iproute2: support latest cake & restore DSCP washing

Support new packet overhead passing paradigm in cake qdisc, also restore
DSCP wash/nowash keywords.

Signed-off-by: Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk>

1 files changed, 75 insertions, 55 deletions

diff --git a/package/network/utils/iproute2/patches/950-add-cake-to-tc.patch b/package/network/utils/iproute2/patches/950-add-cake-to-tc.patch
index ffb3e63e13..d0fc66daaa 100644
--- a/package/network/utils/iproute2/patches/950-add-cake-to-tc.patch
+++ b/package/network/utils/iproute2/patches/950-add-cake-to-tc.patch
@@ -1,6 +1,8 @@
+diff --git a/include/linux/pkt_sched.h b/include/linux/pkt_sched.h
+index 8d2530d..028b83c 100644
 --- a/include/linux/pkt_sched.h
 +++ b/include/linux/pkt_sched.h
-@@ -850,4 +850,57 @@ struct tc_pie_xstats {
+@@ -850,4 +850,59 @@ struct tc_pie_xstats {
  	__u32 maxq;             /* maximum queue size */
  	__u32 ecn_mark;         /* packets marked with ecn*/
  };
@@ -18,6 +20,8 @@
 +	TCA_CAKE_AUTORATE,
 +	TCA_CAKE_MEMORY,
 +	TCA_CAKE_NAT,
++	TCA_CAKE_ETHERNET,
++	TCA_CAKE_WASH,
 +	__TCA_CAKE_MAX
 +};
 +#define TCA_CAKE_MAX	(__TCA_CAKE_MAX - 1)
@@ -58,6 +62,8 @@
 +};
 +
  #endif
+diff --git a/tc/Makefile b/tc/Makefile
+index 56acbaa..d421b8e 100644
 --- a/tc/Makefile
 +++ b/tc/Makefile
 @@ -63,6 +63,7 @@ TCMODULES += q_codel.o
@@ -68,9 +74,12 @@
  TCMODULES += q_hhf.o
  TCMODULES += e_bpf.o
  
