qemu_domain.c 480.1 KB
Newer Older
1
/*
2
 * qemu_domain.c: QEMU domain private state
3
 *
4
 * Copyright (C) 2006-2019 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
 */

#include <config.h>

#include "qemu_domain.h"
J
John Ferlan 已提交
25
#include "qemu_alias.h"
26
#include "qemu_block.h"
27
#include "qemu_cgroup.h"
28
#include "qemu_command.h"
M
Marc-André Lureau 已提交
29
#include "qemu_dbus.h"
30
#include "qemu_process.h"
31
#include "qemu_capabilities.h"
32
#include "qemu_hostdev.h"
33
#include "qemu_migration.h"
34
#include "qemu_migration_params.h"
35
#include "qemu_security.h"
36
#include "qemu_slirp.h"
37
#include "qemu_extdevice.h"
38
#include "qemu_blockjob.h"
39
#include "qemu_checkpoint.h"
40
#include "viralloc.h"
41
#include "virlog.h"
42
#include "virerror.h"
43
#include "c-ctype.h"
44
#include "cpu/cpu.h"
45
#include "viruuid.h"
E
Eric Blake 已提交
46
#include "virfile.h"
47
#include "domain_addr.h"
48
#include "domain_capabilities.h"
49
#include "domain_event.h"
50
#include "virtime.h"
51
#include "virnetdevopenvswitch.h"
52
#include "virstoragefile.h"
53
#include "virstring.h"
54
#include "virthreadjob.h"
55
#include "viratomic.h"
56
#include "virprocess.h"
57
#include "vircrypto.h"
58
#include "virrandom.h"
59
#include "virsystemd.h"
60
#include "secret_util.h"
61
#include "logging/log_manager.h"
62
#include "locking/domain_lock.h"
63
#include "virdomainsnapshotobjlist.h"
64
#include "virdomaincheckpointobjlist.h"
65

66 67 68 69 70
#ifdef MAJOR_IN_MKDEV
# include <sys/mkdev.h>
#elif MAJOR_IN_SYSMACROS
# include <sys/sysmacros.h>
#endif
71
#include <sys/time.h>
72
#include <fcntl.h>
73 74 75
#if defined(HAVE_SYS_MOUNT_H)
# include <sys/mount.h>
#endif
76 77 78
#ifdef WITH_SELINUX
# include <selinux/selinux.h>
#endif
79

80
#include "dosname.h"
81

82 83
#define QEMU_QXL_VGAMEM_DEFAULT 16 * 1024

84 85
#define VIR_FROM_THIS VIR_FROM_QEMU

86 87
VIR_LOG_INIT("qemu.qemu_domain");

88 89
VIR_ENUM_IMPL(qemuDomainJob,
              QEMU_JOB_LAST,
90 91 92 93 94
              "none",
              "query",
              "destroy",
              "suspend",
              "modify",
95
              "abort",
96
              "migration operation",
97 98 99 100
              "none",   /* async job is never stored in job.active */
              "async nested",
);

101 102
VIR_ENUM_IMPL(qemuDomainAgentJob,
              QEMU_AGENT_JOB_LAST,
103 104 105 106 107
              "none",
              "query",
              "modify",
);

108 109
VIR_ENUM_IMPL(qemuDomainAsyncJob,
              QEMU_ASYNC_JOB_LAST,
110 111 112 113 114
              "none",
              "migration out",
              "migration in",
              "save",
              "dump",
115
              "snapshot",
116
              "start",
117 118
);

119 120
VIR_ENUM_IMPL(qemuDomainNamespace,
              QEMU_DOMAIN_NS_LAST,
121 122 123
              "mount",
);

124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154

/**
 * qemuDomainObjFromDomain:
 * @domain: Domain pointer that has to be looked up
 *
 * This function looks up @domain and returns the appropriate virDomainObjPtr
 * that has to be released by calling virDomainObjEndAPI().
 *
 * Returns the domain object with incremented reference counter which is locked
 * on success, NULL otherwise.
 */
virDomainObjPtr
qemuDomainObjFromDomain(virDomainPtr domain)
{
    virDomainObjPtr vm;
    virQEMUDriverPtr driver = domain->conn->privateData;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

    vm = virDomainObjListFindByUUID(driver->domains, domain->uuid);
    if (!vm) {
        virUUIDFormat(domain->uuid, uuidstr);
        virReportError(VIR_ERR_NO_DOMAIN,
                       _("no domain with matching uuid '%s' (%s)"),
                       uuidstr, domain->name);
        return NULL;
    }

    return vm;
}


155
struct _qemuDomainLogContext {
156 157
    virObject parent;

158
    int writefd;
159
    int readfd; /* Only used if manager == NULL */
160
    off_t pos;
161
    ino_t inode; /* Only used if manager != NULL */
162
    char *path;
163
    virLogManagerPtr manager;
164 165
};

166
static virClassPtr qemuDomainLogContextClass;
167
static virClassPtr qemuDomainSaveCookieClass;
168 169

static void qemuDomainLogContextDispose(void *obj);
170
static void qemuDomainSaveCookieDispose(void *obj);
171

172

173
static int
174
qemuDomainOnceInit(void)
175
{
176
    if (!VIR_CLASS_NEW(qemuDomainLogContext, virClassForObject()))
177 178
        return -1;

179
    if (!VIR_CLASS_NEW(qemuDomainSaveCookie, virClassForObject()))
180 181
        return -1;

182 183 184
    return 0;
}

185
VIR_ONCE_GLOBAL_INIT(qemuDomain);
186 187 188 189 190 191 192 193 194 195 196 197 198

static void
qemuDomainLogContextDispose(void *obj)
{
    qemuDomainLogContextPtr ctxt = obj;
    VIR_DEBUG("ctxt=%p", ctxt);

    virLogManagerFree(ctxt->manager);
    VIR_FREE(ctxt->path);
    VIR_FORCE_CLOSE(ctxt->writefd);
    VIR_FORCE_CLOSE(ctxt->readfd);
}

J
Jiri Denemark 已提交
199
const char *
200
qemuDomainAsyncJobPhaseToString(qemuDomainAsyncJob job,
J
Ján Tomko 已提交
201
                                int phase G_GNUC_UNUSED)
J
Jiri Denemark 已提交
202 203 204 205
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
206 207
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
208 209
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
210
    case QEMU_ASYNC_JOB_SNAPSHOT:
211
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
212
    case QEMU_ASYNC_JOB_NONE:
213
        G_GNUC_FALLTHROUGH;
214 215
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
216 217 218 219 220 221
    }

    return "none";
}

int
222
qemuDomainAsyncJobPhaseFromString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
223 224 225 226 227 228 229 230
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
231 232
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
233 234
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
235
    case QEMU_ASYNC_JOB_SNAPSHOT:
236
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
237
    case QEMU_ASYNC_JOB_NONE:
238
        G_GNUC_FALLTHROUGH;
239 240
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
241 242 243 244 245 246 247 248
    }

    if (STREQ(phase, "none"))
        return 0;
    else
        return -1;
}

249

250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281
bool
qemuDomainNamespaceEnabled(virDomainObjPtr vm,
                           qemuDomainNamespace ns)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return priv->namespaces &&
        virBitmapIsBitSet(priv->namespaces, ns);
}


static int
qemuDomainEnableNamespace(virDomainObjPtr vm,
                          qemuDomainNamespace ns)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (!priv->namespaces &&
        !(priv->namespaces = virBitmapNew(QEMU_DOMAIN_NS_LAST)))
        return -1;

    if (virBitmapSetBit(priv->namespaces, ns) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unable to enable namespace: %s"),
                       qemuDomainNamespaceTypeToString(ns));
        return -1;
    }

    return 0;
}


282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297
static void
qemuDomainDisableNamespace(virDomainObjPtr vm,
                           qemuDomainNamespace ns)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (priv->namespaces) {
        ignore_value(virBitmapClearBit(priv->namespaces, ns));
        if (virBitmapIsAllClear(priv->namespaces)) {
            virBitmapFree(priv->namespaces);
            priv->namespaces = NULL;
        }
    }
}


298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317
void
qemuDomainEventEmitJobCompleted(virQEMUDriverPtr driver,
                                virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virObjectEventPtr event;
    virTypedParameterPtr params = NULL;
    int nparams = 0;
    int type;

    if (!priv->job.completed)
        return;

    if (qemuDomainJobInfoToParams(priv->job.completed, &type,
                                  &params, &nparams) < 0) {
        VIR_WARN("Could not get stats for completed job; domain %s",
                 vm->def->name);
    }

    event = virDomainEventJobCompletedNewFromObj(vm, params, nparams);
318
    virObjectEventStateQueue(driver->domainEventState, event);
319 320 321
}


322 323 324 325 326 327 328 329
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

    if (virCondInit(&priv->job.cond) < 0)
        return -1;

330
    if (virCondInit(&priv->job.asyncCond) < 0) {
331
        virCondDestroy(&priv->job.cond);
332 333 334
        return -1;
    }

335 336 337 338 339 340
    return 0;
}

static void
qemuDomainObjResetJob(qemuDomainObjPrivatePtr priv)
{
341
    qemuDomainJobObjPtr job = &priv->job;
342 343

    job->active = QEMU_JOB_NONE;
344
    job->owner = 0;
345
    job->ownerAPI = NULL;
J
Jiri Denemark 已提交
346
    job->started = 0;
347 348
}

349 350 351 352 353 354 355 356 357 358 359 360 361

static void
qemuDomainObjResetAgentJob(qemuDomainObjPrivatePtr priv)
{
    qemuDomainJobObjPtr job = &priv->job;

    job->agentActive = QEMU_AGENT_JOB_NONE;
    job->agentOwner = 0;
    job->agentOwnerAPI = NULL;
    job->agentStarted = 0;
}


362 363 364
static void
qemuDomainObjResetAsyncJob(qemuDomainObjPrivatePtr priv)
{
365
    qemuDomainJobObjPtr job = &priv->job;
366 367

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
368
    job->asyncOwner = 0;
369
    job->asyncOwnerAPI = NULL;
J
Jiri Denemark 已提交
370
    job->asyncStarted = 0;
J
Jiri Denemark 已提交
371
    job->phase = 0;
372
    job->mask = QEMU_JOB_DEFAULT_MASK;
373
    job->abortJob = false;
374
    job->spiceMigration = false;
375
    job->spiceMigrated = false;
376 377
    job->dumpCompleted = false;
    VIR_FREE(job->error);
J
Jiri Denemark 已提交
378
    VIR_FREE(job->current);
379 380
    qemuMigrationParamsFree(job->migParams);
    job->migParams = NULL;
381
    job->apiFlags = 0;
382 383
}

384 385
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
386
                        qemuDomainJobObjPtr job)
387 388 389 390 391
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
392
    job->owner = priv->job.owner;
393
    job->asyncJob = priv->job.asyncJob;
394
    job->asyncOwner = priv->job.asyncOwner;
J
Jiri Denemark 已提交
395
    job->phase = priv->job.phase;
396
    job->migParams = g_steal_pointer(&priv->job.migParams);
397
    job->apiFlags = priv->job.apiFlags;
398 399 400 401 402

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

403 404 405
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
J
Jiri Denemark 已提交
406
    VIR_FREE(priv->job.current);
407
    VIR_FREE(priv->job.completed);
408 409
    virCondDestroy(&priv->job.cond);
    virCondDestroy(&priv->job.asyncCond);
410 411
}

412
static bool
413
qemuDomainTrackJob(qemuDomainJob job)
414 415 416 417
{
    return (QEMU_DOMAIN_TRACK_JOBS & JOB_MASK(job)) != 0;
}

418

J
Jiri Denemark 已提交
419 420 421 422 423 424 425 426 427 428 429
int
qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->started)
        return 0;

    if (virTimeMillisNow(&now) < 0)
        return -1;

430 431 432 433 434 435
    if (now < jobInfo->started) {
        VIR_WARN("Async job starts in the future");
        jobInfo->started = 0;
        return 0;
    }

J
Jiri Denemark 已提交
436 437 438 439
    jobInfo->timeElapsed = now - jobInfo->started;
    return 0;
}

440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456
int
qemuDomainJobInfoUpdateDowntime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->stopped)
        return 0;

    if (virTimeMillisNow(&now) < 0)
        return -1;

    if (now < jobInfo->stopped) {
        VIR_WARN("Guest's CPUs stopped in the future");
        jobInfo->stopped = 0;
        return 0;
    }

457 458
    jobInfo->stats.mig.downtime = now - jobInfo->stopped;
    jobInfo->stats.mig.downtime_set = true;
459 460 461
    return 0;
}

462 463 464 465 466 467 468 469
static virDomainJobType
qemuDomainJobStatusToType(qemuDomainJobStatus status)
{
    switch (status) {
    case QEMU_DOMAIN_JOB_STATUS_NONE:
        break;

    case QEMU_DOMAIN_JOB_STATUS_ACTIVE:
470
    case QEMU_DOMAIN_JOB_STATUS_MIGRATING:
471
    case QEMU_DOMAIN_JOB_STATUS_QEMU_COMPLETED:
472
    case QEMU_DOMAIN_JOB_STATUS_POSTCOPY:
473
    case QEMU_DOMAIN_JOB_STATUS_PAUSED:
474 475 476 477 478 479 480 481 482 483 484 485 486 487 488
        return VIR_DOMAIN_JOB_UNBOUNDED;

    case QEMU_DOMAIN_JOB_STATUS_COMPLETED:
        return VIR_DOMAIN_JOB_COMPLETED;

    case QEMU_DOMAIN_JOB_STATUS_FAILED:
        return VIR_DOMAIN_JOB_FAILED;

    case QEMU_DOMAIN_JOB_STATUS_CANCELED:
        return VIR_DOMAIN_JOB_CANCELLED;
    }

    return VIR_DOMAIN_JOB_NONE;
}

J
Jiri Denemark 已提交
489 490 491 492
int
qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                        virDomainJobInfoPtr info)
{
493
    info->type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
494 495
    info->timeElapsed = jobInfo->timeElapsed;

496 497 498 499 500 501 502 503 504 505 506 507 508
    switch (jobInfo->statsType) {
    case QEMU_DOMAIN_JOB_STATS_TYPE_MIGRATION:
        info->memTotal = jobInfo->stats.mig.ram_total;
        info->memRemaining = jobInfo->stats.mig.ram_remaining;
        info->memProcessed = jobInfo->stats.mig.ram_transferred;
        info->fileTotal = jobInfo->stats.mig.disk_total +
                          jobInfo->mirrorStats.total;
        info->fileRemaining = jobInfo->stats.mig.disk_remaining +
                              (jobInfo->mirrorStats.total -
                               jobInfo->mirrorStats.transferred);
        info->fileProcessed = jobInfo->stats.mig.disk_transferred +
                              jobInfo->mirrorStats.transferred;
        break;
J
Jiri Denemark 已提交
509

510 511 512 513 514 515
    case QEMU_DOMAIN_JOB_STATS_TYPE_SAVEDUMP:
        info->memTotal = jobInfo->stats.mig.ram_total;
        info->memRemaining = jobInfo->stats.mig.ram_remaining;
        info->memProcessed = jobInfo->stats.mig.ram_transferred;
        break;

516
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
517 518 519 520 521
        info->memTotal = jobInfo->stats.dump.total;
        info->memProcessed = jobInfo->stats.dump.completed;
        info->memRemaining = info->memTotal - info->memProcessed;
        break;

522 523 524
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
        break;
    }
J
Jiri Denemark 已提交
525 526 527 528 529 530 531 532

    info->dataTotal = info->memTotal + info->fileTotal;
    info->dataRemaining = info->memRemaining + info->fileRemaining;
    info->dataProcessed = info->memProcessed + info->fileProcessed;

    return 0;
}

533 534 535 536 537 538

static int
qemuDomainMigrationJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                                   int *type,
                                   virTypedParameterPtr *params,
                                   int *nparams)
J
Jiri Denemark 已提交
539
{
540
    qemuMonitorMigrationStats *stats = &jobInfo->stats.mig;
541
    qemuDomainMirrorStatsPtr mirrorStats = &jobInfo->mirrorStats;
J
Jiri Denemark 已提交
542 543 544
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;
545 546
    unsigned long long mirrorRemaining = mirrorStats->total -
                                         mirrorStats->transferred;
J
Jiri Denemark 已提交
547

548 549 550 551 552
    if (virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_OPERATION,
                             jobInfo->operation) < 0)
        goto error;

J
Jiri Denemark 已提交
553 554 555 556 557
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED,
                                jobInfo->timeElapsed) < 0)
        goto error;

558 559 560 561 562 563 564
    if (jobInfo->timeDeltaSet &&
        jobInfo->timeElapsed > jobInfo->timeDelta &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED_NET,
                                jobInfo->timeElapsed - jobInfo->timeDelta) < 0)
        goto error;

565
    if (stats->downtime_set &&
J
Jiri Denemark 已提交
566 567
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME,
568
                                stats->downtime) < 0)
J
Jiri Denemark 已提交
569 570
        goto error;

571
    if (stats->downtime_set &&
572
        jobInfo->timeDeltaSet &&
573
        stats->downtime > jobInfo->timeDelta &&
574 575
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME_NET,
576
                                stats->downtime - jobInfo->timeDelta) < 0)
577 578
        goto error;

579
    if (stats->setup_time_set &&
580 581
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_SETUP_TIME,
582
                                stats->setup_time) < 0)
583 584
        goto error;

J
Jiri Denemark 已提交
585 586
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_TOTAL,
587
                                stats->ram_total +
588 589
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
590 591
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_PROCESSED,
592
                                stats->ram_transferred +
593 594
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
595 596
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_REMAINING,
597
                                stats->ram_remaining +
598 599
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
600 601 602 603
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
604
                                stats->ram_total) < 0 ||
J
Jiri Denemark 已提交
605 606
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
607
                                stats->ram_transferred) < 0 ||
J
Jiri Denemark 已提交
608 609
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
610
                                stats->ram_remaining) < 0)
J
Jiri Denemark 已提交
611 612
        goto error;

613
    if (stats->ram_bps &&
614 615
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_BPS,
616
                                stats->ram_bps) < 0)
617 618
        goto error;

619
    if (stats->ram_duplicate_set) {
J
Jiri Denemark 已提交
620 621
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_CONSTANT,
622
                                    stats->ram_duplicate) < 0 ||
J
Jiri Denemark 已提交
623 624
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL,
625
                                    stats->ram_normal) < 0 ||
J
Jiri Denemark 已提交
626 627
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL_BYTES,
628
                                    stats->ram_normal_bytes) < 0)
J
Jiri Denemark 已提交
629 630 631
            goto error;
    }

632 633 634 635 636
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_DIRTY_RATE,
                                stats->ram_dirty_rate) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_ITERATION,
637 638 639 640
                                stats->ram_iteration) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_POSTCOPY_REQS,
                                stats->ram_postcopy_reqs) < 0)
641 642
        goto error;

643 644 645 646 647 648
    if (stats->ram_page_size > 0 &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PAGE_SIZE,
                                stats->ram_page_size) < 0)
        goto error;

649 650 651 652 653
    /* The remaining stats are disk, mirror, or migration specific
     * so if this is a SAVEDUMP, we can just skip them */
    if (jobInfo->statsType == QEMU_DOMAIN_JOB_STATS_TYPE_SAVEDUMP)
        goto done;

J
Jiri Denemark 已提交
654 655
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_TOTAL,
656 657
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
658 659
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_PROCESSED,
660 661
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
662 663
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_REMAINING,
664 665
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
666 667
        goto error;

668
    if (stats->disk_bps &&
669 670
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_BPS,
671
                                stats->disk_bps) < 0)
672 673
        goto error;

674
    if (stats->xbzrle_set) {
J
Jiri Denemark 已提交
675 676
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE,
677
                                    stats->xbzrle_cache_size) < 0 ||
J
Jiri Denemark 已提交
678 679
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_BYTES,
680
                                    stats->xbzrle_bytes) < 0 ||
J
Jiri Denemark 已提交
681 682
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_PAGES,
683
                                    stats->xbzrle_pages) < 0 ||
J
Jiri Denemark 已提交
684 685
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE_MISSES,
686
                                    stats->xbzrle_cache_miss) < 0 ||
J
Jiri Denemark 已提交
687 688
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_OVERFLOW,
689
                                    stats->xbzrle_overflow) < 0)
J
Jiri Denemark 已提交
690 691 692
            goto error;
    }

693 694 695 696 697 698
    if (stats->cpu_throttle_percentage &&
        virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_AUTO_CONVERGE_THROTTLE,
                             stats->cpu_throttle_percentage) < 0)
        goto error;

699
 done:
700
    *type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
701 702 703 704 705 706 707 708 709 710
    *params = par;
    *nparams = npar;
    return 0;

 error:
    virTypedParamsFree(par, npar);
    return -1;
}


711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753
static int
qemuDomainDumpJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                              int *type,
                              virTypedParameterPtr *params,
                              int *nparams)
{
    qemuMonitorDumpStats *stats = &jobInfo->stats.dump;
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;

    if (virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_OPERATION,
                             jobInfo->operation) < 0)
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED,
                                jobInfo->timeElapsed) < 0)
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
                                stats->total) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
                                stats->completed) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
                                stats->total - stats->completed) < 0)
        goto error;

    *type = qemuDomainJobStatusToType(jobInfo->status);
    *params = par;
    *nparams = npar;
    return 0;

 error:
    virTypedParamsFree(par, npar);
    return -1;
}


754 755 756 757 758 759 760 761
int
qemuDomainJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                          int *type,
                          virTypedParameterPtr *params,
                          int *nparams)
{
    switch (jobInfo->statsType) {
    case QEMU_DOMAIN_JOB_STATS_TYPE_MIGRATION:
762
    case QEMU_DOMAIN_JOB_STATS_TYPE_SAVEDUMP:
763 764
        return qemuDomainMigrationJobInfoToParams(jobInfo, type, params, nparams);

