qemu_capabilities.c 72.9 KB
Newer Older
1 2 3
/*
 * qemu_capabilities.c: QEMU capabilities generation
 *
4
 * Copyright (C) 2006-2012 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * 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
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include "qemu_capabilities.h"
#include "memory.h"
#include "logging.h"
#include "virterror_internal.h"
#include "util.h"
E
Eric Blake 已提交
31
#include "virfile.h"
32 33
#include "virpidfile.h"
#include "virprocess.h"
34 35 36
#include "nodeinfo.h"
#include "cpu/cpu.h"
#include "domain_conf.h"
37
#include "command.h"
38
#include "bitmap.h"
39
#include "virnodesuspend.h"
40
#include "qemu_monitor.h"
41 42 43 44 45

#include <sys/stat.h>
#include <unistd.h>
#include <sys/wait.h>
#include <sys/utsname.h>
46
#include <stdarg.h>
47 48 49

#define VIR_FROM_THIS VIR_FROM_QEMU

50 51 52 53 54 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
/* While not public, these strings must not change. They
 * are used in domain status files which are read on
 * daemon restarts
 */
VIR_ENUM_IMPL(qemuCaps, QEMU_CAPS_LAST,
              "kqemu",  /* 0 */
              "vnc-colon",
              "no-reboot",
              "drive",
              "drive-boot",

              "name", /* 5 */
              "uuid",
              "domid",
              "vnet-hdr",
              "migrate-kvm-stdio",

              "migrate-qemu-tcp", /* 10 */
              "migrate-qemu-exec",
              "drive-cache-v2",
              "kvm",
              "drive-format",

              "vga", /* 15 */
              "0.10",
              "pci-device",
              "mem-path",
              "drive-serial",

              "xen-domid", /* 20 */
              "migrate-qemu-unix",
              "chardev",
              "enable-kvm",
              "monitor-json",

              "balloon", /* 25 */
              "device",
              "sdl",
              "smp-topology",
              "netdev",

              "rtc", /* 30 */
92
              "vhost-net",
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
              "rtc-td-hack",
              "no-hpet",
              "no-kvm-pit",

              "tdf", /* 35 */
              "pci-configfd",
              "nodefconfig",
              "boot-menu",
              "enable-kqemu",

              "fsdev", /* 40 */
              "nesting",
              "name-process",
              "drive-readonly",
              "smbios-type",

              "vga-qxl", /* 45 */
              "spice",
              "vga-none",
              "migrate-qemu-fd",
              "boot-index",

              "hda-duplex", /* 50 */
              "drive-aio",
              "pci-multibus",
              "pci-bootindex",
              "ccid-emulated",

              "ccid-passthru", /* 55 */
              "chardev-spicevmc",
              "device-spicevmc",
              "virtio-tx-alg",
              "device-qxl-vga",
126 127

              "pci-multifunction", /* 60 */
128
              "virtio-blk-pci.ioeventfd",
M
Michal Privoznik 已提交
129
              "sga",
130 131
              "virtio-blk-pci.event_idx",
              "virtio-net-pci.event_idx",
132 133 134 135 136 137 138 139 140 141

              "cache-directsync", /* 65 */
              "piix3-usb-uhci",
              "piix4-usb-uhci",
              "usb-ehci",
              "ich9-usb-ehci1",

              "vt82c686b-usb-uhci", /* 70 */
              "pci-ohci",
              "usb-redir",
M
Marc-André Lureau 已提交
142
              "usb-hub",
143
              "no-shutdown",
144 145

              "cache-unsafe", /* 75 */
146
              "rombar",
J
Jim Fehlig 已提交
147
              "ich9-ahci",
148
              "no-acpi",
149
              "fsdev-readonly",
150

151
              "virtio-blk-pci.scsi", /* 80 */
152
              "blk-sg-io",
O
Osier Yang 已提交
153
              "drive-copy-on-read",
154
              "cpu-host",
155
              "fsdev-writeout",
156 157

              "drive-iotune", /* 85 */
158
              "system_wakeup",
159
              "scsi-disk.channel",
160
              "scsi-block",
161
              "transaction",
162 163 164

              "block-job-sync", /* 90 */
              "block-job-async",
165
              "scsi-cd",
166
              "ide-cd",
167
              "no-user-config",
M
Marc-André Lureau 已提交
168 169

              "hda-micro", /* 95 */
170
              "dump-guest-memory",
G
Gerd Hoffmann 已提交
171
              "nec-usb-xhci",
172
              "virtio-s390",
173
              "balloon-event",
M
Marc-André Lureau 已提交
174

R
Richa Marwaha 已提交
175
              "bridge", /* 100 */
176 177
              "lsi",
              "virtio-scsi-pci",
V
Viktor Mihajlovski 已提交
178
              "blockio",
179
              "disable-s3",
R
Richa Marwaha 已提交
180

181
              "disable-s4", /* 105 */
182
              "usb-redir.filter",
183 184
              "ide-drive.wwn",
              "scsi-disk.wwn",
185
              "seccomp-sandbox",
186 187

              "reboot-timeout", /* 110 */
188
              "dump-guest-core",
189
              "seamless-migration",
190
              "block-commit",
191
              "vnc",
192 193

              "drive-mirror", /* 115 */
194 195
              "usb-redir.bootindex",
              "usb-host.bootindex",
196
              "blockdev-snapshot-sync",
197 198 199 200 201 202
              "qxl",

              "VGA", /* 120 */
              "cirrus-vga",
              "vmware-svga",
              "device-video-primary",
203 204
    );

205 206 207
struct _qemuCaps {
    virObject object;

208 209
    bool usedQMP;

210 211 212
    char *binary;
    time_t mtime;

213
    virBitmapPtr flags;
214 215 216 217 218 219 220 221 222 223 224 225

    unsigned int version;
    unsigned int kvmVersion;

    char *arch;

    size_t ncpuDefinitions;
    char **cpuDefinitions;

    size_t nmachineTypes;
    char **machineTypes;
    char **machineAliases;
226 227
};

228 229 230
struct _qemuCapsCache {
    virMutex lock;
    virHashTablePtr binaries;
231
    char *libDir;
232
    char *runDir;
233 234
    uid_t runUid;
    gid_t runGid;
235 236
};

237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252

static virClassPtr qemuCapsClass;
static void qemuCapsDispose(void *obj);

static int qemuCapsOnceInit(void)
{
    if (!(qemuCapsClass = virClassNew("qemuCaps",
                                      sizeof(qemuCaps),
                                      qemuCapsDispose)))
        return -1;

    return 0;
}

VIR_ONCE_GLOBAL_INIT(qemuCaps)

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
struct _qemuCapsHookData {
    uid_t runUid;
    gid_t runGid;
};
typedef struct _qemuCapsHookData qemuCapsHookData;
typedef qemuCapsHookData *qemuCapsHookDataPtr;

static int qemuCapsHook(void * data)
{
    int ret;
    qemuCapsHookDataPtr hookData = data;

    if (!hookData) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("QEMU uid:gid not specified by caller"));
        ret = -1;
        goto cleanup;
    }

    VIR_DEBUG("Switch QEMU uid:gid to %d:%d",
              hookData->runUid, hookData->runGid);
    ret = virSetUIDGID(hookData->runUid, hookData->runGid);

cleanup:
    return ret;
}

280 281
static virCommandPtr
qemuCapsProbeCommand(const char *qemu,
282 283
                     qemuCapsPtr caps,
                     qemuCapsHookDataPtr hookData)
284 285 286 287 288 289 290 291 292 293 294 295
{
    virCommandPtr cmd = virCommandNew(qemu);

    if (caps) {
        if (qemuCapsGet(caps, QEMU_CAPS_NO_USER_CONFIG))
            virCommandAddArg(cmd, "-no-user-config");
        else if (qemuCapsGet(caps, QEMU_CAPS_NODEFCONFIG))
            virCommandAddArg(cmd, "-nodefconfig");
    }

    virCommandAddEnvPassCommon(cmd);
    virCommandClearCaps(cmd);
296
    virCommandSetPreExecHook(cmd, qemuCapsHook, hookData);
297 298 299 300 301

    return cmd;
}


302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318
static void
qemuSetDefaultMachine(qemuCapsPtr caps,
                      size_t defIdx)
{
    char *name = caps->machineTypes[defIdx];
    char *alias = caps->machineAliases[defIdx];

    memmove(caps->machineTypes + 1,
            caps->machineTypes,
            sizeof(caps->machineTypes[0]) * defIdx);
    memmove(caps->machineAliases + 1,
            caps->machineAliases,
            sizeof(caps->machineAliases[0]) * defIdx);
    caps->machineTypes[0] = name;
    caps->machineAliases[0] = alias;
}

319 320 321 322 323
/* Format is:
 * <machine> <desc> [(default)|(alias of <canonical>)]
 */
static int
qemuCapsParseMachineTypesStr(const char *output,
324
                             qemuCapsPtr caps)
325 326 327
{
    const char *p = output;
    const char *next;
328
    size_t defIdx = 0;
329 330 331

    do {
        const char *t;
332 333
        char *name;
        char *canonical = NULL;
334 335 336 337 338 339 340 341 342 343

        if ((next = strchr(p, '\n')))
            ++next;

        if (STRPREFIX(p, "Supported machines are:"))
            continue;

        if (!(t = strchr(p, ' ')) || (next && t >= next))
            continue;

344
        if (!(name = strndup(p, t - p)))
345 346 347
            goto no_memory;

        p = t;
348
        if ((t = strstr(p, "(default)")) && (!next || t < next))
349
            defIdx = caps->nmachineTypes;
350 351 352 353 354 355

        if ((t = strstr(p, "(alias of ")) && (!next || t < next)) {
            p = t + strlen("(alias of ");
            if (!(t = strchr(p, ')')) || (next && t >= next))
                continue;

356 357
            if (!(canonical = strndup(p, t - p))) {
                VIR_FREE(name);
358
                goto no_memory;
359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374
            }
        }

        if (VIR_REALLOC_N(caps->machineTypes, caps->nmachineTypes + 1) < 0 ||
            VIR_REALLOC_N(caps->machineAliases, caps->nmachineTypes + 1) < 0) {
            VIR_FREE(name);
            VIR_FREE(canonical);
            goto no_memory;
        }
        caps->nmachineTypes++;
        if (canonical) {
            caps->machineTypes[caps->nmachineTypes-1] = canonical;
            caps->machineAliases[caps->nmachineTypes-1] = name;
        } else {
            caps->machineTypes[caps->nmachineTypes-1] = name;
            caps->machineAliases[caps->nmachineTypes-1] = NULL;
375 376 377
        }
    } while ((p = next));

378

379 380
    if (defIdx)
        qemuSetDefaultMachine(caps, defIdx);
381 382 383

    return 0;

384
no_memory:
385 386 387 388
    virReportOOMError();
    return -1;
}

389
static int
390
qemuCapsProbeMachineTypes(qemuCapsPtr caps, qemuCapsHookDataPtr hookData)
391 392
{
    char *output;
393 394
    int ret = -1;
    virCommandPtr cmd;
395
    int status;
396

397 398 399 400
    /* Make sure the binary we are about to try exec'ing exists.
     * Technically we could catch the exec() failure, but that's
     * in a sub-process so it's hard to feed back a useful error.
     */
401 402 403
    if (!virFileIsExecutable(caps->binary)) {
        virReportSystemError(errno, _("Cannot find QEMU binary %s"),
                             caps->binary);
404 405 406
        return -1;
    }

407
    cmd = qemuCapsProbeCommand(caps->binary, caps, hookData);
408
    virCommandAddArgList(cmd, "-M", "?", NULL);
409
    virCommandSetOutputBuffer(cmd, &output);
410

411 412
    /* Ignore failure from older qemu that did not understand '-M ?'.  */
    if (virCommandRun(cmd, &status) < 0)
413 414
        goto cleanup;

