pci.c 56.1 KB
Newer Older
1
/*
E
Eric Blake 已提交
2
 * Copyright (C) 2009-2011 Red Hat, Inc.
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library 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
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
 *
 * Authors:
 *     Mark McLoughlin <markmc@redhat.com>
 */

#include <config.h>

#include "pci.h"

#include <dirent.h>
#include <fcntl.h>
#include <inttypes.h>
#include <limits.h>
#include <stdio.h>
#include <string.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <unistd.h>
35
#include <stdlib.h>
36 37 38

#include "logging.h"
#include "memory.h"
39
#include "command.h"
40
#include "virterror_internal.h"
E
Eric Blake 已提交
41
#include "virfile.h"
42

43 44
/* avoid compilation breakage on some systems */
#ifndef MODPROBE
45
# define MODPROBE "modprobe"
46 47
#endif

48 49 50 51
#define PCI_SYSFS "/sys/bus/pci/"
#define PCI_ID_LEN 10   /* "XXXX XXXX" */
#define PCI_ADDR_LEN 13 /* "XXXX:XX:XX.X" */

52 53 54 55
#define SRIOV_FOUND 0
#define SRIOV_NOT_FOUND 1
#define SRIOV_ERROR -1

56 57 58 59 60 61 62 63
struct _pciDevice {
    unsigned      domain;
    unsigned      bus;
    unsigned      slot;
    unsigned      function;

    char          name[PCI_ADDR_LEN]; /* domain:bus:slot.function */
    char          id[PCI_ID_LEN];     /* product vendor */
E
Eric Blake 已提交
64
    char          *path;
65 66 67 68 69 70 71
    int           fd;

    unsigned      initted;
    unsigned      pcie_cap_pos;
    unsigned      pci_pm_cap_pos;
    unsigned      has_flr : 1;
    unsigned      has_pm_reset : 1;
72
    unsigned      managed : 1;
73 74 75 76 77

    /* used by reattach function */
    unsigned      unbind_from_stub : 1;
    unsigned      remove_slot : 1;
    unsigned      reprobe : 1;
78 79
};

80 81 82 83 84 85
struct _pciDeviceList {
    unsigned count;
    pciDevice **devs;
};


86 87 88
/* For virReportOOMError()  and virReportSystemError() */
#define VIR_FROM_THIS VIR_FROM_NONE

89
#define pciReportError(code, ...)                              \
90
    virReportErrorHelper(VIR_FROM_NONE, code, __FILE__,        \
91
                         __FUNCTION__, __LINE__, __VA_ARGS__)
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106

/* Specifications referenced in comments:
 *  PCI30  - PCI Local Bus Specification 3.0
 *  PCIe20 - PCI Express Base Specification 2.0
 *  BR12   - PCI-to-PCI Bridge Architecture Specification 1.2
 *  PM12   - PCI Bus Power Management Interface Specification 1.2
 *  ECN_AF - Advanced Capabilities for Conventional PCI ECN
 */

/* Type 0 config space header length; PCI30 Section 6.1 Configuration Space Organization */
#define PCI_CONF_LEN            0x100
#define PCI_CONF_HEADER_LEN     0x40

/* PCI30 6.2.1 */
#define PCI_HEADER_TYPE         0x0e    /* Header type */
107 108 109
#define PCI_HEADER_TYPE_BRIDGE 0x1
#define PCI_HEADER_TYPE_MASK   0x7f
#define PCI_HEADER_TYPE_MULTI  0x80
110 111 112 113 114 115 116 117 118

/* PCI30 6.2.1  Device Identification */
#define PCI_CLASS_DEVICE        0x0a    /* Device class */

/* Class Code for bridge; PCI30 D.7  Base Class 06h */
#define PCI_CLASS_BRIDGE_PCI    0x0604

/* PCI30 6.2.3  Device Status */
#define PCI_STATUS              0x06    /* 16 bits */
119
#define PCI_STATUS_CAP_LIST    0x10    /* Support Capability List */
120 121 122 123 124 125 126 127 128 129 130 131 132

/* PCI30 6.7  Capabilities List */
#define PCI_CAPABILITY_LIST     0x34    /* Offset of first capability list entry */

/* PM12 3.2.1  Capability Identifier */
#define PCI_CAP_ID_PM           0x01    /* Power Management */
/* PCI30 H Capability IDs */
#define PCI_CAP_ID_EXP          0x10    /* PCI Express */
/* ECN_AF 6.x.1.1  Capability ID for AF */
#define PCI_CAP_ID_AF           0x13    /* Advanced Features */

/* PCIe20 7.8.3  Device Capabilities Register (Offset 04h) */
#define PCI_EXP_DEVCAP          0x4     /* Device capabilities */
133
#define PCI_EXP_DEVCAP_FLR     (1<<28) /* Function Level Reset */
134 135 136 137 138 139 140

/* Header type 1 BR12 3.2 PCI-to-PCI Bridge Configuration Space Header Format */
#define PCI_PRIMARY_BUS         0x18    /* BR12 3.2.5.2 Primary bus number */
#define PCI_SECONDARY_BUS       0x19    /* BR12 3.2.5.3 Secondary bus number */
#define PCI_SUBORDINATE_BUS     0x1a    /* BR12 3.2.5.4 Highest bus number behind the bridge */
#define PCI_BRIDGE_CONTROL      0x3e
/* BR12 3.2.5.18  Bridge Control Register */
141
#define PCI_BRIDGE_CTL_RESET   0x40    /* Secondary bus reset */
142 143 144

/* PM12 3.2.4  Power Management Control/Status (Offset = 4) */
#define PCI_PM_CTRL                4    /* PM control and status register */
145 146 147 148
#define PCI_PM_CTRL_STATE_MASK    0x3  /* Current power state (D0 to D3) */
#define PCI_PM_CTRL_STATE_D0      0x0  /* D0 state */
#define PCI_PM_CTRL_STATE_D3hot   0x3  /* D3 state */
#define PCI_PM_CTRL_NO_SOFT_RESET 0x8  /* No reset for D3hot->D0 */
149 150 151

/* ECN_AF 6.x.1  Advanced Features Capability Structure */
#define PCI_AF_CAP              0x3     /* Advanced features capabilities */
152
#define PCI_AF_CAP_FLR         0x2     /* Function Level Reset */
153

J
Jiri Denemark 已提交
154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175
#define PCI_EXP_FLAGS           0x2
#define PCI_EXP_FLAGS_TYPE      0x00f0
#define PCI_EXP_TYPE_DOWNSTREAM 0x6

#define PCI_EXT_CAP_BASE          0x100
#define PCI_EXT_CAP_LIMIT         0x1000
#define PCI_EXT_CAP_ID_MASK       0x0000ffff
#define PCI_EXT_CAP_OFFSET_SHIFT  20
#define PCI_EXT_CAP_OFFSET_MASK   0x00000ffc

#define PCI_EXT_CAP_ID_ACS      0x000d
#define PCI_EXT_ACS_CTRL        0x06

#define PCI_EXT_CAP_ACS_SV      0x01
#define PCI_EXT_CAP_ACS_RR      0x04
#define PCI_EXT_CAP_ACS_CR      0x08
#define PCI_EXT_CAP_ACS_UF      0x10
#define PCI_EXT_CAP_ACS_ENABLED (PCI_EXT_CAP_ACS_SV | \
                                 PCI_EXT_CAP_ACS_RR | \
                                 PCI_EXT_CAP_ACS_CR | \
                                 PCI_EXT_CAP_ACS_UF)

176 177 178 179 180 181 182 183 184 185 186
static int
pciOpenConfig(pciDevice *dev)
{
    int fd;

    if (dev->fd > 0)
        return 0;

    fd = open(dev->path, O_RDWR);
    if (fd < 0) {
        char ebuf[1024];
187
        VIR_WARN("Failed to open config space file '%s': %s",
188 189 190 191 192 193 194 195
                 dev->path, virStrerror(errno, ebuf, sizeof(ebuf)));
        return -1;
    }
    VIR_DEBUG("%s %s: opened %s", dev->id, dev->name, dev->path);
    dev->fd = fd;
    return 0;
}

196 197 198 199 200 201
static void
pciCloseConfig(pciDevice *dev)
{
    if (!dev)
        return;

202
    VIR_FORCE_CLOSE(dev->fd);
203 204
}

205 206 207 208 209 210 211 212
static int
pciRead(pciDevice *dev, unsigned pos, uint8_t *buf, unsigned buflen)
{
    memset(buf, 0, buflen);

    if (pciOpenConfig(dev) < 0)
        return -1;