765
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
766 767
        return qemuDomainDumpJobInfoToParams(jobInfo, type, params, nparams);

768
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
769 770 771 772 773 774
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid job statistics type"));
        break;

    default:
        virReportEnumRangeError(qemuDomainJobStatsType, jobInfo->statsType);
775 776 777 778 779 780 781
        break;
    }

    return -1;
}


J
John Ferlan 已提交
782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803
/* qemuDomainGetMasterKeyFilePath:
 * @libDir: Directory path to domain lib files
 *
 * Generate a path to the domain master key file for libDir.
 * It's up to the caller to handle checking if path exists.
 *
 * Returns path to memory containing the name of the file. It is up to the
 * caller to free; otherwise, NULL on failure.
 */
char *
qemuDomainGetMasterKeyFilePath(const char *libDir)
{
    if (!libDir) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid path for master key file"));
        return NULL;
    }
    return virFileBuildPath(libDir, "master-key.aes", NULL);
}


/* qemuDomainWriteMasterKeyFile:
804 805
 * @driver: qemu driver data
 * @vm: Pointer to the vm object
J
John Ferlan 已提交
806 807 808 809 810
 *
 * Get the desired path to the masterKey file and store it in the path.
 *
 * Returns 0 on success, -1 on failure with error message indicating failure
 */
811
int
M
Martin Kletzander 已提交
812 813
qemuDomainWriteMasterKeyFile(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
J
John Ferlan 已提交
814 815 816 817
{
    char *path;
    int fd = -1;
    int ret = -1;
M
Martin Kletzander 已提交
818
    qemuDomainObjPrivatePtr priv = vm->privateData;
J
John Ferlan 已提交
819

820 821 822 823
    /* Only gets filled in if we have the capability */
    if (!priv->masterKey)
        return 0;

J
John Ferlan 已提交
824 825 826 827 828 829 830 831 832 833 834 835 836 837 838
    if (!(path = qemuDomainGetMasterKeyFilePath(priv->libDir)))
        return -1;

    if ((fd = open(path, O_WRONLY|O_TRUNC|O_CREAT, 0600)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to open domain master key file for write"));
        goto cleanup;
    }

    if (safewrite(fd, priv->masterKey, priv->masterKeyLen) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to write master key file for domain"));
        goto cleanup;
    }

839
    if (qemuSecurityDomainSetPathLabel(driver, vm, path, false) < 0)
M
Martin Kletzander 已提交
840 841
        goto cleanup;

J
John Ferlan 已提交
842 843 844 845 846 847 848 849 850 851
    ret = 0;

 cleanup:
    VIR_FORCE_CLOSE(fd);
    VIR_FREE(path);

    return ret;
}


852 853 854 855 856 857 858 859 860
static void
qemuDomainMasterKeyFree(qemuDomainObjPrivatePtr priv)
{
    if (!priv->masterKey)
        return;

    VIR_DISPOSE_N(priv->masterKey, priv->masterKeyLen);
}

J
John Ferlan 已提交
861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958
/* qemuDomainMasterKeyReadFile:
 * @priv: pointer to domain private object
 *
 * Expected to be called during qemuProcessReconnect once the domain
 * libDir has been generated through qemuStateInitialize calling
 * virDomainObjListLoadAllConfigs which will restore the libDir path
 * to the domain private object.
 *
 * This function will get the path to the master key file and if it
 * exists, it will read the contents of the file saving it in priv->masterKey.
 *
 * Once the file exists, the validity checks may cause failures; however,
 * if the file doesn't exist or the capability doesn't exist, we just
 * return (mostly) quietly.
 *
 * Returns 0 on success or lack of capability
 *        -1 on failure with error message indicating failure
 */
int
qemuDomainMasterKeyReadFile(qemuDomainObjPrivatePtr priv)
{
    char *path;
    int fd = -1;
    uint8_t *masterKey = NULL;
    ssize_t masterKeyLen = 0;

    /* If we don't have the capability, then do nothing. */
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

    if (!(path = qemuDomainGetMasterKeyFilePath(priv->libDir)))
        return -1;

    if (!virFileExists(path)) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("domain master key file doesn't exist in %s"),
                       priv->libDir);
        goto error;
    }

    if ((fd = open(path, O_RDONLY)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to open domain master key file for read"));
        goto error;
    }

    if (VIR_ALLOC_N(masterKey, 1024) < 0)
        goto error;

    if ((masterKeyLen = saferead(fd, masterKey, 1024)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("unable to read domain master key file"));
        goto error;
    }

    if (masterKeyLen != QEMU_DOMAIN_MASTER_KEY_LEN) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("invalid master key read, size=%zd"), masterKeyLen);
        goto error;
    }

    ignore_value(VIR_REALLOC_N_QUIET(masterKey, masterKeyLen));

    priv->masterKey = masterKey;
    priv->masterKeyLen = masterKeyLen;

    VIR_FORCE_CLOSE(fd);
    VIR_FREE(path);

    return 0;

 error:
    if (masterKeyLen > 0)
        memset(masterKey, 0, masterKeyLen);
    VIR_FREE(masterKey);

    VIR_FORCE_CLOSE(fd);
    VIR_FREE(path);

    return -1;
}


/* qemuDomainMasterKeyRemove:
 * @priv: Pointer to the domain private object
 *
 * Remove the traces of the master key, clear the heap, clear the file,
 * delete the file.
 */
void
qemuDomainMasterKeyRemove(qemuDomainObjPrivatePtr priv)
{
    char *path = NULL;

    if (!priv->masterKey)
        return;

    /* Clear the contents */
959
    qemuDomainMasterKeyFree(priv);
J
John Ferlan 已提交
960 961 962 963 964 965 966 967 968 969

    /* Delete the master key file */
    path = qemuDomainGetMasterKeyFilePath(priv->libDir);
    unlink(path);

    VIR_FREE(path);
}


/* qemuDomainMasterKeyCreate:
970
 * @vm: Pointer to the domain object
J
John Ferlan 已提交
971 972 973 974 975 976 977 978
 *
 * As long as the underlying qemu has the secret capability,
 * generate and store 'raw' in a file a random 32-byte key to
 * be used as a secret shared with qemu to share sensitive data.
 *
 * Returns: 0 on success, -1 w/ error message on failure
 */
int
979
qemuDomainMasterKeyCreate(virDomainObjPtr vm)
J
John Ferlan 已提交
980
{
M
Martin Kletzander 已提交
981 982
    qemuDomainObjPrivatePtr priv = vm->privateData;

J
John Ferlan 已提交
983 984 985 986
    /* If we don't have the capability, then do nothing. */
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

987
    if (VIR_ALLOC_N(priv->masterKey, QEMU_DOMAIN_MASTER_KEY_LEN) < 0)
988
        return -1;
J
John Ferlan 已提交
989 990
    priv->masterKeyLen = QEMU_DOMAIN_MASTER_KEY_LEN;

991
    if (virRandomBytes(priv->masterKey, priv->masterKeyLen) < 0) {
992 993 994 995
        VIR_DISPOSE_N(priv->masterKey, priv->masterKeyLen);
        return -1;
    }

J
John Ferlan 已提交
996 997 998 999
    return 0;
}


1000
static void
1001
qemuDomainSecretPlainClear(qemuDomainSecretPlainPtr secret)
1002
{
1003 1004
    VIR_FREE(secret->username);
    VIR_DISPOSE_N(secret->secret, secret->secretlen);
1005 1006 1007
}


J
John Ferlan 已提交
1008
static void
1009
qemuDomainSecretAESClear(qemuDomainSecretAESPtr secret,
1010
                         bool keepAlias)
J
John Ferlan 已提交
1011
{
1012
    if (!keepAlias)
1013
        VIR_FREE(secret->alias);
1014

1015 1016 1017
    VIR_FREE(secret->username);
    VIR_FREE(secret->iv);
    VIR_FREE(secret->ciphertext);
J
John Ferlan 已提交
1018 1019 1020
}


1021 1022 1023
static void
qemuDomainSecretInfoClear(qemuDomainSecretInfoPtr secinfo,
                          bool keepAlias)
1024
{
1025
    if (!secinfo)
1026 1027
        return;

1028
    switch ((qemuDomainSecretInfoType) secinfo->type) {
J
John Ferlan 已提交
1029
    case VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN:
1030
        qemuDomainSecretPlainClear(&secinfo->s.plain);
J
John Ferlan 已提交
1031 1032
        break;

1033
    case VIR_DOMAIN_SECRET_INFO_TYPE_AES:
1034
        qemuDomainSecretAESClear(&secinfo->s.aes, keepAlias);
J
John Ferlan 已提交
1035 1036 1037 1038 1039
        break;

    case VIR_DOMAIN_SECRET_INFO_TYPE_LAST:
        break;
    }
1040 1041 1042 1043 1044 1045 1046 1047 1048 1049
}


void
qemuDomainSecretInfoFree(qemuDomainSecretInfoPtr *secinfo)
{
    if (!*secinfo)
        return;

    qemuDomainSecretInfoClear(*secinfo, false);
J
John Ferlan 已提交
1050

1051 1052 1053 1054
    VIR_FREE(*secinfo);
}


1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067
/**
 * qemuDomainSecretInfoDestroy:
 * @secinfo: object to destroy
 *
 * Removes any data unnecessary for further use, but keeps alias allocated.
 */
void
qemuDomainSecretInfoDestroy(qemuDomainSecretInfoPtr secinfo)
{
    qemuDomainSecretInfoClear(secinfo, true);
}


1068
static virClassPtr qemuDomainDiskPrivateClass;
1069
static void qemuDomainDiskPrivateDispose(void *obj);
1070 1071 1072 1073

static int
qemuDomainDiskPrivateOnceInit(void)
{
1074
    if (!VIR_CLASS_NEW(qemuDomainDiskPrivate, virClassForObject()))
1075
        return -1;
1076 1077

    return 0;
1078 1079
}

1080
VIR_ONCE_GLOBAL_INIT(qemuDomainDiskPrivate);
1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095

static virObjectPtr
qemuDomainDiskPrivateNew(void)
{
    qemuDomainDiskPrivatePtr priv;

    if (qemuDomainDiskPrivateInitialize() < 0)
        return NULL;

    if (!(priv = virObjectNew(qemuDomainDiskPrivateClass)))
        return NULL;

    return (virObjectPtr) priv;
}

1096 1097 1098 1099 1100
static void
qemuDomainDiskPrivateDispose(void *obj)
{
    qemuDomainDiskPrivatePtr priv = obj;

1101
    virObjectUnref(priv->migrSource);
1102
    VIR_FREE(priv->qomName);
1103
    VIR_FREE(priv->nodeCopyOnRead);
1104
    virObjectUnref(priv->blockjob);
1105
}
1106

1107 1108 1109 1110 1111 1112
static virClassPtr qemuDomainStorageSourcePrivateClass;
static void qemuDomainStorageSourcePrivateDispose(void *obj);

static int
qemuDomainStorageSourcePrivateOnceInit(void)
{
1113
    if (!VIR_CLASS_NEW(qemuDomainStorageSourcePrivate, virClassForObject()))
1114
        return -1;
1115 1116

    return 0;
1117 1118
}

1119
VIR_ONCE_GLOBAL_INIT(qemuDomainStorageSourcePrivate);
1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145

virObjectPtr
qemuDomainStorageSourcePrivateNew(void)
{
    qemuDomainStorageSourcePrivatePtr priv;

    if (qemuDomainStorageSourcePrivateInitialize() < 0)
        return NULL;

    if (!(priv = virObjectNew(qemuDomainStorageSourcePrivateClass)))
        return NULL;

    return (virObjectPtr) priv;
}


static void
qemuDomainStorageSourcePrivateDispose(void *obj)
{
    qemuDomainStorageSourcePrivatePtr priv = obj;

    qemuDomainSecretInfoFree(&priv->secinfo);
    qemuDomainSecretInfoFree(&priv->encinfo);
}


1146 1147 1148 1149 1150 1151
static virClassPtr qemuDomainVcpuPrivateClass;
static void qemuDomainVcpuPrivateDispose(void *obj);

static int
qemuDomainVcpuPrivateOnceInit(void)
{
1152
    if (!VIR_CLASS_NEW(qemuDomainVcpuPrivate, virClassForObject()))
1153
        return -1;
1154 1155

    return 0;
1156 1157
}

1158
VIR_ONCE_GLOBAL_INIT(qemuDomainVcpuPrivate);
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175

static virObjectPtr
qemuDomainVcpuPrivateNew(void)
{
    qemuDomainVcpuPrivatePtr priv;

    if (qemuDomainVcpuPrivateInitialize() < 0)
        return NULL;

    if (!(priv = virObjectNew(qemuDomainVcpuPrivateClass)))
        return NULL;

    return (virObjectPtr) priv;
}


static void
1176
qemuDomainVcpuPrivateDispose(void *obj)
1177
{
1178 1179 1180 1181
    qemuDomainVcpuPrivatePtr priv = obj;

    VIR_FREE(priv->type);
    VIR_FREE(priv->alias);
1182
    virJSONValueFree(priv->props);
1183 1184 1185 1186
    return;
}


1187 1188
static virClassPtr qemuDomainChrSourcePrivateClass;
static void qemuDomainChrSourcePrivateDispose(void *obj);
1189 1190

static int
1191
qemuDomainChrSourcePrivateOnceInit(void)
1192
{
1193
    if (!VIR_CLASS_NEW(qemuDomainChrSourcePrivate, virClassForObject()))
1194
        return -1;
1195 1196

    return 0;
1197 1198
}

1199
VIR_ONCE_GLOBAL_INIT(qemuDomainChrSourcePrivate);
1200 1201

static virObjectPtr
1202
qemuDomainChrSourcePrivateNew(void)
1203
{
1204
    qemuDomainChrSourcePrivatePtr priv;
1205

1206
    if (qemuDomainChrSourcePrivateInitialize() < 0)
1207 1208
        return NULL;

1209
    if (!(priv = virObjectNew(qemuDomainChrSourcePrivateClass)))
1210 1211 1212 1213 1214 1215 1216
        return NULL;

    return (virObjectPtr) priv;
}


static void
1217
qemuDomainChrSourcePrivateDispose(void *obj)
1218
{
1219
    qemuDomainChrSourcePrivatePtr priv = obj;
1220 1221 1222 1223 1224

    qemuDomainSecretInfoFree(&priv->secinfo);
}


J
Ján Tomko 已提交
1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236
static virClassPtr qemuDomainVsockPrivateClass;
static void qemuDomainVsockPrivateDispose(void *obj);

static int
qemuDomainVsockPrivateOnceInit(void)
{
    if (!VIR_CLASS_NEW(qemuDomainVsockPrivate, virClassForObject()))
        return -1;

    return 0;
}

1237
VIR_ONCE_GLOBAL_INIT(qemuDomainVsockPrivate);
J
Ján Tomko 已提交
1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249

static virObjectPtr
qemuDomainVsockPrivateNew(void)
{
    qemuDomainVsockPrivatePtr priv;

    if (qemuDomainVsockPrivateInitialize() < 0)
        return NULL;

    if (!(priv = virObjectNew(qemuDomainVsockPrivateClass)))
        return NULL;

1250 1251
    priv->vhostfd = -1;

J
Ján Tomko 已提交
1252 1253 1254 1255 1256
    return (virObjectPtr) priv;
}


static void
J
Ján Tomko 已提交
1257
qemuDomainVsockPrivateDispose(void *obj G_GNUC_UNUSED)
J
Ján Tomko 已提交
1258
{
1259 1260 1261
    qemuDomainVsockPrivatePtr priv = obj;

    VIR_FORCE_CLOSE(priv->vhostfd);
J
Ján Tomko 已提交
1262 1263 1264
}


1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276
static virClassPtr qemuDomainGraphicsPrivateClass;
static void qemuDomainGraphicsPrivateDispose(void *obj);

static int
qemuDomainGraphicsPrivateOnceInit(void)
{
    if (!VIR_CLASS_NEW(qemuDomainGraphicsPrivate, virClassForObject()))
        return -1;

    return 0;
}

1277
VIR_ONCE_GLOBAL_INIT(qemuDomainGraphicsPrivate);
1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299

static virObjectPtr
qemuDomainGraphicsPrivateNew(void)
{
    qemuDomainGraphicsPrivatePtr priv;

    if (qemuDomainGraphicsPrivateInitialize() < 0)
        return NULL;

    if (!(priv = virObjectNew(qemuDomainGraphicsPrivateClass)))
        return NULL;

    return (virObjectPtr) priv;
}


static void
qemuDomainGraphicsPrivateDispose(void *obj)
{
    qemuDomainGraphicsPrivatePtr priv = obj;

    VIR_FREE(priv->tlsAlias);
1300
    qemuDomainSecretInfoFree(&priv->secinfo);
1301 1302 1303
}


1304 1305 1306 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 1334 1335 1336
static virClassPtr qemuDomainNetworkPrivateClass;
static void qemuDomainNetworkPrivateDispose(void *obj);


static int
qemuDomainNetworkPrivateOnceInit(void)
{
    if (!VIR_CLASS_NEW(qemuDomainNetworkPrivate, virClassForObject()))
        return -1;

    return 0;
}


VIR_ONCE_GLOBAL_INIT(qemuDomainNetworkPrivate);


static virObjectPtr
qemuDomainNetworkPrivateNew(void)
{
    qemuDomainNetworkPrivatePtr priv;

    if (qemuDomainNetworkPrivateInitialize() < 0)
        return NULL;

    if (!(priv = virObjectNew(qemuDomainNetworkPrivateClass)))
        return NULL;

    return (virObjectPtr) priv;
}


static void
J
Ján Tomko 已提交
1337
qemuDomainNetworkPrivateDispose(void *obj G_GNUC_UNUSED)
1338
{
1339 1340 1341
    qemuDomainNetworkPrivatePtr priv = obj;

    qemuSlirpFree(priv->slirp);
1342 1343 1344
}


1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371
static virClassPtr qemuDomainVideoPrivateClass;
static void qemuDomainVideoPrivateDispose(void *obj);


static int
qemuDomainVideoPrivateOnceInit(void)
{
    if (!VIR_CLASS_NEW(qemuDomainVideoPrivate, virClassForObject()))
        return -1;

    return 0;
}

VIR_ONCE_GLOBAL_INIT(qemuDomainVideoPrivate);


static virObjectPtr
qemuDomainVideoPrivateNew(void)
{
    qemuDomainVideoPrivatePtr priv;

    if (qemuDomainVideoPrivateInitialize() < 0)
        return NULL;

    if (!(priv = virObjectNew(qemuDomainVideoPrivateClass)))
        return NULL;

1372 1373
    priv->vhost_user_fd = -1;

1374 1375 1376 1377 1378
    return (virObjectPtr) priv;
}


static void
1379
qemuDomainVideoPrivateDispose(void *obj)
1380
{
1381 1382 1383
    qemuDomainVideoPrivatePtr priv = obj;

    VIR_FORCE_CLOSE(priv->vhost_user_fd);
1384 1385 1386
}


1387 1388
/* qemuDomainSecretPlainSetup:
 * @secinfo: Pointer to secret info
J
John Ferlan 已提交
1389
 * @usageType: The virSecretUsageType
1390 1391
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1392 1393 1394 1395 1396 1397
 *
 * Taking a secinfo, fill in the plaintext information
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1398
qemuDomainSecretPlainSetup(qemuDomainSecretInfoPtr secinfo,
J
John Ferlan 已提交
1399
                           virSecretUsageType usageType,
1400 1401
                           const char *username,
                           virSecretLookupTypeDefPtr seclookupdef)
1402
{
1403 1404 1405 1406 1407 1408 1409
    virConnectPtr conn;
    int ret = -1;

    conn = virGetConnectSecret();
    if (!conn)
        return -1;

1410
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN;
1411
    secinfo->s.plain.username = g_strdup(username);
1412

1413 1414 1415 1416 1417 1418
    ret = virSecretGetSecretString(conn, seclookupdef, usageType,
                                   &secinfo->s.plain.secret,
                                   &secinfo->s.plain.secretlen);

    virObjectUnref(conn);
    return ret;
1419 1420 1421
}


1422 1423 1424 1425
/* qemuDomainSecretAESSetup:
 * @priv: pointer to domain private object
 * @secinfo: Pointer to secret info
 * @srcalias: Alias of the disk/hostdev used to generate the secret alias
J
John Ferlan 已提交
1426
 * @usageType: The virSecretUsageType
1427 1428
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1429
 * @isLuks: True/False for is for luks (alias generation)
1430 1431 1432 1433 1434 1435
 *
 * Taking a secinfo, fill in the AES specific information using the
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1436
qemuDomainSecretAESSetup(qemuDomainObjPrivatePtr priv,
1437 1438
                         qemuDomainSecretInfoPtr secinfo,
                         const char *srcalias,
J
John Ferlan 已提交
1439
                         virSecretUsageType usageType,
1440
                         const char *username,
1441 1442
                         virSecretLookupTypeDefPtr seclookupdef,
                         bool isLuks)
1443
{
1444
    virConnectPtr conn;
1445 1446 1447 1448 1449 1450 1451 1452
    int ret = -1;
    uint8_t *raw_iv = NULL;
    size_t ivlen = QEMU_DOMAIN_AES_IV_LEN;
    uint8_t *secret = NULL;
    size_t secretlen = 0;
    uint8_t *ciphertext = NULL;
    size_t ciphertextlen = 0;