415
    if (qemuCapsParseMachineTypesStr(output, caps) < 0)
416
        goto cleanup;
417 418 419 420

    ret = 0;

cleanup:
421 422
    VIR_FREE(output);
    virCommandFree(cmd);
423 424 425 426 427 428 429

    return ret;
}


typedef int
(*qemuCapsParseCPUModels)(const char *output,
430
                          qemuCapsPtr caps);
431 432 433 434 435 436 437 438

/* Format:
 *      <arch> <model>
 * qemu-0.13 encloses some model names in []:
 *      <arch> [<model>]
 */
static int
qemuCapsParseX86Models(const char *output,
439
                       qemuCapsPtr caps)
440 441 442
{
    const char *p = output;
    const char *next;
443
    int ret = -1;
444 445 446

    do {
        const char *t;
447
        size_t len;
448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464

        if ((next = strchr(p, '\n')))
            next++;

        if (!(t = strchr(p, ' ')) || (next && t >= next))
            continue;

        if (!STRPREFIX(p, "x86"))
            continue;

        p = t;
        while (*p == ' ')
            p++;

        if (*p == '\0' || *p == '\n')
            continue;

465 466 467 468
        if (VIR_EXPAND_N(caps->cpuDefinitions, caps->ncpuDefinitions, 1) < 0) {
            virReportOOMError();
            goto cleanup;
        }
469

470 471 472 473
        if (next)
            len = next - p - 1;
        else
            len = strlen(p);
474

475 476 477 478
        if (len > 2 && *p == '[' && p[len - 1] == ']') {
            p++;
            len -= 2;
        }
479

480 481 482
        if (!(caps->cpuDefinitions[caps->ncpuDefinitions - 1] = strndup(p, len))) {
            virReportOOMError();
            goto cleanup;
483 484 485
        }
    } while ((p = next));

486
    ret = 0;
487

488 489
cleanup:
    return ret;
490 491
}

P
Prerna Saxena 已提交
492 493 494 495 496
/* ppc64 parser.
 * Format : PowerPC <machine> <description>
 */
static int
qemuCapsParsePPCModels(const char *output,
497
                       qemuCapsPtr caps)
P
Prerna Saxena 已提交
498 499 500
{
    const char *p = output;
    const char *next;
501
    int ret = -1;
P
Prerna Saxena 已提交
502 503 504

    do {
        const char *t;
505
        size_t len;
P
Prerna Saxena 已提交
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525

        if ((next = strchr(p, '\n')))
            next++;

        if (!STRPREFIX(p, "PowerPC "))
            continue;

        /* Skip the preceding sub-string "PowerPC " */
        p += 8;

        /*Malformed string, does not obey the format 'PowerPC <model> <desc>'*/
        if (!(t = strchr(p, ' ')) || (next && t >= next))
            continue;

        if (*p == '\0')
            break;

        if (*p == '\n')
            continue;

526 527 528 529
        if (VIR_EXPAND_N(caps->cpuDefinitions, caps->ncpuDefinitions, 1) < 0) {
            virReportOOMError();
            goto cleanup;
        }
P
Prerna Saxena 已提交
530

531
        len = t - p - 1;
P
Prerna Saxena 已提交
532

533 534 535
        if (!(caps->cpuDefinitions[caps->ncpuDefinitions - 1] = strndup(p, len))) {
            virReportOOMError();
            goto cleanup;
P
Prerna Saxena 已提交
536 537 538
        }
    } while ((p = next));

539
    ret = 0;
P
Prerna Saxena 已提交
540

541 542
cleanup:
    return ret;
P
Prerna Saxena 已提交
543
}
544

545
static int
546
qemuCapsProbeCPUModels(qemuCapsPtr caps, qemuCapsHookDataPtr hookData)
547 548 549 550
{
    char *output = NULL;
    int ret = -1;
    qemuCapsParseCPUModels parse;
551
    virCommandPtr cmd;
552

553 554
    if (STREQ(caps->arch, "i686") ||
        STREQ(caps->arch, "x86_64"))
555
        parse = qemuCapsParseX86Models;
556
    else if (STREQ(caps->arch, "ppc64"))
P
Prerna Saxena 已提交
557
        parse = qemuCapsParsePPCModels;
558
    else {
559 560
        VIR_DEBUG("don't know how to parse %s CPU models",
                  caps->arch);
561 562 563
        return 0;
    }

564
    cmd = qemuCapsProbeCommand(caps->binary, caps, hookData);
565
    virCommandAddArgList(cmd, "-cpu", "?", NULL);
566
    virCommandSetOutputBuffer(cmd, &output);
567

568
    if (virCommandRun(cmd, NULL) < 0)
569 570
        goto cleanup;

571
    if (parse(output, caps) < 0)
572 573 574 575 576 577
        goto cleanup;

    ret = 0;

cleanup:
    VIR_FREE(output);
578
    virCommandFree(cmd);
579 580 581 582 583

    return ret;
}


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
static char *
qemuCapsFindBinaryForArch(const char *hostarch,
                          const char *guestarch)
{
    char *ret;

    if (STREQ(guestarch, "i686")) {
        ret = virFindFileInPath("qemu-system-i386");
        if (ret && !virFileIsExecutable(ret))
            VIR_FREE(ret);

        if (!ret && STREQ(hostarch, "x86_64")) {
            ret = virFindFileInPath("qemu-system-x86_64");
            if (ret && !virFileIsExecutable(ret))
                VIR_FREE(ret);
        }

        if (!ret)
            ret = virFindFileInPath("qemu");
    } else if (STREQ(guestarch, "itanium")) {
        ret = virFindFileInPath("qemu-system-ia64");
    } else {
        char *bin;
        if (virAsprintf(&bin, "qemu-system-%s", guestarch) < 0) {
            virReportOOMError();
            return NULL;
        }
        ret = virFindFileInPath(bin);
        VIR_FREE(bin);
    }
    if (ret && !virFileIsExecutable(ret))
        VIR_FREE(ret);
    return ret;
}

static int
qemuCapsGetArchWordSize(const char *guestarch)
{
    if (STREQ(guestarch, "i686") ||
        STREQ(guestarch, "ppc") ||
        STREQ(guestarch, "sparc") ||
        STREQ(guestarch, "mips") ||
        STREQ(guestarch, "mipsel"))
        return 32;
    return 64;
}

static bool
qemuCapsIsValidForKVM(const char *hostarch,
                      const char *guestarch)
{
    if (STREQ(hostarch, guestarch))
        return true;
    if (STREQ(hostarch, "x86_64") &&
        STREQ(guestarch, "i686"))
        return true;
    return false;
}

643 644
static int
qemuCapsInitGuest(virCapsPtr caps,
645
                  qemuCapsCachePtr cache,
646 647
                  const char *hostarch,
                  const char *guestarch)
648 649 650 651 652 653 654 655
{
    virCapsGuestPtr guest;
    int i;
    int haskvm = 0;
    int haskqemu = 0;
    char *kvmbin = NULL;
    char *binary = NULL;
    virCapsGuestMachinePtr *machines = NULL;
656
    size_t nmachines = 0;
657 658
    qemuCapsPtr qemubinCaps = NULL;
    qemuCapsPtr kvmbinCaps = NULL;
659 660
    int ret = -1;

J
Ján Tomko 已提交
661
    /* Check for existence of base emulator, or alternate base
662 663
     * which can be used with magic cpu choice
     */
664
    binary = qemuCapsFindBinaryForArch(hostarch, guestarch);
665

666
    /* Ignore binary if extracting version info fails */
667 668 669 670 671 672
    if (binary) {
        if (!(qemubinCaps = qemuCapsCacheLookup(cache, binary))) {
            virResetLastError();
            VIR_FREE(binary);
        }
    }
673 674

    /* qemu-kvm/kvm binaries can only be used if
675 676 677 678 679
     *  - host & guest arches match
     * Or
     *  - hostarch is x86_64 and guest arch is i686
     * The latter simply needs "-cpu qemu32"
     */
680
    if (qemuCapsIsValidForKVM(hostarch, guestarch)) {
681 682 683
        const char *const kvmbins[] = { "/usr/libexec/qemu-kvm", /* RHEL */
                                        "qemu-kvm", /* Fedora */
                                        "kvm" }; /* Upstream .spec */
684

685 686
        for (i = 0; i < ARRAY_CARDINALITY(kvmbins); ++i) {
            kvmbin = virFindFileInPath(kvmbins[i]);
687

688 689
            if (!kvmbin)
                continue;
690

691 692
            if (!(kvmbinCaps = qemuCapsCacheLookup(cache, kvmbin))) {
                virResetLastError();
693 694 695
                VIR_FREE(kvmbin);
                continue;
            }
696

697 698
            if (!binary) {
                binary = kvmbin;
699
                qemubinCaps = kvmbinCaps;
700
                kvmbin = NULL;
701
                kvmbinCaps = NULL;
702
            }
703
            break;
704 705 706 707 708 709
        }
    }

    if (!binary)
        return 0;

710
    if (access("/dev/kvm", F_OK) == 0 &&
711 712
        (qemuCapsGet(qemubinCaps, QEMU_CAPS_KVM) ||
         qemuCapsGet(qemubinCaps, QEMU_CAPS_ENABLE_KVM) ||
713 714 715 716
         kvmbin))
        haskvm = 1;

    if (access("/dev/kqemu", F_OK) == 0 &&
717
        qemuCapsGet(qemubinCaps, QEMU_CAPS_KQEMU))
718
        haskqemu = 1;
J
Jiri Denemark 已提交
719

D
Daniel P. Berrange 已提交
720 721
    if (qemuCapsGetMachineTypesCaps(qemubinCaps, &nmachines, &machines) < 0)
        goto error;
722 723 724 725

    /* We register kvm as the base emulator too, since we can
     * just give -no-kvm to disable acceleration if required */
    if ((guest = virCapabilitiesAddGuest(caps,
D
Daniel P. Berrange 已提交
726
                                         "hvm",
727 728
                                         guestarch,
                                         qemuCapsGetArchWordSize(guestarch),
729 730 731 732 733 734 735 736 737 738
                                         binary,
                                         NULL,
                                         nmachines,
                                         machines)) == NULL)
        goto error;

    machines = NULL;
    nmachines = 0;

    if (caps->host.cpu &&
J
Jiri Denemark 已提交
739
        caps->host.cpu->model &&
740
        qemuCapsGetCPUDefinitions(qemubinCaps, NULL) > 0 &&
741 742 743
        !virCapabilitiesAddGuestFeature(guest, "cpuselection", 1, 0))
        goto error;

744
    if (qemuCapsGet(qemubinCaps, QEMU_CAPS_BOOTINDEX) &&
J
Jiri Denemark 已提交
745
        !virCapabilitiesAddGuestFeature(guest, "deviceboot", 1, 0))
746 747
        goto error;

D
Daniel P. Berrange 已提交
748 749 750 751 752 753 754
    if (virCapabilitiesAddGuestDomain(guest,
                                      "qemu",
                                      NULL,
                                      NULL,
                                      0,
                                      NULL) == NULL)
        goto error;
755

D
Daniel P. Berrange 已提交
756 757 758 759 760 761 762 763
    if (haskqemu &&
        virCapabilitiesAddGuestDomain(guest,
                                      "kqemu",
                                      NULL,
                                      NULL,
                                      0,
                                      NULL) == NULL)
        goto error;
764

D
Daniel P. Berrange 已提交
765 766
    if (haskvm) {
        virCapsGuestDomainPtr dom;
767

D
Daniel P. Berrange 已提交
768 769 770
        if (kvmbin &&
            qemuCapsGetMachineTypesCaps(kvmbinCaps, &nmachines, &machines) < 0)
            goto error;
771

D
Daniel P. Berrange 已提交
772 773 774 775 776 777 778 779
        if ((dom = virCapabilitiesAddGuestDomain(guest,
                                                 "kvm",
                                                 kvmbin ? kvmbin : binary,
                                                 NULL,
                                                 nmachines,
                                                 machines)) == NULL) {
            goto error;
        }
780

D
Daniel P. Berrange 已提交
781 782
        machines = NULL;
        nmachines = 0;
783 784 785