213 214
    if (lseek(dev->fd, pos, SEEK_SET) != pos ||
        saferead(dev->fd, buf, buflen) != buflen) {
215
        char ebuf[1024];
216
        VIR_WARN("Failed to read from '%s' : %s", dev->path,
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
                 virStrerror(errno, ebuf, sizeof(ebuf)));
        return -1;
    }
    return 0;
}

static uint8_t
pciRead8(pciDevice *dev, unsigned pos)
{
    uint8_t buf;
    pciRead(dev, pos, &buf, sizeof(buf));
    return buf;
}

static uint16_t
pciRead16(pciDevice *dev, unsigned pos)
{
    uint8_t buf[2];
    pciRead(dev, pos, &buf[0], sizeof(buf));
    return (buf[0] << 0) | (buf[1] << 8);
}

static uint32_t
pciRead32(pciDevice *dev, unsigned pos)
{
    uint8_t buf[4];
    pciRead(dev, pos, &buf[0], sizeof(buf));
    return (buf[0] << 0) | (buf[1] << 8) | (buf[2] << 16) | (buf[3] << 24);
}

static int
pciWrite(pciDevice *dev, unsigned pos, uint8_t *buf, unsigned buflen)
{
    if (pciOpenConfig(dev) < 0)
        return -1;

253 254
    if (lseek(dev->fd, pos, SEEK_SET) != pos ||
        safewrite(dev->fd, buf, buflen) != buflen) {
255
        char ebuf[1024];
256
        VIR_WARN("Failed to write to '%s' : %s", dev->path,
257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276
                 virStrerror(errno, ebuf, sizeof(ebuf)));
        return -1;
    }
    return 0;
}

static void
pciWrite16(pciDevice *dev, unsigned pos, uint16_t val)
{
    uint8_t buf[2] = { (val >> 0), (val >> 8) };
    pciWrite(dev, pos, &buf[0], sizeof(buf));
}

static void
pciWrite32(pciDevice *dev, unsigned pos, uint32_t val)
{
    uint8_t buf[4] = { (val >> 0), (val >> 8), (val >> 16), (val >> 14) };
    pciWrite(dev, pos, &buf[0], sizeof(buf));
}

277
typedef int (*pciIterPredicate)(pciDevice *, pciDevice *, void *);
278 279 280 281 282 283 284

/* Iterate over available PCI devices calling @predicate
 * to compare each one to @dev.
 * Return -1 on error since we don't want to assume it is
 * safe to reset if there is an error.
 */
static int
285
pciIterDevices(pciIterPredicate predicate,
286
               pciDevice *dev,
287 288
               pciDevice **matched,
               void *data)
289 290 291
{
    DIR *dir;
    struct dirent *entry;
292
    int ret = 0;
293
    int rc;
294 295 296 297 298 299 300

    *matched = NULL;

    VIR_DEBUG("%s %s: iterating over " PCI_SYSFS "devices", dev->id, dev->name);

    dir = opendir(PCI_SYSFS "devices");
    if (!dir) {
301
        VIR_WARN("Failed to open " PCI_SYSFS "devices");
302 303 304 305
        return -1;
    }

    while ((entry = readdir(dir))) {
306
        unsigned int domain, bus, slot, function;
307
        pciDevice *check;
308
        char *tmp;
309 310 311 312 313

        /* Ignore '.' and '..' */
        if (entry->d_name[0] == '.')
            continue;

314 315 316 317 318 319 320 321 322
        /* expected format: <domain>:<bus>:<slot>.<function> */
        if (/* domain */
            virStrToLong_ui(entry->d_name, &tmp, 16, &domain) < 0 || *tmp != ':' ||
            /* bus */
            virStrToLong_ui(tmp + 1, &tmp, 16, &bus) < 0 || *tmp != ':' ||
            /* slot */
            virStrToLong_ui(tmp + 1, &tmp, 16, &slot) < 0 || *tmp != '.' ||
            /* function */
            virStrToLong_ui(tmp + 1, NULL, 16, &function) < 0) {
323 324 325 326
            VIR_WARN("Unusual entry in " PCI_SYSFS "devices: %s", entry->d_name);
            continue;
        }

327
        check = pciGetDevice(domain, bus, slot, function);
328
        if (!check) {
329 330 331
            ret = -1;
            break;
        }
332

333 334 335 336 337 338 339 340
        rc = predicate(dev, check, data);
        if (rc < 0) {
            /* the predicate returned an error, bail */
            pciFreeDevice(check);
            ret = -1;
            break;
        }
        else if (rc == 1) {
341 342
            VIR_DEBUG("%s %s: iter matched on %s", dev->id, dev->name, check->name);
            *matched = check;
343
            ret = 1;
344 345
            break;
        }
346

347
        pciFreeDevice(check);
348 349
    }
    closedir(dir);
350
    return ret;
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
}

static uint8_t
pciFindCapabilityOffset(pciDevice *dev, unsigned capability)
{
    uint16_t status;
    uint8_t pos;

    status = pciRead16(dev, PCI_STATUS);
    if (!(status & PCI_STATUS_CAP_LIST))
        return 0;

    pos = pciRead8(dev, PCI_CAPABILITY_LIST);

    /* Zero indicates last capability, capabilities can't
     * be in the config space header and 0xff is returned
     * by the kernel if we don't have access to this region
     *
     * Note: we're not handling loops or extended
     * capabilities here.
     */
    while (pos >= PCI_CONF_HEADER_LEN && pos != 0xff) {
        uint8_t capid = pciRead8(dev, pos);
        if (capid == capability) {
            VIR_DEBUG("%s %s: found cap 0x%.2x at 0x%.2x",
                      dev->id, dev->name, capability, pos);
            return pos;
        }

        pos = pciRead8(dev, pos + 1);
    }

    VIR_DEBUG("%s %s: failed to find cap 0x%.2x", dev->id, dev->name, capability);

    return 0;
}

J
Jiri Denemark 已提交
388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
static unsigned int
pciFindExtendedCapabilityOffset(pciDevice *dev, unsigned capability)
{
    int ttl;
    unsigned int pos;
    uint32_t header;

    /* minimum 8 bytes per capability */
    ttl = (PCI_EXT_CAP_LIMIT - PCI_EXT_CAP_BASE) / 8;
    pos = PCI_EXT_CAP_BASE;

    while (ttl > 0 && pos >= PCI_EXT_CAP_BASE) {
        header = pciRead32(dev, pos);

        if ((header & PCI_EXT_CAP_ID_MASK) == capability)
            return pos;

        pos = (header >> PCI_EXT_CAP_OFFSET_SHIFT) & PCI_EXT_CAP_OFFSET_MASK;
        ttl--;
    }

    return 0;
}

412 413 414 415
/* detects whether this device has FLR.  Returns 0 if the device does
 * not have FLR, 1 if it does, and -1 on error
 */
static int
416 417
pciDetectFunctionLevelReset(pciDevice *dev)
{
M
Mark McLoughlin 已提交
418
    uint32_t caps;
419
    uint8_t pos;
420 421
    char *path;
    int found;
422 423 424 425 426 427 428 429

    /* The PCIe Function Level Reset capability allows
     * individual device functions to be reset without
     * affecting any other functions on the device or
     * any other devices on the bus. This is only common
     * on SR-IOV NICs at the moment.
     */
    if (dev->pcie_cap_pos) {
M
Mark McLoughlin 已提交
430
        caps = pciRead32(dev, dev->pcie_cap_pos + PCI_EXP_DEVCAP);
431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449
        if (caps & PCI_EXP_DEVCAP_FLR) {
            VIR_DEBUG("%s %s: detected PCIe FLR capability", dev->id, dev->name);
            return 1;
        }
    }

    /* The PCI AF Function Level Reset capability is
     * the same thing, except for conventional PCI
     * devices. This is not common yet.
     */
    pos = pciFindCapabilityOffset(dev, PCI_CAP_ID_AF);
    if (pos) {
        caps = pciRead16(dev, pos + PCI_AF_CAP);
        if (caps & PCI_AF_CAP_FLR) {
            VIR_DEBUG("%s %s: detected PCI FLR capability", dev->id, dev->name);
            return 1;
        }
    }

450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468
    /* there are some buggy devices that do support FLR, but forget to
     * advertise that fact in their capabilities.  However, FLR is *required*
     * to be present for virtual functions (VFs), so if we see that this
     * device is a VF, we just assume FLR works
     */

    if (virAsprintf(&path, PCI_SYSFS "devices/%s/physfn", dev->name) < 0) {
        virReportOOMError();
        return -1;
    }

    found = virFileExists(path);
    VIR_FREE(path);
    if (found) {
        VIR_DEBUG("%s %s: buggy device didn't advertise FLR, but is a VF; forcing flr on",
                  dev->id, dev->name);
        return 1;
    }

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
    VIR_DEBUG("%s %s: no FLR capability found", dev->id, dev->name);

    return 0;
}

/* Require the device has the PCI Power Management capability
 * and that a D3hot->D0 transition will results in a full
 * internal reset, not just a soft reset.
 */
static unsigned
pciDetectPowerManagementReset(pciDevice *dev)
{
    if (dev->pci_pm_cap_pos) {
        uint32_t ctl;

