aboutsummaryrefslogtreecommitdiffstats
path: root/frontends/verilog
diff options
context:
space:
mode:
Diffstat (limited to 'frontends/verilog')
0 files changed, 0 insertions, 0 deletions
'>55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762
/* -*-  Mode:C; c-basic-offset:4; tab-width:4 -*-
 ****************************************************************************
 * (c) 2004 - Rolf Neugebauer - Intel Research Cambridge
 * (c) 2004 - Keir Fraser - University of Cambridge
 ****************************************************************************
 * 
 * Description: allows a domain to access devices on the PCI bus
 *
 * A guest OS may be given access to particular devices on the PCI bus.
 * For each domain a list of PCI devices is maintained, describing the
 * access mode for the domain. 
 *
 * Guests can figure out the virtualised PCI space through normal PCI config
 * register access. Some of the accesses, in particular write accesses, are
 * faked. For example the sequence for detecting the IO regions, which requires
 * writes to determine the size of the region, is faked out by a very simple
 * state machine, preventing direct writes to the PCI config registers by a
 * guest.
 */

#include <xen/config.h>
#include <xen/lib.h>
#include <xen/types.h>
#include <xen/sched.h>
#include <xen/pci.h>
#include <xen/irq.h>
#include <xen/event.h>
#include <asm/pci.h>
#include <public/xen.h>
#include <public/physdev.h>

/* Called by PHYSDEV_PCI_INITIALISE_DEVICE to finalise IRQ routing. */
extern void pcibios_enable_irq(struct pci_dev *dev);

#if 0
#define VERBOSE_INFO(_f, _a...) printk( _f , ## _a )
#else
#define VERBOSE_INFO(_f, _a...) ((void)0)
#endif

#ifdef VERBOSE
#define INFO(_f, _a...) printk( _f, ## _a )
#else
#define INFO(_f, _a...) ((void)0)
#endif


#define ACC_READ  1
#define ACC_WRITE 2

/* Upper bounds for PCI-device addressing. */
#define PCI_BUSMAX  255
#define PCI_DEVMAX   31
#define PCI_FUNCMAX   7
#define PCI_REGMAX  255

/* Bit offsets into state. */
#define ST_BASE_ADDRESS  0   /* bits 0-5: are for base address access */
#define ST_ROM_ADDRESS   6   /* bit 6: is for rom address access */    

typedef struct _phys_dev_st {
    int flags;                       /* flags for access etc */
    struct pci_dev *dev;             /* the device */
    struct list_head node;           /* link to the list */
    struct domain *owner;       /* 'owner of this device' */
    int state;                       /* state for various checks */
} phys_dev_t;


/* Find a device on a per-domain device list. */
static phys_dev_t *find_pdev(struct domain *p, struct pci_dev *dev)
{
    phys_dev_t *t, *res = NULL;
    struct list_head *tmp;

    list_for_each(tmp, &p->pcidev_list)
    {
        t = list_entry(tmp,  phys_dev_t, node);
        if ( dev == t->dev )
        {
            res = t;
            break;
        }
    }
    return res;
}

/* Add a device to a per-domain device-access list. */
static void add_dev_to_task(struct domain *p, 
                            struct pci_dev *dev, int acc)
{
    phys_dev_t *pdev;
    
    if ( (pdev = find_pdev(p, dev)) )
    {
        /* Sevice already on list: update access permissions. */
        pdev->flags = acc;
        return;
    }

    if ( (pdev = xmalloc(sizeof(phys_dev_t))) == NULL )
    {
        INFO("Error allocating pdev structure.\n");
        return;
    }
    
    pdev->dev = dev;
    pdev->flags = acc;
    pdev->state = 0;
    list_add(&pdev->node, &p->pcidev_list);

    if ( acc == ACC_WRITE )
        pdev->owner = p;
}

/*
 * physdev_pci_access_modify:
 * Allow/disallow access to a specific PCI device.  Guests should not be
 * allowed to see bridge devices as it needlessly complicates things (one
 * possible exception to this is the AGP bridge).  If the given device is a
 * bridge, then the domain should get access to all the leaf devices below
 * that bridge (XXX this is unimplemented!).
 */
int physdev_pci_access_modify(
    domid_t dom, int bus, int dev, int func, int enable)
{
    struct domain *p;
    struct pci_dev *pdev;
    int i, j, rc = 0;
 
    if ( !IS_PRIV(current) )
        BUG();

    if ( (bus > PCI_BUSMAX) || (dev > PCI_DEVMAX) || (func > PCI_FUNCMAX) )
        return -EINVAL;

    if ( !enable )
    {
        INFO("Disallowing access is not yet supported.\n");
        return -EINVAL;
    }