+diff --git a/tc/q_cake.c b/tc/q_cake.c
+new file mode 100644
+index 0000000..14e5aab
 --- /dev/null
 +++ b/tc/q_cake.c
-@@ -0,0 +1,643 @@
+@@ -0,0 +1,654 @@
 +/*
 + * Common Applications Kept Enhanced  --  CAKE
 + *
@@ -129,6 +138,7 @@
 +	                "                [ besteffort | precedence | diffserv8 | diffserv4* ]\n"
 +	                "                [ flowblind | srchost | dsthost | hosts | flows* | dual-srchost | dual-dsthost | triple-isolate ] [ nat | nonat* ]\n"
 +	                "                [ ptm | atm | noatm* ] [ overhead N | conservative | raw* ]\n"
++	                "                [ wash | nowash* ]\n"
 +	                "                [ memlimit LIMIT ]\n"
 +	                "    (* marks defaults)\n");
 +}
@@ -144,6 +154,8 @@
 +	unsigned memlimit = 0;
 +	int  overhead = 0;
 +	bool overhead_set = false;
++	bool overhead_override = false;
++	int wash = -1;
 +	int flowmode = -1;
 +	int nat = -1;
 +	int atm = -1;
@@ -213,6 +225,11 @@
 +		} else if (strcmp(*argv, "diffserv-llt") == 0) {
 +			diffserv = 5;
 +
++		} else if (strcmp(*argv, "nowash") == 0) {
++			wash = 0;
++		} else if (strcmp(*argv, "wash") == 0) {
++			wash = 1;
++
 +		} else if (strcmp(*argv, "flowblind") == 0) {
 +			flowmode = 0;
 +		} else if (strcmp(*argv, "srchost") == 0) {
@@ -246,6 +263,7 @@
 +			atm = 0;
 +			overhead = 0;
 +			overhead_set = true;
++			overhead_override = true;
 +		} else if (strcmp(*argv, "conservative") == 0) {
 +			/*
 +			 * Deliberately over-estimate overhead:
@@ -256,28 +274,6 @@
 +			overhead = 48;
 +			overhead_set = true;
 +
-+		/*
-+		 * DOCSIS overhead figures courtesy of Greg White @ CableLabs.
-+		 * The "-ip" versions include the Ethernet frame header, in case
-+		 * you are shaping an IP interface instead of an Ethernet one.
-+		 */
-+		} else if (strcmp(*argv, "docsis-downstream-ip") == 0) {
-+			atm = 0;
-+			overhead += 35;
-+			overhead_set = true;
-+		} else if (strcmp(*argv, "docsis-downstream") == 0) {
-+			atm = 0;
-+			overhead += 35 - 14;
-+			overhead_set = true;
-+		} else if (strcmp(*argv, "docsis-upstream-ip") == 0) {
-+			atm = 0;
-+			overhead += 28;
-+			overhead_set = true;
-+		} else if (strcmp(*argv, "docsis-upstream") == 0) {
-+			atm = 0;
-+			overhead += 28 - 14;
-+			overhead_set = true;
-+
 +		/* Various ADSL framing schemes, all over ATM cells */
 +		} else if (strcmp(*argv, "ipoa-vcmux") == 0) {
 +			atm = 1;
@@ -325,38 +321,38 @@
 +
 +		} else if (strcmp(*argv, "via-ethernet") == 0) {
 +			/*
-+			 * The above overheads are relative to an IP packet,
-+			 * but Linux includes Ethernet framing overhead already
-+			 * if we are shaping an Ethernet interface rather than
-+			 * an IP interface.
++			 * We used to use this flag to manually compensate for
++			 * Linux including the Ethernet header on Ethernet-type
++			 * interfaces, but not on IP-type interfaces.
++			 *
++			 * It is no longer needed, because Cake now adjusts for
++			 * that automatically, and is thus ignored.
++			 *
++			 * It would be deleted entirely, but it appears in the
++			 * stats output when the automatic compensation is active.
 +			 */
-+			overhead -= 14;
-+			overhead_set = true;
 +
-+		/* Additional Ethernet-related overheads used by some ISPs */
-+		} else if (strcmp(*argv, "ether-phy") == 0) {
-+			/* ethernet pre-amble & interframe gap 20 bytes
-+			 * Linux will have already accounted for MACs & frame type 14 bytes
-+			 * you probably want to add an FCS as well*/
-+			overhead += 20;
-+			overhead_set = true;
-+		} else if (strcmp(*argv, "ether-all") == 0) {