1453 1454 1455 1456
    conn = virGetConnectSecret();
    if (!conn)
        return -1;

1457
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_AES;
1458
    secinfo->s.aes.username = g_strdup(username);
1459

1460
    if (!(secinfo->s.aes.alias = qemuDomainGetSecretAESAlias(srcalias, isLuks)))
1461
        goto cleanup;
1462

1463 1464 1465
    if (VIR_ALLOC_N(raw_iv, ivlen) < 0)
        goto cleanup;

1466
    /* Create a random initialization vector */
1467
    if (virRandomBytes(raw_iv, ivlen) < 0)
1468
        goto cleanup;
1469 1470

    /* Encode the IV and save that since qemu will need it */
1471
    secinfo->s.aes.iv = g_base64_encode(raw_iv, ivlen);
1472 1473

    /* Grab the unencoded secret */
J
John Ferlan 已提交
1474
    if (virSecretGetSecretString(conn, seclookupdef, usageType,
1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487
                                 &secret, &secretlen) < 0)
        goto cleanup;

    if (virCryptoEncryptData(VIR_CRYPTO_CIPHER_AES256CBC,
                             priv->masterKey, QEMU_DOMAIN_MASTER_KEY_LEN,
                             raw_iv, ivlen, secret, secretlen,
                             &ciphertext, &ciphertextlen) < 0)
        goto cleanup;

    /* Clear out the secret */
    memset(secret, 0, secretlen);

    /* Now encode the ciphertext and store to be passed to qemu */
1488 1489
    secinfo->s.aes.ciphertext = g_base64_encode(ciphertext,
                                                ciphertextlen);
1490 1491 1492 1493 1494 1495 1496

    ret = 0;

 cleanup:
    VIR_DISPOSE_N(raw_iv, ivlen);
    VIR_DISPOSE_N(secret, secretlen);
    VIR_DISPOSE_N(ciphertext, ciphertextlen);
1497
    virObjectUnref(conn);
1498 1499 1500 1501
    return ret;
}


1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517
/**
 * qemuDomainSupportsEncryptedSecret:
 * @priv: qemu domain private data
 *
 * Returns true if libvirt can use encrypted 'secret' objects with VM which
 * @priv belongs to.
 */
bool
qemuDomainSupportsEncryptedSecret(qemuDomainObjPrivatePtr priv)
{
    return virCryptoHaveCipher(VIR_CRYPTO_CIPHER_AES256CBC) &&
           virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET) &&
           priv->masterKey;
}


1518
/* qemuDomainSecretInfoNewPlain:
1519
 * @usageType: Secret usage type
1520 1521
 * @username: username
 * @lookupDef: lookup def describing secret
1522
 *
1523
 * Helper function to create a secinfo to be used for secinfo consumers. This
1524
 * sets up a 'plain' (unencrypted) secret for legacy consumers.
1525 1526 1527 1528 1529
 *
 * Returns @secinfo on success, NULL on failure. Caller is responsible
 * to eventually free @secinfo.
 */
static qemuDomainSecretInfoPtr
1530
qemuDomainSecretInfoNewPlain(virSecretUsageType usageType,
1531
                             const char *username,
1532
                             virSecretLookupTypeDefPtr lookupDef)
1533 1534 1535 1536 1537 1538
{
    qemuDomainSecretInfoPtr secinfo = NULL;

    if (VIR_ALLOC(secinfo) < 0)
        return NULL;

1539 1540 1541
    if (qemuDomainSecretPlainSetup(secinfo, usageType, username, lookupDef) < 0) {
        qemuDomainSecretInfoFree(&secinfo);
        return NULL;
1542 1543 1544 1545 1546 1547
    }

    return secinfo;
}


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 1582 1583 1584 1585 1586 1587 1588 1589 1590
/* qemuDomainSecretInfoNew:
 * @priv: pointer to domain private object
 * @srcAlias: Alias base to use for TLS object
 * @usageType: Secret usage type
 * @username: username
 * @looupDef: lookup def describing secret
 * @isLuks: boolean for luks lookup
 *
 * Helper function to create a secinfo to be used for secinfo consumers. This
 * sets up encrypted data to be used with qemu's 'secret' object.
 *
 * Returns @secinfo on success, NULL on failure. Caller is responsible
 * to eventually free @secinfo.
 */
static qemuDomainSecretInfoPtr
qemuDomainSecretInfoNew(qemuDomainObjPrivatePtr priv,
                        const char *srcAlias,
                        virSecretUsageType usageType,
                        const char *username,
                        virSecretLookupTypeDefPtr lookupDef,
                        bool isLuks)
{
    qemuDomainSecretInfoPtr secinfo = NULL;

    if (!qemuDomainSupportsEncryptedSecret(priv)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("encrypted secrets are not supported"));
        return NULL;
    }

    if (VIR_ALLOC(secinfo) < 0)
        return NULL;

    if (qemuDomainSecretAESSetup(priv, secinfo, srcAlias, usageType, username,
                                 lookupDef, isLuks) < 0) {
        qemuDomainSecretInfoFree(&secinfo);
        return NULL;
    }

    return secinfo;
}


1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601
/**
 * qemuDomainSecretInfoTLSNew:
 * @priv: pointer to domain private object
 * @srcAlias: Alias base to use for TLS object
 * @secretUUID: Provide a secretUUID value to look up/create the secretInfo
 *
 * Using the passed @secretUUID, generate a seclookupdef that can be used
 * to generate the returned qemuDomainSecretInfoPtr for a TLS based secret.
 *
 * Returns qemuDomainSecretInfoPtr or NULL on error.
 */
1602
qemuDomainSecretInfoPtr
1603
qemuDomainSecretInfoTLSNew(qemuDomainObjPrivatePtr priv,
1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616
                           const char *srcAlias,
                           const char *secretUUID)
{
    virSecretLookupTypeDef seclookupdef = {0};

    if (virUUIDParse(secretUUID, seclookupdef.u.uuid) < 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("malformed TLS secret uuid '%s' provided"),
                       secretUUID);
        return NULL;
    }
    seclookupdef.type = VIR_SECRET_LOOKUP_TYPE_UUID;

1617 1618 1619
    return qemuDomainSecretInfoNew(priv, srcAlias,
                                   VIR_SECRET_USAGE_TYPE_TLS, NULL,
                                   &seclookupdef, false);
1620 1621 1622
}


1623 1624 1625
void
qemuDomainSecretDiskDestroy(virDomainDiskDefPtr disk)
{
1626 1627
    qemuDomainStorageSourcePrivatePtr srcPriv;
    virStorageSourcePtr n;
1628

1629 1630 1631 1632 1633 1634
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if ((srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(n))) {
            qemuDomainSecretInfoDestroy(srcPriv->secinfo);
            qemuDomainSecretInfoDestroy(srcPriv->encinfo);
        }
    }
1635 1636 1637
}


1638
bool
1639
qemuDomainStorageSourceHasAuth(virStorageSourcePtr src)
1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651
{
    if (!virStorageSourceIsEmpty(src) &&
        virStorageSourceGetActualType(src) == VIR_STORAGE_TYPE_NETWORK &&
        src->auth &&
        (src->protocol == VIR_STORAGE_NET_PROTOCOL_ISCSI ||
         src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD))
        return true;

    return false;
}


1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663
bool
qemuDomainDiskHasEncryptionSecret(virStorageSourcePtr src)
{
    if (!virStorageSourceIsEmpty(src) && src->encryption &&
        src->encryption->format == VIR_STORAGE_ENCRYPTION_FORMAT_LUKS &&
        src->encryption->nsecrets > 0)
        return true;

    return false;
}


1664 1665 1666 1667 1668 1669
/**
 * qemuDomainSecretStorageSourcePrepare:
 * @priv: domain private object
 * @src: storage source struct to setup
 * @authalias: prefix of the alias for secret holding authentication data
 * @encalias: prefix of the alias for secret holding encryption password
1670
 *
1671 1672 1673 1674
 * Prepares data necessary for encryption and authentication of @src. The two
 * alias prefixes are provided since in the backing chain authentication belongs
 * to the storage protocol data whereas encryption is relevant to the format
 * driver in qemu. The two will have different node names.
1675
 *
1676
 * Returns 0 on success; -1 on error while reporting an libvirt error.
1677
 */
1678
static int
1679
qemuDomainSecretStorageSourcePrepare(qemuDomainObjPrivatePtr priv,
1680 1681 1682
                                     virStorageSourcePtr src,
                                     const char *authalias,
                                     const char *encalias)
1683
{
1684
    qemuDomainStorageSourcePrivatePtr srcPriv;
1685
    bool iscsiHasPS = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET);
1686
    bool hasAuth = qemuDomainStorageSourceHasAuth(src);
1687 1688 1689 1690
    bool hasEnc = qemuDomainDiskHasEncryptionSecret(src);

    if (!hasAuth && !hasEnc)
        return 0;
1691

1692
    if (!(src->privateData = qemuDomainStorageSourcePrivateNew()))
1693 1694
        return -1;

1695
    srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
1696

1697
    if (hasAuth) {
J
John Ferlan 已提交
1698
        virSecretUsageType usageType = VIR_SECRET_USAGE_TYPE_ISCSI;
1699

1700
        if (src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD)
J
John Ferlan 已提交
1701
            usageType = VIR_SECRET_USAGE_TYPE_CEPH;
1702

1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717
        if (!qemuDomainSupportsEncryptedSecret(priv) ||
            (src->protocol == VIR_STORAGE_NET_PROTOCOL_ISCSI && !iscsiHasPS)) {
            srcPriv->secinfo = qemuDomainSecretInfoNewPlain(usageType,
                                                            src->auth->username,
                                                            &src->auth->seclookupdef);
        } else {
            srcPriv->secinfo = qemuDomainSecretInfoNew(priv, authalias,
                                                       usageType,
                                                       src->auth->username,
                                                       &src->auth->seclookupdef,
                                                       false);
        }

        if (!srcPriv->secinfo)
            return -1;
1718 1719
    }

1720
    if (hasEnc) {
1721
        if (!(srcPriv->encinfo =
1722 1723 1724 1725
              qemuDomainSecretInfoNew(priv, encalias,
                                      VIR_SECRET_USAGE_TYPE_VOLUME, NULL,
                                      &src->encryption->secrets[0]->seclookupdef,
                                      true)))
1726
              return -1;
1727 1728
    }

1729 1730 1731 1732
    return 0;
}


1733 1734 1735
void
qemuDomainSecretHostdevDestroy(virDomainHostdevDefPtr hostdev)
{
1736
    qemuDomainStorageSourcePrivatePtr srcPriv;
1737

1738 1739 1740
    if (virHostdevIsSCSIDevice(hostdev)) {
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1741

1742 1743 1744 1745 1746 1747
        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI) {
            srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(iscsisrc->src);
            if (srcPriv && srcPriv->secinfo)
                qemuDomainSecretInfoFree(&srcPriv->secinfo);
        }
    }
1748 1749 1750 1751
}


/* qemuDomainSecretHostdevPrepare:
J
John Ferlan 已提交
1752
 * @priv: pointer to domain private object
1753 1754 1755 1756 1757 1758 1759
 * @hostdev: Pointer to a hostdev definition
 *
 * For the right host device, generate the qemuDomainSecretInfo structure.
 *
 * Returns 0 on success, -1 on failure
 */
int
1760
qemuDomainSecretHostdevPrepare(qemuDomainObjPrivatePtr priv,
1761 1762
                               virDomainHostdevDefPtr hostdev)
{
1763
    if (virHostdevIsSCSIDevice(hostdev)) {
1764 1765
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1766
        virStorageSourcePtr src = iscsisrc->src;
1767 1768

        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI &&
1769
            src->auth) {
1770 1771
            if (qemuDomainSecretStorageSourcePrepare(priv, src,
                                                     hostdev->info->alias, NULL) < 0)
1772 1773 1774 1775 1776 1777 1778 1779
                return -1;
        }
    }

    return 0;
}


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
void
qemuDomainSecretChardevDestroy(virDomainChrSourceDefPtr dev)
{
    qemuDomainChrSourcePrivatePtr chrSourcePriv =
        QEMU_DOMAIN_CHR_SOURCE_PRIVATE(dev);

    if (!chrSourcePriv || !chrSourcePriv->secinfo)
        return;

    qemuDomainSecretInfoFree(&chrSourcePriv->secinfo);
}


/* qemuDomainSecretChardevPrepare:
 * @cfg: Pointer to driver config object
 * @priv: pointer to domain private object
 * @chrAlias: Alias of the chr device
 * @dev: Pointer to a char source definition
 *
 * For a TCP character device, generate a qemuDomainSecretInfo to be used
 * by the command line code to generate the secret for the tls-creds to use.
 *
 * Returns 0 on success, -1 on failure
 */
int
1805
qemuDomainSecretChardevPrepare(virQEMUDriverConfigPtr cfg,
1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820
                               qemuDomainObjPrivatePtr priv,
                               const char *chrAlias,
                               virDomainChrSourceDefPtr dev)
{
    char *charAlias = NULL;

    if (dev->type != VIR_DOMAIN_CHR_TYPE_TCP)
        return 0;

    if (dev->data.tcp.haveTLS == VIR_TRISTATE_BOOL_YES &&
        cfg->chardevTLSx509secretUUID) {
        qemuDomainChrSourcePrivatePtr chrSourcePriv =
            QEMU_DOMAIN_CHR_SOURCE_PRIVATE(dev);

        if (!(charAlias = qemuAliasChardevFromDevAlias(chrAlias)))
1821
            return -1;
1822

1823
        chrSourcePriv->secinfo =
1824
            qemuDomainSecretInfoTLSNew(priv, charAlias,
1825
                                       cfg->chardevTLSx509secretUUID);
1826
        VIR_FREE(charAlias);
1827 1828 1829

        if (!chrSourcePriv->secinfo)
            return -1;
1830 1831 1832 1833 1834 1835
    }

    return 0;
}


1836 1837 1838 1839 1840 1841 1842 1843 1844
static void
qemuDomainSecretGraphicsDestroy(virDomainGraphicsDefPtr graphics)
{
    qemuDomainGraphicsPrivatePtr gfxPriv = QEMU_DOMAIN_GRAPHICS_PRIVATE(graphics);

    if (!gfxPriv)
        return;

    VIR_FREE(gfxPriv->tlsAlias);
1845
    qemuDomainSecretInfoFree(&gfxPriv->secinfo);
1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865
}


static int
qemuDomainSecretGraphicsPrepare(virQEMUDriverConfigPtr cfg,
                                qemuDomainObjPrivatePtr priv,
                                virDomainGraphicsDefPtr graphics)
{
    virQEMUCapsPtr qemuCaps = priv->qemuCaps;
    qemuDomainGraphicsPrivatePtr gfxPriv = QEMU_DOMAIN_GRAPHICS_PRIVATE(graphics);

    if (graphics->type != VIR_DOMAIN_GRAPHICS_TYPE_VNC)
        return 0;

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_TLS_CREDS_X509))
        return 0;

    if (!cfg->vncTLS)
        return 0;

1866
    gfxPriv->tlsAlias = g_strdup("vnc-tls-creds0");
1867

1868 1869 1870 1871 1872 1873 1874
    if (cfg->vncTLSx509secretUUID) {
        gfxPriv->secinfo = qemuDomainSecretInfoTLSNew(priv, gfxPriv->tlsAlias,
                                                      cfg->vncTLSx509secretUUID);
        if (!gfxPriv->secinfo)
            return -1;
    }

1875 1876 1877 1878
    return 0;
}


1879 1880 1881
/* qemuDomainSecretDestroy:
 * @vm: Domain object
 *
1882
 * Removes all unnecessary data which was needed to generate 'secret' objects.
1883 1884 1885 1886 1887 1888 1889 1890
 */
void
qemuDomainSecretDestroy(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++)
        qemuDomainSecretDiskDestroy(vm->def->disks[i]);
1891 1892 1893

    for (i = 0; i < vm->def->nhostdevs; i++)
        qemuDomainSecretHostdevDestroy(vm->def->hostdevs[i]);
1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919

    for (i = 0; i < vm->def->nserials; i++)
        qemuDomainSecretChardevDestroy(vm->def->serials[i]->source);

    for (i = 0; i < vm->def->nparallels; i++)
        qemuDomainSecretChardevDestroy(vm->def->parallels[i]->source);

    for (i = 0; i < vm->def->nchannels; i++)
        qemuDomainSecretChardevDestroy(vm->def->channels[i]->source);

    for (i = 0; i < vm->def->nconsoles; i++)
        qemuDomainSecretChardevDestroy(vm->def->consoles[i]->source);

    for (i = 0; i < vm->def->nsmartcards; i++) {
        if (vm->def->smartcards[i]->type ==
            VIR_DOMAIN_SMARTCARD_TYPE_PASSTHROUGH)
            qemuDomainSecretChardevDestroy(vm->def->smartcards[i]->data.passthru);
    }

    for (i = 0; i < vm->def->nrngs; i++) {
        if (vm->def->rngs[i]->backend == VIR_DOMAIN_RNG_BACKEND_EGD)
            qemuDomainSecretChardevDestroy(vm->def->rngs[i]->source.chardev);
    }

    for (i = 0; i < vm->def->nredirdevs; i++)
        qemuDomainSecretChardevDestroy(vm->def->redirdevs[i]->source);
1920 1921 1922

    for (i = 0; i < vm->def->ngraphics; i++)
        qemuDomainSecretGraphicsDestroy(vm->def->graphics[i]);
1923 1924 1925 1926
}


/* qemuDomainSecretPrepare:
1927
 * @driver: Pointer to driver object
1928 1929 1930
 * @vm: Domain object
 *
 * For any objects that may require an auth/secret setup, create a
1931
 * qemuDomainSecretInfo and save it in the appropriate place within
1932 1933 1934 1935 1936 1937 1938
 * the private structures. This will be used by command line build
 * code in order to pass the secret along to qemu in order to provide
 * the necessary authentication data.
 *
 * Returns 0 on success, -1 on failure with error message set
 */
int
1939
qemuDomainSecretPrepare(virQEMUDriverPtr driver,
1940 1941
                        virDomainObjPtr vm)
{
J
John Ferlan 已提交
1942
    qemuDomainObjPrivatePtr priv = vm->privateData;
1943
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1944
    size_t i;
1945
    int ret = -1;
1946

1947
    /* disk secrets are prepared when preparing disks */
1948

1949
    for (i = 0; i < vm->def->nhostdevs; i++) {
1950
        if (qemuDomainSecretHostdevPrepare(priv,
J
John Ferlan 已提交
1951
                                           vm->def->hostdevs[i]) < 0)
1952
            goto cleanup;
1953 1954
    }

1955
    for (i = 0; i < vm->def->nserials; i++) {
1956
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1957 1958 1959 1960 1961 1962
                                           vm->def->serials[i]->info.alias,
                                           vm->def->serials[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nparallels; i++) {
1963
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1964 1965 1966 1967 1968 1969
                                           vm->def->parallels[i]->info.alias,
                                           vm->def->parallels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nchannels; i++) {
1970
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1971 1972 1973 1974 1975 1976
                                           vm->def->channels[i]->info.alias,
                                           vm->def->channels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nconsoles; i++) {
1977
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1978 1979 1980 1981 1982 1983 1984 1985
                                           vm->def->consoles[i]->info.alias,
                                           vm->def->consoles[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nsmartcards; i++)
        if (vm->def->smartcards[i]->type ==
            VIR_DOMAIN_SMARTCARD_TYPE_PASSTHROUGH &&
1986
            qemuDomainSecretChardevPrepare(cfg, priv,
1987 1988 1989 1990 1991 1992
                                           vm->def->smartcards[i]->info.alias,
                                           vm->def->smartcards[i]->data.passthru) < 0)
            goto cleanup;

    for (i = 0; i < vm->def->nrngs; i++) {
        if (vm->def->rngs[i]->backend == VIR_DOMAIN_RNG_BACKEND_EGD &&
1993
            qemuDomainSecretChardevPrepare(cfg, priv,
1994 1995 1996 1997 1998 1999
                                           vm->def->rngs[i]->info.alias,
                                           vm->def->rngs[i]->source.chardev) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nredirdevs; i++) {
2000
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2001 2002 2003 2004 2005
                                           vm->def->redirdevs[i]->info.alias,
                                           vm->def->redirdevs[i]->source) < 0)
            goto cleanup;
    }

2006 2007 2008 2009 2010
    for (i = 0; i < vm->def->ngraphics; i++) {
        if (qemuDomainSecretGraphicsPrepare(cfg, priv, vm->def->graphics[i]) < 0)
            goto cleanup;
    }

2011 2012 2013 2014 2015
    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
2016 2017 2018
}


2019
/* This is the old way of setting up per-domain directories */
2020
static void
2021 2022 2023 2024 2025 2026
qemuDomainSetPrivatePathsOld(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

2027 2028
    if (!priv->libDir)
        priv->libDir = g_strdup_printf("%s/domain-%s", cfg->libDir, vm->def->name);
2029

2030 2031 2032
    if (!priv->channelTargetDir)
        priv->channelTargetDir = g_strdup_printf("%s/domain-%s",
                                                 cfg->channelTargetDir, vm->def->name);
2033 2034 2035 2036 2037 2038

    virObjectUnref(cfg);
}


int
2039 2040
qemuDomainSetPrivatePaths(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
2041
{
2042 2043
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainObjPrivatePtr priv = vm->privateData;
2044
    char *domname = virDomainDefGetShortName(vm->def);
2045
    int ret = -1;
2046

M
Martin Kletzander 已提交
2047 2048 2049
    if (!domname)
        goto cleanup;