        /* require the NO_SOFT_RESET bit is clear */
        ctl = pciRead32(dev, dev->pci_pm_cap_pos + PCI_PM_CTRL);
        if (!(ctl & PCI_PM_CTRL_NO_SOFT_RESET)) {
            VIR_DEBUG("%s %s: detected PM reset capability", dev->id, dev->name);
            return 1;
        }
    }

    VIR_DEBUG("%s %s: no PM reset capability found", dev->id, dev->name);

    return 0;
}

497
/* Any active devices on the same domain/bus ? */
498
static int
499
pciSharesBusWithActive(pciDevice *dev, pciDevice *check, void *data)
500
{
501
    pciDeviceList *inactiveDevs = data;
502

503
    /* Different domain, different bus, or simply identical device */
504 505
    if (dev->domain != check->domain ||
        dev->bus != check->bus ||
506 507
        (dev->slot == check->slot &&
         dev->function == check->function))
508 509
        return 0;

510 511
    /* same bus, but inactive, i.e. about to be assigned to guest */
    if (inactiveDevs && pciDeviceListFind(inactiveDevs, check))
512
        return 0;
513

514
    return 1;
515 516
}

517
static pciDevice *
518
pciBusContainsActiveDevices(pciDevice *dev,
519
                            pciDeviceList *inactiveDevs)
520 521
{
    pciDevice *active = NULL;
522
    if (pciIterDevices(pciSharesBusWithActive,
523
                       dev, &active, inactiveDevs) < 0)
524 525 526 527 528
        return NULL;
    return active;
}

/* Is @check the parent of @dev ? */
529
static int
530
pciIsParent(pciDevice *dev, pciDevice *check, void *data)
531 532 533
{
    uint16_t device_class;
    uint8_t header_type, secondary, subordinate;
534
    pciDevice **best = data;
535

536
    if (dev->domain != check->domain)
537 538 539
        return 0;

    /* Is it a bridge? */
540
    device_class = pciRead16(check, PCI_CLASS_DEVICE);
541 542 543 544
    if (device_class != PCI_CLASS_BRIDGE_PCI)
        return 0;

    /* Is it a plane? */
545
    header_type = pciRead8(check, PCI_HEADER_TYPE);
546 547 548
    if ((header_type & PCI_HEADER_TYPE_MASK) != PCI_HEADER_TYPE_BRIDGE)
        return 0;

549 550
    secondary   = pciRead8(check, PCI_SECONDARY_BUS);
    subordinate = pciRead8(check, PCI_SUBORDINATE_BUS);
551

552
    VIR_DEBUG("%s %s: found parent device %s", dev->id, dev->name, check->name);
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
    /* if the secondary bus exactly equals the device's bus, then we found
     * the direct parent.  No further work is necessary
     */
    if (dev->bus == secondary)
        return 1;

    /* otherwise, SRIOV allows VFs to be on different busses then their PFs.
     * In this case, what we need to do is look for the "best" match; i.e.
     * the most restrictive match that still satisfies all of the conditions.
     */
    if (dev->bus > secondary && dev->bus <= subordinate) {
        if (*best == NULL) {
            *best = pciGetDevice(check->domain, check->bus, check->slot,
                                 check->function);
            if (*best == NULL)
                return -1;
        }
        else {
            /* OK, we had already recorded a previous "best" match for the
             * parent.  See if the current device is more restrictive than the
             * best, and if so, make it the new best
             */
            if (secondary > pciRead8(*best, PCI_SECONDARY_BUS)) {
                pciFreeDevice(*best);
                *best = pciGetDevice(check->domain, check->bus, check->slot,
                                     check->function);
                if (*best == NULL)
                    return -1;
            }
        }
    }

    return 0;
587 588
}

589 590
static int
pciGetParentDevice(pciDevice *dev, pciDevice **parent)
591
{
592 593 594 595 596 597 598 599 600 601
    pciDevice *best = NULL;
    int ret;

    *parent = NULL;
    ret = pciIterDevices(pciIsParent, dev, parent, &best);
    if (ret == 1)
        pciFreeDevice(best);
    else if (ret == 0)
        *parent = best;
    return ret;
602 603 604 605 606 607
}

/* Secondary Bus Reset is our sledgehammer - it resets all
 * devices behind a bus.
 */
static int
608
pciTrySecondaryBusReset(pciDevice *dev,
609
                        pciDeviceList *inactiveDevs)
610
{
611
    pciDevice *parent, *conflict;
612 613 614 615 616 617 618 619 620
    uint8_t config_space[PCI_CONF_LEN];
    uint16_t ctl;
    int ret = -1;

    /* For now, we just refuse to do a secondary bus reset
     * if there are other devices/functions behind the bus.
     * In future, we could allow it so long as those devices
     * are not in use by the host or other guests.
     */
621
    if ((conflict = pciBusContainsActiveDevices(dev, inactiveDevs))) {
622
        pciReportError(VIR_ERR_INTERNAL_ERROR,
623 624
                       _("Active %s devices on bus with %s, not doing bus reset"),
                       conflict->name, dev->name);
625 626 627 628
        return -1;
    }

    /* Find the parent bus */
629 630
    if (pciGetParentDevice(dev, &parent) < 0)
        return -1;
631
    if (!parent) {
632
        pciReportError(VIR_ERR_INTERNAL_ERROR,
633 634
                       _("Failed to find parent device for %s"),
                       dev->name);
635 636 637 638 639 640 641 642 643 644
        return -1;
    }

    VIR_DEBUG("%s %s: doing a secondary bus reset", dev->id, dev->name);

    /* Save and restore the device's config space; we only do this
     * for the supplied device since we refuse to do a reset if there
     * are multiple devices/functions
     */
    if (pciRead(dev, 0, config_space, PCI_CONF_LEN) < 0) {
645
        pciReportError(VIR_ERR_INTERNAL_ERROR,
646
                       _("Failed to read PCI config space for %s"),
647
                       dev->name);
648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663
        goto out;
    }

    /* Read the control register, set the reset flag, wait 200ms,
     * unset the reset flag and wait 200ms.
     */
    ctl = pciRead16(dev, PCI_BRIDGE_CONTROL);

    pciWrite16(parent, PCI_BRIDGE_CONTROL, ctl | PCI_BRIDGE_CTL_RESET);

    usleep(200 * 1000); /* sleep 200ms */

    pciWrite16(parent, PCI_BRIDGE_CONTROL, ctl);

    usleep(200 * 1000); /* sleep 200ms */

664
    if (pciWrite(dev, 0, config_space, PCI_CONF_LEN) < 0) {
665
        pciReportError(VIR_ERR_INTERNAL_ERROR,
666 667 668 669
                       _("Failed to restore PCI config space for %s"),
                       dev->name);
        goto out;
    }
670 671
    ret = 0;
out:
672
    pciFreeDevice(parent);
673 674 675 676 677 678 679 680
    return ret;
}

/* Power management reset attempts to reset a device using a
 * D-state transition from D3hot to D0. Note, in detect_pm_reset()
 * above we require the device supports a full internal reset.
 */
static int
681
pciTryPowerManagementReset(pciDevice *dev)
682 683 684 685 686 687 688 689 690
{
    uint8_t config_space[PCI_CONF_LEN];
    uint32_t ctl;

    if (!dev->pci_pm_cap_pos)
        return -1;

    /* Save and restore the device's config space. */
    if (pciRead(dev, 0, &config_space[0], PCI_CONF_LEN) < 0) {
691
        pciReportError(VIR_ERR_INTERNAL_ERROR,
692
                       _("Failed to read PCI config space for %s"),
693
                       dev->name);
694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709
        return -1;
    }

    VIR_DEBUG("%s %s: doing a power management reset", dev->id, dev->name);

    ctl = pciRead32(dev, dev->pci_pm_cap_pos + PCI_PM_CTRL);
    ctl &= ~PCI_PM_CTRL_STATE_MASK;

    pciWrite32(dev, dev->pci_pm_cap_pos + PCI_PM_CTRL, ctl|PCI_PM_CTRL_STATE_D3hot);

    usleep(10 * 1000); /* sleep 10ms */

    pciWrite32(dev, dev->pci_pm_cap_pos + PCI_PM_CTRL, ctl|PCI_PM_CTRL_STATE_D0);

    usleep(10 * 1000); /* sleep 10ms */

710
    if (pciWrite(dev, 0, &config_space[0], PCI_CONF_LEN) < 0) {
711
        pciReportError(VIR_ERR_INTERNAL_ERROR,
712 713 714 715
                       _("Failed to restore PCI config space for %s"),
                       dev->name);
        return -1;
    }
716 717 718 719 720

    return 0;
}

static int
721
pciInitDevice(pciDevice *dev)
722
{
723 724
    int flr;