    INFO("physdev_pci_access_modify: %02x:%02x:%02x\n", bus, dev, func);

    if ( (p = find_domain_by_id(dom)) == NULL ) 
        return -ESRCH;

    /* Make the domain privileged. */
    set_bit(DF_PHYSDEV, &p->flags);
	/* FIXME: MAW for now make the domain REALLY privileged so that it
	 * can run a backend driver (hw access should work OK otherwise) */
	set_bit(DF_PRIVILEGED, &p->flags);

    /* Grant write access to the specified device. */
    if ( (pdev = pci_find_slot(bus, PCI_DEVFN(dev, func))) == NULL )
    {
        INFO("  dev does not exist\n");
        rc = -ENODEV;
        goto out;
    }
    add_dev_to_task(p, pdev, ACC_WRITE);

    INFO("  add RW %02x:%02x:%02x\n", pdev->bus->number,
         PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn));

    /* Is the device a bridge or cardbus? */
    if ( pdev->hdr_type != PCI_HEADER_TYPE_NORMAL )
        INFO("XXX can't give access to bridge devices yet\n");

    /* Now, setup access to the IO ports and memory regions for the device. */

    if ( p->io_bitmap == NULL )
    {
        if ( (p->io_bitmap = xmalloc(IO_BITMAP_BYTES)) == NULL )
        {
            rc = -ENOMEM;
            goto out;
        }
        memset(p->io_bitmap, 0xFF, IO_BITMAP_BYTES);

        p->io_bitmap_sel = ~0ULL;
    }

    for ( i = 0; i < DEVICE_COUNT_RESOURCE; i++ )
    {
        struct resource *r = &pdev->resource[i];
        
        if ( r->flags & IORESOURCE_IO )
        {
            /* Give the domain access to the IO ports it needs.  Currently,
             * this will allow all processes in that domain access to those
             * ports as well.  This will do for now, since driver domains don't
             * run untrusted processes! */
            INFO("Giving domain %u IO resources (%lx - %lx) "
                 "for device %s\n", dom, r->start, r->end, pdev->slot_name);
            for ( j = r->start; j < r->end + 1; j++ )
            {
                clear_bit(j, p->io_bitmap);
                /* Record that we cleared a bit using bit n of the selector:
                 * n = (j / (4 bytes in a word * 8 bits in a byte))
                 *     / number of words per selector bit
                 */
                clear_bit((j / (8 * 4)) / IOBMP_SELBIT_LWORDS,
                          &p->io_bitmap_sel);
            }
        }

        /* rights to IO memory regions are checked when the domain maps them */
    }
 out:
    put_domain(p);
    return rc;
}

/* Check if a domain controls a device with IO memory within frame @pfn.
 * Returns: 1 if the domain should be allowed to map @pfn, 0 otherwise.  */
int domain_iomem_in_pfn(struct domain *p, unsigned long pfn)
{
    int ret = 0;
    struct list_head *l;

    VERBOSE_INFO("Checking if physdev-capable domain %u needs access to "
                 "pfn %08lx\n", p->id, pfn);
    
    spin_lock(&p->pcidev_lock);

    list_for_each(l, &p->pcidev_list)
    {
        int i;
        phys_dev_t *phys_dev = list_entry(l, phys_dev_t, node);
        struct pci_dev *pci_dev = phys_dev->dev;

        for ( i = 0; (i < DEVICE_COUNT_RESOURCE) && (ret == 0); i++ )
        {
            struct resource *r = &pci_dev->resource[i];
            
            if ( r->flags & IORESOURCE_MEM )
                if ( (r->start >> PAGE_SHIFT) == pfn
                     || (r->end >> PAGE_SHIFT) == pfn
                     || ((r->start >> PAGE_SHIFT < pfn)
                         && (r->end >> PAGE_SHIFT > pfn)) )
                    ret = 1;
        }

        if ( ret != 0 ) break;
    }
    
    spin_unlock(&p->pcidev_lock);