2050 2051
    if (!priv->libDir)
        priv->libDir = g_strdup_printf("%s/domain-%s", cfg->libDir, domname);
2052

2053 2054 2055
    if (!priv->channelTargetDir)
        priv->channelTargetDir = g_strdup_printf("%s/domain-%s",
                                                 cfg->channelTargetDir, domname);
2056

2057 2058 2059
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
M
Martin Kletzander 已提交
2060
    VIR_FREE(domname);
2061
    return ret;
2062 2063 2064
}


M
Marc-André Lureau 已提交
2065 2066
static void
dbusVMStateHashFree(void *opaque,
J
Ján Tomko 已提交
2067
                    const void *name G_GNUC_UNUSED)
M
Marc-André Lureau 已提交
2068 2069 2070 2071 2072
{
    qemuDBusVMStateFree(opaque);
}


2073
static void *
2074
qemuDomainObjPrivateAlloc(void *opaque)
2075 2076 2077 2078 2079 2080
{
    qemuDomainObjPrivatePtr priv;

    if (VIR_ALLOC(priv) < 0)
        return NULL;

2081 2082 2083
    if (qemuDomainObjInitJob(priv) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to init qemu driver mutexes"));
2084
        goto error;
2085
    }
2086

2087
    if (!(priv->devs = virChrdevAlloc()))
2088 2089
        goto error;

2090 2091 2092
    if (!(priv->blockjobs = virHashCreate(5, virObjectFreeHashData)))
        goto error;

M
Marc-André Lureau 已提交
2093 2094 2095
    if (!(priv->dbusVMStates = virHashCreate(5, dbusVMStateHashFree)))
        goto error;

2096
    priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
2097
    priv->driver = opaque;
2098

2099
    return priv;
2100

2101
 error:
2102 2103
    VIR_FREE(priv);
    return NULL;
2104 2105
}

2106 2107 2108 2109 2110 2111 2112 2113 2114
/**
 * qemuDomainObjPrivateDataClear:
 * @priv: domain private data
 *
 * Clears private data entries, which are not necessary or stale if the VM is
 * not running.
 */
void
qemuDomainObjPrivateDataClear(qemuDomainObjPrivatePtr priv)
2115
{
2116 2117 2118
    virStringListFree(priv->qemuDevices);
    priv->qemuDevices = NULL;

2119
    virCgroupFree(&priv->cgroup);
2120 2121 2122 2123 2124

    virPerfFree(priv->perf);
    priv->perf = NULL;

    VIR_FREE(priv->machineName);
2125

2126
    virObjectUnref(priv->qemuCaps);
2127
    priv->qemuCaps = NULL;
2128

2129
    VIR_FREE(priv->pidfile);
2130

2131 2132 2133
    VIR_FREE(priv->libDir);
    VIR_FREE(priv->channelTargetDir);

2134 2135
    priv->memPrealloc = false;

2136 2137 2138 2139 2140 2141 2142
    /* remove automatic pinning data */
    virBitmapFree(priv->autoNodeset);
    priv->autoNodeset = NULL;
    virBitmapFree(priv->autoCpuset);
    priv->autoCpuset = NULL;

    /* remove address data */
2143
    virDomainPCIAddressSetFree(priv->pciaddrs);
2144
    priv->pciaddrs = NULL;
J
Ján Tomko 已提交
2145
    virDomainUSBAddressSetFree(priv->usbaddrs);
2146 2147 2148 2149 2150 2151 2152 2153
    priv->usbaddrs = NULL;

    virCPUDefFree(priv->origCPU);
    priv->origCPU = NULL;

    /* clear previously used namespaces */
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
2154

2155 2156
    priv->rememberOwner = false;

2157
    priv->reconnectBlockjobs = VIR_TRISTATE_BOOL_ABSENT;
2158
    priv->allowReboot = VIR_TRISTATE_BOOL_ABSENT;
2159 2160 2161

    virBitmapFree(priv->migrationCaps);
    priv->migrationCaps = NULL;
2162 2163 2164

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
2165 2166

    virHashRemoveAll(priv->blockjobs);
M
Marc-André Lureau 已提交
2167
    virHashRemoveAll(priv->dbusVMStates);
2168 2169 2170 2171 2172 2173 2174 2175 2176 2177
}


static void
qemuDomainObjPrivateFree(void *data)
{
    qemuDomainObjPrivatePtr priv = data;

    qemuDomainObjPrivateDataClear(priv);

2178
    virObjectUnref(priv->monConfig);
2179
    qemuDomainObjFreeJob(priv);
2180
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
2181
    VIR_FREE(priv->origname);
2182

2183
    virChrdevFree(priv->devs);
2184

2185 2186
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
2187
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
2188 2189
        qemuMonitorClose(priv->mon);
    }
D
Daniel P. Berrange 已提交
2190 2191 2192 2193
    if (priv->agent) {
        VIR_ERROR(_("Unexpected QEMU agent still active during domain deletion"));
        qemuAgentClose(priv->agent);
    }
2194
    VIR_FREE(priv->cleanupCallbacks);
2195 2196

    qemuDomainSecretInfoFree(&priv->migSecinfo);
2197
    qemuDomainMasterKeyFree(priv);
2198

2199
    virHashFree(priv->blockjobs);
M
Marc-André Lureau 已提交
2200
    virHashFree(priv->dbusVMStates);
2201

2202 2203 2204 2205
    VIR_FREE(priv);
}


2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220
static int
qemuStorageSourcePrivateDataAssignSecinfo(qemuDomainSecretInfoPtr *secinfo,
                                          char **alias)
{
    if (!*alias)
        return 0;

    if (!*secinfo) {
        if (VIR_ALLOC(*secinfo) < 0)
            return -1;

        (*secinfo)->type = VIR_DOMAIN_SECRET_INFO_TYPE_AES;
    }

    if ((*secinfo)->type == VIR_DOMAIN_SECRET_INFO_TYPE_AES)
2221
        (*secinfo)->s.aes.alias = g_steal_pointer(&*alias);
2222 2223 2224 2225 2226

    return 0;
}


2227 2228 2229 2230
static int
qemuStorageSourcePrivateDataParse(xmlXPathContextPtr ctxt,
                                  virStorageSourcePtr src)
{
2231 2232 2233 2234 2235
    qemuDomainStorageSourcePrivatePtr priv;
    char *authalias = NULL;
    char *encalias = NULL;
    int ret = -1;

2236 2237
    src->nodestorage = virXPathString("string(./nodenames/nodename[@type='storage']/@name)", ctxt);
    src->nodeformat = virXPathString("string(./nodenames/nodename[@type='format']/@name)", ctxt);
2238
    src->tlsAlias = virXPathString("string(./objects/TLSx509/@alias)", ctxt);
2239

2240 2241 2242
    if (src->pr)
        src->pr->mgralias = virXPathString("string(./reservations/@mgralias)", ctxt);

2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259
    authalias = virXPathString("string(./objects/secret[@type='auth']/@alias)", ctxt);
    encalias = virXPathString("string(./objects/secret[@type='encryption']/@alias)", ctxt);

    if (authalias || encalias) {
        if (!src->privateData &&
            !(src->privateData = qemuDomainStorageSourcePrivateNew()))
            goto cleanup;

        priv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);

        if (qemuStorageSourcePrivateDataAssignSecinfo(&priv->secinfo, &authalias) < 0)
            goto cleanup;

        if (qemuStorageSourcePrivateDataAssignSecinfo(&priv->encinfo, &encalias) < 0)
            goto cleanup;
    }

2260
    if (virStorageSourcePrivateDataParseRelPath(ctxt, src) < 0)
2261
        goto cleanup;
2262

2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284
    ret = 0;

 cleanup:
    VIR_FREE(authalias);
    VIR_FREE(encalias);

    return ret;
}


static void
qemuStorageSourcePrivateDataFormatSecinfo(virBufferPtr buf,
                                          qemuDomainSecretInfoPtr secinfo,
                                          const char *type)
{
    if (!secinfo ||
        secinfo->type != VIR_DOMAIN_SECRET_INFO_TYPE_AES ||
        !secinfo->s.aes.alias)
        return;

    virBufferAsprintf(buf, "<secret type='%s' alias='%s'/>\n",
                      type, secinfo->s.aes.alias);
2285 2286 2287 2288 2289 2290 2291
}


static int
qemuStorageSourcePrivateDataFormat(virStorageSourcePtr src,
                                   virBufferPtr buf)
{
2292
    g_auto(virBuffer) tmp = VIR_BUFFER_INITIALIZER;
2293 2294 2295
    qemuDomainStorageSourcePrivatePtr srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
    int ret = -1;

2296 2297 2298 2299 2300 2301 2302 2303 2304
    if (src->nodestorage || src->nodeformat) {
        virBufferAddLit(buf, "<nodenames>\n");
        virBufferAdjustIndent(buf, 2);
        virBufferEscapeString(buf, "<nodename type='storage' name='%s'/>\n", src->nodestorage);
        virBufferEscapeString(buf, "<nodename type='format' name='%s'/>\n", src->nodeformat);
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</nodenames>\n");
    }

2305 2306 2307
    if (src->pr)
        virBufferAsprintf(buf, "<reservations mgralias='%s'/>\n", src->pr->mgralias);

2308
    if (virStorageSourcePrivateDataFormatRelPath(src, buf) < 0)
2309
        goto cleanup;
2310

2311 2312 2313 2314 2315 2316 2317
    virBufferSetChildIndent(&tmp, buf);

    if (srcPriv) {
        qemuStorageSourcePrivateDataFormatSecinfo(&tmp, srcPriv->secinfo, "auth");
        qemuStorageSourcePrivateDataFormatSecinfo(&tmp, srcPriv->encinfo, "encryption");
    }

2318 2319 2320
    if (src->tlsAlias)
        virBufferAsprintf(&tmp, "<TLSx509 alias='%s'/>\n", src->tlsAlias);

2321
    virXMLFormatElement(buf, "objects", NULL, &tmp);
2322 2323 2324 2325 2326

    ret = 0;

 cleanup:
    return ret;
2327 2328 2329
}


2330 2331 2332 2333 2334 2335 2336
static int
qemuDomainDiskPrivateParse(xmlXPathContextPtr ctxt,
                           virDomainDiskDefPtr disk)
{
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);

    priv->qomName = virXPathString("string(./qom/@name)", ctxt);
2337
    priv->nodeCopyOnRead = virXPathString("string(./nodenames/nodename[@type='copyOnRead']/@name)", ctxt);
2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350

    return 0;
}


static int
qemuDomainDiskPrivateFormat(virDomainDiskDefPtr disk,
                            virBufferPtr buf)
{
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);

    virBufferEscapeString(buf, "<qom name='%s'/>\n", priv->qomName);

2351 2352 2353 2354 2355 2356 2357 2358 2359
    if (priv->nodeCopyOnRead) {
        virBufferAddLit(buf, "<nodenames>\n");
        virBufferAdjustIndent(buf, 2);
        virBufferEscapeString(buf, "<nodename type='copyOnRead' name='%s'/>\n",
                              priv->nodeCopyOnRead);
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</nodenames>\n");
    }

2360 2361 2362 2363
    return 0;
}


2364 2365
static void
qemuDomainObjPrivateXMLFormatVcpus(virBufferPtr buf,
2366
                                   virDomainDefPtr def)
2367 2368
{
    size_t i;
2369 2370 2371
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    pid_t tid;
2372 2373 2374 2375

    virBufferAddLit(buf, "<vcpus>\n");
    virBufferAdjustIndent(buf, 2);

2376 2377 2378 2379 2380 2381 2382 2383 2384
    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(def, i);
        tid = QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid;

        if (!vcpu->online || tid == 0)
            continue;

        virBufferAsprintf(buf, "<vcpu id='%zu' pid='%d'/>\n", i, tid);
    }
2385 2386 2387 2388 2389 2390

    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</vcpus>\n");
}


2391
static int
2392
qemuDomainObjPrivateXMLFormatAutomaticPlacement(virBufferPtr buf,
2393
                                                qemuDomainObjPrivatePtr priv)
2394 2395
{
    char *nodeset = NULL;
2396
    char *cpuset = NULL;
2397 2398
    int ret = -1;

2399
    if (!priv->autoNodeset && !priv->autoCpuset)
2400 2401
        return 0;

2402 2403
    if (priv->autoNodeset &&
        !((nodeset = virBitmapFormat(priv->autoNodeset))))
2404 2405
        goto cleanup;

2406 2407 2408 2409 2410 2411 2412 2413
    if (priv->autoCpuset &&
        !((cpuset = virBitmapFormat(priv->autoCpuset))))
        goto cleanup;

    virBufferAddLit(buf, "<numad");
    virBufferEscapeString(buf, " nodeset='%s'", nodeset);
    virBufferEscapeString(buf, " cpuset='%s'", cpuset);
    virBufferAddLit(buf, "/>\n");
2414 2415 2416 2417 2418

    ret = 0;

 cleanup:
    VIR_FREE(nodeset);
2419
    VIR_FREE(cpuset);
2420 2421 2422 2423
    return ret;
}


2424 2425 2426 2427 2428 2429 2430
typedef struct qemuDomainPrivateBlockJobFormatData {
    virDomainXMLOptionPtr xmlopt;
    virBufferPtr buf;
} qemuDomainPrivateBlockJobFormatData;


static int
2431 2432 2433 2434 2435
qemuDomainObjPrivateXMLFormatBlockjobFormatSource(virBufferPtr buf,
                                                  const char *element,
                                                  virStorageSourcePtr src,
                                                  virDomainXMLOptionPtr xmlopt,
                                                  bool chain)
2436
{
2437 2438
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    g_auto(virBuffer) childBuf = VIR_BUFFER_INITIALIZER;
2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449
    unsigned int xmlflags = VIR_DOMAIN_DEF_FORMAT_STATUS;

    virBufferSetChildIndent(&childBuf, buf);

    virBufferAsprintf(&attrBuf, " type='%s' format='%s'",
                      virStorageTypeToString(src->type),
                      virStorageFileFormatTypeToString(src->format));

    if (virDomainDiskSourceFormat(&childBuf, src, "source", 0, true, xmlflags, xmlopt) < 0)
        return -1;

2450 2451
    if (chain &&
        virDomainDiskBackingStoreFormat(&childBuf, src, xmlopt, xmlflags) < 0)
2452 2453
        return -1;

2454
    virXMLFormatElement(buf, element, &attrBuf, &childBuf);
2455 2456 2457 2458 2459

    return 0;
}


2460 2461
static int
qemuDomainObjPrivateXMLFormatBlockjobIterator(void *payload,
J
Ján Tomko 已提交
2462
                                              const void *name G_GNUC_UNUSED,
2463
                                              void *opaque)
2464
{
2465 2466 2467
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    g_auto(virBuffer) childBuf = VIR_BUFFER_INITIALIZER;
    g_auto(virBuffer) chainsBuf = VIR_BUFFER_INITIALIZER;
2468 2469 2470
    qemuBlockJobDataPtr job = payload;
    const char *state = qemuBlockjobStateTypeToString(job->state);
    const char *newstate = NULL;
2471
    struct qemuDomainPrivateBlockJobFormatData *data = opaque;
2472 2473 2474 2475

    if (job->newstate != -1)
        newstate = qemuBlockjobStateTypeToString(job->newstate);

2476 2477
    virBufferSetChildIndent(&childBuf, data->buf);
    virBufferSetChildIndent(&chainsBuf, &childBuf);
2478 2479 2480 2481 2482 2483 2484

    virBufferEscapeString(&attrBuf, " name='%s'", job->name);
    virBufferEscapeString(&attrBuf, " type='%s'", qemuBlockjobTypeToString(job->type));
    virBufferEscapeString(&attrBuf, " state='%s'", state);
    virBufferEscapeString(&attrBuf, " newstate='%s'", newstate);
    virBufferEscapeString(&childBuf, "<errmsg>%s</errmsg>", job->errmsg);

2485
    if (job->disk) {
2486 2487 2488 2489
        virBufferEscapeString(&childBuf, "<disk dst='%s'", job->disk->dst);
        if (job->mirrorChain)
            virBufferAddLit(&childBuf, " mirror='yes'");
        virBufferAddLit(&childBuf, "/>\n");
2490 2491
    } else {
        if (job->chain &&
2492 2493 2494 2495 2496
            qemuDomainObjPrivateXMLFormatBlockjobFormatSource(&chainsBuf,
                                                              "disk",
                                                              job->chain,
                                                              data->xmlopt,
                                                              true) < 0)
2497
            return -1;
2498

2499
        if (job->mirrorChain &&
2500 2501 2502 2503 2504
            qemuDomainObjPrivateXMLFormatBlockjobFormatSource(&chainsBuf,
                                                              "mirror",
                                                              job->mirrorChain,
                                                              data->xmlopt,
                                                              true) < 0)
2505 2506
            return -1;

2507
        virXMLFormatElement(&childBuf, "chains", NULL, &chainsBuf);
2508 2509
    }

2510 2511 2512 2513 2514 2515 2516 2517
    switch ((qemuBlockJobType) job->type) {
        case QEMU_BLOCKJOB_TYPE_PULL:
            if (job->data.pull.base)
                virBufferAsprintf(&childBuf, "<base node='%s'/>\n", job->data.pull.base->nodeformat);
            break;

        case QEMU_BLOCKJOB_TYPE_COMMIT:
        case QEMU_BLOCKJOB_TYPE_ACTIVE_COMMIT:
2518 2519 2520 2521 2522 2523
            if (job->data.commit.base)
                virBufferAsprintf(&childBuf, "<base node='%s'/>\n", job->data.commit.base->nodeformat);
            if (job->data.commit.top)
                virBufferAsprintf(&childBuf, "<top node='%s'/>\n", job->data.commit.top->nodeformat);
            if (job->data.commit.topparent)
                virBufferAsprintf(&childBuf, "<topparent node='%s'/>\n", job->data.commit.topparent->nodeformat);
2524 2525
            break;

2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538
        case QEMU_BLOCKJOB_TYPE_CREATE:
            if (job->data.create.storage)
                virBufferAddLit(&childBuf, "<create mode='storage'/>\n");

            if (job->data.create.src &&
                qemuDomainObjPrivateXMLFormatBlockjobFormatSource(&childBuf,
                                                                  "src",
                                                                  job->data.create.src,
                                                                  data->xmlopt,
                                                                  false) < 0)
                return -1;
            break;

2539
        case QEMU_BLOCKJOB_TYPE_COPY:
2540 2541 2542 2543
            if (job->data.copy.shallownew)
                virBufferAddLit(&attrBuf, " shallownew='yes'");
            break;

2544 2545 2546 2547 2548 2549
        case QEMU_BLOCKJOB_TYPE_NONE:
        case QEMU_BLOCKJOB_TYPE_INTERNAL:
        case QEMU_BLOCKJOB_TYPE_LAST:
            break;
    }

2550 2551
    virXMLFormatElement(data->buf, "blockjob", &attrBuf, &childBuf);
    return 0;
2552 2553 2554
}


2555 2556 2557 2558
static int
qemuDomainObjPrivateXMLFormatBlockjobs(virBufferPtr buf,
                                       virDomainObjPtr vm)
{
2559
    qemuDomainObjPrivatePtr priv = vm->privateData;
2560 2561
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    g_auto(virBuffer) childBuf = VIR_BUFFER_INITIALIZER;
2562
    bool bj = qemuDomainHasBlockjob(vm, false);
2563 2564
    struct qemuDomainPrivateBlockJobFormatData iterdata = { priv->driver->xmlopt,
                                                            &childBuf };
2565 2566 2567 2568

    virBufferAsprintf(&attrBuf, " active='%s'",
                      virTristateBoolTypeToString(virTristateBoolFromBool(bj)));

2569 2570 2571 2572 2573
    virBufferSetChildIndent(&childBuf, buf);

    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV) &&
        virHashForEach(priv->blockjobs,
                       qemuDomainObjPrivateXMLFormatBlockjobIterator,
2574
                       &iterdata) < 0)
2575 2576
        return -1;

2577 2578
    virXMLFormatElement(buf, "blockjobs", &attrBuf, &childBuf);
    return 0;
2579 2580 2581
}


2582
void
2583 2584 2585 2586 2587 2588 2589 2590 2591
qemuDomainObjPrivateXMLFormatAllowReboot(virBufferPtr buf,
                                         virTristateBool allowReboot)
{
    virBufferAsprintf(buf, "<allowReboot value='%s'/>\n",
                      virTristateBoolTypeToString(allowReboot));

}


2592 2593 2594 2595 2596 2597 2598 2599 2600
static void
qemuDomainObjPrivateXMLFormatPR(virBufferPtr buf,
                                qemuDomainObjPrivatePtr priv)
{
    if (priv->prDaemonRunning)
        virBufferAddLit(buf, "<prDaemon/>\n");
}


2601 2602
static int
qemuDomainObjPrivateXMLFormatNBDMigrationSource(virBufferPtr buf,
2603 2604
                                                virStorageSourcePtr src,
                                                virDomainXMLOptionPtr xmlopt)
2605
{
2606 2607
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    g_auto(virBuffer) childBuf = VIR_BUFFER_INITIALIZER;
2608 2609 2610 2611 2612 2613 2614 2615
    int ret = -1;

    virBufferSetChildIndent(&childBuf, buf);

    virBufferAsprintf(&attrBuf, " type='%s' format='%s'",
                      virStorageTypeToString(src->type),
                      virStorageFileFormatTypeToString(src->format));

2616
    if (virDomainDiskSourceFormat(&childBuf, src, "source", 0, false,
2617
                                  VIR_DOMAIN_DEF_FORMAT_STATUS, xmlopt) < 0)
2618 2619
        goto cleanup;