    }

786 787 788 789 790 791 792 793 794 795
    if ((STREQ(guestarch, "i686") ||
         STREQ(guestarch, "x86_64")) &&
        (virCapabilitiesAddGuestFeature(guest, "acpi", 1, 1) == NULL ||
         virCapabilitiesAddGuestFeature(guest, "apic", 1, 0) == NULL))
        goto error;

    if (STREQ(guestarch, "i686") &&
        (virCapabilitiesAddGuestFeature(guest, "pae", 1, 0) == NULL ||
         virCapabilitiesAddGuestFeature(guest, "nonpae", 1, 0) == NULL))
        goto error;
796 797 798 799

    ret = 0;

cleanup:
800 801
    VIR_FREE(binary);
    VIR_FREE(kvmbin);
802 803
    virObjectUnref(qemubinCaps);
    virObjectUnref(kvmbinCaps);
804 805 806 807 808 809 810 811 812 813 814 815

    return ret;

error:
    virCapabilitiesFreeMachines(machines, nmachines);

    goto cleanup;
}


static int
qemuCapsInitCPU(virCapsPtr caps,
816
                const char *arch)
817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835
{
    virCPUDefPtr cpu = NULL;
    union cpuData *data = NULL;
    virNodeInfo nodeinfo;
    int ret = -1;

    if (VIR_ALLOC(cpu) < 0
        || !(cpu->arch = strdup(arch))) {
        virReportOOMError();
        goto error;
    }

    if (nodeGetInfo(NULL, &nodeinfo))
        goto error;

    cpu->type = VIR_CPU_TYPE_HOST;
    cpu->sockets = nodeinfo.sockets;
    cpu->cores = nodeinfo.cores;
    cpu->threads = nodeinfo.threads;
J
Jiri Denemark 已提交
836
    caps->host.cpu = cpu;
837 838 839

    if (!(data = cpuNodeData(arch))
        || cpuDecode(cpu, data, NULL, 0, NULL) < 0)
J
Jiri Denemark 已提交
840
        goto cleanup;
841 842 843 844 845 846 847 848 849 850 851 852 853 854

    ret = 0;

cleanup:
    cpuDataFree(arch, data);

    return ret;

error:
    virCPUDefFree(cpu);
    goto cleanup;
}


855 856
static int qemuDefaultConsoleType(const char *ostype ATTRIBUTE_UNUSED,
                                  const char *arch)
857
{
858 859 860 861
    if (STRPREFIX(arch, "s390"))
        return VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_VIRTIO;
    else
        return VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_SERIAL;
862 863 864
}


865
virCapsPtr qemuCapsInit(qemuCapsCachePtr cache)
866 867 868 869
{
    struct utsname utsname;
    virCapsPtr caps;
    int i;
870 871 872 873 874 875 876
    const char *const arches[] = {
        "i686", "x86_64", "arm",
        "microblaze", "microblazeel",
        "mips", "mipsel", "sparc",
        "ppc", "ppc64", "itanium",
        "s390x"
    };
877 878

    /* Really, this never fails - look at the man-page. */
879
    uname(&utsname);
880 881 882

    if ((caps = virCapabilitiesNew(utsname.machine,
                                   1, 1)) == NULL)
883
        goto error;
884 885 886 887 888 889 890 891 892 893

    /* Using KVM's mac prefix for QEMU too */
    virCapabilitiesSetMacPrefix(caps, (unsigned char[]){ 0x52, 0x54, 0x00 });

    /* Some machines have problematic NUMA toplogy causing
     * unexpected failures. We don't want to break the QEMU
     * driver in this scenario, so log errors & carry on
     */
    if (nodeCapsInitNUMA(caps) < 0) {
        virCapabilitiesFreeNUMAInfo(caps);
894
        VIR_WARN("Failed to query host NUMA topology, disabling NUMA capabilities");
895 896
    }

897 898
    if (qemuCapsInitCPU(caps, utsname.machine) < 0)
        VIR_WARN("Failed to get host CPU");
899

900 901
    /* Add the power management features of the host */

902
    if (virNodeSuspendGetTargetMask(&caps->host.powerMgmt) < 0)
903 904
        VIR_WARN("Failed to get host power management capabilities");

905 906 907 908
    virCapabilitiesAddHostMigrateTransport(caps,
                                           "tcp");

    /* First the pure HVM guests */
909
    for (i = 0 ; i < ARRAY_CARDINALITY(arches) ; i++)
910
        if (qemuCapsInitGuest(caps, cache,
911
                              utsname.machine,
912 913
                              arches[i]) < 0)
            goto error;
914 915 916 917

    /* QEMU Requires an emulator in the XML */
    virCapabilitiesSetEmulatorRequired(caps);

918
    caps->defaultConsoleTargetType = qemuDefaultConsoleType;
919 920 921

    return caps;

922
error:
923 924 925 926 927
    virCapabilitiesFree(caps);
    return NULL;
}


928
static int
929 930 931
qemuCapsComputeCmdFlags(const char *help,
                        unsigned int version,
                        unsigned int is_kvm,
932
                        unsigned int kvm_version,
933
                        qemuCapsPtr caps,
934
                        bool check_yajl ATTRIBUTE_UNUSED)
935 936
{
    const char *p;
R
Richa Marwaha 已提交
937
    const char *fsdev, *netdev;
938 939

    if (strstr(help, "-no-kqemu"))
940
        qemuCapsSet(caps, QEMU_CAPS_KQEMU);
941
    if (strstr(help, "-enable-kqemu"))
942
        qemuCapsSet(caps, QEMU_CAPS_ENABLE_KQEMU);
943
    if (strstr(help, "-no-kvm"))
944
        qemuCapsSet(caps, QEMU_CAPS_KVM);
945
    if (strstr(help, "-enable-kvm"))
946
        qemuCapsSet(caps, QEMU_CAPS_ENABLE_KVM);
947
    if (strstr(help, "-no-reboot"))
948
        qemuCapsSet(caps, QEMU_CAPS_NO_REBOOT);
949
    if (strstr(help, "-name")) {
950
        qemuCapsSet(caps, QEMU_CAPS_NAME);
951
        if (strstr(help, ",process="))
952
            qemuCapsSet(caps, QEMU_CAPS_NAME_PROCESS);
953 954
    }
    if (strstr(help, "-uuid"))
955
        qemuCapsSet(caps, QEMU_CAPS_UUID);
956
    if (strstr(help, "-xen-domid"))
957
        qemuCapsSet(caps, QEMU_CAPS_XEN_DOMID);
958
    else if (strstr(help, "-domid"))
959
        qemuCapsSet(caps, QEMU_CAPS_DOMID);
960
    if (strstr(help, "-drive")) {
961 962
        const char *cache = strstr(help, "cache=");

963
        qemuCapsSet(caps, QEMU_CAPS_DRIVE);
964 965
        if (cache && (p = strchr(cache, ']'))) {
            if (memmem(cache, p - cache, "on|off", sizeof("on|off") - 1) == NULL)
966
                qemuCapsSet(caps, QEMU_CAPS_DRIVE_CACHE_V2);
967
            if (memmem(cache, p - cache, "directsync", sizeof("directsync") - 1))
968
                qemuCapsSet(caps, QEMU_CAPS_DRIVE_CACHE_DIRECTSYNC);
969
            if (memmem(cache, p - cache, "unsafe", sizeof("unsafe") - 1))
970
                qemuCapsSet(caps, QEMU_CAPS_DRIVE_CACHE_UNSAFE);
971
        }
972
        if (strstr(help, "format="))
973
            qemuCapsSet(caps, QEMU_CAPS_DRIVE_FORMAT);
974
        if (strstr(help, "readonly="))
975
            qemuCapsSet(caps, QEMU_CAPS_DRIVE_READONLY);
976
        if (strstr(help, "aio=threads|native"))
977
            qemuCapsSet(caps, QEMU_CAPS_DRIVE_AIO);
O
Osier Yang 已提交
978
        if (strstr(help, "copy-on-read=on|off"))
979
            qemuCapsSet(caps, QEMU_CAPS_DRIVE_COPY_ON_READ);
980
        if (strstr(help, "bps="))
981
            qemuCapsSet(caps, QEMU_CAPS_DRIVE_IOTUNE);
982 983 984 985
    }
    if ((p = strstr(help, "-vga")) && !strstr(help, "-std-vga")) {
        const char *nl = strstr(p, "\n");

986
        qemuCapsSet(caps, QEMU_CAPS_VGA);
987 988

        if (strstr(p, "|qxl"))
989
            qemuCapsSet(caps, QEMU_CAPS_VGA_QXL);
990
        if ((p = strstr(p, "|none")) && p < nl)
991
            qemuCapsSet(caps, QEMU_CAPS_VGA_NONE);
992 993
    }
    if (strstr(help, "-spice"))
994
        qemuCapsSet(caps, QEMU_CAPS_SPICE);
995 996
    if (strstr(help, "-vnc"))
        qemuCapsSet(caps, QEMU_CAPS_VNC);
997 998
    if (strstr(help, "seamless-migration="))
        qemuCapsSet(caps, QEMU_CAPS_SEAMLESS_MIGRATION);
999
    if (strstr(help, "boot=on"))
1000
        qemuCapsSet(caps, QEMU_CAPS_DRIVE_BOOT);
1001
    if (strstr(help, "serial=s"))
1002
        qemuCapsSet(caps, QEMU_CAPS_DRIVE_SERIAL);
1003
    if (strstr(help, "-pcidevice"))
1004
        qemuCapsSet(caps, QEMU_CAPS_PCIDEVICE);
1005
    if (strstr(help, "-mem-path"))
1006
        qemuCapsSet(caps, QEMU_CAPS_MEM_PATH);
1007
    if (strstr(help, "-chardev")) {
1008
        qemuCapsSet(caps, QEMU_CAPS_CHARDEV);
1009
        if (strstr(help, "-chardev spicevmc"))
1010
            qemuCapsSet(caps, QEMU_CAPS_CHARDEV_SPICEVMC);
1011
    }
1012
    if (strstr(help, "-balloon"))
1013
        qemuCapsSet(caps, QEMU_CAPS_BALLOON);
1014
    if (strstr(help, "-device")) {
1015
        qemuCapsSet(caps, QEMU_CAPS_DEVICE);
1016 1017 1018 1019
        /*
         * When -device was introduced, qemu already supported drive's
         * readonly option but didn't advertise that.
         */
1020
        qemuCapsSet(caps, QEMU_CAPS_DRIVE_READONLY);
1021 1022
    }
    if (strstr(help, "-nodefconfig"))
1023
        qemuCapsSet(caps, QEMU_CAPS_NODEFCONFIG);
1024
    if (strstr(help, "-no-user-config"))
1025
        qemuCapsSet(caps, QEMU_CAPS_NO_USER_CONFIG);
1026 1027
    /* The trailing ' ' is important to avoid a bogus match */
    if (strstr(help, "-rtc "))
1028
        qemuCapsSet(caps, QEMU_CAPS_RTC);
1029 1030
    /* to wit */
    if (strstr(help, "-rtc-td-hack"))
1031
        qemuCapsSet(caps, QEMU_CAPS_RTC_TD_HACK);
1032
    if (strstr(help, "-no-hpet"))
1033
        qemuCapsSet(caps, QEMU_CAPS_NO_HPET);
1034
    if (strstr(help, "-no-acpi"))
1035
        qemuCapsSet(caps, QEMU_CAPS_NO_ACPI);
1036
    if (strstr(help, "-no-kvm-pit-reinjection"))
1037
        qemuCapsSet(caps, QEMU_CAPS_NO_KVM_PIT);
1038
    if (strstr(help, "-tdf"))
1039
        qemuCapsSet(caps, QEMU_CAPS_TDF);
1040
    if (strstr(help, "-enable-nesting"))
1041
        qemuCapsSet(caps, QEMU_CAPS_NESTING);
1042
    if (strstr(help, ",menu=on"))
1043
        qemuCapsSet(caps, QEMU_CAPS_BOOT_MENU);
1044 1045
    if (strstr(help, ",reboot-timeout=rb_time"))
        qemuCapsSet(caps, QEMU_CAPS_REBOOT_TIMEOUT);
1046
    if ((fsdev = strstr(help, "-fsdev"))) {
1047
        qemuCapsSet(caps, QEMU_CAPS_FSDEV);
1048
        if (strstr(fsdev, "readonly"))
1049
            qemuCapsSet(caps, QEMU_CAPS_FSDEV_READONLY);
1050
        if (strstr(fsdev, "writeout"))
1051
            qemuCapsSet(caps, QEMU_CAPS_FSDEV_WRITEOUT);
1052
    }
1053
    if (strstr(help, "-smbios type"))
1054
        qemuCapsSet(caps, QEMU_CAPS_SMBIOS_TYPE);
1055 1056
    if (strstr(help, "-sandbox"))
        qemuCapsSet(caps, QEMU_CAPS_SECCOMP_SANDBOX);
1057