725
    if (pciOpenConfig(dev) < 0) {
726
        virReportSystemError(errno,
727 728 729 730 731 732 733
                             _("Failed to open config space file '%s'"),
                             dev->path);
        return -1;
    }

    dev->pcie_cap_pos   = pciFindCapabilityOffset(dev, PCI_CAP_ID_EXP);
    dev->pci_pm_cap_pos = pciFindCapabilityOffset(dev, PCI_CAP_ID_PM);
734
    flr = pciDetectFunctionLevelReset(dev);
735
    if (flr < 0)
736 737
        return flr;
    dev->has_flr        = flr;
738 739 740 741 742 743
    dev->has_pm_reset   = pciDetectPowerManagementReset(dev);
    dev->initted        = 1;
    return 0;
}

int
744
pciResetDevice(pciDevice *dev,
745 746
               pciDeviceList *activeDevs,
               pciDeviceList *inactiveDevs)
747 748 749
{
    int ret = -1;

750
    if (activeDevs && pciDeviceListFind(activeDevs, dev)) {
751
        pciReportError(VIR_ERR_INTERNAL_ERROR,
752 753 754 755
                       _("Not resetting active device %s"), dev->name);
        return -1;
    }

756
    if (!dev->initted && pciInitDevice(dev) < 0)
757 758 759 760 761 762 763 764
        return -1;

    /* KVM will perform FLR when starting and stopping
     * a guest, so there is no need for us to do it here.
     */
    if (dev->has_flr)
        return 0;

765 766 767 768 769
    /* If the device supports PCI power management reset,
     * that's the next best thing because it only resets
     * the function, not the whole device.
     */
    if (dev->has_pm_reset)
770
        ret = pciTryPowerManagementReset(dev);
771

772
    /* Bus reset is not an option with the root bus */
773
    if (ret < 0 && dev->bus != 0)
774
        ret = pciTrySecondaryBusReset(dev, inactiveDevs);
775

776 777
    if (ret < 0) {
        virErrorPtr err = virGetLastError();
778
        pciReportError(VIR_ERR_INTERNAL_ERROR,
779 780 781 782 783
                       _("Unable to reset PCI device %s: %s"),
                       dev->name,
                       err ? err->message : _("no FLR, PM reset or bus reset available"));
    }

784 785 786
    return ret;
}

787

788 789
static int
pciDriverDir(char **buffer, const char *driver)
790
{
791 792 793 794 795 796 797 798
    VIR_FREE(*buffer);

    if (virAsprintf(buffer, PCI_SYSFS "drivers/%s", driver) < 0) {
        virReportOOMError();
        return -1;
    }

    return 0;
799 800
}

801 802
static int
pciDriverFile(char **buffer, const char *driver, const char *file)
803
{
804 805 806 807 808 809 810 811
    VIR_FREE(*buffer);

    if (virAsprintf(buffer, PCI_SYSFS "drivers/%s/%s", driver, file) < 0) {
        virReportOOMError();
        return -1;
    }

    return 0;
812
}
813

814 815
static int
pciDeviceFile(char **buffer, const char *device, const char *file)
816
{
817 818 819 820 821 822 823 824
    VIR_FREE(*buffer);

    if (virAsprintf(buffer, PCI_SYSFS "devices/%s/%s", device, file) < 0) {
        virReportOOMError();
        return -1;
    }

    return 0;
825
}
826 827


828
static const char *
829
pciFindStubDriver(void)
830
{
831
    char *drvpath = NULL;
832 833 834
    int probed = 0;

recheck:
835 836 837 838 839 840
    if (pciDriverDir(&drvpath, "pci-stub") < 0) {
        return NULL;
    }

    if (virFileExists(drvpath)) {
        VIR_FREE(drvpath);
841
        return "pci-stub";
842 843 844 845 846 847 848 849
    }

    if (pciDriverDir(&drvpath, "pciback") < 0) {
        return NULL;
    }

    if (virFileExists(drvpath)) {
        VIR_FREE(drvpath);
850
        return "pciback";
851 852 853
    }

    VIR_FREE(drvpath);
854 855 856 857 858 859 860 861 862 863 864 865

    if (!probed) {
        const char *const stubprobe[] = { MODPROBE, "pci-stub", NULL };
        const char *const backprobe[] = { MODPROBE, "pciback", NULL };

        probed = 1;
        /*
         * Probing for pci-stub will succeed regardless of whether
         * on native or Xen kernels.
         * On Xen though, we want to prefer pciback, so probe
         * for that first, because that will only work on Xen
         */
866 867
        if (virRun(backprobe, NULL) < 0 &&
            virRun(stubprobe, NULL) < 0) {
868
            char ebuf[1024];
869
            VIR_WARN("failed to load pci-stub or pciback drivers: %s",
870
                     virStrerror(errno, ebuf, sizeof ebuf));
871
            return NULL;
872
        }
873 874

        goto recheck;
875 876
    }

877 878 879
    return NULL;
}

880 881 882 883 884 885 886
static int
pciUnbindDeviceFromStub(pciDevice *dev, const char *driver)
{
    int result = -1;
    char *drvdir = NULL;
    char *path = NULL;

E
Eric Blake 已提交
887 888 889
    if (pciDriverDir(&drvdir, driver) < 0)
        goto cleanup;

890 891 892
    if (!dev->unbind_from_stub)
        goto remove_slot;

893 894
    /* If the device is bound to stub, unbind it.
     */
E
Eric Blake 已提交
895
    if (pciDeviceFile(&path, dev->name, "driver") < 0)
896 897 898 899 900 901 902 903 904
        goto cleanup;

    if (virFileExists(drvdir) && virFileLinkPointsTo(path, drvdir)) {
        if (pciDriverFile(&path, driver, "unbind") < 0) {
            goto cleanup;
        }

        if (virFileWriteStr(path, dev->name, 0) < 0) {
            virReportSystemError(errno,
905
                                 _("Failed to unbind PCI device '%s' from %s"),
906 907 908 909
                                 dev->name, driver);
            goto cleanup;
        }
    }
910 911 912 913 914
    dev->unbind_from_stub = 0;

remove_slot:
    if (!dev->remove_slot)
        goto reprobe;
915 916 917 918 919 920 921 922

    /* Xen's pciback.ko wants you to use remove_slot on the specific device */
    if (pciDriverFile(&path, driver, "remove_slot") < 0) {
        goto cleanup;
    }

    if (virFileExists(path) && virFileWriteStr(path, dev->name, 0) < 0) {
        virReportSystemError(errno,
923
                             _("Failed to remove slot for PCI device '%s' from %s"),
924 925 926
                             dev->name, driver);
        goto cleanup;
    }
927 928 929 930 931 932 933
    dev->remove_slot = 0;

reprobe:
    if (!dev->reprobe) {
        result = 0;
        goto cleanup;
    }
934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955

    /* Trigger a re-probe of the device is not in the stub's dynamic
     * ID table. If the stub is available, but 'remove_id' isn't
     * available, then re-probing would just cause the device to be
     * re-bound to the stub.
     */
    if (pciDriverFile(&path, driver, "remove_id") < 0) {
        goto cleanup;
    }

    if (!virFileExists(drvdir) || virFileExists(path)) {
        if (virFileWriteStr(PCI_SYSFS "drivers_probe", dev->name, 0) < 0) {
            virReportSystemError(errno,
                                 _("Failed to trigger a re-probe for PCI device '%s'"),
                                 dev->name);
            goto cleanup;
        }
    }

    result = 0;

cleanup:
956 957 958 959 960
    /* do not do it again */
    dev->unbind_from_stub = 0;
    dev->remove_slot = 0;
    dev->reprobe = 0;

961 962 963 964 965 966
    VIR_FREE(drvdir);
    VIR_FREE(path);

    return result;
}

967 968

static int
969
pciBindDeviceToStub(pciDevice *dev, const char *driver)
970
{
971 972 973
    int result = -1;
    char *drvdir = NULL;
    char *path = NULL;
974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989
    int reprobe = 0;

    /* check whether the device is already bound to a driver */
    if (pciDriverDir(&drvdir, driver) < 0 ||
        pciDeviceFile(&path, dev->name, "driver") < 0) {
        goto cleanup;
    }

    if (virFileExists(path)) {
        if (virFileLinkPointsTo(path, drvdir)) {
            /* The device is already bound to pci-stub */
            result = 0;
            goto cleanup;
        }
        reprobe = 1;
    }
990 991 992 993 994 995 996 997 998

    /* Add the PCI device ID to the stub's dynamic ID table;
     * this is needed to allow us to bind the device to the stub.
     * Note: if the device is not currently bound to any driver,
     * stub will immediately be bound to the device. Also, note
     * that if a new device with this ID is hotplugged, or if a probe
     * is triggered for such a device, it will also be immediately
     * bound by the stub.
     */
999
    if (pciDriverFile(&path, driver, "new_id") < 0) {
1000
        goto cleanup;
1001 1002
    }

1003
    if (virFileWriteStr(path, dev->id, 0) < 0) {
1004
        virReportSystemError(errno,
1005 1006
                             _("Failed to add PCI device ID '%s' to %s"),
                             dev->id, driver);
1007
        goto cleanup;
1008 1009
    }