++		} else if (strcmp(*argv, "ethernet") == 0) {
 +			/* ethernet pre-amble & interframe gap & FCS
-+			 * Linux will have already accounted for MACs & frame type 14 bytes
-+			 * you may need to add vlan tag*/
-+			overhead += 24;
++			 * you may need to add vlan tag */
++			overhead += 38;
 +			overhead_set = true;
 +
-+		} else if (strcmp(*argv, "ether-fcs") == 0) {
-+			/* Frame Check Sequence */
-+			/* we ignore the minimum frame size, because IP packets usually meet it */
-+			overhead += 4;
-+			overhead_set = true;
++		/* Additional Ethernet-related overhead used by some ISPs */
 +		} else if (strcmp(*argv, "ether-vlan") == 0) {
 +			/* 802.1q VLAN tag - may be repeated */
 +			overhead += 4;
 +			overhead_set = true;
 +
++		/*
++		 * DOCSIS cable shapers account for Ethernet frame with FCS,
++		 * but not interframe gap nor preamble.
++		 */
++		} else if (strcmp(*argv, "docsis") == 0) {
++			atm = 0;
++			overhead += 18;
++			overhead_set = true;
++
 +		} else if (strcmp(*argv, "overhead") == 0) {
 +			char* p = NULL;
 +			NEXT_ARG();
@@ -397,6 +393,10 @@
 +		addattr_l(n, 1024, TCA_CAKE_FLOW_MODE, &flowmode, sizeof(flowmode));
 +	if (overhead_set)
 +		addattr_l(n, 1024, TCA_CAKE_OVERHEAD, &overhead, sizeof(overhead));
++	if (overhead_override) {
++		unsigned zero = 0;
++		addattr_l(n, 1024, TCA_CAKE_ETHERNET, &zero, sizeof(zero));
++	}
 +	if (interval)
 +		addattr_l(n, 1024, TCA_CAKE_RTT, &interval, sizeof(interval));
 +	if (target)
@@ -407,6 +407,8 @@
 +		addattr_l(n, 1024, TCA_CAKE_MEMORY, &memlimit, sizeof(memlimit));
 +	if (nat != -1)
 +		addattr_l(n, 1024, TCA_CAKE_NAT, &nat, sizeof(nat));
++	if (wash != -1)
++		addattr_l(n, 1024, TCA_CAKE_WASH, &wash, sizeof(wash));
 +
 +	tail->rta_len = (void *) NLMSG_TAIL(n) - (void *) tail;
 +	return 0;
@@ -422,9 +424,11 @@
 +	unsigned interval = 0;
 +	unsigned memlimit = 0;
 +	int overhead = 0;
++	int ethernet = 0;
 +	int atm = 0;
 +	int nat = 0;
 +	int autorate = 0;
++	int wash = 0;
 +	SPRINT_BUF(b1);
 +	SPRINT_BUF(b2);
 +
@@ -511,6 +515,10 @@
 +		if(nat)
 +			fprintf(f, "nat ");
 +	}
++	if (tb[TCA_CAKE_WASH] &&
++	    RTA_PAYLOAD(tb[TCA_CAKE_WASH]) >= sizeof(__u32)) {
++		wash = rta_getattr_u32(tb[TCA_CAKE_WASH]);
++	}
 +	if (tb[TCA_CAKE_ATM] &&
 +	    RTA_PAYLOAD(tb[TCA_CAKE_ATM]) >= sizeof(__u32)) {
 +		atm = rta_getattr_u32(tb[TCA_CAKE_ATM]);
@@ -519,26 +527,38 @@
 +	    RTA_PAYLOAD(tb[TCA_CAKE_OVERHEAD]) >= sizeof(__u32)) {
 +		overhead = rta_getattr_u32(tb[TCA_CAKE_OVERHEAD]);
 +	}
++	if (tb[TCA_CAKE_ETHERNET] &&
++	    RTA_PAYLOAD(tb[TCA_CAKE_ETHERNET]) >= sizeof(__u32)) {
++		ethernet = rta_getattr_u32(tb[TCA_CAKE_ETHERNET]);
++	}
 +	if (tb[TCA_CAKE_RTT] &&
 +	    RTA_PAYLOAD(tb[TCA_CAKE_RTT]) >= sizeof(__u32)) {
 +		interval = rta_getattr_u32(tb[TCA_CAKE_RTT]);
 +	}
 +
++	if (wash)
++		fprintf(f,"wash ");
++
 +	if (interval)
 +		fprintf(f, "rtt %s ", sprint_time(interval, b2));
 +
-+	if (atm == 1)
-+		fprintf(f, "atm ");
-+	else if (atm == 2)
-+		fprintf(f, "ptm ");
-+	else if (overhead)
-+		fprintf(f, "noatm ");
++	if (!atm && overhead == ethernet) {
++		fprintf(f, "raw ");
++	} else {
++		if (atm == 1)
++			fprintf(f, "atm ");
++		else if (atm == 2)
++			fprintf(f, "ptm ");
++		else
++			fprintf(f, "noatm ");
 +
-+	if (overhead || atm)
 +		fprintf(f, "overhead %d ", overhead);
 +
-+	if (!atm && !overhead)
-+		fprintf(f, "raw ");
++		// This is actually the *amount* of automatic compensation, but we only report
++		// its presence as a boolean for now.
++		if (ethernet)
++			fprintf(f, "via-ethernet ");
++	}
 +
 +	if (memlimit)
 +		fprintf(f, "memlimit %s", sprint_size(memlimit, b1));