE +#include <getopt.h> +#include <string.h> +#include <poll.h> +#include <sys/eventfd.h> +#include <stdlib.h> +#include <assert.h> +#include <unistd.h> +#include <sys/ioctl.h> +#include <sys/stat.h> +#include <sys/types.h> +#include <fcntl.h> +#include <linux/vhost.h> +#include <linux/virtio.h> +#include <linux/virtio_ring.h> +#include "../../drivers/vhost/test.h" + +struct vq_info { + int kick; + int call; + int num; + int idx; + void *ring; + /* copy used for control */ + struct vring vring; + struct virtqueue *vq; +}; + +struct vdev_info { + struct virtio_device vdev; + int control; + struct pollfd fds[1]; + struct vq_info vqs[1]; + int nvqs; + void *buf; + size_t buf_size; + struct vhost_memory *mem; +}; + +void vq_notify(struct virtqueue *vq) +{ + struct vq_info *info = vq->priv; + unsigned long long v = 1; + int r; + r = write(info->kick, &v, sizeof v); + assert(r == sizeof v); +} + +void vq_callback(struct virtqueue *vq) +{ +} + + +void vhost_vq_setup(struct vdev_info *dev, struct vq_info *info) +{ + struct vhost_vring_state state = { .index = info->idx }; + struct vhost_vring_file file = { .index = info->idx }; + unsigned long long features = dev->vdev.features[0]; + struct vhost_vring_addr addr = { + .index = info->idx, + .desc_user_addr = (uint64_t)(unsigned long)info->vring.desc, + .avail_user_addr = (uint64_t)(unsigned long)info->vring.avail, + .used_user_addr = (uint64_t)(unsigned long)info->vring.used, + }; + int r; + r = ioctl(dev->control, VHOST_SET_FEATURES, &features); + assert(r >= 0); + state.num = info->vring.num; + r = ioctl(dev->control, VHOST_SET_VRING_NUM, &state); + assert(r >= 0); + state.num = 0; + r = ioctl(dev->control, VHOST_SET_VRING_BASE, &state); + assert(r >= 0); + r = ioctl(dev->control, VHOST_SET_VRING_ADDR, &addr); + assert(r >= 0); + file.fd = info->kick; + r = ioctl(dev->control, VHOST_SET_VRING_KICK, &file); + assert(r >= 0); + file.fd = info->call; + r = ioctl(dev->control, VHOST_SET_VRING_CALL, &file); + assert(r >= 0); +} + +static void vq_info_add(struct vdev_info *dev, int num) +{ + struct vq_info *info = &dev->vqs[dev->nvqs]; + int r; + info->idx = dev->nvqs; + info->kick = eventfd(0, EFD_NONBLOCK); + info->call = eventfd(0, EFD_NONBLOCK); + r = posix_memalign(&info->ring, 4096, vring_size(num, 4096)); + assert(r >= 0); + memset(info->ring, 0, vring_size(num, 4096)); + vring_init(&info->vring, num, info->ring, 4096); + info->vq = vring_new_virtqueue(info->vring.num, 4096, &dev->vdev, info->ring, + vq_notify, vq_callback, "test"); + assert(info->vq); + info->vq->priv = info; + vhost_vq_setup(dev, info); + dev->fds[info->idx].fd = info->call; + dev->fds[info->idx].events = POLLIN; + dev->nvqs++; +} + +static void vdev_info_init(struct vdev_info* dev, unsigned long long features) +{ + int r; + memset(dev, 0, sizeof *dev); + dev->vdev.features[0] = features; + dev->vdev.features[1] = features >> 32; + dev->buf_size = 1024; + dev->buf = malloc(dev->buf_size); + assert(dev->buf); + dev->control = open("/dev/vhost-test", O_RDWR); + assert(dev->control >= 0); + r = ioctl(dev->control, VHOST_SET_OWNER, NULL); + assert(r >= 0); + dev->mem = malloc(offsetof(struct vhost_memory, regions) + + sizeof dev->mem->regions[0]); + assert(dev->mem); + memset(dev->mem, 0, offsetof(struct vhost_memory, regions) + + sizeof dev->mem->regions[0]); + dev->mem->nregions = 1; + dev->mem->regions[0].guest_phys_addr = (long)dev->buf; + dev->mem->regions[0].userspace_addr = (long)dev->buf; + dev->mem->regions[0].memory_size = dev->buf_size; + r = ioctl(dev->control, VHOST_SET_MEM_TABLE, dev->mem); + assert(r >= 0); +} + +/* TODO: this is pretty bad: we get a cache line bounce + * for the wait queue on poll and another one on read, + * plus the read which is there just to clear the + * current state. */ +static void wait_for_interrupt(struct vdev_info *dev) +{ + int i; + unsigned long long val; + poll(dev->fds, dev->nvqs, -1); + for (i = 0; i < dev->nvqs; ++i) + if (dev->fds[i].revents & POLLIN) { + read(dev->fds[i].fd, &val, sizeof val); + } +} + +static void run_test(struct vdev_info *dev, struct vq_info *vq, int bufs) +{ + struct scatterlist sl; + long started = 0, completed = 0; + long completed_before; + int r, test = 1; + unsigned len; + long long spurious = 0; + r = ioctl(dev->control, VHOST_TEST_RUN, &test); + assert(r >= 0); + for (;;) { + virtqueue_disable_cb(vq->vq); + completed_before = completed; + do { + if (started < bufs) { + sg_init_one(&sl, dev->buf, dev->buf_size); + r = virtqueue_add_buf(vq->vq, &sl, 1, 0, + dev->buf + started); + if (likely(r >= 0)) { + ++started; + virtqueue_kick(vq->vq); + } + } else + r = -1; + + /* Flush out completed bufs if any */ + if (virtqueue_get_buf(vq->vq, &len)) { + ++completed; + r = 0; + } + + } while (r >= 0); + if (completed == completed_before) + ++spurious; + assert(completed <= bufs); + assert(started <= bufs); + if (completed == bufs) + break; + if (virtqueue_enable_cb(vq->vq)) { + wait_for_interrupt(dev); + } + } + test = 0; + r = ioctl(dev->control, VHOST_TEST_RUN, &test); + assert(r >= 0); + fprintf(stderr, "spurious wakeus: 0x%llx\n", spurious); +} + +const char optstring[] = "h"; +const struct option longopts[] = { + { + .name = "help", + .val = 'h', + }, + { + .name = "event-idx", + .val = 'E', + }, + { + .name = "no-event-idx", + .val = 'e', + }, + { + .name = "indirect", + .val = 'I', + }, + { + .name = "no-indirect", + .val = 'i', + }, + { + } +}; + +static void help() +{ + fprintf(stderr, "Usage: virtio_test [--help]" + " [--no-indirect]" + " [--no-event-idx]" + "\n"); +} + +int main(int argc, char **argv) +{ + struct vdev_info dev; + unsigned long long features = (1ULL << VIRTIO_RING_F_INDIRECT_DESC) | + (1ULL << VIRTIO_RING_F_EVENT_IDX); + int o; + + for (;;) { + o = getopt_long(argc, argv, optstring, longopts, NULL); + switch (o) { + case -1: + goto done; + case '?': + help(); + exit(2); + case 'e': + features &= ~(1ULL << VIRTIO_RING_F_EVENT_IDX); + break; + case 'h': + help(); + goto done; + case 'i': + features &= ~(1ULL << VIRTIO_RING_F_INDIRECT_DESC); + break; + default: + assert(0); + break; + } + } + +done: + vdev_info_init(&dev, features); + vq_info_add(&dev, 256); + run_test(&dev, &dev.vqs[0], 0x100000); + return 0; +} |