1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023
    /* check whether the device is bound to pci-stub when we write dev->id to
     * new_id.
     */
    if (pciDriverDir(&drvdir, driver) < 0 ||
        pciDeviceFile(&path, dev->name, "driver") < 0) {
        goto remove_id;
    }

    if (virFileLinkPointsTo(path, drvdir)) {
        dev->unbind_from_stub = 1;
        dev->remove_slot = 1;
        goto remove_id;
    }

1024 1025 1026 1027 1028
    /* If the device is already bound to a driver, unbind it.
     * Note, this will have rather unpleasant side effects if this
     * PCI device happens to be IDE controller for the disk hosting
     * your root filesystem.
     */
1029 1030 1031 1032
    if (pciDeviceFile(&path, dev->name, "driver/unbind") < 0) {
        goto cleanup;
    }

1033 1034 1035 1036 1037 1038 1039 1040
    if (virFileExists(path)) {
        if (virFileWriteStr(path, dev->name, 0) < 0) {
            virReportSystemError(errno,
                                 _("Failed to unbind PCI device '%s'"),
                                 dev->name);
            goto cleanup;
        }
        dev->reprobe = reprobe;
1041 1042
    }

1043 1044
    /* If the device isn't already bound to pci-stub, try binding it now.
     */
1045 1046
    if (pciDriverDir(&drvdir, driver) < 0 ||
        pciDeviceFile(&path, dev->name, "driver") < 0) {
1047
        goto remove_id;
1048 1049
    }

1050 1051
    if (!virFileLinkPointsTo(path, drvdir)) {
        /* Xen's pciback.ko wants you to use new_slot first */
1052
        if (pciDriverFile(&path, driver, "new_slot") < 0) {
1053
            goto remove_id;
1054 1055
        }

1056
        if (virFileExists(path) && virFileWriteStr(path, dev->name, 0) < 0) {
1057
            virReportSystemError(errno,
1058 1059
                                 _("Failed to add slot for PCI device '%s' to %s"),
                                 dev->name, driver);
1060
            goto remove_id;
1061
        }
1062
        dev->remove_slot = 1;
1063 1064

        if (pciDriverFile(&path, driver, "bind") < 0) {
1065
            goto remove_id;
1066 1067
        }

1068
        if (virFileWriteStr(path, dev->name, 0) < 0) {
1069
            virReportSystemError(errno,
1070 1071
                                 _("Failed to bind PCI device '%s' to %s"),
                                 dev->name, driver);
1072
            goto remove_id;
1073
        }
1074
        dev->unbind_from_stub = 1;
1075 1076
    }

1077
remove_id:
1078 1079 1080
    /* If 'remove_id' exists, remove the device id from pci-stub's dynamic
     * ID table so that 'drivers_probe' works below.
     */
1081
    if (pciDriverFile(&path, driver, "remove_id") < 0) {
E
Eric Blake 已提交
1082
        /* We do not remove PCI ID from pci-stub, and we cannot reprobe it */
1083 1084 1085 1086 1087
        if (dev->reprobe) {
            VIR_WARN("Could not remove PCI ID '%s' from %s, and the device "
                     "cannot be probed again.", dev->id, driver);
        }
        dev->reprobe = 0;
1088 1089 1090
        goto cleanup;
    }

1091
    if (virFileExists(path) && virFileWriteStr(path, dev->id, 0) < 0) {
1092
        virReportSystemError(errno,
1093 1094
                             _("Failed to remove PCI ID '%s' from %s"),
                             dev->id, driver);
1095

E
Eric Blake 已提交
1096
        /* remove PCI ID from pci-stub failed, and we cannot reprobe it */
1097 1098 1099 1100 1101
        if (dev->reprobe) {
            VIR_WARN("Failed to remove PCI ID '%s' from %s, and the device "
                     "cannot be probed again.", dev->id, driver);
        }
        dev->reprobe = 0;
1102
        goto cleanup;
1103 1104
    }

1105 1106 1107 1108 1109 1110
    result = 0;

cleanup:
    VIR_FREE(drvdir);
    VIR_FREE(path);

1111 1112 1113 1114
    if (result < 0) {
        pciUnbindDeviceFromStub(dev, driver);
    }

1115
    return result;
1116 1117 1118
}

int
1119
pciDettachDevice(pciDevice *dev, pciDeviceList *activeDevs)
1120
{
1121
    const char *driver = pciFindStubDriver();
1122
    if (!driver) {
1123
        pciReportError(VIR_ERR_INTERNAL_ERROR, "%s",
1124 1125 1126 1127
                       _("cannot find any PCI stub module"));
        return -1;
    }

1128 1129 1130 1131 1132 1133
    if (activeDevs && pciDeviceListFind(activeDevs, dev)) {
        pciReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Not detaching active device %s"), dev->name);
        return -1;
    }

1134
    return pciBindDeviceToStub(dev, driver);
1135 1136 1137
}

int
1138
pciReAttachDevice(pciDevice *dev, pciDeviceList *activeDevs)
1139
{
1140
    const char *driver = pciFindStubDriver();
1141
    if (!driver) {
1142
        pciReportError(VIR_ERR_INTERNAL_ERROR, "%s",
1143 1144 1145 1146
                       _("cannot find any PCI stub module"));
        return -1;
    }

1147 1148 1149 1150 1151 1152
    if (activeDevs && pciDeviceListFind(activeDevs, dev)) {
        pciReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Not reattaching active device %s"), dev->name);
        return -1;
    }

1153
    return pciUnbindDeviceFromStub(dev, driver);
1154 1155
}

1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189
/* Certain hypervisors (like qemu/kvm) map the PCI bar(s) on
 * the host when doing device passthrough.  This can lead to a race
 * condition where the hypervisor is still cleaning up the device while
 * libvirt is trying to re-attach it to the host device driver.  To avoid
 * this situation, we look through /proc/iomem, and if the hypervisor is
 * still holding onto the bar (denoted by the string in the matcher variable),
 * then we can wait around a bit for that to clear up.
 *
 * A typical /proc/iomem looks like this (snipped for brevity):
 * 00010000-0008efff : System RAM
 * 0008f000-0008ffff : reserved
 * ...
 * 00100000-cc9fcfff : System RAM
 *   00200000-00483d3b : Kernel code
 *   00483d3c-005c88df : Kernel data
 * cc9fd000-ccc71fff : ACPI Non-volatile Storage
 * ...
 * d0200000-d02fffff : PCI Bus #05
 *   d0200000-d021ffff : 0000:05:00.0
 *     d0200000-d021ffff : e1000e
 *   d0220000-d023ffff : 0000:05:00.0
 *     d0220000-d023ffff : e1000e
 * ...
 * f0000000-f0003fff : 0000:00:1b.0
 *   f0000000-f0003fff : kvm_assigned_device
 *
 * Returns 0 if we are clear to continue, and 1 if the hypervisor is still
 * holding onto the resource.
 */
int
pciWaitForDeviceCleanup(pciDevice *dev, const char *matcher)
{
    FILE *fp;
    char line[160];
1190
    char *tmp;
1191
    unsigned long long start, end;
1192
    unsigned int domain, bus, slot, function;
1193 1194 1195 1196 1197 1198 1199 1200 1201 1202
    int in_matching_device;
    int ret;
    size_t match_depth;

    fp = fopen("/proc/iomem", "r");
    if (!fp) {
        /* If we failed to open iomem, we just basically ignore the error.  The
         * unbind might succeed anyway, and besides, it's very likely we have
         * no way to report the error
         */
1203
        VIR_DEBUG("Failed to open /proc/iomem, trying to continue anyway");
1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219
        return 0;
    }

    ret = 0;
    in_matching_device = 0;
    match_depth = 0;
    while (fgets(line, sizeof(line), fp) != 0) {
        /* the logic here is a bit confusing.  For each line, we look to
         * see if it matches the domain:bus:slot.function we were given.
         * If this line matches the DBSF, then any subsequent lines indented
         * by 2 spaces are the PCI regions for this device.  It's also
         * possible that none of the PCI regions are currently mapped, in
         * which case we have no indented regions.  This code handles all
         * of these situations
         */
        if (in_matching_device && (strspn(line, " ") == (match_depth + 2))) {
1220 1221 1222 1223 1224 1225
            /* expected format: <start>-<end> : <suffix> */
            if (/* start */
                virStrToLong_ull(line, &tmp, 16, &start) < 0 || *tmp != '-' ||
                /* end */
                virStrToLong_ull(tmp + 1, &tmp, 16, &end) < 0 ||
                (tmp = STRSKIP(tmp, " : ")) == NULL)
1226 1227
                continue;

1228
            if (STRPREFIX(tmp, matcher)) {
1229 1230 1231 1232 1233 1234 1235
                ret = 1;
                break;
            }
        }
        else {
            in_matching_device = 0;