2620
    virXMLFormatElement(buf, "migrationSource", &attrBuf, &childBuf);
2621 2622 2623 2624 2625 2626 2627 2628 2629

    ret = 0;

 cleanup:
    return ret;
}


static int
2630 2631 2632
qemuDomainObjPrivateXMLFormatNBDMigration(virBufferPtr buf,
                                          virDomainObjPtr vm)
{
2633
    qemuDomainObjPrivatePtr priv = vm->privateData;
2634 2635
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    g_auto(virBuffer) childBuf = VIR_BUFFER_INITIALIZER;
2636 2637 2638
    size_t i;
    virDomainDiskDefPtr disk;
    qemuDomainDiskPrivatePtr diskPriv;
2639
    int ret = -1;
2640 2641 2642 2643

    for (i = 0; i < vm->def->ndisks; i++) {
        disk = vm->def->disks[i];
        diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
2644 2645 2646 2647

        virBufferSetChildIndent(&childBuf, buf);

        virBufferAsprintf(&attrBuf, " dev='%s' migrating='%s'",
2648
                          disk->dst, diskPriv->migrating ? "yes" : "no");
2649 2650 2651

        if (diskPriv->migrSource &&
            qemuDomainObjPrivateXMLFormatNBDMigrationSource(&childBuf,
2652 2653
                                                            diskPriv->migrSource,
                                                            priv->driver->xmlopt) < 0)
2654 2655
            goto cleanup;

2656
        virXMLFormatElement(buf, "disk", &attrBuf, &childBuf);
2657
    }
2658 2659 2660 2661 2662

    ret = 0;

 cleanup:
    return ret;
2663 2664 2665
}


2666
static int
2667 2668 2669 2670
qemuDomainObjPrivateXMLFormatJob(virBufferPtr buf,
                                 virDomainObjPtr vm,
                                 qemuDomainObjPrivatePtr priv)
{
2671 2672
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    g_auto(virBuffer) childBuf = VIR_BUFFER_INITIALIZER;
2673
    qemuDomainJob job = priv->job.active;
2674
    int ret = -1;
2675 2676

    if (!qemuDomainTrackJob(job))
2677
        job = QEMU_JOB_NONE;
2678

2679 2680
    if (job == QEMU_JOB_NONE &&
        priv->job.asyncJob == QEMU_ASYNC_JOB_NONE)
2681 2682 2683
        return 0;

    virBufferSetChildIndent(&childBuf, buf);
2684

2685
    virBufferAsprintf(&attrBuf, " type='%s' async='%s'",
2686 2687
                      qemuDomainJobTypeToString(job),
                      qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
2688

2689
    if (priv->job.phase) {
2690
        virBufferAsprintf(&attrBuf, " phase='%s'",
2691 2692 2693
                          qemuDomainAsyncJobPhaseToString(priv->job.asyncJob,
                                                          priv->job.phase));
    }
2694

2695 2696 2697
    if (priv->job.asyncJob != QEMU_ASYNC_JOB_NONE)
        virBufferAsprintf(&attrBuf, " flags='0x%lx'", priv->job.apiFlags);

2698 2699 2700
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_OUT &&
        qemuDomainObjPrivateXMLFormatNBDMigration(&childBuf, vm) < 0)
        goto cleanup;
2701

2702 2703 2704
    if (priv->job.migParams)
        qemuMigrationParamsFormat(&childBuf, priv->job.migParams);

2705
    virXMLFormatElement(buf, "job", &attrBuf, &childBuf);
2706 2707 2708 2709 2710

    ret = 0;

 cleanup:
    return ret;
2711 2712 2713
}


2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729
static bool
qemuDomainHasSlirp(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->nnets; i++) {
        virDomainNetDefPtr net = vm->def->nets[i];

        if (QEMU_DOMAIN_NETWORK_PRIVATE(net)->slirp)
            return true;
    }

    return false;
}


2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747
static bool
qemuDomainGetSlirpHelperOk(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->nnets; i++) {
        virDomainNetDefPtr net = vm->def->nets[i];

        /* if there is a builtin slirp, prevent slirp-helper */
        if (net->type == VIR_DOMAIN_NET_TYPE_USER &&
            !QEMU_DOMAIN_NETWORK_PRIVATE(net)->slirp)
            return false;
    }

    return true;
}


2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788
static int
qemuDomainObjPrivateXMLFormatSlirp(virBufferPtr buf,
                                   virDomainObjPtr vm)
{
    size_t i;

    if (!qemuDomainHasSlirp(vm))
        return 0;

    virBufferAddLit(buf, "<slirp>\n");
    virBufferAdjustIndent(buf, 2);

    for (i = 0; i < vm->def->nnets; i++) {
        virDomainNetDefPtr net = vm->def->nets[i];
        qemuSlirpPtr slirp = QEMU_DOMAIN_NETWORK_PRIVATE(net)->slirp;
        size_t j;

        if (!slirp)
            continue;

        virBufferAsprintf(buf, "<helper alias='%s' pid='%d'>\n",
                          net->info.alias, slirp->pid);

        virBufferAdjustIndent(buf, 2);
        for (j = 0; j < QEMU_SLIRP_FEATURE_LAST; j++) {
            if (qemuSlirpHasFeature(slirp, j)) {
                virBufferAsprintf(buf, "<feature name='%s'/>\n",
                                  qemuSlirpFeatureTypeToString(j));
            }
        }
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</helper>\n");
    }

    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</slirp>\n");


    return 0;
}

2789
static int
2790 2791
qemuDomainObjPrivateXMLFormat(virBufferPtr buf,
                              virDomainObjPtr vm)
2792
{
2793
    qemuDomainObjPrivatePtr priv = vm->privateData;
2794 2795 2796 2797
    const char *monitorpath;

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
2798
        switch (priv->monConfig->type) {
2799
        case VIR_DOMAIN_CHR_TYPE_UNIX:
2800
            monitorpath = priv->monConfig->data.nix.path;
2801 2802 2803
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
2804
            monitorpath = priv->monConfig->data.file.path;
2805 2806 2807
            break;
        }

2808
        virBufferEscapeString(buf, "<monitor path='%s'", monitorpath);
2809
        virBufferAsprintf(buf, " type='%s'/>\n",
2810
                          virDomainChrTypeToString(priv->monConfig->type));
2811 2812
    }

2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823
    if (priv->namespaces) {
        ssize_t ns = -1;

        virBufferAddLit(buf, "<namespaces>\n");
        virBufferAdjustIndent(buf, 2);
        while ((ns = virBitmapNextSetBit(priv->namespaces, ns)) >= 0)
            virBufferAsprintf(buf, "<%s/>\n", qemuDomainNamespaceTypeToString(ns));
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</namespaces>\n");
    }

2824
    qemuDomainObjPrivateXMLFormatVcpus(buf, vm->def);
2825

2826
    if (priv->qemuCaps) {
2827
        size_t i;
2828 2829
        virBufferAddLit(buf, "<qemuCaps>\n");
        virBufferAdjustIndent(buf, 2);
2830
        for (i = 0; i < QEMU_CAPS_LAST; i++) {
2831
            if (virQEMUCapsGet(priv->qemuCaps, i)) {
2832
                virBufferAsprintf(buf, "<flag name='%s'/>\n",
2833
                                  virQEMUCapsTypeToString(i));
2834 2835
            }
        }
2836 2837
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</qemuCaps>\n");
2838 2839
    }

2840
    if (priv->lockState)
2841
        virBufferAsprintf(buf, "<lockstate>%s</lockstate>\n", priv->lockState);
2842

2843 2844
    if (qemuDomainObjPrivateXMLFormatJob(buf, vm, priv) < 0)
        return -1;
2845

2846
    if (priv->fakeReboot)
2847
        virBufferAddLit(buf, "<fakereboot/>\n");
2848

2849 2850
    if (priv->qemuDevices && *priv->qemuDevices) {
        char **tmp = priv->qemuDevices;
2851 2852
        virBufferAddLit(buf, "<devices>\n");
        virBufferAdjustIndent(buf, 2);
2853
        while (*tmp) {
2854
            virBufferAsprintf(buf, "<device alias='%s'/>\n", *tmp);
2855 2856
            tmp++;
        }
2857 2858
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</devices>\n");
2859 2860
    }

2861
    if (qemuDomainObjPrivateXMLFormatAutomaticPlacement(buf, priv) < 0)
2862
        return -1;
2863

2864 2865 2866 2867 2868
    /* Various per-domain paths */
    virBufferEscapeString(buf, "<libDir path='%s'/>\n", priv->libDir);
    virBufferEscapeString(buf, "<channelTargetDir path='%s'/>\n",
                          priv->channelTargetDir);

2869
    virCPUDefFormatBufFull(buf, priv->origCPU, NULL);
2870

2871
    if (priv->chardevStdioLogd)
2872
        virBufferAddLit(buf, "<chardevStdioLogd/>\n");
2873

2874 2875 2876
    if (priv->rememberOwner)
        virBufferAddLit(buf, "<rememberOwner/>\n");

2877 2878
    qemuDomainObjPrivateXMLFormatAllowReboot(buf, priv->allowReboot);

2879 2880
    qemuDomainObjPrivateXMLFormatPR(buf, priv);

2881 2882 2883
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV))
        virBufferAsprintf(buf, "<nodename index='%llu'/>\n", priv->nodenameindex);

2884 2885 2886
    if (priv->memPrealloc)
        virBufferAddLit(buf, "<memPrealloc/>\n");

2887 2888 2889
    if (qemuDomainObjPrivateXMLFormatBlockjobs(buf, vm) < 0)
        return -1;

2890 2891 2892
    if (qemuDomainObjPrivateXMLFormatSlirp(buf, vm) < 0)
        return -1;

2893 2894 2895
    return 0;
}

2896 2897 2898 2899

static int
qemuDomainObjPrivateXMLParseVcpu(xmlNodePtr node,
                                 unsigned int idx,
2900
                                 virDomainDefPtr def)
2901
{
2902
    virDomainVcpuDefPtr vcpu;
2903
    char *idstr;
2904
    char *pidstr;
2905
    unsigned int tmp;
2906 2907
    int ret = -1;

2908 2909
    idstr = virXMLPropString(node, "id");

2910 2911
    if (idstr &&
        (virStrToLong_uip(idstr, NULL, 10, &idx) < 0)) {
2912
        virReportError(VIR_ERR_INTERNAL_ERROR,
2913 2914 2915 2916 2917 2918
                       _("cannot parse vcpu index '%s'"), idstr);
        goto cleanup;
    }
    if (!(vcpu = virDomainDefGetVcpu(def, idx))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("invalid vcpu index '%u'"), idx);
2919
        goto cleanup;
2920 2921
    }

2922 2923 2924
    if (!(pidstr = virXMLPropString(node, "pid")))
        goto cleanup;

2925
    if (virStrToLong_uip(pidstr, NULL, 10, &tmp) < 0)
2926 2927
        goto cleanup;

2928 2929
    QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid = tmp;

2930 2931 2932
    ret = 0;

 cleanup:
2933
    VIR_FREE(idstr);
2934 2935 2936 2937 2938
    VIR_FREE(pidstr);
    return ret;
}


2939 2940 2941 2942 2943 2944 2945
static int
qemuDomainObjPrivateXMLParseAutomaticPlacement(xmlXPathContextPtr ctxt,
                                               qemuDomainObjPrivatePtr priv,
                                               virQEMUDriverPtr driver)
{
    virCapsPtr caps = NULL;
    char *nodeset;
2946
    char *cpuset;
2947 2948
    int nodesetSize = 0;
    size_t i;
2949 2950 2951
    int ret = -1;

    nodeset = virXPathString("string(./numad/@nodeset)", ctxt);
2952
    cpuset = virXPathString("string(./numad/@cpuset)", ctxt);
2953

2954
    if (!nodeset && !cpuset)
2955 2956 2957 2958 2959
        return 0;

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

2960 2961 2962 2963 2964 2965
    /* Figure out how big the nodeset bitmap needs to be.
     * This is necessary because NUMA node IDs are not guaranteed to
     * start from 0 or be densely allocated */
    for (i = 0; i < caps->host.nnumaCell; i++)
        nodesetSize = MAX(nodesetSize, caps->host.numaCell[i]->num + 1);

2966
    if (nodeset &&
2967
        virBitmapParse(nodeset, &priv->autoNodeset, nodesetSize) < 0)
2968 2969
        goto cleanup;

2970 2971 2972 2973 2974 2975 2976 2977 2978 2979
    if (cpuset) {
        if (virBitmapParse(cpuset, &priv->autoCpuset, VIR_DOMAIN_CPUMASK_LEN) < 0)
            goto cleanup;
    } else {
        /* autoNodeset is present in this case, since otherwise we wouldn't
         * reach this code */
        if (!(priv->autoCpuset = virCapabilitiesGetCpusForNodemask(caps,
                                                                   priv->autoNodeset)))
            goto cleanup;
    }
2980 2981 2982 2983 2984 2985

    ret = 0;

 cleanup:
    virObjectUnref(caps);
    VIR_FREE(nodeset);
2986
    VIR_FREE(cpuset);
2987 2988 2989 2990 2991

    return ret;
}


2992 2993 2994 2995 2996 2997 2998
static virStorageSourcePtr
qemuDomainObjPrivateXMLParseBlockjobChain(xmlNodePtr node,
                                          xmlXPathContextPtr ctxt,
                                          virDomainXMLOptionPtr xmlopt)

{
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
2999 3000 3001
    g_autofree char *format = NULL;
    g_autofree char *type = NULL;
    g_autofree char *index = NULL;
3002
    g_autoptr(virStorageSource) src = NULL;
3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025
    xmlNodePtr sourceNode;
    unsigned int xmlflags = VIR_DOMAIN_DEF_PARSE_STATUS;

    ctxt->node = node;

    if (!(type = virXMLPropString(ctxt->node, "type")) ||
        !(format = virXMLPropString(ctxt->node, "format")) ||
        !(index = virXPathString("string(./source/@index)", ctxt)) ||
        !(sourceNode = virXPathNode("./source", ctxt))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing job chain data"));
        return NULL;
    }

    if (!(src = virDomainStorageSourceParseBase(type, format, index)))
        return NULL;

    if (virDomainStorageSourceParse(sourceNode, ctxt, src, xmlflags, xmlopt) < 0)
        return NULL;

    if (virDomainDiskBackingStoreParse(ctxt, src, xmlflags, xmlopt) < 0)
        return NULL;

J
Ján Tomko 已提交
3026
    return g_steal_pointer(&src);
3027 3028 3029
}


3030 3031 3032 3033 3034 3035
static void
qemuDomainObjPrivateXMLParseBlockjobNodename(qemuBlockJobDataPtr job,
                                             const char *xpath,
                                             virStorageSourcePtr *src,
                                             xmlXPathContextPtr ctxt)
{
3036
    g_autofree char *nodename = NULL;
3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063

    *src = NULL;

    if (!(nodename = virXPathString(xpath, ctxt)))
        return;

    if (job->disk &&
        (*src = virStorageSourceFindByNodeName(job->disk->src, nodename, NULL)))
        return;

    if (job->chain &&
        (*src = virStorageSourceFindByNodeName(job->chain, nodename, NULL)))
        return;

    if (job->mirrorChain &&
        (*src = virStorageSourceFindByNodeName(job->mirrorChain, nodename, NULL)))
        return;

    /* the node was in the XML but was not found in the job definitions */
    VIR_DEBUG("marking block job '%s' as invalid: node name '%s' missing",
              job->name, nodename);
    job->invalidData = true;
}


static void
qemuDomainObjPrivateXMLParseBlockjobDataSpecific(qemuBlockJobDataPtr job,
3064 3065
                                                 xmlXPathContextPtr ctxt,
                                                 virDomainXMLOptionPtr xmlopt)
3066
{
3067 3068
    g_autofree char *createmode = NULL;
    g_autofree char *shallownew = NULL;
3069 3070
    xmlNodePtr tmp;

3071 3072 3073 3074 3075 3076 3077 3078 3079 3080
    switch ((qemuBlockJobType) job->type) {
        case QEMU_BLOCKJOB_TYPE_PULL:
            qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                         "string(./base/@node)",
                                                         &job->data.pull.base,
                                                         ctxt);
            /* base is not present if pulling everything */
            break;

        case QEMU_BLOCKJOB_TYPE_COMMIT:
3081 3082 3083 3084 3085 3086 3087 3088
            qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                         "string(./topparent/@node)",
                                                         &job->data.commit.topparent,
                                                         ctxt);

            if (!job->data.commit.topparent)
                goto broken;

3089
            G_GNUC_FALLTHROUGH;
3090
        case QEMU_BLOCKJOB_TYPE_ACTIVE_COMMIT:
3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103
            qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                         "string(./top/@node)",
                                                         &job->data.commit.top,
                                                         ctxt);
            qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                         "string(./base/@node)",
                                                         &job->data.commit.base,
                                                         ctxt);
            if (!job->data.commit.top ||
                !job->data.commit.base)
                goto broken;
            break;

3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116
        case QEMU_BLOCKJOB_TYPE_CREATE:
            if (!(tmp = virXPathNode("./src", ctxt)) ||
                !(job->data.create.src = qemuDomainObjPrivateXMLParseBlockjobChain(tmp, ctxt, xmlopt)))
                goto broken;

            if ((createmode = virXPathString("string(./create/@mode)", ctxt))) {
                if (STRNEQ(createmode, "storage"))
                    goto broken;

                job->data.create.storage = true;
            }
            break;

3117
        case QEMU_BLOCKJOB_TYPE_COPY:
3118 3119 3120 3121 3122 3123 3124 3125
            if ((shallownew =  virXPathString("string(./@shallownew)", ctxt))) {
                if (STRNEQ(shallownew, "yes"))
                    goto broken;

                job->data.copy.shallownew = true;
            }
            break;

3126 3127 3128 3129 3130 3131 3132
        case QEMU_BLOCKJOB_TYPE_NONE:
        case QEMU_BLOCKJOB_TYPE_INTERNAL:
        case QEMU_BLOCKJOB_TYPE_LAST:
            break;
    }

    return;
3133 3134 3135 3136

 broken:
    VIR_DEBUG("marking block job '%s' as invalid: malformed job data", job->name);
    job->invalidData = true;
3137 3138 3139
}


3140
static int
3141 3142
qemuDomainObjPrivateXMLParseBlockjobData(virDomainObjPtr vm,
                                         xmlNodePtr node,
3143 3144
                                         xmlXPathContextPtr ctxt,
                                         virDomainXMLOptionPtr xmlopt)
3145 3146 3147
{
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
    virDomainDiskDefPtr disk = NULL;
3148
    g_autoptr(qemuBlockJobData) job = NULL;
3149 3150
    g_autofree char *name = NULL;
    g_autofree char *typestr = NULL;
3151
    int type;
3152
    g_autofree char *statestr = NULL;
3153
    int state = QEMU_BLOCKJOB_STATE_FAILED;
3154 3155 3156
    g_autofree char *diskdst = NULL;
    g_autofree char *newstatestr = NULL;
    g_autofree char *mirror = NULL;
3157 3158
    int newstate = -1;
    bool invalidData = false;
3159
    xmlNodePtr tmp;
3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187

    ctxt->node = node;

    if (!(name = virXPathString("string(./@name)", ctxt))) {
        VIR_WARN("malformed block job data for vm '%s'", vm->def->name);
        return 0;
    }

    /* if the job name is known we need to register such a job so that we can
     * clean it up */
    if (!(typestr = virXPathString("string(./@type)", ctxt)) ||
        (type = qemuBlockjobTypeFromString(typestr)) < 0) {
        type = QEMU_BLOCKJOB_TYPE_NONE;
        invalidData = true;
    }

    if (!(job = qemuBlockJobDataNew(type, name)))
        return -1;

    if (!(statestr = virXPathString("string(./@state)", ctxt)) ||
        (state = qemuBlockjobStateTypeFromString(statestr)) < 0)
        invalidData = true;

    if ((newstatestr = virXPathString("string(./@newstate)", ctxt)) &&
        (newstate = qemuBlockjobStateTypeFromString(newstatestr)) < 0)
        invalidData = true;

    if ((diskdst = virXPathString("string(./disk/@dst)", ctxt)) &&
3188
        !(disk = virDomainDiskByTarget(vm->def, diskdst)))
3189 3190
        invalidData = true;

3191 3192 3193 3194
    if ((mirror = virXPathString("string(./disk/@mirror)", ctxt)) &&
        STRNEQ(mirror, "yes"))
        invalidData = true;

3195 3196 3197 3198 3199 3200 3201 3202 3203 3204
    if (!disk && !invalidData) {
        if ((tmp = virXPathNode("./chains/disk", ctxt)) &&
            !(job->chain = qemuDomainObjPrivateXMLParseBlockjobChain(tmp, ctxt, xmlopt)))
            invalidData = true;

        if ((tmp = virXPathNode("./chains/mirror", ctxt)) &&
            !(job->mirrorChain = qemuDomainObjPrivateXMLParseBlockjobChain(tmp, ctxt, xmlopt)))
            invalidData = true;
    }

3205 3206 3207 3208 3209 3210 3211
    if (mirror) {
        if (disk)
            job->mirrorChain = virObjectRef(disk->mirror);
        else
            invalidData = true;
    }

3212 3213 3214 3215
    job->state = state;
    job->newstate = newstate;
    job->errmsg = virXPathString("string(./errmsg)", ctxt);
    job->invalidData = invalidData;
3216 3217
    job->disk = disk;

3218
    qemuDomainObjPrivateXMLParseBlockjobDataSpecific(job, ctxt, xmlopt);
3219