    VERBOSE_INFO("Domain %u %s mapping of pfn %08lx\n",
                 p->id, ret ? "allowed" : "disallowed", pfn);

    return ret;
}

/* check if a domain has general access to a device */
inline static int check_dev_acc (struct domain *p,
                                 int bus, int dev, int func,
                                 phys_dev_t **pdev) 
{
    struct pci_dev *target_dev;
    phys_dev_t     *target_pdev;
    unsigned int    target_devfn;

    *pdev = NULL;

     if ( !IS_CAPABLE_PHYSDEV(p) )
         return -EPERM; /* no pci access permission */

    if ( bus > PCI_BUSMAX || dev > PCI_DEVMAX || func > PCI_FUNCMAX )
        return -EINVAL;

    VERBOSE_INFO("b=%x d=%x f=%x ", bus, dev, func);

    /* check target device */
    target_devfn = PCI_DEVFN(dev, func);
    target_dev   = pci_find_slot(bus, target_devfn);
    if ( !target_dev )
    {
        VERBOSE_INFO("target does not exist\n");
        return -ENODEV;
    }

    /* check access */
    target_pdev = find_pdev(p, target_dev);
    if ( !target_pdev )
    {
        VERBOSE_INFO("dom has no access to target\n");
        return -EPERM;
    }

    *pdev = target_pdev;
    return 0;
}


/*
 * Base address registers contain the base address for IO regions.
 * The length can be determined by writing all 1s to the register and
 * reading the value again. The device will zero the lower unused bits.
 * 
 * to work out the length of the io region a device probe typically does:
 * 1) a = read_base_addr_reg()
 * 2) write_base_addr_reg(0xffffffff)
 * 3) b = read_base_addr_reg()  [device zeros lower bits]
 * 4) write_base_addr_reg(a)    [restore original value]
 * this function fakes out step 2-4. *no* writes are made to the device.
 * 
 * phys_dev_t contains a bit field (a bit for each base address register).
 * if the bit for a register is set the guest had writen all 1s to the 
 * register and subsequent read request need to fake out the b.
 * if the guest restores the original value (step 4 above) the bit is
 * cleared again. If the guest attempts to "restores" a wrong value an
 * error is flagged.
 */
static int do_base_address_access(phys_dev_t *pdev, int acc, int idx, 
                                  int len, u32 *val)
{
    int st_bit, reg = PCI_BASE_ADDRESS_0 + (idx*4), ret = -EINVAL;
    struct pci_dev *dev = pdev->dev;
    u32 orig_val, sz;
    struct resource *res;

    if ( len != sizeof(u32) )
    {
        /* This isn't illegal, but there doesn't seem to be a very good reason
         * to do it for normal devices (bridges are another matter).  Since it
         * would complicate the code below, we don't support this for now. */

        /* We could set *val to some value but the guest may well be in trouble
         * anyway if this write fails.  Hopefully the printk will give us a
         * clue what went wrong. */
        INFO("Guest %u attempting sub-dword %s to BASE_ADDRESS %d\n",
             pdev->owner->id, (acc == ACC_READ) ? "read" : "write", idx);
        
        return -EPERM;
    }

    st_bit = idx + ST_BASE_ADDRESS;
    res    = &(pdev->dev->resource[idx]);

    if ( acc == ACC_WRITE )
    {
        if ( (*val == 0xffffffff) || 
             ((res->flags & IORESOURCE_IO) && (*val == 0xffff)) )
        {
            /* Set bit and return. */
            set_bit(st_bit, &pdev->state);
            ret = 0;
        }
        else
        {
            /* Assume guest wants to set the base address. */
            clear_bit(st_bit, &pdev->state);