1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249
            /* expected format: <start>-<end> : <domain>:<bus>:<slot>.<function> */
            if (/* start */
                virStrToLong_ull(line, &tmp, 16, &start) < 0 || *tmp != '-' ||
                /* end */
                virStrToLong_ull(tmp + 1, &tmp, 16, &end) < 0 ||
                (tmp = STRSKIP(tmp, " : ")) == NULL ||
                /* domain */
                virStrToLong_ui(tmp, &tmp, 16, &domain) < 0 || *tmp != ':' ||
                /* bus */
                virStrToLong_ui(tmp + 1, &tmp, 16, &bus) < 0 || *tmp != ':' ||
                /* slot */
                virStrToLong_ui(tmp + 1, &tmp, 16, &slot) < 0 || *tmp != '.' ||
                /* function */
                virStrToLong_ui(tmp + 1, &tmp, 16, &function) < 0 || *tmp != '\n')
1250 1251 1252 1253 1254 1255 1256 1257 1258 1259
                continue;

            if (domain != dev->domain || bus != dev->bus || slot != dev->slot ||
                function != dev->function)
                continue;
            in_matching_device = 1;
            match_depth = strspn(line, " ");
        }
    }

E
Eric Blake 已提交
1260
    VIR_FORCE_FCLOSE(fp);
1261 1262 1263 1264

    return ret;
}

1265 1266 1267
static char *
pciReadDeviceID(pciDevice *dev, const char *id_name)
{
1268
    char *path = NULL;
1269 1270
    char *id_str;

1271 1272 1273
    if (pciDeviceFile(&path, dev->name, id_name) < 0) {
        return NULL;
    }
1274 1275

    /* ID string is '0xNNNN\n' ... i.e. 7 bytes */
1276 1277
    if (virFileReadAll(path, 7, &id_str) < 0) {
        VIR_FREE(path);
1278
        return NULL;
1279 1280 1281
    }

    VIR_FREE(path);
1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295

    /* Check for 0x suffix */
    if (id_str[0] != '0' || id_str[1] != 'x') {
        VIR_FREE(id_str);
        return NULL;
    }

    /* Chop off the newline; we know the string is 7 bytes */
    id_str[6] = '\0';

    return id_str;
}

pciDevice *
1296
pciGetDevice(unsigned domain,
1297 1298 1299 1300 1301
             unsigned bus,
             unsigned slot,
             unsigned function)
{
    pciDevice *dev;
E
Eric Blake 已提交
1302 1303
    char *vendor = NULL;
    char *product = NULL;
1304 1305

    if (VIR_ALLOC(dev) < 0) {
1306
        virReportOOMError();
1307 1308 1309
        return NULL;
    }

1310
    dev->fd       = -1;
1311 1312 1313 1314 1315
    dev->domain   = domain;
    dev->bus      = bus;
    dev->slot     = slot;
    dev->function = function;

E
Eric Blake 已提交
1316 1317 1318 1319 1320 1321
    if (snprintf(dev->name, sizeof(dev->name), "%.4x:%.2x:%.2x.%.1x",
                 dev->domain, dev->bus, dev->slot,
                 dev->function) >= sizeof(dev->name)) {
        pciReportError(VIR_ERR_INTERNAL_ERROR,
                       _("dev->name buffer overflow: %.4x:%.2x:%.2x.%.1x"),
                       dev->domain, dev->bus, dev->slot, dev->function);
E
Eric Blake 已提交
1322
        goto error;
E
Eric Blake 已提交
1323 1324 1325 1326
    }
    if (virAsprintf(&dev->path, PCI_SYSFS "devices/%s/config",
                    dev->name) < 0) {
        virReportOOMError();
E
Eric Blake 已提交
1327
        goto error;
E
Eric Blake 已提交
1328
    }
1329

1330 1331 1332 1333
    if (access(dev->path, F_OK) != 0) {
        virReportSystemError(errno,
                             _("Device %s not found: could not access %s"),
                             dev->name, dev->path);
E
Eric Blake 已提交
1334
        goto error;
1335 1336
    }

1337 1338 1339 1340
    vendor  = pciReadDeviceID(dev, "vendor");
    product = pciReadDeviceID(dev, "device");

    if (!vendor || !product) {
1341
        pciReportError(VIR_ERR_INTERNAL_ERROR,
1342 1343
                       _("Failed to read product/vendor ID for %s"),
                       dev->name);
E
Eric Blake 已提交
1344
        goto error;
1345 1346 1347
    }

    /* strings contain '0x' prefix */
E
Eric Blake 已提交
1348 1349 1350 1351 1352
    if (snprintf(dev->id, sizeof(dev->id), "%s %s", &vendor[2],
                 &product[2]) >= sizeof(dev->id)) {
        pciReportError(VIR_ERR_INTERNAL_ERROR,
                       _("dev->id buffer overflow: %s %s"),
                       &vendor[2], &product[2]);
E
Eric Blake 已提交
1353
        goto error;
E
Eric Blake 已提交
1354
    }
1355 1356 1357

    VIR_DEBUG("%s %s: initialized", dev->id, dev->name);

E
Eric Blake 已提交
1358 1359 1360
cleanup:
    VIR_FREE(product);
    VIR_FREE(vendor);
1361
    return dev;
E
Eric Blake 已提交
1362 1363 1364 1365 1366

error:
    pciFreeDevice(dev);
    dev = NULL;
    goto cleanup;
1367 1368 1369
}

void
1370
pciFreeDevice(pciDevice *dev)
1371
{
1372 1373
    if (!dev)
        return;
1374
    VIR_DEBUG("%s %s: freeing", dev->id, dev->name);
1375
    pciCloseConfig(dev);
E
Eric Blake 已提交
1376
    VIR_FREE(dev->path);
1377 1378
    VIR_FREE(dev);
}
1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389

void pciDeviceSetManaged(pciDevice *dev, unsigned managed)
{
    dev->managed = !!managed;
}

unsigned pciDeviceGetManaged(pciDevice *dev)
{
    return dev->managed;
}

1390 1391 1392 1393 1394 1395 1396 1397
void pciDeviceReAttachInit(pciDevice *pci)
{
    pci->unbind_from_stub = 1;
    pci->remove_slot = 1;
    pci->reprobe = 1;
}


1398
pciDeviceList *
1399
pciDeviceListNew(void)
1400 1401 1402 1403
{
    pciDeviceList *list;

    if (VIR_ALLOC(list) < 0) {
1404
        virReportOOMError();
1405 1406 1407 1408 1409 1410 1411
        return NULL;
    }

    return list;
}

void
1412
pciDeviceListFree(pciDeviceList *list)
1413 1414 1415 1416 1417 1418 1419
{
    int i;

    if (!list)
        return;

    for (i = 0; i < list->count; i++) {
1420
        pciFreeDevice(list->devs[i]);
1421 1422 1423 1424 1425 1426 1427 1428 1429
        list->devs[i] = NULL;
    }

    list->count = 0;
    VIR_FREE(list->devs);
    VIR_FREE(list);
}

int
1430
pciDeviceListAdd(pciDeviceList *list,
1431 1432 1433
                 pciDevice *dev)
{
    if (pciDeviceListFind(list, dev)) {
1434
        pciReportError(VIR_ERR_INTERNAL_ERROR,
1435 1436 1437 1438 1439
                       _("Device %s is already in use"), dev->name);
        return -1;
    }

    if (VIR_REALLOC_N(list->devs, list->count+1) < 0) {
1440
        virReportOOMError();
1441 1442 1443 1444 1445 1446 1447 1448
        return -1;
    }

    list->devs[list->count++] = dev;

    return 0;
}

1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462
pciDevice *
pciDeviceListGet(pciDeviceList *list,
                 int idx)
{
    if (idx >= list->count)
        return NULL;
    if (idx < 0)
        return NULL;

    return list->devs[idx];
}

int
pciDeviceListCount(pciDeviceList *list)
1463
{
1464 1465 1466 1467
    return list->count;
}

pciDevice *
1468
pciDeviceListSteal(pciDeviceList *list,
1469 1470 1471
                   pciDevice *dev)
{
    pciDevice *ret = NULL;
1472 1473 1474 1475 1476 1477 1478 1479 1480
    int i;

    for (i = 0; i < list->count; i++) {
        if (list->devs[i]->domain   != dev->domain ||
            list->devs[i]->bus      != dev->bus    ||
            list->devs[i]->slot     != dev->slot   ||
            list->devs[i]->function != dev->function)
            continue;