R
Richa Marwaha 已提交
1058
    if ((netdev = strstr(help, "-netdev"))) {
1059 1060
        /* Disable -netdev on 0.12 since although it exists,
         * the corresponding netdev_add/remove monitor commands
1061 1062
         * do not, and we need them to be able to do hotplug.
         * But see below about RHEL build. */
R
Richa Marwaha 已提交
1063 1064
        if (version >= 13000) {
            if (strstr(netdev, "bridge"))
1065 1066
                qemuCapsSet(caps, QEMU_CAPS_NETDEV_BRIDGE);
           qemuCapsSet(caps, QEMU_CAPS_NETDEV);
R
Richa Marwaha 已提交
1067
        }
1068 1069 1070
    }

    if (strstr(help, "-sdl"))
1071
        qemuCapsSet(caps, QEMU_CAPS_SDL);
1072 1073 1074
    if (strstr(help, "cores=") &&
        strstr(help, "threads=") &&
        strstr(help, "sockets="))
1075
        qemuCapsSet(caps, QEMU_CAPS_SMP_TOPOLOGY);
1076 1077

    if (version >= 9000)
1078
        qemuCapsSet(caps, QEMU_CAPS_VNC_COLON);
1079 1080

    if (is_kvm && (version >= 10000 || kvm_version >= 74))
1081
        qemuCapsSet(caps, QEMU_CAPS_VNET_HDR);
1082

1083
    if (strstr(help, ",vhost=")) {
1084
        qemuCapsSet(caps, QEMU_CAPS_VHOST_NET);
1085 1086
    }

1087 1088
    /* Do not use -no-shutdown if qemu doesn't support it or SIGTERM handling
     * is most likely buggy when used with -no-shutdown (which applies for qemu
1089
     * 0.14.* and 0.15.0)
1090
     */
1091
    if (strstr(help, "-no-shutdown") && (version < 14000 || version > 15000))
1092
        qemuCapsSet(caps, QEMU_CAPS_NO_SHUTDOWN);
1093

1094 1095 1096
    if (strstr(help, "dump-guest-core=on|off"))
        qemuCapsSet(caps, QEMU_CAPS_DUMP_GUEST_CORE);

1097 1098 1099 1100
    /*
     * Handling of -incoming arg with varying features
     *  -incoming tcp    (kvm >= 79, qemu >= 0.10.0)
     *  -incoming exec   (kvm >= 80, qemu >= 0.10.0)
1101 1102
     *  -incoming unix   (qemu >= 0.12.0)
     *  -incoming fd     (qemu >= 0.12.0)
1103 1104 1105 1106 1107 1108 1109
     *  -incoming stdio  (all earlier kvm)
     *
     * NB, there was a pre-kvm-79 'tcp' support, but it
     * was broken, because it blocked the monitor console
     * while waiting for data, so pretend it doesn't exist
     */
    if (version >= 10000) {
1110 1111
        qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_TCP);
        qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_EXEC);
1112
        if (version >= 12000) {
1113 1114
            qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_UNIX);
            qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_FD);
1115
        }
1116
    } else if (kvm_version >= 79) {
1117
        qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_TCP);
1118
        if (kvm_version >= 80)
1119
            qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_EXEC);
1120
    } else if (kvm_version > 0) {
1121
        qemuCapsSet(caps, QEMU_CAPS_MIGRATE_KVM_STDIO);
1122 1123 1124
    }

    if (version >= 10000)
1125
        qemuCapsSet(caps, QEMU_CAPS_0_10);
1126

1127
    if (version >= 11000)
1128
        qemuCapsSet(caps, QEMU_CAPS_VIRTIO_BLK_SG_IO);
1129

1130 1131 1132
    /* While JSON mode was available in 0.12.0, it was too
     * incomplete to contemplate using. The 0.13.0 release
     * is good enough to use, even though it lacks one or
1133 1134 1135 1136 1137
     * two features. This is also true of versions of qemu
     * built for RHEL, labeled 0.12.1, but with extra text
     * in the help output that mentions that features were
     * backported for libvirt. The benefits of JSON mode now
     * outweigh the downside.
1138
     */
E
Eric Blake 已提交
1139
#if HAVE_YAJL
1140
    if (version >= 13000) {
1141
        qemuCapsSet(caps, QEMU_CAPS_MONITOR_JSON);
1142 1143
    } else if (version >= 12000 &&
               strstr(help, "libvirt")) {
1144 1145
        qemuCapsSet(caps, QEMU_CAPS_MONITOR_JSON);
        qemuCapsSet(caps, QEMU_CAPS_NETDEV);
1146
    }
1147 1148 1149 1150 1151 1152 1153
#else
    /* Starting with qemu 0.15 and newer, upstream qemu no longer
     * promises to keep the human interface stable, but requests that
     * we use QMP (the JSON interface) for everything.  If the user
     * forgot to include YAJL libraries when building their own
     * libvirt but is targetting a newer qemu, we are better off
     * telling them to recompile (the spec file includes the
1154 1155
     * dependency, so distros won't hit this).  This check is
     * also in configure.ac (see $with_yajl).  */
1156 1157 1158
    if (version >= 15000 ||
        (version >= 12000 && strstr(help, "libvirt"))) {
        if (check_yajl) {
1159 1160 1161
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("this qemu binary requires libvirt to be "
                             "compiled with yajl"));
1162 1163
            return -1;
        }
1164
        qemuCapsSet(caps, QEMU_CAPS_NETDEV);
1165
    }
E
Eric Blake 已提交
1166
#endif
1167 1168

    if (version >= 13000)
1169
        qemuCapsSet(caps, QEMU_CAPS_PCI_MULTIFUNCTION);
1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181

    /* Although very new versions of qemu advertise the presence of
     * the rombar option in the output of "qemu -device pci-assign,?",
     * this advertisement was added to the code long after the option
     * itself. According to qemu developers, though, rombar is
     * available in all qemu binaries from release 0.12 onward.
     * Setting the capability this way makes it available in more
     * cases where it might be needed, and shouldn't cause any false
     * positives (in the case that it did, qemu would produce an error
     * log and refuse to start, so it would be immediately obvious).
     */
    if (version >= 12000)
1182
        qemuCapsSet(caps, QEMU_CAPS_PCI_ROMBAR);
1183 1184

    if (version >= 11000)
1185
        qemuCapsSet(caps, QEMU_CAPS_CPU_HOST);
1186 1187 1188

    if (version >= 1002000)
        qemuCapsSet(caps, QEMU_CAPS_DEVICE_VIDEO_PRIMARY);
1189
    return 0;
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
}

/* We parse the output of 'qemu -help' to get the QEMU
 * version number. The first bit is easy, just parse
 * 'QEMU PC emulator version x.y.z'
 * or
 * 'QEMU emulator version x.y.z'.
 *
 * With qemu-kvm, however, that is followed by a string
 * in parenthesis as follows:
 *  - qemu-kvm-x.y.z in stable releases
 *  - kvm-XX for kvm versions up to kvm-85
 *  - qemu-kvm-devel-XX for kvm version kvm-86 and later
 *
 * For qemu-kvm versions before 0.10.z, we need to detect
 * the KVM version number for some features. With 0.10.z
 * and later, we just need the QEMU version number and
 * whether it is KVM QEMU or mainline QEMU.
 */
#define QEMU_VERSION_STR_1  "QEMU emulator version"
#define QEMU_VERSION_STR_2  "QEMU PC emulator version"
#define QEMU_KVM_VER_PREFIX "(qemu-kvm-"
#define KVM_VER_PREFIX      "(kvm-"

#define SKIP_BLANKS(p) do { while ((*(p) == ' ') || (*(p) == '\t')) (p)++; } while (0)

int qemuCapsParseHelpStr(const char *qemu,
                         const char *help,
1218
                         qemuCapsPtr caps,
1219 1220
                         unsigned int *version,
                         unsigned int *is_kvm,
1221 1222
                         unsigned int *kvm_version,
                         bool check_yajl)
1223 1224 1225
{
    unsigned major, minor, micro;
    const char *p = help;
1226
    char *strflags;
1227

1228
    *version = *is_kvm = *kvm_version = 0;
1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245

    if (STRPREFIX(p, QEMU_VERSION_STR_1))
        p += strlen(QEMU_VERSION_STR_1);
    else if (STRPREFIX(p, QEMU_VERSION_STR_2))
        p += strlen(QEMU_VERSION_STR_2);
    else
        goto fail;

    SKIP_BLANKS(p);

    major = virParseNumber(&p);
    if (major == -1 || *p != '.')
        goto fail;

    ++p;

    minor = virParseNumber(&p);
J
Jiri Denemark 已提交
1246
    if (minor == -1)
1247 1248
        goto fail;

J
Jiri Denemark 已提交
1249 1250 1251 1252 1253 1254 1255 1256
    if (*p != '.') {
        micro = 0;
    } else {
        ++p;
        micro = virParseNumber(&p);
        if (micro == -1)
            goto fail;
    }
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277

    SKIP_BLANKS(p);

    if (STRPREFIX(p, QEMU_KVM_VER_PREFIX)) {
        *is_kvm = 1;
        p += strlen(QEMU_KVM_VER_PREFIX);
    } else if (STRPREFIX(p, KVM_VER_PREFIX)) {
        int ret;

        *is_kvm = 1;
        p += strlen(KVM_VER_PREFIX);

        ret = virParseNumber(&p);
        if (ret == -1)
            goto fail;

        *kvm_version = ret;
    }

    *version = (major * 1000 * 1000) + (minor * 1000) + micro;

1278
    if (qemuCapsComputeCmdFlags(help, *version, *is_kvm, *kvm_version,
1279
                                caps, check_yajl) < 0)
1280
        goto cleanup;
1281

1282
    strflags = virBitmapString(caps->flags);
1283 1284 1285
    VIR_DEBUG("Version %u.%u.%u, cooked version %u, flags %s",
              major, minor, micro, *version, NULLSTR(strflags));
    VIR_FREE(strflags);
1286 1287 1288 1289 1290 1291 1292 1293 1294 1295

    if (*kvm_version)
        VIR_DEBUG("KVM version %d detected", *kvm_version);
    else if (*is_kvm)
        VIR_DEBUG("qemu-kvm version %u.%u.%u detected", major, minor, micro);

    return 0;

fail:
    p = strchr(help, '\n');
1296 1297
    if (!p)
        p = strchr(help, '\0');
1298

1299 1300 1301
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("cannot parse %s version number in '%.*s'"),
                   qemu, (int) (p - help), help);
1302