            /* check if guest tries to restore orig value */
            ret = pci_read_config_dword(dev, reg, &orig_val);
            if ( (ret == 0) && (*val != orig_val) ) 
            {
                INFO("Guest attempting update to BASE_ADDRESS %d\n", idx);
                ret = -EPERM;
            }
        }
        VERBOSE_INFO("fixed pci write: %02x:%02x:%02x reg=0x%02x len=0x%02x"
                     " val=0x%08x %x\n", 
                     dev->bus->number, PCI_SLOT(dev->devfn), 
                     PCI_FUNC(dev->devfn), reg, len, *val, pdev->state);
    }
    else if ( acc == ACC_READ )
    {
        ret = pci_read_config_dword(dev, reg, val);
        if ( (ret == 0) && test_bit(st_bit, &pdev->state) )
        {
            /* Cook the value. */
            sz  = res->end - res->start;
            if ( res->flags & IORESOURCE_MEM )
            {
                /* this is written out explicitly for clarity */
                *val = 0xffffffff;
                /* bit    0 = 0 */
                /* bit  21  = memory type */
                /* bit 3    = prefetchable */
                /* bit 4-31 width */
                sz   = sz >> 4; /* size in blocks of 16 byte */
                sz   = ~sz;     /* invert */
                *val = *val & (sz << 4); /* and in the size */
                /* use read values for low 4 bits */
                *val = *val | (orig_val & 0xf);
            }
            else if ( res->flags & IORESOURCE_IO )
            {
                *val = 0x0000ffff;
                /* bit 10 = 01 */
                /* bit 2-31 width */
                sz   = sz >> 2; /* size in dwords */
                sz   = ~sz & 0x0000ffff;
                *val = *val & (sz << 2);
                *val = *val | 0x1;
            }
        }
        VERBOSE_INFO("fixed pci read: %02x:%02x:%02x reg=0x%02x len=0x%02x"
                     " val=0x%08x %x\n", 
                     dev->bus->number, PCI_SLOT(dev->devfn), 
                     PCI_FUNC(dev->devfn), reg, len, *val, pdev->state);
    }

    return ret;
}


static int do_rom_address_access(phys_dev_t *pdev, int acc, int len, u32 *val)
{
    int st_bit, ret = -EINVAL;
    struct pci_dev *dev = pdev->dev;
    u32 orig_val, sz;
    struct resource *res;

    if ( len != sizeof(u32) )
    {
        INFO("Guest attempting sub-dword %s to ROM_ADDRESS\n", 
             (acc == ACC_READ) ? "read" : "write");
        return -EPERM;
    }

    st_bit = ST_ROM_ADDRESS;
    res = &(pdev->dev->resource[PCI_ROM_RESOURCE]);

    if ( acc == ACC_WRITE )
    {
        if ( (*val == 0xffffffff) || (*val == 0xfffffffe) )
        {
            /* NB. 0xffffffff would be unusual, but we trap it anyway. */
            set_bit(st_bit, &pdev->state);
            ret = 0;
        }
        else
        {
            /* Assume guest wants simply to set the base address. */
            clear_bit(st_bit, &pdev->state);
            
            /* Check if guest tries to restore the original value. */
            ret = pci_read_config_dword(dev, PCI_ROM_ADDRESS, &orig_val);
            if ( (ret == 0) && (*val != orig_val) ) 
            {
                if ( (*val != 0x00000000) )
                {
                    INFO("caution: guest tried to change rom address.\n");
                    ret = -EPERM;
                }
                else
                {
                    INFO("guest disabled rom access for %02x:%02x:%02x\n",
                         dev->bus->number, PCI_SLOT(dev->devfn), 
                         PCI_FUNC(dev->devfn));
                }
            }
        }
        VERBOSE_INFO("fixed pci write: %02x:%02x:%02x reg=0x%02x len=0x%02x"
                     " val=0x%08x %x\n", 
                     dev->bus->number, PCI_SLOT(dev->devfn), 
                     PCI_FUNC(dev->devfn), PCI_ROM_ADDRESS, len, *val, pdev->state);
    }
    else if ( acc == ACC_READ )
    {
        ret = pci_read_config_dword(dev, PCI_ROM_ADDRESS, val);
        if ( (ret == 0) && test_bit(st_bit, &pdev->state) )
        {
            /* Cook the value. */
            sz  = res->end - res->start;
            *val = 0xffffffff;
            /* leave bit 0 untouched */
            /* bit 1-10 reserved, harwired to 0 */
            sz = sz >> 11; /* size is in 2KB blocks */
            sz = ~sz;
            *val = *val & (sz << 11);
            *val = *val | (orig_val & 0x1);
        }
        VERBOSE_INFO("fixed pci read: %02x:%02x:%02x reg=0x%02x len=0x%02x"
                     " val=0x%08x %x\n", 
                     dev->bus->number, PCI_SLOT(dev->devfn), 
                     PCI_FUNC(dev->devfn), PCI_ROM_ADDRESS, len, *val, pdev->state);
    }

    return ret;