3220
    if (qemuBlockJobRegister(job, vm, disk, false) < 0)
3221 3222 3223 3224 3225 3226 3227 3228 3229
        return -1;

    return 0;
}


static int
qemuDomainObjPrivateXMLParseBlockjobs(virDomainObjPtr vm,
                                      qemuDomainObjPrivatePtr priv,
3230 3231
                                      xmlXPathContextPtr ctxt)
{
3232
    g_autofree xmlNodePtr *nodes = NULL;
3233
    ssize_t nnodes = 0;
3234
    g_autofree char *active = NULL;
3235
    int tmp;
3236
    size_t i;
3237 3238 3239 3240 3241

    if ((active = virXPathString("string(./blockjobs/@active)", ctxt)) &&
        (tmp = virTristateBoolTypeFromString(active)) > 0)
        priv->reconnectBlockjobs = tmp;

3242 3243 3244 3245 3246
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if ((nnodes = virXPathNodeSet("./blockjobs/blockjob", ctxt, &nodes)) < 0)
            return -1;

        for (i = 0; i < nnodes; i++) {
3247 3248
            if (qemuDomainObjPrivateXMLParseBlockjobData(vm, nodes[i], ctxt,
                                                         priv->driver->xmlopt) < 0)
3249 3250 3251 3252
                return -1;
        }
    }

3253 3254 3255 3256
    return 0;
}


3257
int
3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281
qemuDomainObjPrivateXMLParseAllowReboot(xmlXPathContextPtr ctxt,
                                        virTristateBool *allowReboot)
{
    int ret = -1;
    int val;
    char *valStr;

    if ((valStr = virXPathString("string(./allowReboot/@value)", ctxt))) {
        if ((val = virTristateBoolTypeFromString(valStr)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("invalid allowReboot value '%s'"), valStr);
            goto cleanup;
        }
        *allowReboot = val;
    }

    ret = 0;

 cleanup:
    VIR_FREE(valStr);
    return ret;
}


3282 3283 3284 3285 3286 3287 3288 3289
static void
qemuDomainObjPrivateXMLParsePR(xmlXPathContextPtr ctxt,
                               bool *prDaemonRunning)
{
    *prDaemonRunning = virXPathBoolean("boolean(./prDaemon)", ctxt) > 0;
}


3290 3291 3292
static int
qemuDomainObjPrivateXMLParseJobNBDSource(xmlNodePtr node,
                                         xmlXPathContextPtr ctxt,
3293 3294
                                         virDomainDiskDefPtr disk,
                                         virDomainXMLOptionPtr xmlopt)
3295
{
3296
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3297
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
3298 3299
    g_autofree char *format = NULL;
    g_autofree char *type = NULL;
3300
    g_autoptr(virStorageSource) migrSource = NULL;
3301
    xmlNodePtr sourceNode;
3302 3303 3304

    ctxt->node = node;

3305 3306
    if (!(ctxt->node = virXPathNode("./migrationSource", ctxt)))
        return 0;
3307 3308 3309 3310

    if (!(type = virXMLPropString(ctxt->node, "type"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing storage source type"));
3311
        return -1;
3312 3313 3314 3315
    }

    if (!(format = virXMLPropString(ctxt->node, "format"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
3316
                       _("missing storage source format"));
3317
        return -1;
3318 3319
    }

3320
    if (!(migrSource = virDomainStorageSourceParseBase(type, format, NULL)))
3321
        return -1;
3322

3323 3324 3325 3326 3327
    /* newer libvirt uses the <source> subelement instead of formatting the
     * source directly into <migrationSource> */
    if ((sourceNode = virXPathNode("./source", ctxt)))
        ctxt->node = sourceNode;

3328
    if (virDomainStorageSourceParse(ctxt->node, ctxt, migrSource,
3329
                                    VIR_DOMAIN_DEF_PARSE_STATUS, xmlopt) < 0)
3330
        return -1;
3331

3332
    diskPriv->migrSource = g_steal_pointer(&migrSource);
3333
    return 0;
3334 3335 3336
}


3337 3338 3339 3340 3341 3342
static int
qemuDomainObjPrivateXMLParseJobNBD(virDomainObjPtr vm,
                                   qemuDomainObjPrivatePtr priv,
                                   xmlXPathContextPtr ctxt)
{
    xmlNodePtr *nodes = NULL;
3343
    char *dst = NULL;
3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359
    size_t i;
    int n;
    int ret = -1;

    if ((n = virXPathNodeSet("./disk[@migrating='yes']", ctxt, &nodes)) < 0)
        goto cleanup;

    if (n > 0) {
        if (priv->job.asyncJob != QEMU_ASYNC_JOB_MIGRATION_OUT) {
            VIR_WARN("Found disks marked for migration but we were not "
                     "migrating");
            n = 0;
        }
        for (i = 0; i < n; i++) {
            virDomainDiskDefPtr disk;

3360
            if ((dst = virXMLPropString(nodes[i], "dev")) &&
3361
                (disk = virDomainDiskByTarget(vm->def, dst))) {
3362
                QEMU_DOMAIN_DISK_PRIVATE(disk)->migrating = true;
3363 3364

                if (qemuDomainObjPrivateXMLParseJobNBDSource(nodes[i], ctxt,
3365 3366
                                                             disk,
                                                             priv->driver->xmlopt) < 0)
3367 3368 3369
                    goto cleanup;
            }

3370 3371 3372 3373 3374 3375 3376 3377
            VIR_FREE(dst);
        }
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
3378
    VIR_FREE(dst);
3379 3380 3381 3382
    return ret;
}


3383 3384 3385 3386 3387
static int
qemuDomainObjPrivateXMLParseJob(virDomainObjPtr vm,
                                qemuDomainObjPrivatePtr priv,
                                xmlXPathContextPtr ctxt)
{
3388
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3389 3390 3391
    char *tmp = NULL;
    int ret = -1;

3392 3393 3394 3395 3396 3397
    if (!(ctxt->node = virXPathNode("./job[1]", ctxt))) {
        ret = 0;
        goto cleanup;
    }

    if ((tmp = virXPathString("string(@type)", ctxt))) {
3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408
        int type;

        if ((type = qemuDomainJobTypeFromString(tmp)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown job type %s"), tmp);
            goto cleanup;
        }
        VIR_FREE(tmp);
        priv->job.active = type;
    }

3409
    if ((tmp = virXPathString("string(@async)", ctxt))) {
3410 3411 3412 3413 3414 3415 3416 3417 3418 3419
        int async;

        if ((async = qemuDomainAsyncJobTypeFromString(tmp)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown async job type %s"), tmp);
            goto cleanup;
        }
        VIR_FREE(tmp);
        priv->job.asyncJob = async;

3420
        if ((tmp = virXPathString("string(@phase)", ctxt))) {
3421 3422 3423 3424 3425 3426 3427 3428 3429 3430
            priv->job.phase = qemuDomainAsyncJobPhaseFromString(async, tmp);
            if (priv->job.phase < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Unknown job phase %s"), tmp);
                goto cleanup;
            }
            VIR_FREE(tmp);
        }
    }

3431 3432 3433 3434 3435
    if (virXPathULongHex("string(@flags)", ctxt, &priv->job.apiFlags) == -2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Invalid job flags"));
        goto cleanup;
    }

3436
    if (qemuDomainObjPrivateXMLParseJobNBD(vm, priv, ctxt) < 0)
3437 3438
        goto cleanup;

3439 3440 3441
    if (qemuMigrationParamsParse(ctxt, &priv->job.migParams) < 0)
        goto cleanup;

3442 3443 3444 3445 3446 3447 3448 3449
    ret = 0;

 cleanup:
    VIR_FREE(tmp);
    return ret;
}


3450 3451 3452 3453 3454 3455
static int
qemuDomainObjPrivateXMLParseSlirpFeatures(xmlNodePtr featuresNode,
                                          xmlXPathContextPtr ctxt,
                                          qemuSlirpPtr slirp)
{
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3456
    g_autofree xmlNodePtr *nodes = NULL;
3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468
    size_t i;
    int n;

    ctxt->node = featuresNode;

    if ((n = virXPathNodeSet("./feature", ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse slirp-helper features"));
        return -1;
    }

    for (i = 0; i < n; i++) {
3469
        g_autofree char *str = virXMLPropString(nodes[i], "name");
3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488
        int feature;

        if (!str)
            continue;

        feature = qemuSlirpFeatureTypeFromString(str);
        if (feature < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown slirp feature %s"), str);
            return -1;
        }

        qemuSlirpSetFeature(slirp, feature);
    }

    return 0;
}


3489
static int
3490
qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt,
3491
                             virDomainObjPtr vm,
3492
                             virDomainDefParserConfigPtr config)
3493
{
3494
    qemuDomainObjPrivatePtr priv = vm->privateData;
3495
    virQEMUDriverPtr driver = config->priv;
3496
    char *monitorpath;
3497
    char *tmp = NULL;
3498 3499
    int n;
    size_t i;
3500
    xmlNodePtr *nodes = NULL;
3501
    xmlNodePtr node = NULL;
3502
    virQEMUCapsPtr qemuCaps = NULL;
3503

3504
    if (!(priv->monConfig = virDomainChrSourceDefNew(NULL)))
3505 3506 3507 3508
        goto error;

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
3509 3510
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("no monitor path"));
3511 3512 3513 3514 3515
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
3516
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
3517
    else
3518
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
3519 3520
    VIR_FREE(tmp);

3521
    switch (priv->monConfig->type) {
3522
    case VIR_DOMAIN_CHR_TYPE_PTY:
3523
        priv->monConfig->data.file.path = monitorpath;
3524 3525
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
3526
        priv->monConfig->data.nix.path = monitorpath;
3527 3528 3529
        break;
    default:
        VIR_FREE(monitorpath);
3530 3531 3532
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported monitor type '%s'"),
                       virDomainChrTypeToString(priv->monConfig->type));
3533 3534 3535
        goto error;
    }

3536 3537 3538 3539
    if ((node = virXPathNode("./namespaces", ctxt))) {
        xmlNodePtr next;

        for (next = node->children; next; next = next->next) {
3540
            int ns = qemuDomainNamespaceTypeFromString((const char *)next->name);
3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559

            if (ns < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("malformed namespace name: %s"),
                               next->name);
                goto error;
            }

            if (qemuDomainEnableNamespace(vm, ns) < 0)
                goto error;
        }
    }

    if (priv->namespaces &&
        virBitmapIsAllClear(priv->namespaces)) {
        virBitmapFree(priv->namespaces);
        priv->namespaces = NULL;
    }

3560 3561
    priv->rememberOwner = virXPathBoolean("count(./rememberOwner) > 0", ctxt);

3562
    if ((n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes)) < 0)
3563 3564
        goto error;

3565
    for (i = 0; i < n; i++) {
3566
        if (qemuDomainObjPrivateXMLParseVcpu(nodes[i], i, vm->def) < 0)
3567
            goto error;
3568
    }
3569
    VIR_FREE(nodes);
3570

3571
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
3572 3573
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse qemu capabilities flags"));
3574 3575 3576
        goto error;
    }
    if (n > 0) {
3577
        if (!(qemuCaps = virQEMUCapsNew()))
3578 3579
            goto error;

3580
        for (i = 0; i < n; i++) {
3581 3582
            char *str = virXMLPropString(nodes[i], "name");
            if (str) {
3583
                int flag = virQEMUCapsTypeFromString(str);
3584
                if (flag < 0) {
3585 3586
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Unknown qemu capabilities flag %s"), str);
3587
                    VIR_FREE(str);
3588 3589
                    goto error;
                }
3590
                VIR_FREE(str);
3591
                virQEMUCapsSet(qemuCaps, flag);
3592 3593 3594
            }
        }

3595
        priv->qemuCaps = g_steal_pointer(&qemuCaps);
3596 3597 3598
    }
    VIR_FREE(nodes);

3599
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
3600

3601
    if (qemuDomainObjPrivateXMLParseJob(vm, priv, ctxt) < 0)
3602 3603
        goto error;

3604 3605
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626
    if ((n = virXPathNodeSet("./devices/device", ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse qemu device list"));
        goto error;
    }
    if (n > 0) {
        /* NULL-terminated list */
        if (VIR_ALLOC_N(priv->qemuDevices, n + 1) < 0)
            goto error;

        for (i = 0; i < n; i++) {
            priv->qemuDevices[i] = virXMLPropString(nodes[i], "alias");
            if (!priv->qemuDevices[i]) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("failed to parse qemu device list"));
                goto error;
            }
        }
    }
    VIR_FREE(nodes);

3627 3628 3629 3630 3631 3632
    if ((n = virXPathNodeSet("./slirp/helper", ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse slirp helper list"));
        goto error;
    }
    for (i = 0; i < n; i++) {
3633 3634
        g_autofree char *alias = virXMLPropString(nodes[i], "alias");
        g_autofree char *pid = virXMLPropString(nodes[i], "pid");
J
Ján Tomko 已提交
3635
        g_autoptr(qemuSlirp) slirp = qemuSlirpNew();
3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651
        virDomainDeviceDef dev;

        if (!alias || !pid || !slirp ||
            virStrToLong_i(pid, NULL, 10, &slirp->pid) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("failed to parse slirp helper list"));
            goto error;
        }

        if (virDomainDefFindDevice(vm->def, alias, &dev, true) < 0 ||
            dev.type != VIR_DOMAIN_DEVICE_NET)
            goto error;

        if (qemuDomainObjPrivateXMLParseSlirpFeatures(nodes[i], ctxt, slirp) < 0)
            goto error;

3652
        QEMU_DOMAIN_NETWORK_PRIVATE(dev.data.net)->slirp = g_steal_pointer(&slirp);
3653 3654 3655
    }
    VIR_FREE(nodes);

3656
    if (qemuDomainObjPrivateXMLParseAutomaticPlacement(ctxt, priv, driver) < 0)
3657 3658
        goto error;

3659 3660 3661 3662 3663 3664
    if ((tmp = virXPathString("string(./libDir/@path)", ctxt)))
        priv->libDir = tmp;
    if ((tmp = virXPathString("string(./channelTargetDir/@path)", ctxt)))
        priv->channelTargetDir = tmp;
    tmp = NULL;

3665
    qemuDomainSetPrivatePathsOld(driver, vm);
3666

3667 3668 3669
    if (virCPUDefParseXML(ctxt, "./cpu", VIR_CPU_TYPE_GUEST, &priv->origCPU) < 0)
        goto error;

3670 3671 3672
    priv->chardevStdioLogd = virXPathBoolean("boolean(./chardevStdioLogd)",
                                             ctxt) == 1;

3673 3674
    qemuDomainObjPrivateXMLParseAllowReboot(ctxt, &priv->allowReboot);

3675 3676
    qemuDomainObjPrivateXMLParsePR(ctxt, &priv->prDaemonRunning);

3677
    if (qemuDomainObjPrivateXMLParseBlockjobs(vm, priv, ctxt) < 0)
3678 3679
        goto error;

3680 3681 3682 3683 3684 3685 3686 3687
    qemuDomainStorageIdReset(priv);
    if (virXPathULongLong("string(./nodename/@index)", ctxt,
                          &priv->nodenameindex) == -2) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("failed to parse node name index"));
        goto error;
    }

3688 3689
    priv->memPrealloc = virXPathBoolean("boolean(./memPrealloc)", ctxt) == 1;

3690 3691
    return 0;

3692
 error:
3693
    VIR_FREE(nodes);
3694
    VIR_FREE(tmp);
3695 3696
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
3697
    virObjectUnref(priv->monConfig);
3698
    priv->monConfig = NULL;
3699
    virStringListFree(priv->qemuDevices);
3700
    priv->qemuDevices = NULL;
3701
    virObjectUnref(qemuCaps);
3702 3703 3704 3705
    return -1;
}


3706 3707 3708 3709 3710 3711 3712 3713 3714
static void *
qemuDomainObjPrivateXMLGetParseOpaque(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return priv->qemuCaps;
}


3715 3716 3717
virDomainXMLPrivateDataCallbacks virQEMUDriverPrivateDataCallbacks = {
    .alloc = qemuDomainObjPrivateAlloc,
    .free = qemuDomainObjPrivateFree,
3718
    .diskNew = qemuDomainDiskPrivateNew,
3719 3720
    .diskParse = qemuDomainDiskPrivateParse,
    .diskFormat = qemuDomainDiskPrivateFormat,
3721
    .vcpuNew = qemuDomainVcpuPrivateNew,
3722
    .chrSourceNew = qemuDomainChrSourcePrivateNew,
J
Ján Tomko 已提交
3723
    .vsockNew = qemuDomainVsockPrivateNew,
3724
    .graphicsNew = qemuDomainGraphicsPrivateNew,
3725
    .networkNew = qemuDomainNetworkPrivateNew,
3726
    .videoNew = qemuDomainVideoPrivateNew,
3727 3728
    .parse = qemuDomainObjPrivateXMLParse,
    .format = qemuDomainObjPrivateXMLFormat,
3729
    .getParseOpaque = qemuDomainObjPrivateXMLGetParseOpaque,
3730 3731
    .storageParse = qemuStorageSourcePrivateDataParse,
    .storageFormat = qemuStorageSourcePrivateDataFormat,
3732 3733 3734
};


3735 3736 3737 3738 3739 3740
static void
qemuDomainXmlNsDefFree(qemuDomainXmlNsDefPtr def)
{
    if (!def)
        return;

3741 3742 3743
    virStringListFreeCount(def->args, def->num_args);
    virStringListFreeCount(def->env_name, def->num_env);
    virStringListFreeCount(def->env_value, def->num_env);
3744 3745
    virStringListFreeCount(def->capsadd, def->ncapsadd);
    virStringListFreeCount(def->capsdel, def->ncapsdel);
3746

3747 3748 3749 3750
    VIR_FREE(def);
}


3751 3752 3753
static void
qemuDomainDefNamespaceFree(void *nsdata)
{
3754
    qemuDomainXmlNsDefPtr cmd = nsdata;
3755

3756
    qemuDomainXmlNsDefFree(cmd);
3757 3758
}

3759 3760 3761 3762 3763

static int
qemuDomainDefNamespaceParseCommandlineArgs(qemuDomainXmlNsDefPtr nsdef,
                                           xmlXPathContextPtr ctxt)
{
3764
    g_autofree xmlNodePtr *nodes = NULL;
3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788
    ssize_t nnodes;
    size_t i;

    if ((nnodes = virXPathNodeSet("./qemu:commandline/qemu:arg", ctxt, &nodes)) < 0)
        return -1;

    if (nnodes == 0)
        return 0;

    if (VIR_ALLOC_N(nsdef->args, nnodes) < 0)
        return -1;

    for (i = 0; i < nnodes; i++) {
        if (!(nsdef->args[nsdef->num_args++] = virXMLPropString(nodes[i], "value"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("No qemu command-line argument specified"));
            return -1;
        }
    }

    return 0;
}


3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811
static int
qemuDomainDefNamespaceParseCommandlineEnvNameValidate(const char *envname)
{
    if (!c_isalpha(envname[0]) && envname[0] != '_') {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Invalid environment name, it must begin with a letter or underscore"));
        return -1;
    }

    if (strspn(envname, "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789_") != strlen(envname)) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Invalid environment name, it must contain only alphanumerics and underscore"));
        return -1;
    }

    return 0;
}


static int
qemuDomainDefNamespaceParseCommandlineEnv(qemuDomainXmlNsDefPtr nsdef,
                                          xmlXPathContextPtr ctxt)
{
3812
    g_autofree xmlNodePtr *nodes = NULL;
3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844
    ssize_t nnodes;
    size_t i;

    if ((nnodes = virXPathNodeSet("./qemu:commandline/qemu:env", ctxt, &nodes)) < 0)
        return -1;

    if (nnodes == 0)
        return 0;

    if (VIR_ALLOC_N(nsdef->env_name, nnodes) < 0 ||
        VIR_ALLOC_N(nsdef->env_value, nnodes) < 0)
        return -1;

    for (i = 0; i < nnodes; i++) {
        if (!(nsdef->env_name[nsdef->num_env] = virXMLPropString(nodes[i], "name"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("No qemu environment name specified"));
            return -1;
        }

        if (qemuDomainDefNamespaceParseCommandlineEnvNameValidate(nsdef->env_name[nsdef->num_env]) < 0)
            return -1;

        nsdef->env_value[nsdef->num_env] = virXMLPropString(nodes[i], "value");
        /* a NULL value for command is allowed, since it might be empty */
        nsdef->num_env++;
    }

    return 0;
}


3845 3846 3847 3848
static int
qemuDomainDefNamespaceParseCaps(qemuDomainXmlNsDefPtr nsdef,
                                xmlXPathContextPtr ctxt)
{
3849
    g_autofree xmlNodePtr *nodesadd = NULL;
3850
    ssize_t nnodesadd;
3851
    g_autofree xmlNodePtr *nodesdel = NULL;
3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888
    ssize_t nnodesdel;
    size_t i;

    if ((nnodesadd = virXPathNodeSet("./qemu:capabilities/qemu:add", ctxt, &nodesadd)) < 0 ||
        (nnodesdel = virXPathNodeSet("./qemu:capabilities/qemu:del", ctxt, &nodesdel)) < 0)
        return -1;

    if (nnodesadd > 0) {
        if (VIR_ALLOC_N(nsdef->capsadd, nnodesadd) < 0)
            return -1;

        for (i = 0; i < nnodesadd; i++) {
            if (!(nsdef->capsadd[nsdef->ncapsadd++] = virXMLPropString(nodesadd[i], "capability"))) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing capability name"));
                return -1;
            }
        }
    }

    if (nnodesdel > 0) {
        if (VIR_ALLOC_N(nsdef->capsdel, nnodesdel) < 0)
            return -1;

        for (i = 0; i < nnodesdel; i++) {
            if (!(nsdef->capsdel[nsdef->ncapsdel++] = virXMLPropString(nodesdel[i], "capability"))) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing capability name"));
                return -1;
            }
        }
    }

    return 0;
}


3889
static int
3890
qemuDomainDefNamespaceParse(xmlXPathContextPtr ctxt,
3891 3892
                            void **data)
{
3893 3894
    qemuDomainXmlNsDefPtr nsdata = NULL;
    int ret = -1;
3895