1303
cleanup:
1304 1305 1306
    return -1;
}

1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333

struct qemuCapsStringFlags {
    const char *value;
    int flag;
};


struct qemuCapsStringFlags qemuCapsObjectTypes[] = {
    { "hda-duplex", QEMU_CAPS_HDA_DUPLEX },
    { "hda-micro", QEMU_CAPS_HDA_MICRO },
    { "ccid-card-emulated", QEMU_CAPS_CCID_EMULATED },
    { "ccid-card-passthru", QEMU_CAPS_CCID_PASSTHRU },
    { "piix3-usb-uhci", QEMU_CAPS_PIIX3_USB_UHCI },
    { "piix4-usb-uhci", QEMU_CAPS_PIIX4_USB_UHCI },
    { "usb-ehci", QEMU_CAPS_USB_EHCI },
    { "ich9-usb-ehci1", QEMU_CAPS_ICH9_USB_EHCI1 },
    { "vt82c686b-usb-uhci", QEMU_CAPS_VT82C686B_USB_UHCI },
    { "pci-ohci", QEMU_CAPS_PCI_OHCI },
    { "nec-usb-xhci", QEMU_CAPS_NEC_USB_XHCI },
    { "usb-redir", QEMU_CAPS_USB_REDIR },
    { "usb-hub", QEMU_CAPS_USB_HUB },
    { "ich9-ahci", QEMU_CAPS_ICH9_AHCI },
    { "virtio-blk-s390", QEMU_CAPS_VIRTIO_S390 },
    { "lsi53c895a", QEMU_CAPS_SCSI_LSI },
    { "virtio-scsi-pci", QEMU_CAPS_VIRTIO_SCSI_PCI },
    { "spicevmc", QEMU_CAPS_DEVICE_SPICEVMC },
    { "qxl-vga", QEMU_CAPS_DEVICE_QXL_VGA },
1334
    { "qxl", QEMU_CAPS_DEVICE_QXL },
1335 1336 1337 1338
    { "sga", QEMU_CAPS_SGA },
    { "scsi-block", QEMU_CAPS_SCSI_BLOCK },
    { "scsi-cd", QEMU_CAPS_SCSI_CD },
    { "ide-cd", QEMU_CAPS_IDE_CD },
1339 1340 1341
    { "VGA", QEMU_CAPS_DEVICE_VGA },
    { "cirrus-vga", QEMU_CAPS_DEVICE_CIRRUS_VGA },
    { "vmware-svga", QEMU_CAPS_DEVICE_VMWARE_SVGA },
1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359
};


static struct qemuCapsStringFlags qemuCapsObjectPropsVirtioBlk[] = {
    { "multifunction", QEMU_CAPS_PCI_MULTIFUNCTION },
    { "bootindex", QEMU_CAPS_BOOTINDEX },
    { "ioeventfd", QEMU_CAPS_VIRTIO_IOEVENTFD },
    { "event_idx", QEMU_CAPS_VIRTIO_BLK_EVENT_IDX },
    { "scsi", QEMU_CAPS_VIRTIO_BLK_SCSI },
    { "logical_block_size", QEMU_CAPS_BLOCKIO },
};

static struct qemuCapsStringFlags qemuCapsObjectPropsVirtioNet[] = {
    { "tx", QEMU_CAPS_VIRTIO_TX_ALG },
    { "event_idx", QEMU_CAPS_VIRTIO_NET_EVENT_IDX },
};

static struct qemuCapsStringFlags qemuCapsObjectPropsPciAssign[] = {
1360
    { "rombar", QEMU_CAPS_PCI_ROMBAR },
1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380
    { "configfd", QEMU_CAPS_PCI_CONFIGFD },
    { "bootindex", QEMU_CAPS_PCI_BOOTINDEX },
};

static struct qemuCapsStringFlags qemuCapsObjectPropsScsiDisk[] = {
    { "channel", QEMU_CAPS_SCSI_DISK_CHANNEL },
    { "wwn", QEMU_CAPS_SCSI_DISK_WWN },
};

static struct qemuCapsStringFlags qemuCapsObjectPropsIDEDrive[] = {
    { "wwn", QEMU_CAPS_IDE_DRIVE_WWN },
};

static struct qemuCapsStringFlags qemuCapsObjectPropsPixx4PM[] = {
    { "disable_s3", QEMU_CAPS_DISABLE_S3 },
    { "disable_s4", QEMU_CAPS_DISABLE_S4 },
};

static struct qemuCapsStringFlags qemuCapsObjectPropsUsbRedir[] = {
    { "filter", QEMU_CAPS_USB_REDIR_FILTER },
1381 1382 1383 1384 1385
    { "bootindex", QEMU_CAPS_USB_REDIR_BOOTINDEX },
};

static struct qemuCapsStringFlags qemuCapsObjectPropsUsbHost[] = {
    { "bootindex", QEMU_CAPS_USB_HOST_BOOTINDEX },
1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400
};

struct qemuCapsObjectTypeProps {
    const char *type;
    struct qemuCapsStringFlags *props;
    size_t nprops;
};

static struct qemuCapsObjectTypeProps qemuCapsObjectProps[] = {
    { "virtio-blk-pci", qemuCapsObjectPropsVirtioBlk,
      ARRAY_CARDINALITY(qemuCapsObjectPropsVirtioBlk) },
    { "virtio-net-pci", qemuCapsObjectPropsVirtioNet,
      ARRAY_CARDINALITY(qemuCapsObjectPropsVirtioNet) },
    { "pci-assign", qemuCapsObjectPropsPciAssign,
      ARRAY_CARDINALITY(qemuCapsObjectPropsPciAssign) },
1401 1402
    { "kvm-pci-assign", qemuCapsObjectPropsPciAssign,
      ARRAY_CARDINALITY(qemuCapsObjectPropsPciAssign) },
1403 1404 1405 1406 1407 1408 1409 1410
    { "scsi-disk", qemuCapsObjectPropsScsiDisk,
      ARRAY_CARDINALITY(qemuCapsObjectPropsScsiDisk) },
    { "ide-drive", qemuCapsObjectPropsIDEDrive,
      ARRAY_CARDINALITY(qemuCapsObjectPropsIDEDrive) },
    { "PIIX4_PM", qemuCapsObjectPropsPixx4PM,
      ARRAY_CARDINALITY(qemuCapsObjectPropsPixx4PM) },
    { "usb-redir", qemuCapsObjectPropsUsbRedir,
      ARRAY_CARDINALITY(qemuCapsObjectPropsUsbRedir) },
1411 1412
    { "usb-host", qemuCapsObjectPropsUsbHost,
      ARRAY_CARDINALITY(qemuCapsObjectPropsUsbHost) },
1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581
};


static void
qemuCapsProcessStringFlags(qemuCapsPtr caps,
                           size_t nflags,
                           struct qemuCapsStringFlags *flags,
                           size_t nvalues,
                           char *const*values)
{
    size_t i, j;
    for (i = 0 ; i < nflags ; i++) {
        for (j = 0 ; j < nvalues ; j++) {
            if (STREQ(values[j], flags[i].value)) {
                qemuCapsSet(caps, flags[i].flag);
                break;
            }
        }
    }
}


static void
qemuCapsFreeStringList(size_t len,
                       char **values)
{
    size_t i;
    for (i = 0 ; i < len ; i++)
        VIR_FREE(values[i]);
    VIR_FREE(values);
}


#define OBJECT_TYPE_PREFIX "name \""

static int
qemuCapsParseDeviceStrObjectTypes(const char *str,
                                  char ***types)
{
    const char *tmp = str;
    int ret = -1;
    size_t ntypelist = 0;
    char **typelist = NULL;

    *types = NULL;

    while ((tmp = strstr(tmp, OBJECT_TYPE_PREFIX))) {
        char *end;
        tmp += strlen(OBJECT_TYPE_PREFIX);
        end = strstr(tmp, "\"");
        if (!end) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("Malformed QEMU device list string, missing quote"));
            goto cleanup;
        }

        if (VIR_EXPAND_N(typelist, ntypelist, 1) < 0) {
            virReportOOMError();
            goto cleanup;
        }
        if (!(typelist[ntypelist-1] = strndup(tmp, end-tmp))) {
            virReportOOMError();
            goto cleanup;
        }
    }

    *types = typelist;
    ret = ntypelist;

cleanup:
    if (ret < 0)
        qemuCapsFreeStringList(ntypelist, typelist);
    return ret;
}


static int
qemuCapsParseDeviceStrObjectProps(const char *str,
                                  const char *type,
                                  char ***props)
{
    const char *tmp = str;
    int ret = -1;
    size_t nproplist = 0;
    char **proplist = NULL;

    VIR_DEBUG("Extract type %s", type);
    *props = NULL;

    while ((tmp = strchr(tmp, '\n'))) {
        char *end;
        tmp += 1;

        if (*tmp == '\0')
            break;

        if (STRPREFIX(tmp, OBJECT_TYPE_PREFIX))
            continue;

        if (!STRPREFIX(tmp, type))
            continue;

        tmp += strlen(type);
        if (*tmp != '.')
            continue;
        tmp++;

        end = strstr(tmp, "=");
        if (!end) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("Malformed QEMU device list string, missing '='"));
            goto cleanup;
        }
        if (VIR_EXPAND_N(proplist, nproplist, 1) < 0) {
            virReportOOMError();
            goto cleanup;
        }
        if (!(proplist[nproplist-1] = strndup(tmp, end-tmp))) {
            virReportOOMError();
            goto cleanup;
        }
    }

    *props = proplist;
    ret = nproplist;

cleanup:
    if (ret < 0)
        qemuCapsFreeStringList(nproplist, proplist);
    return ret;
}


int
qemuCapsParseDeviceStr(qemuCapsPtr caps, const char *str)
{
    int nvalues;
    char **values;
    size_t i;

    if ((nvalues = qemuCapsParseDeviceStrObjectTypes(str, &values)) < 0)
        return -1;
    qemuCapsProcessStringFlags(caps,
                               ARRAY_CARDINALITY(qemuCapsObjectTypes),
                               qemuCapsObjectTypes,
                               nvalues, values);
    qemuCapsFreeStringList(nvalues, values);

    for (i = 0 ; i < ARRAY_CARDINALITY(qemuCapsObjectProps); i++) {
        const char *type = qemuCapsObjectProps[i].type;
        if ((nvalues = qemuCapsParseDeviceStrObjectProps(str,
                                                         type,
                                                         &values)) < 0)
            return -1;
        qemuCapsProcessStringFlags(caps,
                                   qemuCapsObjectProps[i].nprops,
                                   qemuCapsObjectProps[i].props,
                                   nvalues, values);
        qemuCapsFreeStringList(nvalues, values);
    }

    /* Prefer -chardev spicevmc (detected earlier) over -device spicevmc */
    if (qemuCapsGet(caps, QEMU_CAPS_CHARDEV_SPICEVMC))
        qemuCapsClear(caps, QEMU_CAPS_DEVICE_SPICEVMC);

    return 0;
}


E
Eric Blake 已提交
1582 1583
static int
qemuCapsExtractDeviceStr(const char *qemu,
1584 1585
                         qemuCapsPtr caps,
                         qemuCapsHookDataPtr hookData)
1586
{
E
Eric Blake 已提交
1587
    char *output = NULL;
1588
    virCommandPtr cmd;
E
Eric Blake 已提交
1589
    int ret = -1;
1590

E
Eric Blake 已提交
1591 1592
    /* Cram together all device-related queries into one invocation;
     * the output format makes it possible to distinguish what we
1593 1594
     * need.  With qemu 0.13.0 and later, unrecognized '-device
     * bogus,?' cause an error in isolation, but are silently ignored
1595
     * in combination with '-device ?'.  Upstream qemu 0.12.x doesn't
1596 1597
     * understand '-device name,?', and always exits with status 1 for
     * the simpler '-device ?', so this function is really only useful
1598
     * if -help includes "device driver,?".  */
1599
    cmd = qemuCapsProbeCommand(qemu, caps, hookData);
1600 1601 1602 1603 1604 1605
    virCommandAddArgList(cmd,
                         "-device", "?",
                         "-device", "pci-assign,?",
                         "-device", "virtio-blk-pci,?",
                         "-device", "virtio-net-pci,?",
                         "-device", "scsi-disk,?",
1606
                         "-device", "PIIX4_PM,?",
1607
                         "-device", "usb-redir,?",
1608
                         "-device", "ide-drive,?",
1609
                         "-device", "usb-host,?",
1610
                         NULL);
1611
    /* qemu -help goes to stdout, but qemu -device ? goes to stderr.  */
E
Eric Blake 已提交
1612
    virCommandSetErrorBuffer(cmd, &output);
1613