1481
        ret = list->devs[i];
1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493

        if (i != --list->count)
            memmove(&list->devs[i],
                    &list->devs[i+1],
                    sizeof(*list->devs) * (list->count-i));

        if (VIR_REALLOC_N(list->devs, list->count) < 0) {
            ; /* not fatal */
        }

        break;
    }
1494 1495 1496 1497
    return ret;
}

void
1498
pciDeviceListDel(pciDeviceList *list,
1499 1500
                 pciDevice *dev)
{
1501
    pciDevice *ret = pciDeviceListSteal(list, dev);
1502
    if (ret)
1503
        pciFreeDevice(ret);
1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518
}

pciDevice *
pciDeviceListFind(pciDeviceList *list, pciDevice *dev)
{
    int i;

    for (i = 0; i < list->count; i++)
        if (list->devs[i]->domain   == dev->domain &&
            list->devs[i]->bus      == dev->bus    &&
            list->devs[i]->slot     == dev->slot   &&
            list->devs[i]->function == dev->function)
            return list->devs[i];
    return NULL;
}
1519 1520


1521
int pciDeviceFileIterate(pciDevice *dev,
1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532
                         pciDeviceFileActor actor,
                         void *opaque)
{
    char *pcidir = NULL;
    char *file = NULL;
    DIR *dir = NULL;
    int ret = -1;
    struct dirent *ent;

    if (virAsprintf(&pcidir, "/sys/bus/pci/devices/%04x:%02x:%02x.%x",
                    dev->domain, dev->bus, dev->slot, dev->function) < 0) {
1533
        virReportOOMError();
1534 1535 1536 1537
        goto cleanup;
    }

    if (!(dir = opendir(pcidir))) {
1538
        virReportSystemError(errno,
1539 1540 1541 1542 1543 1544
                             _("cannot open %s"), pcidir);
        goto cleanup;
    }

    while ((ent = readdir(dir)) != NULL) {
        /* Device assignment requires:
A
Alex Williamson 已提交
1545 1546
         *   $PCIDIR/config, $PCIDIR/resource, $PCIDIR/resourceNNN,
         *   $PCIDIR/rom, $PCIDIR/reset
1547 1548 1549
         */
        if (STREQ(ent->d_name, "config") ||
            STRPREFIX(ent->d_name, "resource") ||
A
Alex Williamson 已提交
1550 1551
            STREQ(ent->d_name, "rom") ||
            STREQ(ent->d_name, "reset")) {
1552
            if (virAsprintf(&file, "%s/%s", pcidir, ent->d_name) < 0) {
1553
                virReportOOMError();
1554 1555
                goto cleanup;
            }
1556
            if ((actor)(dev, file, opaque) < 0)
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571
                goto cleanup;

            VIR_FREE(file);
        }
    }

    ret = 0;

cleanup:
    if (dir)
        closedir(dir);
    VIR_FREE(file);
    VIR_FREE(pcidir);
    return ret;
}
J
Jiri Denemark 已提交
1572 1573

static int
1574
pciDeviceDownstreamLacksACS(pciDevice *dev)
J
Jiri Denemark 已提交
1575 1576 1577 1578 1579
{
    uint16_t flags;
    uint16_t ctrl;
    unsigned int pos;

1580
    if (!dev->initted && pciInitDevice(dev) < 0)
J
Jiri Denemark 已提交
1581 1582 1583
        return -1;

    pos = dev->pcie_cap_pos;
1584
    if (!pos || pciRead16(dev, PCI_CLASS_DEVICE) != PCI_CLASS_BRIDGE_PCI)
J
Jiri Denemark 已提交
1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
        return 0;

    flags = pciRead16(dev, pos + PCI_EXP_FLAGS);
    if (((flags & PCI_EXP_FLAGS_TYPE) >> 4) != PCI_EXP_TYPE_DOWNSTREAM)
        return 0;

    pos = pciFindExtendedCapabilityOffset(dev, PCI_EXT_CAP_ID_ACS);
    if (!pos) {
        VIR_DEBUG("%s %s: downstream port lacks ACS", dev->id, dev->name);
        return 1;
    }

    ctrl = pciRead16(dev, pos + PCI_EXT_ACS_CTRL);
    if ((ctrl & PCI_EXT_CAP_ACS_ENABLED) != PCI_EXT_CAP_ACS_ENABLED) {
        VIR_DEBUG("%s %s: downstream port has ACS disabled",
                  dev->id, dev->name);
        return 1;
    }

    return 0;
}

static int
1608
pciDeviceIsBehindSwitchLackingACS(pciDevice *dev)
J
Jiri Denemark 已提交
1609 1610 1611
{
    pciDevice *parent;

1612 1613
    if (pciGetParentDevice(dev, &parent) < 0)
        return -1;
1614 1615 1616 1617 1618 1619 1620 1621
    if (!parent) {
        /* if we have no parent, and this is the root bus, ACS doesn't come
         * into play since devices on the root bus can't P2P without going
         * through the root IOMMU.
         */
        if (dev->bus == 0)
            return 0;
        else {
1622
            pciReportError(VIR_ERR_INTERNAL_ERROR,
1623 1624 1625 1626
                           _("Failed to find parent device for %s"),
                           dev->name);
            return -1;
        }
J
Jiri Denemark 已提交
1627 1628 1629 1630 1631 1632 1633 1634 1635
    }

    /* XXX we should rather fail when we can't find device's parent and
     * stop the loop when we get to root instead of just stopping when no
     * parent can be found
     */
    do {
        pciDevice *tmp;
        int acs;
1636
        int ret;
J
Jiri Denemark 已提交
1637

1638
        acs = pciDeviceDownstreamLacksACS(parent);
J
Jiri Denemark 已提交
1639 1640

        if (acs) {
1641
            pciFreeDevice(parent);
J
Jiri Denemark 已提交
1642 1643 1644 1645 1646 1647 1648
            if (acs < 0)
                return -1;
            else
                return 1;
        }

        tmp = parent;
1649
        ret = pciGetParentDevice(parent, &parent);
1650
        pciFreeDevice(tmp);
1651 1652
        if (ret < 0)
            return -1;
J
Jiri Denemark 已提交
1653 1654 1655 1656 1657
    } while (parent);

    return 0;
}

1658
int pciDeviceIsAssignable(pciDevice *dev,
J
Jiri Denemark 已提交
1659 1660 1661 1662 1663 1664 1665 1666 1667
                          int strict_acs_check)
{
    int ret;

    /* XXX This could be a great place to actually check that a non-managed
     * device isn't in use, e.g. by checking that device is either un-bound
     * or bound to a stub driver.
     */

1668
    ret = pciDeviceIsBehindSwitchLackingACS(dev);
J
Jiri Denemark 已提交
1669 1670 1671 1672 1673 1674 1675 1676
    if (ret < 0)
        return 0;

    if (ret) {
        if (!strict_acs_check) {
            VIR_DEBUG("%s %s: strict ACS check disabled; device assignment allowed",
                      dev->id, dev->name);
        } else {
1677
            pciReportError(VIR_ERR_INTERNAL_ERROR,
J
Jiri Denemark 已提交
1678 1679 1680 1681 1682 1683 1684 1685 1686
                           _("Device %s is behind a switch lacking ACS and "
                             "cannot be assigned"),
                           dev->name);
            return 0;
        }
    }

    return 1;
}
1687

1688 1689
#ifdef __linux__

1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702
/*
 * returns 1 if equal and 0 if not
 */
static int
pciConfigAddressEqual(struct pci_config_address *bdf1,
                      struct pci_config_address *bdf2)
{
    return ((bdf1->domain == bdf2->domain) &&
            (bdf1->bus == bdf2->bus) &&
            (bdf1->slot == bdf2->slot) &&
            (bdf1->function == bdf2->function));
}

1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905
static int
logStrToLong_ui(char const *s,
                char **end_ptr,
                int base,
                unsigned int *result)
{
    int ret = 0;

    ret = virStrToLong_ui(s, end_ptr, base, result);
    if (ret != 0) {
        VIR_ERROR(_("Failed to convert '%s' to unsigned int"), s);
    } else {
        VIR_DEBUG("Converted '%s' to unsigned int %u", s, *result);
    }

    return ret;
}

static int
pciParsePciConfigAddress(char *address,
                         struct pci_config_address *bdf)
{
    char *p = NULL;
    int ret = -1;

    if ((address == NULL) || (logStrToLong_ui(address, &p, 16,
                                              &bdf->domain) == -1)) {
        goto out;
    }

    if ((p == NULL) || (logStrToLong_ui(p+1, &p, 16,
                                        &bdf->bus) == -1)) {
        goto out;
    }

    if ((p == NULL) || (logStrToLong_ui(p+1, &p, 16,
                                        &bdf->slot) == -1)) {
        goto out;
    }

    if ((p == NULL) || (logStrToLong_ui(p+1, &p, 16,
                                        &bdf->function) == -1)) {
        goto out;
    }

    ret = 0;

out:
    return ret;
}

static int
pciGetPciConfigAddressFromSysfsDeviceLink(const char *device_link,
                                          struct pci_config_address **bdf)
{
    char *config_address = NULL;
    char *device_path = NULL;
    char errbuf[64];
    int ret = -1;