3896
    if (VIR_ALLOC(nsdata) < 0)
3897 3898
        return -1;

3899
    if (qemuDomainDefNamespaceParseCommandlineArgs(nsdata, ctxt) < 0 ||
3900 3901
        qemuDomainDefNamespaceParseCommandlineEnv(nsdata, ctxt) < 0 ||
        qemuDomainDefNamespaceParseCaps(nsdata, ctxt) < 0)
3902
        goto cleanup;
3903

3904 3905
    if (nsdata->num_args > 0 || nsdata->num_env > 0 ||
        nsdata->ncapsadd > 0 || nsdata->ncapsdel > 0)
3906
        *data = g_steal_pointer(&nsdata);
3907

3908
    ret = 0;
3909

3910 3911 3912
 cleanup:
    qemuDomainDefNamespaceFree(nsdata);
    return ret;
3913 3914
}

3915 3916 3917 3918

static void
qemuDomainDefNamespaceFormatXMLCommandline(virBufferPtr buf,
                                           qemuDomainXmlNsDefPtr cmd)
3919
{
3920
    size_t i;
3921 3922

    if (!cmd->num_args && !cmd->num_env)
3923
        return;
3924

3925 3926 3927
    virBufferAddLit(buf, "<qemu:commandline>\n");
    virBufferAdjustIndent(buf, 2);

3928
    for (i = 0; i < cmd->num_args; i++)
3929
        virBufferEscapeString(buf, "<qemu:arg value='%s'/>\n",
3930 3931
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
3932
        virBufferAsprintf(buf, "<qemu:env name='%s'", cmd->env_name[i]);
3933 3934 3935 3936 3937
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }

3938 3939
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:commandline>\n");
3940 3941 3942
}


3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965
static void
qemuDomainDefNamespaceFormatXMLCaps(virBufferPtr buf,
                                    qemuDomainXmlNsDefPtr xmlns)
{
    size_t i;

    if (!xmlns->ncapsadd && !xmlns->ncapsdel)
        return;

    virBufferAddLit(buf, "<qemu:capabilities>\n");
    virBufferAdjustIndent(buf, 2);

    for (i = 0; i < xmlns->ncapsadd; i++)
        virBufferEscapeString(buf, "<qemu:add capability='%s'/>\n", xmlns->capsadd[i]);

    for (i = 0; i < xmlns->ncapsdel; i++)
        virBufferEscapeString(buf, "<qemu:del capability='%s'/>\n", xmlns->capsdel[i]);

    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:capabilities>\n");
}


3966 3967 3968 3969 3970 3971 3972
static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainXmlNsDefPtr cmd = nsdata;

    qemuDomainDefNamespaceFormatXMLCommandline(buf, cmd);
3973
    qemuDomainDefNamespaceFormatXMLCaps(buf, cmd);
3974

3975 3976 3977 3978
    return 0;
}


3979
virXMLNamespace virQEMUDriverDomainXMLNamespace = {
3980 3981 3982
    .parse = qemuDomainDefNamespaceParse,
    .free = qemuDomainDefNamespaceFree,
    .format = qemuDomainDefNamespaceFormatXML,
3983
    .prefix = "qemu",
3984
    .uri = "http://libvirt.org/schemas/domain/qemu/1.0",
3985
};
3986

3987

P
Pavel Hrdina 已提交
3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006
static int
qemuDomainDefAddImplicitInputDevice(virDomainDef *def)
{
    if (ARCH_IS_X86(def->os.arch)) {
        if (virDomainDefMaybeAddInput(def,
                                      VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                      VIR_DOMAIN_INPUT_BUS_PS2) < 0)
            return -1;

        if (virDomainDefMaybeAddInput(def,
                                      VIR_DOMAIN_INPUT_TYPE_KBD,
                                      VIR_DOMAIN_INPUT_BUS_PS2) < 0)
            return -1;
    }

    return 0;
}


4007
static int
4008 4009
qemuDomainDefAddDefaultDevices(virDomainDefPtr def,
                               virQEMUCapsPtr qemuCaps)
4010
{
4011
    bool addDefaultUSB = true;
4012
    int usbModel = -1; /* "default for machinetype" */
4013
    int pciRoot;       /* index within def->controllers */
4014
    bool addImplicitSATA = false;
4015
    bool addPCIRoot = false;
L
Laine Stump 已提交
4016
    bool addPCIeRoot = false;
4017
    bool addDefaultMemballoon = true;
4018 4019
    bool addDefaultUSBKBD = false;
    bool addDefaultUSBMouse = false;
4020
    bool addPanicDevice = false;
4021
    int ret = -1;
4022

P
Pavel Hrdina 已提交
4023 4024 4025 4026
    /* add implicit input devices */
    if (qemuDomainDefAddImplicitInputDevice(def) < 0)
        goto cleanup;

4027 4028 4029 4030
    /* Add implicit PCI root controller if the machine has one */
    switch (def->os.arch) {
    case VIR_ARCH_I686:
    case VIR_ARCH_X86_64:
L
Laine Stump 已提交
4031
        if (STREQ(def->os.machine, "isapc")) {
4032
            addDefaultUSB = false;
4033
            break;
4034
        }
4035
        if (qemuDomainIsQ35(def)) {
4036 4037
            addPCIeRoot = true;
            addImplicitSATA = true;
4038

4039 4040 4041
            /* Prefer adding a USB3 controller if supported, fall back
             * to USB2 if there is no USB3 available, and if that's
             * unavailable don't add anything.
4042
             */
4043 4044 4045
            if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_QEMU_XHCI))
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_QEMU_XHCI;
            else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_NEC_USB_XHCI))
4046 4047
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
            else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_USB_EHCI1))
4048 4049 4050
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_ICH9_EHCI1;
            else
                addDefaultUSB = false;
4051
            break;
L
Laine Stump 已提交
4052
        }
4053
        if (qemuDomainIsI440FX(def))
4054
            addPCIRoot = true;
4055 4056
        break;

4057 4058 4059 4060 4061 4062 4063
    case VIR_ARCH_ARMV6L:
        addDefaultUSB = false;
        addDefaultMemballoon = false;
        if (STREQ(def->os.machine, "versatilepb"))
            addPCIRoot = true;
        break;

4064
    case VIR_ARCH_ARMV7L:
4065
    case VIR_ARCH_AARCH64:
4066 4067
        addDefaultUSB = false;
        addDefaultMemballoon = false;
4068
        if (qemuDomainIsARMVirt(def))
4069 4070
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
        break;
4071

4072
    case VIR_ARCH_PPC64:
4073
    case VIR_ARCH_PPC64LE:
4074 4075 4076
        addPCIRoot = true;
        addDefaultUSBKBD = true;
        addDefaultUSBMouse = true;
4077 4078 4079
        /* For pSeries guests, the firmware provides the same
         * functionality as the pvpanic device, so automatically
         * add the definition if not already present */
4080
        if (qemuDomainIsPSeries(def))
4081
            addPanicDevice = true;
4082 4083
        break;

4084 4085 4086 4087 4088 4089 4090
    case VIR_ARCH_ALPHA:
    case VIR_ARCH_PPC:
    case VIR_ARCH_PPCEMB:
    case VIR_ARCH_SH4:
    case VIR_ARCH_SH4EB:
        addPCIRoot = true;
        break;
4091

4092 4093 4094
    case VIR_ARCH_RISCV32:
    case VIR_ARCH_RISCV64:
        addDefaultUSB = false;
4095 4096
        if (qemuDomainIsRISCVVirt(def))
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
4097 4098
        break;

4099 4100 4101
    case VIR_ARCH_S390:
    case VIR_ARCH_S390X:
        addDefaultUSB = false;
4102
        addPanicDevice = true;
4103
        addPCIRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_ZPCI);
4104
        break;
4105 4106 4107 4108 4109 4110

    case VIR_ARCH_SPARC:
    case VIR_ARCH_SPARC64:
        addPCIRoot = true;
        break;

4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130
    case VIR_ARCH_ARMV7B:
    case VIR_ARCH_CRIS:
    case VIR_ARCH_ITANIUM:
    case VIR_ARCH_LM32:
    case VIR_ARCH_M68K:
    case VIR_ARCH_MICROBLAZE:
    case VIR_ARCH_MICROBLAZEEL:
    case VIR_ARCH_MIPS:
    case VIR_ARCH_MIPSEL:
    case VIR_ARCH_MIPS64:
    case VIR_ARCH_MIPS64EL:
    case VIR_ARCH_OR32:
    case VIR_ARCH_PARISC:
    case VIR_ARCH_PARISC64:
    case VIR_ARCH_PPCLE:
    case VIR_ARCH_UNICORE32:
    case VIR_ARCH_XTENSA:
    case VIR_ARCH_XTENSAEB:
    case VIR_ARCH_NONE:
    case VIR_ARCH_LAST:
4131 4132 4133 4134
    default:
        break;
    }

4135
    if (addDefaultUSB &&
4136 4137
        virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_USB, 0) < 0 &&
        virDomainDefAddUSBController(def, 0, usbModel) < 0)
4138
        goto cleanup;
4139

4140 4141 4142
    if (addImplicitSATA &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_SATA, 0, -1) < 0)
4143
        goto cleanup;
4144

4145 4146
    pciRoot = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

4147 4148 4149
    /* NB: any machine that sets addPCIRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
     */
4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164
    if (addPCIRoot) {
        if (pciRoot >= 0) {
            if (def->controllers[pciRoot]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT) {
                virReportError(VIR_ERR_XML_ERROR,
                               _("The PCI controller with index='0' must be "
                                 "model='pci-root' for this machine type, "
                                 "but model='%s' was found instead"),
                               virDomainControllerModelPCITypeToString(def->controllers[pciRoot]->model));
                goto cleanup;
            }
        } else if (!virDomainDefAddController(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
                                              VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)) {
            goto cleanup;
        }
    }
4165

4166 4167 4168
    /* When a machine has a pcie-root, make sure that there is always
     * a dmi-to-pci-bridge controller added as bus 1, and a pci-bridge
     * as bus 2, so that standard PCI devices can be connected
4169 4170 4171
     *
     * NB: any machine that sets addPCIeRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
4172 4173
     */
    if (addPCIeRoot) {
4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186
        if (pciRoot >= 0) {
            if (def->controllers[pciRoot]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT) {
                virReportError(VIR_ERR_XML_ERROR,
                               _("The PCI controller with index='0' must be "
                                 "model='pcie-root' for this machine type, "
                                 "but model='%s' was found instead"),
                               virDomainControllerModelPCITypeToString(def->controllers[pciRoot]->model));
                goto cleanup;
            }
        } else if (!virDomainDefAddController(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
                                             VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT)) {
            goto cleanup;
        }
4187
    }
4188

4189
    if (addDefaultMemballoon && !def->memballoon) {
4190 4191
        virDomainMemballoonDefPtr memballoon;
        if (VIR_ALLOC(memballoon) < 0)
4192
            goto cleanup;
4193 4194 4195 4196 4197

        memballoon->model = VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO;
        def->memballoon = memballoon;
    }

4198 4199 4200 4201
    if (STRPREFIX(def->os.machine, "s390-virtio") &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_S390) && def->memballoon)
        def->memballoon->model = VIR_DOMAIN_MEMBALLOON_MODEL_NONE;

4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221
    if (addDefaultUSBMouse) {
        bool hasUSBTablet = false;
        size_t j;

        for (j = 0; j < def->ninputs; j++) {
            if (def->inputs[j]->type == VIR_DOMAIN_INPUT_TYPE_TABLET &&
                def->inputs[j]->bus == VIR_DOMAIN_INPUT_BUS_USB) {
                hasUSBTablet = true;
                break;
            }
        }

        /* Historically, we have automatically added USB keyboard and
         * mouse to some guests. While the former device is generally
         * safe to have, adding the latter is undesiderable if a USB
         * tablet is already present in the guest */
        if (hasUSBTablet)
            addDefaultUSBMouse = false;
    }

4222 4223 4224 4225 4226
    if (addDefaultUSBKBD &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_KBD,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
4227
        goto cleanup;
4228 4229 4230 4231 4232 4233

    if (addDefaultUSBMouse &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
4234
        goto cleanup;
4235

D
Dmitry Andreev 已提交
4236 4237 4238 4239
    if (addPanicDevice) {
        size_t j;
        for (j = 0; j < def->npanics; j++) {
            if (def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT ||
4240 4241 4242 4243
                (ARCH_IS_PPC64(def->os.arch) &&
                     def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_PSERIES) ||
                (ARCH_IS_S390(def->os.arch) &&
                     def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_S390))
D
Dmitry Andreev 已提交
4244 4245
                break;
        }
4246

D
Dmitry Andreev 已提交
4247 4248 4249 4250 4251 4252 4253 4254 4255
        if (j == def->npanics) {
            virDomainPanicDefPtr panic;
            if (VIR_ALLOC(panic) < 0 ||
                VIR_APPEND_ELEMENT_COPY(def->panics,
                                        def->npanics, panic) < 0) {
                VIR_FREE(panic);
                goto cleanup;
            }
        }
4256 4257
    }

4258 4259 4260 4261 4262 4263
    ret = 0;
 cleanup:
    return ret;
}


A
Andrea Bolognani 已提交
4264 4265 4266
/**
 * qemuDomainDefEnableDefaultFeatures:
 * @def: domain definition
4267
 * @qemuCaps: QEMU capabilities
A
Andrea Bolognani 已提交
4268 4269 4270 4271 4272
 *
 * Make sure that features that should be enabled by default are actually
 * enabled and configure default values related to those features.
 */
static void
4273 4274
qemuDomainDefEnableDefaultFeatures(virDomainDefPtr def,
                                   virQEMUCapsPtr qemuCaps)
A
Andrea Bolognani 已提交
4275
{
4276
    /* The virt machine type always uses GIC: if the relevant information
4277 4278
     * was not included in the domain XML, we need to choose a suitable
     * GIC version ourselves */
4279
    if ((def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ABSENT &&
4280
         qemuDomainIsARMVirt(def)) ||
4281 4282 4283
        (def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ON &&
         def->gic_version == VIR_GIC_VERSION_NONE)) {
        virGICVersion version;
4284 4285 4286 4287 4288

        VIR_DEBUG("Looking for usable GIC version in domain capabilities");
        for (version = VIR_GIC_VERSION_LAST - 1;
             version > VIR_GIC_VERSION_NONE;
             version--) {
4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303

            /* We want to use the highest available GIC version for guests;
             * however, the emulated GICv3 is currently lacking a MSI controller,
             * making it unsuitable for the pure PCIe topology we aim for.
             *
             * For that reason, we skip this step entirely for TCG guests,
             * and rely on the code below to pick the default version, GICv2,
             * which supports all the features we need.
             *
             * See https://bugzilla.redhat.com/show_bug.cgi?id=1414081 */
            if (version == VIR_GIC_VERSION_3 &&
                def->virtType == VIR_DOMAIN_VIRT_QEMU) {
                continue;
            }

4304 4305 4306 4307 4308 4309 4310 4311
            if (virQEMUCapsSupportsGICVersion(qemuCaps,
                                              def->virtType,
                                              version)) {
                VIR_DEBUG("Using GIC version %s",
                          virGICVersionTypeToString(version));
                def->gic_version = version;
                break;
            }
4312 4313
        }

4314 4315 4316 4317 4318 4319 4320
        /* Use the default GIC version (GICv2) as a last-ditch attempt
         * if no match could be found above */
        if (def->gic_version == VIR_GIC_VERSION_NONE) {
            VIR_DEBUG("Using GIC version 2 (default)");
            def->gic_version = VIR_GIC_VERSION_2;
        }

4321 4322 4323
        /* Even if we haven't found a usable GIC version in the domain
         * capabilities, we still want to enable this */
        def->features[VIR_DOMAIN_FEATURE_GIC] = VIR_TRISTATE_SWITCH_ON;
4324
    }
A
Andrea Bolognani 已提交
4325 4326 4327
}


4328 4329 4330 4331 4332 4333 4334 4335 4336 4337
static int
qemuCanonicalizeMachine(virDomainDefPtr def, virQEMUCapsPtr qemuCaps)
{
    const char *canon;

    if (!(canon = virQEMUCapsGetCanonicalMachine(qemuCaps, def->os.machine)))
        return 0;

    if (STRNEQ(canon, def->os.machine)) {
        char *tmp;
4338
        tmp = g_strdup(canon);
4339 4340 4341 4342 4343 4344 4345 4346
        VIR_FREE(def->os.machine);
        def->os.machine = tmp;
    }

    return 0;
}


4347
static int
4348 4349 4350 4351 4352
qemuDomainRecheckInternalPaths(virDomainDefPtr def,
                               virQEMUDriverConfigPtr cfg,
                               unsigned int flags)
{
    size_t i = 0;
4353
    size_t j = 0;
4354 4355 4356 4357

    for (i = 0; i < def->ngraphics; ++i) {
        virDomainGraphicsDefPtr graphics = def->graphics[i];

4358 4359 4360 4361 4362 4363
        for (j = 0; j < graphics->nListens; ++j) {
            virDomainGraphicsListenDefPtr glisten =  &graphics->listens[j];

            /* This will happen only if we parse XML from old libvirts where
             * unix socket was available only for VNC graphics.  In this
             * particular case we should follow the behavior and if we remove
4364
             * the auto-generated socket based on config option from qemu.conf
4365 4366 4367 4368
             * we need to change the listen type to address. */
            if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
                glisten->type == VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_SOCKET &&
                glisten->socket &&
4369
                !glisten->autoGenerated &&
4370 4371 4372 4373 4374 4375 4376
                STRPREFIX(glisten->socket, cfg->libDir)) {
                if (flags & VIR_DOMAIN_DEF_PARSE_INACTIVE) {
                    VIR_FREE(glisten->socket);
                    glisten->type = VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_ADDRESS;
                } else {
                    glisten->fromConfig = true;
                }
4377
            }
4378 4379
        }
    }
4380 4381

    return 0;
4382 4383 4384
}


4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454
static int
qemuDomainDefVcpusPostParse(virDomainDefPtr def)
{
    unsigned int maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    virDomainVcpuDefPtr prevvcpu;
    size_t i;
    bool has_order = false;

    /* vcpu 0 needs to be present, first, and non-hotpluggable */
    vcpu = virDomainDefGetVcpu(def, 0);
    if (!vcpu->online) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("vcpu 0 can't be offline"));
        return -1;
    }
    if (vcpu->hotpluggable == VIR_TRISTATE_BOOL_YES) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("vcpu0 can't be hotpluggable"));
        return -1;
    }
    if (vcpu->order != 0 && vcpu->order != 1) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("vcpu0 must be enabled first"));
        return -1;
    }

    if (vcpu->order != 0)
        has_order = true;

    prevvcpu = vcpu;

    /* all online vcpus or non online vcpu need to have order set */
    for (i = 1; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(def, i);

        if (vcpu->online &&
            (vcpu->order != 0) != has_order) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("all vcpus must have either set or unset order"));
            return -1;
        }

        /* few conditions for non-hotpluggable (thus online) vcpus */
        if (vcpu->hotpluggable == VIR_TRISTATE_BOOL_NO) {
            /* they can be ordered only at the beginning */
            if (prevvcpu->hotpluggable == VIR_TRISTATE_BOOL_YES) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("online non-hotpluggable vcpus need to be "
                                 "ordered prior to hotplugable vcpus"));
                return -1;
            }

            /* they need to be in order (qemu doesn't support any order yet).
             * Also note that multiple vcpus may share order on some platforms */
            if (prevvcpu->order > vcpu->order) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("online non-hotpluggable vcpus must be ordered "
                                 "in ascending order"));
                return -1;
            }
        }

        prevvcpu = vcpu;
    }

    return 0;
}


4455 4456 4457
static int
qemuDomainDefCPUPostParse(virDomainDefPtr def)
{
A
Andrea Bolognani 已提交
4458 4459 4460 4461
    virCPUFeatureDefPtr sveFeature = NULL;
    bool sveVectorLengthsProvided = false;
    size_t i;

4462 4463 4464
    if (!def->cpu)
        return 0;