1614
    if (virCommandRun(cmd, NULL) < 0)
1615 1616
        goto cleanup;

1617
    ret = qemuCapsParseDeviceStr(caps, output);
1618 1619

cleanup:
E
Eric Blake 已提交
1620
    VIR_FREE(output);
1621
    virCommandFree(cmd);
E
Eric Blake 已提交
1622 1623 1624 1625
    return ret;
}


1626
static void
1627
uname_normalize(struct utsname *ut)
1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639
{
    uname(ut);

    /* Map i386, i486, i586 to i686.  */
    if (ut->machine[0] == 'i' &&
        ut->machine[1] != '\0' &&
        ut->machine[2] == '8' &&
        ut->machine[3] == '6' &&
        ut->machine[4] == '\0')
        ut->machine[1] = '6';
}

1640 1641 1642
int qemuCapsGetDefaultVersion(virCapsPtr caps,
                              qemuCapsCachePtr capsCache,
                              unsigned int *version)
1643 1644 1645
{
    const char *binary;
    struct utsname ut;
1646
    qemuCapsPtr qemucaps;
1647 1648 1649 1650 1651 1652 1653 1654 1655

    if (*version > 0)
        return 0;

    uname_normalize(&ut);
    if ((binary = virCapabilitiesDefaultGuestEmulator(caps,
                                                      "hvm",
                                                      ut.machine,
                                                      "qemu")) == NULL) {
1656 1657
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Cannot find suitable emulator for %s"), ut.machine);
1658 1659 1660
        return -1;
    }

1661
    if (!(qemucaps = qemuCapsCacheLookup(capsCache, binary)))
1662 1663
        return -1;

1664 1665
    *version = qemuCapsGetVersion(qemucaps);
    virObjectUnref(qemucaps);
1666 1667
    return 0;
}
1668 1669


1670 1671 1672


qemuCapsPtr
1673 1674
qemuCapsNew(void)
{
1675
    qemuCapsPtr caps;
1676

1677 1678 1679 1680 1681 1682
    if (qemuCapsInitialize() < 0)
        return NULL;

    if (!(caps = virObjectNew(qemuCapsClass)))
        return NULL;

1683
    if (!(caps->flags = virBitmapNew(QEMU_CAPS_LAST)))
1684
        goto no_memory;
1685 1686

    return caps;
1687 1688 1689 1690 1691

no_memory:
    virReportOOMError();
    virObjectUnref(caps);
    return NULL;
1692 1693 1694
}


1695 1696 1697 1698 1699 1700 1701 1702 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
qemuCapsPtr qemuCapsNewCopy(qemuCapsPtr caps)
{
    qemuCapsPtr ret = qemuCapsNew();
    size_t i;

    if (!ret)
        return NULL;

    virBitmapCopy(ret->flags, caps->flags);

    ret->version = caps->version;
    ret->kvmVersion = caps->kvmVersion;

    if (caps->arch &&
        !(ret->arch = strdup(caps->arch)))
        goto no_memory;

    if (VIR_ALLOC_N(ret->cpuDefinitions, caps->ncpuDefinitions) < 0)
        goto no_memory;
    ret->ncpuDefinitions = caps->ncpuDefinitions;
    for (i = 0 ; i < caps->ncpuDefinitions ; i++) {
        if (!(ret->cpuDefinitions[i] = strdup(caps->cpuDefinitions[i])))
            goto no_memory;
    }

    if (VIR_ALLOC_N(ret->machineTypes, caps->nmachineTypes) < 0)
        goto no_memory;
    if (VIR_ALLOC_N(ret->machineAliases, caps->nmachineTypes) < 0)
        goto no_memory;
    ret->nmachineTypes = caps->nmachineTypes;
    for (i = 0 ; i < caps->nmachineTypes ; i++) {
        if (!(ret->machineTypes[i] = strdup(caps->machineTypes[i])))
            goto no_memory;
        if (caps->machineAliases[i] &&
            !(ret->machineAliases[i] = strdup(caps->machineAliases[i])))
            goto no_memory;
    }

    return ret;

no_memory:
    virReportOOMError();
    virObjectUnref(ret);
    return NULL;
}


1742 1743 1744
void qemuCapsDispose(void *obj)
{
    qemuCapsPtr caps = obj;
1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759
    size_t i;

    VIR_FREE(caps->arch);

    for (i = 0 ; i < caps->nmachineTypes ; i++) {
        VIR_FREE(caps->machineTypes[i]);
        VIR_FREE(caps->machineAliases[i]);
    }
    VIR_FREE(caps->machineTypes);
    VIR_FREE(caps->machineAliases);

    for (i = 0 ; i < caps->ncpuDefinitions ; i++) {
        VIR_FREE(caps->cpuDefinitions[i]);
    }
    VIR_FREE(caps->cpuDefinitions);
1760 1761

    virBitmapFree(caps->flags);
1762 1763

    VIR_FREE(caps->binary);
1764 1765
}

1766
void
1767
qemuCapsSet(qemuCapsPtr caps,
1768 1769
            enum qemuCapsFlags flag)
{
1770
    ignore_value(virBitmapSetBit(caps->flags, flag));
1771 1772 1773 1774
}


void
1775
qemuCapsSetList(qemuCapsPtr caps, ...)
1776 1777 1778 1779 1780 1781
{
    va_list list;
    int flag;

    va_start(list, caps);
    while ((flag = va_arg(list, int)) < QEMU_CAPS_LAST)
1782
        ignore_value(virBitmapSetBit(caps->flags, flag));
1783
    va_end(list);
1784 1785 1786 1787
}


void
1788
qemuCapsClear(qemuCapsPtr caps,
1789 1790
              enum qemuCapsFlags flag)
{
1791 1792 1793 1794 1795 1796 1797
    ignore_value(virBitmapClearBit(caps->flags, flag));
}


char *qemuCapsFlagsString(qemuCapsPtr caps)
{
    return virBitmapString(caps->flags);
1798 1799 1800 1801
}


bool
1802
qemuCapsGet(qemuCapsPtr caps,
1803 1804
            enum qemuCapsFlags flag)
{
1805 1806
    bool b;

1807
    if (!caps || virBitmapGetBit(caps->flags, flag, &b) < 0)
1808 1809 1810
        return false;
    else
        return b;
1811
}
1812 1813


1814 1815 1816 1817 1818
const char *qemuCapsGetBinary(qemuCapsPtr caps)
{
    return caps->binary;
}

1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836
const char *qemuCapsGetArch(qemuCapsPtr caps)
{
    return caps->arch;
}


unsigned int qemuCapsGetVersion(qemuCapsPtr caps)
{
    return caps->version;
}


unsigned int qemuCapsGetKVMVersion(qemuCapsPtr caps)
{
    return caps->kvmVersion;
}


1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854
int qemuCapsAddCPUDefinition(qemuCapsPtr caps,
                             const char *name)
{
    char *tmp = strdup(name);
    if (!tmp) {
        virReportOOMError();
        return -1;
    }
    if (VIR_EXPAND_N(caps->cpuDefinitions, caps->ncpuDefinitions, 1) < 0) {
        VIR_FREE(tmp);
        virReportOOMError();
        return -1;
    }
    caps->cpuDefinitions[caps->ncpuDefinitions-1] = tmp;
    return 0;
}


1855 1856 1857
size_t qemuCapsGetCPUDefinitions(qemuCapsPtr caps,
                                 char ***names)
{
1858 1859
    if (names)
        *names = caps->cpuDefinitions;
1860 1861 1862 1863 1864 1865 1866
    return caps->ncpuDefinitions;
}


size_t qemuCapsGetMachineTypes(qemuCapsPtr caps,
                               char ***names)
{
1867 1868
    if (names)
        *names = caps->machineTypes;
1869 1870 1871
    return caps->nmachineTypes;
}

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 1906 1907 1908 1909 1910
int qemuCapsGetMachineTypesCaps(qemuCapsPtr caps,
                                size_t *nmachines,
                                virCapsGuestMachinePtr **machines)
{
    size_t i;

    *nmachines = 0;
    *machines = NULL;
    if (VIR_ALLOC_N(*machines, caps->nmachineTypes) < 0)
        goto no_memory;
    *nmachines = caps->nmachineTypes;

    for (i = 0 ; i < caps->nmachineTypes ; i++) {
        virCapsGuestMachinePtr mach;
        if (VIR_ALLOC(mach) < 0)
            goto no_memory;
        if (caps->machineAliases[i]) {
            if (!(mach->name = strdup(caps->machineAliases[i])))
                goto no_memory;
            if (!(mach->canonical = strdup(caps->machineTypes[i])))
                goto no_memory;
        } else {
            if (!(mach->name = strdup(caps->machineTypes[i])))
                goto no_memory;
        }
        (*machines)[i] = mach;
    }

    return 0;

no_memory:
    virCapabilitiesFreeMachines(*machines, *nmachines);
    *nmachines = 0;
    *machines = NULL;
    return -1;
}



1911 1912 1913 1914 1915 1916

const char *qemuCapsGetCanonicalMachine(qemuCapsPtr caps,
                                        const char *name)
{
    size_t i;

1917 1918 1919
    if (!name)
        return NULL;

1920 1921 1922 1923 1924 1925 1926 1927 1928
    for (i = 0 ; i < caps->nmachineTypes ; i++) {
        if (!caps->machineAliases[i])
            continue;
        if (STREQ(caps->machineAliases[i], name))
            return caps->machineTypes[i];
    }

    return name;
}
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
static int
qemuCapsProbeQMPCommands(qemuCapsPtr caps,
                         qemuMonitorPtr mon)
{
    char **commands = NULL;
    int ncommands;
    size_t i;

    if ((ncommands = qemuMonitorGetCommands(mon, &commands)) < 0)
        return -1;

    for (i = 0 ; i < ncommands ; i++) {
        char *name = commands[i];
        if (STREQ(name, "system_wakeup"))
            qemuCapsSet(caps, QEMU_CAPS_WAKEUP);
        else if (STREQ(name, "transaction"))
            qemuCapsSet(caps, QEMU_CAPS_TRANSACTION);
        else if (STREQ(name, "block_job_cancel"))
            qemuCapsSet(caps, QEMU_CAPS_BLOCKJOB_SYNC);
        else if (STREQ(name, "block-job-cancel"))
            qemuCapsSet(caps, QEMU_CAPS_BLOCKJOB_ASYNC);
        else if (STREQ(name, "dump-guest-memory"))
            qemuCapsSet(caps, QEMU_CAPS_DUMP_GUEST_MEMORY);
1954 1955 1956 1957
        else if (STREQ(name, "query-spice"))
            qemuCapsSet(caps, QEMU_CAPS_SPICE);
        else if (STREQ(name, "query-kvm"))
            qemuCapsSet(caps, QEMU_CAPS_KVM);
1958 1959
        else if (STREQ(name, "block-commit"))
            qemuCapsSet(caps, QEMU_CAPS_BLOCK_COMMIT);
1960 1961
        else if (STREQ(name, "query-vnc"))
            qemuCapsSet(caps, QEMU_CAPS_VNC);
1962 1963
        else if (STREQ(name, "drive-mirror"))
            qemuCapsSet(caps, QEMU_CAPS_DRIVE_MIRROR);
1964 1965
        else if (STREQ(name, "blockdev-snapshot-sync"))
            qemuCapsSet(caps, QEMU_CAPS_DISK_SNAPSHOT);
1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989
        VIR_FREE(name);
    }
    VIR_FREE(commands);

    return 0;
}


static int
qemuCapsProbeQMPEvents(qemuCapsPtr caps,
                       qemuMonitorPtr mon)
{
    char **events = NULL;
    int nevents;
    size_t i;

    if ((nevents = qemuMonitorGetEvents(mon, &events)) < 0)
        return -1;

    for (i = 0 ; i < nevents ; i++) {
        char *name = events[i];

        if (STREQ(name, "BALLOON_CHANGE"))
            qemuCapsSet(caps, QEMU_CAPS_BALLOON_EVENT);
1990 1991
        if (STREQ(name, "SPICE_MIGRATE_COMPLETED"))
            qemuCapsSet(caps, QEMU_CAPS_SEAMLESS_MIGRATION);
1992 1993 1994 1995 1996 1997 1998 1999
        VIR_FREE(name);
    }
    VIR_FREE(events);

    return 0;
}


2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044
static int
qemuCapsProbeQMPObjects(qemuCapsPtr caps,
                        qemuMonitorPtr mon)
{
    int nvalues;
    char **values;
    size_t i;

    if ((nvalues = qemuMonitorGetObjectTypes(mon, &values)) < 0)
        return -1;
    qemuCapsProcessStringFlags(caps,
                               ARRAY_CARDINALITY(qemuCapsObjectTypes),
                               qemuCapsObjectTypes,
                               nvalues, values);
    qemuCapsFreeStringList(nvalues, values);

    for (i = 0 ; i < ARRAY_CARDINALITY(qemuCapsObjectProps); i++) {
        const char *type = qemuCapsObjectProps[i].type;
        if ((nvalues = qemuMonitorGetObjectProps(mon,
                                                 type,
                                                 &values)) < 0)
            return -1;
        qemuCapsProcessStringFlags(caps,
                                   qemuCapsObjectProps[i].nprops,
                                   qemuCapsObjectProps[i].props,
                                   nvalues, values);
        qemuCapsFreeStringList(nvalues, values);
    }