}

/*
 * Handle a PCI config space read access if the domain has access privileges.
 */
static long pci_cfgreg_read(int bus, int dev, int func, int reg,
                            int len, u32 *val)
{
    int ret;
    phys_dev_t *pdev;

    if ( (ret = check_dev_acc(current, bus, dev, func, &pdev)) != 0 )
    {
        /* PCI spec states that reads from non-existent devices should return
         * all 1s.  In this case the domain has no read access, which should
         * also look like the device is non-existent. */
        *val = 0xFFFFFFFF;
        return ret; /* KAF: error return seems to matter on my test machine. */
    }

    /* Fake out read requests for some registers. */
    switch ( reg )
    {
    case PCI_BASE_ADDRESS_0:
        ret = do_base_address_access(pdev, ACC_READ, 0, len, val);
        break;

    case PCI_BASE_ADDRESS_1:
        ret = do_base_address_access(pdev, ACC_READ, 1, len, val);
        break;

    case PCI_BASE_ADDRESS_2:
        ret = do_base_address_access(pdev, ACC_READ, 2, len, val);
        break;

    case PCI_BASE_ADDRESS_3:
        ret = do_base_address_access(pdev, ACC_READ, 3, len, val);
        break;

    case PCI_BASE_ADDRESS_4:
        ret = do_base_address_access(pdev, ACC_READ, 4, len, val);
        break;

    case PCI_BASE_ADDRESS_5:
        ret = do_base_address_access(pdev, ACC_READ, 5, len, val);
        break;

    case PCI_ROM_ADDRESS:
        ret = do_rom_address_access(pdev, ACC_READ, len, val);
        break;        

    case PCI_INTERRUPT_LINE:
        *val = pdev->dev->irq;
        ret = 0;
        break;

    default:
        ret = pci_config_read(0, bus, dev, func, reg, len, val);        
        VERBOSE_INFO("pci read : %02x:%02x:%02x reg=0x%02x len=0x%02x "
                     "val=0x%08x\n", bus, dev, func, reg, len, *val);
        break;
    }

    return ret;
}


/*
 * Handle a PCI config space write access if the domain has access privileges.
 */
static long pci_cfgreg_write(int bus, int dev, int func, int reg,
                             int len, u32 val)
{
    int ret;
    phys_dev_t *pdev;

    if ( (ret = check_dev_acc(current, bus, dev, func, &pdev)) != 0 )
        return ret;

    /* special treatment for some registers */
    switch (reg)
    {
    case PCI_BASE_ADDRESS_0:
        ret = do_base_address_access(pdev, ACC_WRITE, 0, len, &val);
        break;

    case PCI_BASE_ADDRESS_1:
        ret = do_base_address_access(pdev, ACC_WRITE, 1, len, &val);
        break;

    case PCI_BASE_ADDRESS_2:
        ret = do_base_address_access(pdev, ACC_WRITE, 2, len, &val);
        break;

    case PCI_BASE_ADDRESS_3:
        ret = do_base_address_access(pdev, ACC_WRITE, 3, len, &val);
        break;

    case PCI_BASE_ADDRESS_4:
        ret = do_base_address_access(pdev, ACC_WRITE, 4, len, &val);
        break;

    case PCI_BASE_ADDRESS_5:
        ret = do_base_address_access(pdev, ACC_WRITE, 5, len, &val);
        break;

    case PCI_ROM_ADDRESS:
        ret = do_rom_address_access(pdev, ACC_WRITE, len, &val);
        break;        

    default:
        if ( pdev->flags != ACC_WRITE ) 
        {
            INFO("pci write not allowed %02x:%02x:%02x: "
                 "reg=0x%02x len=0x%02x val=0x%08x\n",
                 bus, dev, func, reg, len, val);
            ret = -EPERM;
        }
        else
        {
            ret = pci_config_write(0, bus, dev, func, reg, len, val);
            VERBOSE_INFO("pci write: %02x:%02x:%02x reg=0x%02x len=0x%02x "
                         "val=0x%08x\n", bus, dev, func, reg, len, val);
        }
        break;
    }

    return ret;
}


static long pci_probe_root_buses(u32 *busmask)
{
    phys_dev_t *pdev;
    struct list_head *tmp;

    memset(busmask, 0, 256/8);

    list_for_each ( tmp, &current->pcidev_list )
    {
        pdev = list_entry(tmp, phys_dev_t, node);
        set_bit(pdev->dev->bus->number, busmask);
    }