    VIR_DEBUG("Attempting to resolve device path from device link '%s'",
              device_link);

    if (!virFileExists(device_link)) {
        VIR_DEBUG("sysfs_path '%s' does not exist", device_link);
        return ret;
    }

    device_path = canonicalize_file_name (device_link);
    if (device_path == NULL) {
        memset(errbuf, '\0', sizeof(errbuf));
        pciReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to resolve device link '%s': '%s'"),
                       device_link, virStrerror(errno, errbuf,
                       sizeof(errbuf)));
        return ret;
    }

    config_address = basename(device_path);
    if (VIR_ALLOC(*bdf) != 0) {
        virReportOOMError();
        goto out;
    }

    if (pciParsePciConfigAddress(config_address, *bdf) != 0) {
        pciReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to parse PCI config address '%s'"),
                       config_address);
        VIR_FREE(*bdf);
        goto out;
    }

    VIR_DEBUG("pci_config_address %.4x:%.2x:%.2x.%.1x",
              (*bdf)->domain,
              (*bdf)->bus,
              (*bdf)->slot,
              (*bdf)->function);

    ret = 0;

out:
    VIR_FREE(device_path);

    return ret;
}

/*
 * Returns Physical function given a virtual function
 */
int
pciGetPhysicalFunction(const char *vf_sysfs_path,
                       struct pci_config_address **physical_function)
{
    int ret = -1;
    char *device_link = NULL;

    VIR_DEBUG("Attempting to get SR IOV physical function for device "
              "with sysfs path '%s'", vf_sysfs_path);

    if (virBuildPath(&device_link, vf_sysfs_path, "physfn") == -1) {
        virReportOOMError();
        return ret;
    } else {
        ret = pciGetPciConfigAddressFromSysfsDeviceLink(device_link,
                                                        physical_function);
    }

    VIR_FREE(device_link);

    return ret;
}

/*
 * Returns virtual functions of a physical function
 */
int
pciGetVirtualFunctions(const char *sysfs_path,
                       struct pci_config_address ***virtual_functions,
                       unsigned int *num_virtual_functions)
{
    int ret = -1;
    DIR *dir = NULL;
    struct dirent *entry = NULL;
    char *device_link = NULL;
    char errbuf[64];

    VIR_DEBUG("Attempting to get SR IOV virtual functions for device"
              "with sysfs path '%s'", sysfs_path);

    dir = opendir(sysfs_path);
    if (dir == NULL) {
        memset(errbuf, '\0', sizeof(errbuf));
        pciReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to open dir '%s': '%s'"),
                       sysfs_path, virStrerror(errno, errbuf,
                       sizeof(errbuf)));
        return ret;
    }

    *virtual_functions = NULL;
    *num_virtual_functions = 0;
    while ((entry = readdir(dir))) {
        if (STRPREFIX(entry->d_name, "virtfn")) {

            if (virBuildPath(&device_link, sysfs_path, entry->d_name) == -1) {
                virReportOOMError();
                goto out;
            }

            VIR_DEBUG("Number of virtual functions: %d",
                      *num_virtual_functions);
            if (VIR_REALLOC_N(*virtual_functions,
                (*num_virtual_functions) + 1) != 0) {
                virReportOOMError();
                VIR_FREE(device_link);
                goto out;
            }

            if (pciGetPciConfigAddressFromSysfsDeviceLink(device_link,
                &((*virtual_functions)[*num_virtual_functions])) !=
                SRIOV_FOUND) {
                /* We should not get back SRIOV_NOT_FOUND in this
                 * case, so if we do, it's an error. */
                pciReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Failed to get SR IOV function from device "
                               "link '%s'"), device_link);
                VIR_FREE(device_link);
                goto out;
            } else {
                (*num_virtual_functions)++;
            }
            VIR_FREE(device_link);
        }
    }

    ret = 0;

out:
    if (dir)
        closedir(dir);

    return ret;
}
1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967

/*
 * Returns 1 if vf device is a virtual function, 0 if not, -1 on error
 */
int
pciDeviceIsVirtualFunction(const char *vf_sysfs_device_link)
{
    char *vf_sysfs_physfn_link = NULL;
    int ret = -1;

    if (virAsprintf(&vf_sysfs_physfn_link, "%s/physfn",
        vf_sysfs_device_link) < 0) {
        virReportOOMError();
        return ret;
    }

    ret = virFileExists(vf_sysfs_physfn_link);

    VIR_FREE(vf_sysfs_physfn_link);

    return ret;
}

/*
 * Returns the sriov virtual function index of vf given its pf
 */
int
pciGetVirtualFunctionIndex(const char *pf_sysfs_device_link,
                           const char *vf_sysfs_device_link,
                           int *vf_index)
{
    int ret = -1, i;
    unsigned int num_virt_fns = 0;
    struct pci_config_address *vf_bdf = NULL;
    struct pci_config_address **virt_fns = NULL;

    if (pciGetPciConfigAddressFromSysfsDeviceLink(vf_sysfs_device_link,
        &vf_bdf))
        return ret;

    if (pciGetVirtualFunctions(pf_sysfs_device_link, &virt_fns,
        &num_virt_fns)) {
        pciReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Error getting physical function's '%s' "
                      "virtual_functions"), pf_sysfs_device_link);
        goto out;
    }

    for (i = 0; i < num_virt_fns; i++) {
         if (pciConfigAddressEqual(vf_bdf, virt_fns[i])) {
             *vf_index = i;
             ret = 0;
             break;
         }
    }

out:

    /* free virtual functions */
    for (i = 0; i < num_virt_fns; i++)
         VIR_FREE(virt_fns[i]);

A
ajia@redhat.com 已提交
1968
    VIR_FREE(virt_fns);
1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
    VIR_FREE(vf_bdf);

    return ret;
}

/*
 * Returns the network device name of a pci device
 */
int
pciDeviceNetName(char *device_link_sysfs_path, char **netname)
{
     char *pcidev_sysfs_net_path = NULL;
     int ret = -1;
     DIR *dir = NULL;
     struct dirent *entry = NULL;

     if (virBuildPath(&pcidev_sysfs_net_path, device_link_sysfs_path,
         "net") == -1) {
         virReportOOMError();
         return -1;
     }

     dir = opendir(pcidev_sysfs_net_path);
     if (dir == NULL)
         goto out;

     while ((entry = readdir(dir))) {
            if (STREQ(entry->d_name, ".") ||
                STREQ(entry->d_name, ".."))
                continue;

            /* Assume a single directory entry */
            *netname = strdup(entry->d_name);
            ret = 0;
            break;
     }

     closedir(dir);

out:
     VIR_FREE(pcidev_sysfs_net_path);

     return ret;
}
2013 2014
#else
int
2015 2016
pciGetPhysicalFunction(const char *vf_sysfs_path ATTRIBUTE_UNUSED,
              struct pci_config_address **physical_function ATTRIBUTE_UNUSED)
2017 2018 2019 2020 2021 2022 2023
{
    pciReportError(VIR_ERR_INTERNAL_ERROR, _("pciGetPhysicalFunction is not "
                   "supported on non-linux platforms"));
    return -1;
}

int
2024 2025 2026
pciGetVirtualFunctions(const char *sysfs_path ATTRIBUTE_UNUSED,
             struct pci_config_address ***virtual_functions ATTRIBUTE_UNUSED,
             unsigned int *num_virtual_functions ATTRIBUTE_UNUSED)
2027 2028 2029 2030 2031
{
    pciReportError(VIR_ERR_INTERNAL_ERROR, _("pciGetVirtualFunctions is not "
                   "supported on non-linux platforms"));
    return -1;
}
2032 2033

int
2034
pciDeviceIsVirtualFunction(const char *vf_sysfs_device_link ATTRIBUTE_UNUSED)
2035 2036 2037 2038 2039 2040 2041
{
    pciReportError(VIR_ERR_INTERNAL_ERROR, _("pciDeviceIsVirtualFunction is "
                   "not supported on non-linux platforms"));
    return -1;
}

int
2042 2043 2044
pciGetVirtualFunctionIndex(const char *pf_sysfs_device_link ATTRIBUTE_UNUSED,
                           const char *vf_sysfs_device_link ATTRIBUTE_UNUSED,
                           int *vf_index ATTRIBUTE_UNUSED)
2045 2046 2047 2048 2049 2050 2051 2052
{
    pciReportError(VIR_ERR_INTERNAL_ERROR, _("pciGetVirtualFunctionIndex is "
                   "not supported on non-linux platforms"));
    return -1;

}

int
2053 2054
pciDeviceNetName(char *device_link_sysfs_path ATTRIBUTE_UNUSED,
                 char **netname ATTRIBUTE_UNUSED)
2055 2056 2057 2058 2059
{
    pciReportError(VIR_ERR_INTERNAL_ERROR, _("pciDeviceNetName is not "
                   "supported on non-linux platforms"));
    return -1;
}
2060
#endif /* __linux__ */