    /* Prefer -chardev spicevmc (detected earlier) over -device spicevmc */
    if (qemuCapsGet(caps, QEMU_CAPS_CHARDEV_SPICEVMC))
        qemuCapsClear(caps, QEMU_CAPS_DEVICE_SPICEVMC);

    return 0;
}


static int
qemuCapsProbeQMPMachineTypes(qemuCapsPtr caps,
                             qemuMonitorPtr mon)
{
    qemuMonitorMachineInfoPtr *machines = NULL;
    int nmachines = 0;
    int ret = -1;
    size_t i;
2045
    size_t defIdx = 0;
2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060

    if ((nmachines = qemuMonitorGetMachines(mon, &machines)) < 0)
        goto cleanup;

    if (VIR_ALLOC_N(caps->machineTypes, nmachines) < 0) {
        virReportOOMError();
        goto cleanup;
    }
    if (VIR_ALLOC_N(caps->machineAliases, nmachines) < 0) {
        virReportOOMError();
        goto cleanup;
    }

    for (i = 0 ; i < nmachines ; i++) {
        if (machines[i]->alias) {
2061
            if (!(caps->machineAliases[i] = strdup(machines[i]->alias))) {
2062 2063 2064 2065
                virReportOOMError();
                goto cleanup;
            }
        }
2066 2067 2068 2069
        if (!(caps->machineTypes[i] = strdup(machines[i]->name))) {
            virReportOOMError();
            goto cleanup;
        }
2070 2071
        if (machines[i]->isDefault)
            defIdx = i;
2072
    }
2073 2074 2075 2076
    caps->nmachineTypes = nmachines;

    if (defIdx)
        qemuSetDefaultMachine(caps, defIdx);
2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104

    ret = 0;

cleanup:
    for (i = 0 ; i < nmachines ; i++)
        qemuMonitorMachineInfoFree(machines[i]);
    VIR_FREE(machines);
    return ret;
}


static int
qemuCapsProbeQMPCPUDefinitions(qemuCapsPtr caps,
                               qemuMonitorPtr mon)
{
    int ncpuDefinitions;
    char **cpuDefinitions;

    if ((ncpuDefinitions = qemuMonitorGetCPUDefinitions(mon, &cpuDefinitions)) < 0)
        return -1;

    caps->ncpuDefinitions = ncpuDefinitions;
    caps->cpuDefinitions = cpuDefinitions;

    return 0;
}


2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131
static int
qemuCapsProbeQMPKVMState(qemuCapsPtr caps,
                         qemuMonitorPtr mon)
{
    bool enabled = false;
    bool present = false;

    if (!qemuCapsGet(caps, QEMU_CAPS_KVM))
        return 0;

    if (qemuMonitorGetKVMState(mon, &enabled, &present) < 0)
        return -1;

    /* The QEMU_CAPS_KVM flag was initially set according to the QEMU
     * reporting the recognition of 'query-kvm' QMP command, but the
     * flag means whether the KVM is enabled by default and should be
     * disabled in case we want SW emulated machine, so let's fix that
     * if it's true. */
    if (!enabled) {
        qemuCapsClear(caps, QEMU_CAPS_KVM);
        qemuCapsSet(caps, QEMU_CAPS_ENABLE_KVM);
    }

    return 0;
}


2132 2133 2134 2135 2136
int qemuCapsProbeQMP(qemuCapsPtr caps,
                     qemuMonitorPtr mon)
{
    VIR_DEBUG("caps=%p mon=%p", caps, mon);

2137 2138 2139
    if (caps->usedQMP)
        return 0;

2140 2141 2142 2143 2144 2145 2146 2147 2148 2149
    if (qemuCapsProbeQMPCommands(caps, mon) < 0)
        return -1;

    if (qemuCapsProbeQMPEvents(caps, mon) < 0)
        return -1;

    return 0;
}


2150 2151
#define QEMU_SYSTEM_PREFIX "qemu-system-"

2152
static int
2153
qemuCapsInitHelp(qemuCapsPtr caps, uid_t runUid, gid_t runGid)
2154
{
2155
    virCommandPtr cmd = NULL;
2156 2157
    unsigned int is_kvm;
    char *help = NULL;
2158 2159 2160
    int ret = -1;
    const char *tmp;
    struct utsname ut;
2161
    qemuCapsHookData hookData;
2162

2163
    VIR_DEBUG("caps=%p", caps);
2164

2165
    tmp = strstr(caps->binary, QEMU_SYSTEM_PREFIX);
2166 2167
    if (tmp) {
        tmp += strlen(QEMU_SYSTEM_PREFIX);
2168 2169 2170 2171

        /* For historical compat we use 'itanium' as arch name */
        if (STREQ(tmp, "ia64"))
            tmp = "itanium";
2172 2173
        else if (STREQ(tmp, "i386"))
            tmp = "i686";
2174 2175 2176 2177
    } else {
        uname_normalize(&ut);
        tmp = ut.machine;
    }
2178 2179 2180
    if (!(caps->arch = strdup(tmp))) {
        virReportOOMError();
        goto cleanup;
2181 2182
    }

2183 2184 2185
    hookData.runUid = runUid;
    hookData.runGid = runGid;
    cmd = qemuCapsProbeCommand(caps->binary, NULL, &hookData);
2186 2187 2188 2189
    virCommandAddArgList(cmd, "-help", NULL);
    virCommandSetOutputBuffer(cmd, &help);

    if (virCommandRun(cmd, NULL) < 0)
2190
        goto cleanup;
2191

2192 2193
    if (qemuCapsParseHelpStr(caps->binary,
                             help, caps,
2194 2195 2196 2197
                             &caps->version,
                             &is_kvm,
                             &caps->kvmVersion,
                             false) < 0)
2198
        goto cleanup;
2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214

    /* Currently only x86_64 and i686 support PCI-multibus. */
    if (STREQLEN(caps->arch, "x86_64", 6) ||
        STREQLEN(caps->arch, "i686", 4)) {
        qemuCapsSet(caps, QEMU_CAPS_PCI_MULTIBUS);
    }

    /* S390 and probably other archs do not support no-acpi -
       maybe the qemu option parsing should be re-thought. */
    if (STRPREFIX(caps->arch, "s390"))
        qemuCapsClear(caps, QEMU_CAPS_NO_ACPI);

    /* qemuCapsExtractDeviceStr will only set additional caps if qemu
     * understands the 0.13.0+ notion of "-device driver,".  */
    if (qemuCapsGet(caps, QEMU_CAPS_DEVICE) &&
        strstr(help, "-device driver,?") &&
2215
        qemuCapsExtractDeviceStr(caps->binary, caps, &hookData) < 0)
2216
        goto cleanup;
2217

2218
    if (qemuCapsProbeCPUModels(caps, &hookData) < 0)
2219
        goto cleanup;
2220

2221
    if (qemuCapsProbeMachineTypes(caps, &hookData) < 0)
2222
        goto cleanup;
2223

2224
    ret = 0;
2225
cleanup:
2226
    virCommandFree(cmd);
2227
    VIR_FREE(help);
2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297
    return ret;
}


static void qemuCapsMonitorNotify(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                  virDomainObjPtr vm ATTRIBUTE_UNUSED)
{
}

static qemuMonitorCallbacks callbacks = {
    .eofNotify = qemuCapsMonitorNotify,
    .errorNotify = qemuCapsMonitorNotify,
};


/* Capabilities that we assume are always enabled
 * for QEMU >= 1.2.0
 */
static void
qemuCapsInitQMPBasic(qemuCapsPtr caps)
{
    qemuCapsSet(caps, QEMU_CAPS_VNC_COLON);
    qemuCapsSet(caps, QEMU_CAPS_NO_REBOOT);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE);
    qemuCapsSet(caps, QEMU_CAPS_NAME);
    qemuCapsSet(caps, QEMU_CAPS_UUID);
    qemuCapsSet(caps, QEMU_CAPS_VNET_HDR);
    qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_TCP);
    qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_EXEC);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE_CACHE_V2);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE_FORMAT);
    qemuCapsSet(caps, QEMU_CAPS_VGA);
    qemuCapsSet(caps, QEMU_CAPS_0_10);
    qemuCapsSet(caps, QEMU_CAPS_MEM_PATH);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE_SERIAL);
    qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_UNIX);
    qemuCapsSet(caps, QEMU_CAPS_CHARDEV);
    qemuCapsSet(caps, QEMU_CAPS_MONITOR_JSON);
    qemuCapsSet(caps, QEMU_CAPS_BALLOON);
    qemuCapsSet(caps, QEMU_CAPS_DEVICE);
    qemuCapsSet(caps, QEMU_CAPS_SDL);
    qemuCapsSet(caps, QEMU_CAPS_SMP_TOPOLOGY);
    qemuCapsSet(caps, QEMU_CAPS_NETDEV);
    qemuCapsSet(caps, QEMU_CAPS_RTC);
    qemuCapsSet(caps, QEMU_CAPS_VHOST_NET);
    qemuCapsSet(caps, QEMU_CAPS_NO_HPET);
    qemuCapsSet(caps, QEMU_CAPS_NODEFCONFIG);
    qemuCapsSet(caps, QEMU_CAPS_BOOT_MENU);
    qemuCapsSet(caps, QEMU_CAPS_FSDEV);
    qemuCapsSet(caps, QEMU_CAPS_NAME_PROCESS);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE_READONLY);
    qemuCapsSet(caps, QEMU_CAPS_SMBIOS_TYPE);
    qemuCapsSet(caps, QEMU_CAPS_VGA_NONE);
    qemuCapsSet(caps, QEMU_CAPS_MIGRATE_QEMU_FD);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE_AIO);
    qemuCapsSet(caps, QEMU_CAPS_CHARDEV_SPICEVMC);
    qemuCapsSet(caps, QEMU_CAPS_DEVICE_QXL_VGA);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE_CACHE_DIRECTSYNC);
    qemuCapsSet(caps, QEMU_CAPS_NO_SHUTDOWN);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE_CACHE_UNSAFE);
    qemuCapsSet(caps, QEMU_CAPS_NO_ACPI);
    qemuCapsSet(caps, QEMU_CAPS_FSDEV_READONLY);
    qemuCapsSet(caps, QEMU_CAPS_VIRTIO_BLK_SG_IO);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE_COPY_ON_READ);
    qemuCapsSet(caps, QEMU_CAPS_CPU_HOST);
    qemuCapsSet(caps, QEMU_CAPS_FSDEV_WRITEOUT);
    qemuCapsSet(caps, QEMU_CAPS_DRIVE_IOTUNE);
    qemuCapsSet(caps, QEMU_CAPS_WAKEUP);
    qemuCapsSet(caps, QEMU_CAPS_NO_USER_CONFIG);
    qemuCapsSet(caps, QEMU_CAPS_NETDEV_BRIDGE);
2298
    qemuCapsSet(caps, QEMU_CAPS_SECCOMP_SANDBOX);
2299 2300 2301 2302 2303
}


static int
qemuCapsInitQMP(qemuCapsPtr caps,
2304
                const char *libDir,
2305 2306 2307
                const char *runDir,
                uid_t runUid,
                gid_t runGid)
2308 2309 2310 2311 2312
{
    int ret = -1;
    virCommandPtr cmd = NULL;
    qemuMonitorPtr mon = NULL;
    int major, minor, micro;
2313
    char *package = NULL;
2314 2315 2316 2317
    int status = 0;
    virDomainChrSourceDef config;
    char *monarg = NULL;
    char *monpath = NULL;
2318
    char *pidfile = NULL;
2319
    qemuCapsHookData hookData;
2320