    return 0;
}


/*
 * Demuxing hypercall.
 */
long do_physdev_op(physdev_op_t *uop)
{
    phys_dev_t  *pdev;
    physdev_op_t op;
    long         ret;
    int          irq;

    if ( unlikely(copy_from_user(&op, uop, sizeof(op)) != 0) )
        return -EFAULT;

    switch ( op.cmd )
    {
    case PHYSDEVOP_PCI_CFGREG_READ:
        ret = pci_cfgreg_read(op.u.pci_cfgreg_read.bus,
                              op.u.pci_cfgreg_read.dev, 
                              op.u.pci_cfgreg_read.func,
                              op.u.pci_cfgreg_read.reg, 
                              op.u.pci_cfgreg_read.len,
                              &op.u.pci_cfgreg_read.value);
        break;

    case PHYSDEVOP_PCI_CFGREG_WRITE:
        ret = pci_cfgreg_write(op.u.pci_cfgreg_write.bus,
                               op.u.pci_cfgreg_write.dev, 
                               op.u.pci_cfgreg_write.func,
                               op.u.pci_cfgreg_write.reg, 
                               op.u.pci_cfgreg_write.len,
                               op.u.pci_cfgreg_write.value);
        break;

    case PHYSDEVOP_PCI_INITIALISE_DEVICE:
        if ( (ret = check_dev_acc(current, 
                                  op.u.pci_initialise_device.bus, 
                                  op.u.pci_initialise_device.dev, 
                                  op.u.pci_initialise_device.func, 
                                  &pdev)) == 0 )
            pcibios_enable_irq(pdev->dev);
        break;

    case PHYSDEVOP_PCI_PROBE_ROOT_BUSES:
        ret = pci_probe_root_buses(op.u.pci_probe_root_buses.busmask);
        break;

    case PHYSDEVOP_IRQ_UNMASK_NOTIFY:
        ret = pirq_guest_unmask(current);
        break;

    case PHYSDEVOP_IRQ_STATUS_QUERY:
        irq = op.u.irq_status_query.irq;
        ret = -EINVAL;
        if ( (irq < 0) || (irq >= NR_IRQS) )
            break;
        op.u.irq_status_query.flags = 0;
        /* Edge-triggered interrupts don't need an explicit unmask downcall. */
        if ( strstr(irq_desc[irq].handler->typename, "edge") == NULL )
            op.u.irq_status_query.flags |= PHYSDEVOP_IRQ_NEEDS_UNMASK_NOTIFY;
        ret = 0;
        break;

    default:
        ret = -EINVAL;
        break;
    }

    copy_to_user(uop, &op, sizeof(op));
    return ret;
}

/* Test if boot params specify this device should NOT be visible to DOM0
 * (e.g. so that another domain can control it instead) */
int pcidev_dom0_hidden(struct pci_dev *dev)
{
    extern char opt_physdev_dom0_hide[];
    char cmp[10] = "(.......)";
    
    strncpy(&cmp[1], dev->slot_name, 7);

    if ( strstr(opt_physdev_dom0_hide, dev->slot_name) == NULL )
        return 0;
    
    return 1;
}


/* Domain 0 has read access to all devices. */
void physdev_init_dom0(struct domain *p)
{
    struct pci_dev *dev;
    phys_dev_t *pdev;

    INFO("Give DOM0 read access to all PCI devices\n");

    pci_for_each_dev(dev)
    {
        if ( pcidev_dom0_hidden(dev) )
        {            
            printk("Hiding PCI device %s from DOM0\n", dev->slot_name);
            continue;
        }

        /* Skip bridges and other peculiarities for now.
         *
         * Note that this can prevent the guest from detecting devices
         * with fn>0 on slots where the fn=0 device is a bridge.  We
         * can identify such slots by looking at the multifunction bit
         * (top bit of hdr_type, masked out in dev->hdr_type).
         *
         * In Linux2.4 we find all devices because the detection code
         * scans all functions if the read of the fn=0 device's header
         * type fails.
         *
         * In Linux2.6 we set pcibios_scan_all_fns().
         */
        if ( dev->hdr_type != PCI_HEADER_TYPE_NORMAL )
            continue;
        pdev = xmalloc(sizeof(phys_dev_t));
        pdev->dev = dev;
        pdev->flags = ACC_WRITE;
        pdev->state = 0;
        pdev->owner = p;
        list_add(&pdev->node, &p->pcidev_list);
    }

    set_bit(DF_PHYSDEV, &p->flags);
}