2321 2322 2323
    /* the ".sock" sufix is important to avoid a possible clash with a qemu
     * domain called "capabilities"
     */
2324 2325 2326 2327 2328 2329 2330 2331 2332
    if (virAsprintf(&monpath, "%s/%s", libDir, "capabilities.monitor.sock") < 0) {
        virReportOOMError();
        goto cleanup;
    }
    if (virAsprintf(&monarg, "unix:%s,server,nowait", monpath) < 0) {
        virReportOOMError();
        goto cleanup;
    }

2333 2334 2335 2336 2337 2338 2339 2340
    /* ".pidfile" suffix is used rather than ".pid" to avoid a possible clash
     * with a qemu domain called "capabilities"
     */
    if (virAsprintf(&pidfile, "%s/%s", runDir, "capabilities.pidfile") < 0) {
        virReportOOMError();
        goto cleanup;
    }

2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357
    memset(&config, 0, sizeof(config));
    config.type = VIR_DOMAIN_CHR_TYPE_UNIX;
    config.data.nix.path = monpath;
    config.data.nix.listen = false;

    VIR_DEBUG("Try to get caps via QMP caps=%p", caps);

    cmd = virCommandNewArgList(caps->binary,
                               "-S",
                               "-no-user-config",
                               "-nodefaults",
                               "-nographic",
                               "-M", "none",
                               "-qmp", monarg,
                               NULL);
    virCommandAddEnvPassCommon(cmd);
    virCommandClearCaps(cmd);
2358 2359 2360 2361 2362
    hookData.runUid = runUid;
    hookData.runGid = runGid;
    virCommandSetPreExecHook(cmd, qemuCapsHook, &hookData);
    virCommandSetPidFile(cmd, pidfile);
    virCommandDaemonize(cmd);
2363 2364 2365 2366 2367 2368 2369 2370 2371 2372

    if (virCommandRun(cmd, &status) < 0)
        goto cleanup;

    if (status != 0) {
        ret = 0;
        VIR_DEBUG("QEMU %s exited with status %d", caps->binary, status);
        goto cleanup;
    }

2373 2374
    if (!(mon = qemuMonitorOpen(NULL, &config, true, &callbacks))) {
        ret = 0;
2375
        goto cleanup;
2376
    }
2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400

    qemuMonitorLock(mon);

    if (qemuMonitorSetCapabilities(mon) < 0) {
        virErrorPtr err = virGetLastError();
        VIR_DEBUG("Failed to set monitor capabilities %s",
                  err ? err->message : "<unknown problem>");
        ret = 0;
        goto cleanup;
    }

    if (qemuMonitorGetVersion(mon,
                              &major, &minor, &micro,
                              &package) < 0) {
        virErrorPtr err = virGetLastError();
        VIR_DEBUG("Failed to query monitor version %s",
                  err ? err->message : "<unknown problem>");
        ret = 0;
        goto cleanup;
    }

    VIR_DEBUG("Got version %d.%d.%d (%s)",
              major, minor, micro, NULLSTR(package));

2401
    if (major < 1 || (major == 1 && minor < 2)) {
2402 2403 2404 2405 2406
        VIR_DEBUG("Not new enough for QMP capabilities detection");
        ret = 0;
        goto cleanup;
    }

2407
    caps->version = major * 1000000 + minor * 1000 + micro;
2408 2409 2410 2411 2412 2413 2414
    caps->usedQMP = true;

    qemuCapsInitQMPBasic(caps);

    if (!(caps->arch = qemuMonitorGetTargetArch(mon)))
        goto cleanup;

2415 2416 2417 2418 2419 2420 2421 2422
    /* Map i386, i486, i586 to i686.  */
    if (caps->arch[0] == 'i' &&
        caps->arch[1] != '\0' &&
        caps->arch[2] == '8' &&
        caps->arch[3] == '6' &&
        caps->arch[4] == '\0')
        caps->arch[1] = '6';

2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443
    /* Currently only x86_64 and i686 support PCI-multibus. */
    if (STREQLEN(caps->arch, "x86_64", 6) ||
        STREQLEN(caps->arch, "i686", 4)) {
        qemuCapsSet(caps, QEMU_CAPS_PCI_MULTIBUS);
    }

    /* S390 and probably other archs do not support no-acpi -
       maybe the qemu option parsing should be re-thought. */
    if (STRPREFIX(caps->arch, "s390"))
        qemuCapsClear(caps, QEMU_CAPS_NO_ACPI);

    if (qemuCapsProbeQMPCommands(caps, mon) < 0)
        goto cleanup;
    if (qemuCapsProbeQMPEvents(caps, mon) < 0)
        goto cleanup;
    if (qemuCapsProbeQMPObjects(caps, mon) < 0)
        goto cleanup;
    if (qemuCapsProbeQMPMachineTypes(caps, mon) < 0)
        goto cleanup;
    if (qemuCapsProbeQMPCPUDefinitions(caps, mon) < 0)
        goto cleanup;
2444 2445
    if (qemuCapsProbeQMPKVMState(caps, mon) < 0)
        goto cleanup;
2446 2447 2448 2449 2450 2451 2452 2453

    ret = 0;

cleanup:
    if (mon)
        qemuMonitorUnlock(mon);
    qemuMonitorClose(mon);
    virCommandAbort(cmd);
2454
    virCommandFree(cmd);
2455 2456
    VIR_FREE(monarg);
    VIR_FREE(monpath);
2457
    VIR_FREE(package);
2458 2459 2460 2461 2462 2463 2464

    if (pidfile) {
        char ebuf[1024];
        pid_t pid;
        int rc;

        if ((rc = virPidFileReadPath(pidfile, &pid)) < 0) {
E
Eric Blake 已提交
2465
            VIR_DEBUG("Failed to read pidfile %s: %s",
2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476
                      pidfile, virStrerror(-rc, ebuf, sizeof(ebuf)));
        } else {
            VIR_DEBUG("Killing QMP caps process %lld", (long long) pid);
            if (virProcessKill(pid, SIGKILL) < 0 && errno != ESRCH)
                VIR_ERROR(_("Failed to kill process %lld: %s"),
                          (long long) pid,
                          virStrerror(errno, ebuf, sizeof(ebuf)));
        }
        unlink(pidfile);
        VIR_FREE(pidfile);
    }
2477 2478 2479 2480 2481
    return ret;
}


qemuCapsPtr qemuCapsNewForBinary(const char *binary,
2482
                                 const char *libDir,
2483 2484 2485
                                 const char *runDir,
                                 uid_t runUid,
                                 gid_t runGid)
2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512
{
    qemuCapsPtr caps = qemuCapsNew();
    struct stat sb;
    int rv;

    if (!(caps->binary = strdup(binary)))
        goto no_memory;

    /* We would also want to check faccessat if we cared about ACLs,
     * but we don't.  */
    if (stat(binary, &sb) < 0) {
        virReportSystemError(errno, _("Cannot check QEMU binary %s"),
                             binary);
        goto error;
    }
    caps->mtime = sb.st_mtime;

    /* Make sure the binary we are about to try exec'ing exists.
     * Technically we could catch the exec() failure, but that's
     * in a sub-process so it's hard to feed back a useful error.
     */
    if (!virFileIsExecutable(binary)) {
        virReportSystemError(errno, _("QEMU binary %s is not executable"),
                             binary);
        goto error;
    }

2513
    if ((rv = qemuCapsInitQMP(caps, libDir, runDir, runUid, runGid)) < 0)
2514 2515 2516
        goto error;

    if (!caps->usedQMP &&
2517
        qemuCapsInitHelp(caps, runUid, runGid) < 0)
2518 2519
        goto error;

2520 2521 2522 2523 2524 2525 2526
    return caps;

no_memory:
    virReportOOMError();
error:
    virObjectUnref(caps);
    caps = NULL;
2527
    return NULL;
2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542
}


bool qemuCapsIsValid(qemuCapsPtr caps)
{
    struct stat sb;

    if (!caps->binary)
        return true;

    if (stat(caps->binary, &sb) < 0)
        return false;

    return sb.st_mtime == caps->mtime;
}
2543 2544 2545 2546 2547 2548 2549 2550 2551 2552


static void
qemuCapsHashDataFree(void *payload, const void *key ATTRIBUTE_UNUSED)
{
    virObjectUnref(payload);
}


qemuCapsCachePtr
2553 2554
qemuCapsCacheNew(const char *libDir, const char *runDir,
                 uid_t runUid, gid_t runGid)
2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571
{
    qemuCapsCachePtr cache;

    if (VIR_ALLOC(cache) < 0) {
        virReportOOMError();
        return NULL;
    }

    if (virMutexInit(&cache->lock) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Unable to initialize mutex"));
        VIR_FREE(cache);
        return NULL;
    }

    if (!(cache->binaries = virHashCreate(10, qemuCapsHashDataFree)))
        goto error;
2572 2573
    if (!(cache->libDir = strdup(libDir)) ||
        !(cache->runDir = strdup(runDir))) {
2574 2575 2576
        virReportOOMError();
        goto error;
    }
2577

2578 2579 2580
    cache->runUid = runUid;
    cache->runGid = runGid;

2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604
    return cache;

error:
    qemuCapsCacheFree(cache);
    return NULL;
}


qemuCapsPtr
qemuCapsCacheLookup(qemuCapsCachePtr cache, const char *binary)
{
    qemuCapsPtr ret = NULL;
    virMutexLock(&cache->lock);
    ret = virHashLookup(cache->binaries, binary);
    if (ret &&
        !qemuCapsIsValid(ret)) {
        VIR_DEBUG("Cached capabilities %p no longer valid for %s",
                  ret, binary);
        virHashRemoveEntry(cache->binaries, binary);
        ret = NULL;
    }
    if (!ret) {
        VIR_DEBUG("Creating capabilities for %s",
                  binary);
2605 2606
        ret = qemuCapsNewForBinary(binary, cache->libDir, cache->runDir,
                                   cache->runUid, cache->runGid);
2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643
        if (ret) {
            VIR_DEBUG("Caching capabilities %p for %s",
                      ret, binary);
            if (virHashAddEntry(cache->binaries, binary, ret) < 0) {
                virObjectUnref(ret);
                ret = NULL;
            }
        }
    }
    VIR_DEBUG("Returning caps %p for %s", ret, binary);
    virObjectRef(ret);
    virMutexUnlock(&cache->lock);
    return ret;
}


qemuCapsPtr
qemuCapsCacheLookupCopy(qemuCapsCachePtr cache, const char *binary)
{
    qemuCapsPtr caps = qemuCapsCacheLookup(cache, binary);
    qemuCapsPtr ret;

    if (!caps)
        return NULL;

    ret = qemuCapsNewCopy(caps);
    virObjectUnref(caps);
    return ret;
}


void
qemuCapsCacheFree(qemuCapsCachePtr cache)
{
    if (!cache)
        return;

2644
    VIR_FREE(cache->libDir);
2645
    VIR_FREE(cache->runDir);
2646 2647 2648 2649
    virHashFree(cache->binaries);
    virMutexDestroy(&cache->lock);
    VIR_FREE(cache);
}