4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518
    if (def->cpu->cache) {
        virCPUCacheDefPtr cache = def->cpu->cache;

        if (!ARCH_IS_X86(def->os.arch)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("CPU cache specification is not supported "
                             "for '%s' architecture"),
                           virArchToString(def->os.arch));
            return -1;
        }

        switch (cache->mode) {
        case VIR_CPU_CACHE_MODE_EMULATE:
            if (cache->level != 3) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("CPU cache mode '%s' can only be used with "
                                 "level='3'"),
                               virCPUCacheModeTypeToString(cache->mode));
                return -1;
            }
            break;

        case VIR_CPU_CACHE_MODE_PASSTHROUGH:
            if (def->cpu->mode != VIR_CPU_MODE_HOST_PASSTHROUGH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("CPU cache mode '%s' can only be used with "
                                 "'%s' CPUs"),
                               virCPUCacheModeTypeToString(cache->mode),
                               virCPUModeTypeToString(VIR_CPU_MODE_HOST_PASSTHROUGH));
                return -1;
            }

            if (cache->level != -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("unsupported CPU cache level for mode '%s'"),
                               virCPUCacheModeTypeToString(cache->mode));
                return -1;
            }
            break;

        case VIR_CPU_CACHE_MODE_DISABLE:
            if (cache->level != -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("unsupported CPU cache level for mode '%s'"),
                               virCPUCacheModeTypeToString(cache->mode));
                return -1;
            }
            break;

        case VIR_CPU_CACHE_MODE_LAST:
            break;
        }
    }

A
Andrea Bolognani 已提交
4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551
    for (i = 0; i < def->cpu->nfeatures; i++) {
        virCPUFeatureDefPtr feature = &def->cpu->features[i];

        if (STREQ(feature->name, "sve")) {
            sveFeature = feature;
        } else if (STRPREFIX(feature->name, "sve")) {
            sveVectorLengthsProvided = true;
        }
    }

    if (sveVectorLengthsProvided) {
        if (sveFeature) {
            if (sveFeature->policy == VIR_CPU_FEATURE_DISABLE ||
                sveFeature->policy == VIR_CPU_FEATURE_FORBID) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("SVE disabled, but SVE vector lengths provided"));
                return -1;
            } else {
                sveFeature->policy = VIR_CPU_FEATURE_REQUIRE;
            }
        } else {
            if (VIR_RESIZE_N(def->cpu->features, def->cpu->nfeatures_max,
                             def->cpu->nfeatures, 1) < 0) {
                return -1;
            }

            def->cpu->features[def->cpu->nfeatures].name = g_strdup("sve");
            def->cpu->features[def->cpu->nfeatures].policy = VIR_CPU_FEATURE_REQUIRE;

            def->cpu->nfeatures++;
        }
    }

4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584
    /* Nothing to be done if only CPU topology is specified. */
    if (def->cpu->mode == VIR_CPU_MODE_CUSTOM &&
        !def->cpu->model)
        return 0;

    if (def->cpu->check != VIR_CPU_CHECK_DEFAULT)
        return 0;

    switch ((virCPUMode) def->cpu->mode) {
    case VIR_CPU_MODE_HOST_PASSTHROUGH:
        def->cpu->check = VIR_CPU_CHECK_NONE;
        break;

    case VIR_CPU_MODE_HOST_MODEL:
        def->cpu->check = VIR_CPU_CHECK_PARTIAL;
        break;

    case VIR_CPU_MODE_CUSTOM:
        /* Custom CPUs in TCG mode are not compared to host CPU by default. */
        if (def->virtType == VIR_DOMAIN_VIRT_QEMU)
            def->cpu->check = VIR_CPU_CHECK_NONE;
        else
            def->cpu->check = VIR_CPU_CHECK_PARTIAL;
        break;

    case VIR_CPU_MODE_LAST:
        break;
    }

    return 0;
}


4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616
static int
qemuDomainDefTsegPostParse(virDomainDefPtr def,
                           virQEMUCapsPtr qemuCaps)
{
    if (def->features[VIR_DOMAIN_FEATURE_SMM] != VIR_TRISTATE_SWITCH_ON)
        return 0;

    if (!def->tseg_specified)
        return 0;

    if (!qemuDomainIsQ35(def)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("SMM TSEG is only supported with q35 machine type"));
        return -1;
    }

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MCH_EXTENDED_TSEG_MBYTES)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Setting TSEG size is not supported with this QEMU binary"));
        return -1;
    }

    if (def->tseg_size & ((1 << 20) - 1)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("SMM TSEG size must be divisible by 1 MiB"));
        return -1;
    }

    return 0;
}


4617 4618 4619
static int
qemuDomainDefPostParseBasic(virDomainDefPtr def,
                            virCapsPtr caps,
J
Ján Tomko 已提交
4620
                            void *opaque G_GNUC_UNUSED)
4621 4622 4623 4624
{
    /* check for emulator and create a default one if needed */
    if (!def->emulator &&
        !(def->emulator = virDomainDefGetDefaultEmulator(def, caps)))
4625
        return 1;
4626 4627 4628 4629 4630

    return 0;
}


4631 4632
static int
qemuDomainDefPostParse(virDomainDefPtr def,
J
Ján Tomko 已提交
4633
                       virCapsPtr caps G_GNUC_UNUSED,
4634
                       unsigned int parseFlags,
4635
                       void *opaque,
4636
                       void *parseOpaque)
4637 4638
{
    virQEMUDriverPtr driver = opaque;
4639
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4640 4641 4642
    /* Note that qemuCaps may be NULL when this function is called. This
     * function shall not fail in that case. It will be re-run on VM startup
     * with the capabilities populated. */
4643
    virQEMUCapsPtr qemuCaps = parseOpaque;
4644 4645 4646 4647 4648
    int ret = -1;

    if (def->os.bootloader || def->os.bootloaderArgs) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bootloader is not supported by QEMU"));
4649
        goto cleanup;
4650 4651
    }

4652 4653 4654
    if (!def->os.machine) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing machine type"));
4655
        goto cleanup;
4656 4657
    }

4658
    qemuDomainNVRAMPathGenerate(cfg, def);
4659

4660 4661 4662
    if (qemuDomainDefAddDefaultDevices(def, qemuCaps) < 0)
        goto cleanup;

4663 4664 4665
    if (qemuCanonicalizeMachine(def, qemuCaps) < 0)
        goto cleanup;

4666
    qemuDomainDefEnableDefaultFeatures(def, qemuCaps);
A
Andrea Bolognani 已提交
4667

4668 4669
    if (qemuDomainRecheckInternalPaths(def, cfg, parseFlags) < 0)
        goto cleanup;
4670

4671
    if (qemuSecurityVerify(driver->securityManager, def) < 0)
4672 4673
        goto cleanup;

4674 4675 4676
    if (qemuDomainDefVcpusPostParse(def) < 0)
        goto cleanup;

4677 4678 4679
    if (qemuDomainDefCPUPostParse(def) < 0)
        goto cleanup;

4680 4681 4682
    if (qemuDomainDefTsegPostParse(def, qemuCaps) < 0)
        goto cleanup;

4683 4684
    ret = 0;
 cleanup:
4685
    virObjectUnref(cfg);
4686
    return ret;
4687 4688
}

4689

4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719
/**
 * qemuDomainDefGetVcpuHotplugGranularity:
 * @def: domain definition
 *
 * With QEMU 2.7 and newer, vCPUs can only be hotplugged in groups that
 * respect the guest's hotplug granularity; because of that, QEMU will
 * not allow guests to start unless the initial number of vCPUs is a
 * multiple of the hotplug granularity.
 *
 * Returns the vCPU hotplug granularity.
 */
static unsigned int
qemuDomainDefGetVcpuHotplugGranularity(const virDomainDef *def)
{
    /* If the guest CPU topology has not been configured, assume we
     * can hotplug vCPUs one at a time */
    if (!def->cpu || def->cpu->sockets == 0)
        return 1;

    /* For pSeries guests, hotplug can only be performed one core
     * at a time, so the vCPU hotplug granularity is the number
     * of threads per core */
    if (qemuDomainIsPSeries(def))
        return def->cpu->threads;

    /* In all other cases, we can hotplug vCPUs one at a time */
    return 1;
}


4720 4721 4722
#define QEMU_MAX_VCPUS_WITHOUT_EIM 255


4723
static int
4724 4725
qemuDomainDefValidateFeatures(const virDomainDef *def,
                              virQEMUCapsPtr qemuCaps)
4726
{
4727
    size_t i;
4728

4729 4730 4731 4732 4733
    for (i = 0; i < VIR_DOMAIN_FEATURE_LAST; i++) {
        const char *featureName = virDomainFeatureTypeToString(i);

        switch ((virDomainFeature) i) {
        case VIR_DOMAIN_FEATURE_IOAPIC:
4734
            if (def->features[i] != VIR_DOMAIN_IOAPIC_NONE &&
4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746
                !ARCH_IS_X86(def->os.arch)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("The '%s' feature is not supported for "
                                 "architecture '%s' or machine type '%s'"),
                               featureName,
                               virArchToString(def->os.arch),
                               def->os.machine);
                return -1;
            }
            break;

        case VIR_DOMAIN_FEATURE_HPT:
4747
        case VIR_DOMAIN_FEATURE_HTM:
4748
        case VIR_DOMAIN_FEATURE_NESTED_HV:
4749
        case VIR_DOMAIN_FEATURE_CCF_ASSIST:
4750
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761
                !qemuDomainIsPSeries(def)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("The '%s' feature is not supported for "
                                 "architecture '%s' or machine type '%s'"),
                               featureName,
                               virArchToString(def->os.arch),
                               def->os.machine);
                return -1;
            }
            break;

4762 4763
        case VIR_DOMAIN_FEATURE_GIC:
            if (def->features[i] == VIR_TRISTATE_SWITCH_ON &&
4764
                !qemuDomainIsARMVirt(def)) {
4765 4766 4767 4768 4769 4770 4771 4772 4773 4774
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("The '%s' feature is not supported for "
                                 "architecture '%s' or machine type '%s'"),
                               featureName,
                               virArchToString(def->os.arch),
                               def->os.machine);
                return -1;
            }
            break;

4775 4776
        case VIR_DOMAIN_FEATURE_SMM:
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
4777
                !virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT)) {
4778 4779 4780 4781 4782 4783
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("smm is not available with this QEMU binary"));
                return -1;
            }
            break;

4784 4785 4786 4787 4788 4789 4790 4791 4792 4793
        case VIR_DOMAIN_FEATURE_KVM:
            if (def->kvm_features[VIR_DOMAIN_KVM_DEDICATED] == VIR_TRISTATE_SWITCH_ON &&
                (!def->cpu || def->cpu->mode != VIR_CPU_MODE_HOST_PASSTHROUGH)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("kvm-hint-dedicated=on is only applicable "
                                 "for cpu host-passthrough"));
                return -1;
            }
            break;

4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805
        case VIR_DOMAIN_FEATURE_ACPI:
        case VIR_DOMAIN_FEATURE_APIC:
        case VIR_DOMAIN_FEATURE_PAE:
        case VIR_DOMAIN_FEATURE_HAP:
        case VIR_DOMAIN_FEATURE_VIRIDIAN:
        case VIR_DOMAIN_FEATURE_PRIVNET:
        case VIR_DOMAIN_FEATURE_HYPERV:
        case VIR_DOMAIN_FEATURE_PVSPINLOCK:
        case VIR_DOMAIN_FEATURE_CAPABILITIES:
        case VIR_DOMAIN_FEATURE_PMU:
        case VIR_DOMAIN_FEATURE_VMPORT:
        case VIR_DOMAIN_FEATURE_VMCOREINFO:
4806
        case VIR_DOMAIN_FEATURE_MSRS:
4807 4808 4809
        case VIR_DOMAIN_FEATURE_LAST:
            break;
        }
4810 4811 4812 4813 4814 4815
    }

    return 0;
}


4816
static int
M
Marc-André Lureau 已提交
4817 4818
qemuDomainDefValidateMemory(const virDomainDef *def,
                            virQEMUCapsPtr qemuCaps)
4819 4820
{
    const long system_page_size = virGetSystemPageSizeKB();
4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838
    const virDomainMemtune *mem = &def->mem;

    if (mem->nhugepages == 0)
        return 0;

    if (mem->allocation == VIR_DOMAIN_MEMORY_ALLOCATION_ONDEMAND) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("hugepages are not allowed with memory "
                         "allocation ondemand"));
        return -1;
    }

    if (mem->source == VIR_DOMAIN_MEMORY_SOURCE_ANONYMOUS) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("hugepages are not allowed with anonymous "
                         "memory source"));
        return -1;
    }
4839

M
Marc-André Lureau 已提交
4840 4841 4842
    if (mem->source == VIR_DOMAIN_MEMORY_SOURCE_MEMFD &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4843
                       _("hugepages is not supported with memfd memory source"));
M
Marc-André Lureau 已提交
4844 4845 4846
        return -1;
    }

4847 4848
    /* We can't guarantee any other mem.access
     * if no guest NUMA nodes are defined. */
4849
    if (mem->hugepages[0].size != system_page_size &&
4850
        virDomainNumaGetNodeCount(def->numa) == 0 &&
4851 4852
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_DEFAULT &&
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_PRIVATE) {
4853 4854 4855
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("memory access mode '%s' not supported "
                         "without guest numa node"),
4856
                       virDomainMemoryAccessTypeToString(mem->access));
4857 4858 4859 4860 4861 4862 4863
        return -1;
    }

    return 0;
}


4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886
static int
qemuDomainValidateCpuCount(const virDomainDef *def,
                            virQEMUCapsPtr qemuCaps)
{
    unsigned int maxCpus = virQEMUCapsGetMachineMaxCpus(qemuCaps, def->os.machine);

    if (virDomainDefGetVcpus(def) == 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Domain requires at least 1 vCPU"));
        return -1;
    }

    if (maxCpus > 0 && virDomainDefGetVcpusMax(def) > maxCpus) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Maximum CPUs greater than specified machine "
                         "type limit %u"), maxCpus);
        return -1;
    }

    return 0;
}


4887 4888
static int
qemuDomainDefValidate(const virDomainDef *def,
J
Ján Tomko 已提交
4889
                      virCapsPtr caps G_GNUC_UNUSED,
4890
                      void *opaque)
4891
{
4892
    virQEMUDriverPtr driver = opaque;
4893
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4894 4895
    virQEMUCapsPtr qemuCaps = NULL;
    int ret = -1;
4896
    size_t i;
4897

4898
    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
4899 4900 4901
                                            def->emulator)))
        goto cleanup;

4902 4903 4904
    if (def->mem.min_guarantee) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Parameter 'min_guarantee' not supported by QEMU."));
4905
        goto cleanup;
4906 4907 4908
    }

    /* On x86, UEFI requires ACPI */
4909 4910 4911
    if ((def->os.firmware == VIR_DOMAIN_OS_DEF_FIRMWARE_EFI ||
         (def->os.loader &&
          def->os.loader->type == VIR_DOMAIN_LOADER_TYPE_PFLASH)) &&
4912 4913 4914 4915 4916 4917 4918 4919 4920 4921
        ARCH_IS_X86(def->os.arch) &&
        def->features[VIR_DOMAIN_FEATURE_ACPI] != VIR_TRISTATE_SWITCH_ON) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("UEFI requires ACPI on this architecture"));
        goto cleanup;
    }

    /* On aarch64, ACPI requires UEFI */
    if (def->features[VIR_DOMAIN_FEATURE_ACPI] == VIR_TRISTATE_SWITCH_ON &&
        def->os.arch == VIR_ARCH_AARCH64 &&
4922 4923 4924
        (def->os.firmware != VIR_DOMAIN_OS_DEF_FIRMWARE_EFI &&
         (!def->os.loader ||
          def->os.loader->type != VIR_DOMAIN_LOADER_TYPE_PFLASH))) {
4925 4926 4927
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("ACPI requires UEFI on this architecture"));
        goto cleanup;
4928 4929
    }

M
Michal Privoznik 已提交
4930 4931 4932 4933 4934
    if (def->os.loader &&
        def->os.loader->secure == VIR_TRISTATE_BOOL_YES) {
        /* These are the QEMU implementation limitations. But we
         * have to live with them for now. */

4935
        if (!qemuDomainIsQ35(def)) {
M
Michal Privoznik 已提交
4936 4937
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot is supported with q35 machine types only"));
4938
            goto cleanup;
M
Michal Privoznik 已提交
4939 4940 4941 4942 4943 4944 4945 4946
        }

        /* Now, technically it is possible to have secure boot on
         * 32bits too, but that would require some -cpu xxx magic
         * too. Not worth it unless we are explicitly asked. */
        if (def->os.arch != VIR_ARCH_X86_64) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot is supported for x86_64 architecture only"));
4947
            goto cleanup;
M
Michal Privoznik 已提交
4948 4949
        }

4950 4951 4952
        /* SMM will be enabled by qemuFirmwareFillDomain() if needed. */
        if (def->os.firmware == VIR_DOMAIN_OS_DEF_FIRMWARE_NONE &&
            def->features[VIR_DOMAIN_FEATURE_SMM] != VIR_TRISTATE_SWITCH_ON) {
M
Michal Privoznik 已提交
4953 4954
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot requires SMM feature enabled"));
4955
            goto cleanup;
M
Michal Privoznik 已提交
4956 4957 4958
        }
    }

4959 4960 4961 4962 4963
    /* QEMU 2.7 (detected via the availability of query-hotpluggable-cpus)
     * enforces stricter rules than previous versions when it comes to guest
     * CPU topology. Verify known constraints are respected */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS)) {
        unsigned int topologycpus;
4964
        unsigned int granularity;
4965
        unsigned int numacpus;
4966 4967 4968 4969

        /* Starting from QEMU 2.5, max vCPU count and overall vCPU topology
         * must agree. We only actually enforce this with QEMU 2.7+, due
         * to the capability check above */
4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982
        if (virDomainDefGetVcpusTopology(def, &topologycpus) == 0) {
            if (topologycpus != virDomainDefGetVcpusMax(def)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("CPU topology doesn't match maximum vcpu count"));
                goto cleanup;
            }

            numacpus = virDomainNumaGetCPUCountTotal(def->numa);
            if ((numacpus != 0) && (topologycpus != numacpus)) {
                VIR_WARN("CPU topology doesn't match numa CPU count; "
                         "partial NUMA mapping is obsoleted and will "
                         "be removed in future");
            }
4983
        }
4984 4985 4986 4987 4988 4989 4990 4991 4992 4993

        /* vCPU hotplug granularity must be respected */
        granularity = qemuDomainDefGetVcpuHotplugGranularity(def);
        if ((virDomainDefGetVcpus(def) % granularity) != 0) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("vCPUs count must be a multiple of the vCPU "
                             "hotplug granularity (%u)"),
                           granularity);
            goto cleanup;
        }
4994 4995
    }

4996
    if (qemuDomainValidateCpuCount(def, qemuCaps) < 0)
4997 4998
        goto cleanup;

4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016
    if (ARCH_IS_X86(def->os.arch) &&
        virDomainDefGetVcpusMax(def) > QEMU_MAX_VCPUS_WITHOUT_EIM) {
        if (!qemuDomainIsQ35(def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("more than %d vCPUs are only supported on "
                             "q35-based machine types"),
                           QEMU_MAX_VCPUS_WITHOUT_EIM);
            goto cleanup;
        }
        if (!def->iommu || def->iommu->eim != VIR_TRISTATE_SWITCH_ON) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("more than %d vCPUs require extended interrupt "
                             "mode enabled on the iommu device"),
                           QEMU_MAX_VCPUS_WITHOUT_EIM);
            goto cleanup;
        }
    }

B
Bing Niu 已提交
5017
    if (def->nresctrls &&
5018 5019 5020 5021 5022 5023
        def->virtType != VIR_DOMAIN_VIRT_KVM) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("cachetune is only supported for KVM domains"));
        goto cleanup;
    }

5024
    if (qemuDomainDefValidateFeatures(def, qemuCaps) < 0)
5025 5026
        goto cleanup;

M
Marc-André Lureau 已提交
5027
    if (qemuDomainDefValidateMemory(def, qemuCaps) < 0)
5028 5029
        goto cleanup;

5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041
    if (cfg->vncTLS && cfg->vncTLSx509secretUUID &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_TLS_CREDS_X509)) {
        for (i = 0; i < def->ngraphics; i++) {
            if (def->graphics[i]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("encrypted VNC TLS keys are not supported with "
                                 "this QEMU binary"));
                goto cleanup;
            }
        }
    }

5042 5043 5044 5045
    ret = 0;

 cleanup:
    virObjectUnref(qemuCaps);
5046
    virObjectUnref(cfg);
5047
    return ret;
5048 5049 5050
}


5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074
static bool
qemuDomainNetSupportsCoalesce(virDomainNetType type)
{
    switch (type) {
    case VIR_DOMAIN_NET_TYPE_NETWORK:
    case VIR_DOMAIN_NET_TYPE_BRIDGE:
        return true;
    case VIR_DOMAIN_NET_TYPE_VHOSTUSER:
    case VIR_DOMAIN_NET_TYPE_ETHERNET:
    case VIR_DOMAIN_NET_TYPE_DIRECT:
    case VIR_DOMAIN_NET_TYPE_HOSTDEV:
    case VIR_DOMAIN_NET_TYPE_USER:
    case VIR_DOMAIN_NET_TYPE_SERVER:
    case VIR_DOMAIN_NET_TYPE_CLIENT:
    case VIR_DOMAIN_NET_TYPE_MCAST:
    case VIR_DOMAIN_NET_TYPE_INTERNAL:
    case VIR_DOMAIN_NET_TYPE_UDP:
    case VIR_DOMAIN_NET_TYPE_LAST:
        break;
    }
    return false;
}


5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121
static int
qemuDomainChrSourceReconnectDefValidate(const virDomainChrSourceReconnectDef *def)
{
    if (def->enabled == VIR_TRISTATE_BOOL_YES &&
        def->timeout == 0) {
        virReportError(VIR_ERR_INVALID_ARG, "%s",
                       _("chardev reconnect source timeout cannot be '0'"));
        return -1;
    }

    return 0;
}


static int
qemuDomainChrSourceDefValidate(const virDomainChrSourceDef *def)
{
    switch ((virDomainChrType)def->type) {
    case VIR_DOMAIN_CHR_TYPE_TCP:
        if (qemuDomainChrSourceReconnectDefValidate(&def->data.tcp.reconnect) < 0)
            return -1;
        break;

    case VIR_DOMAIN_CHR_TYPE_UNIX:
        if (qemuDomainChrSourceReconnectDefValidate(&def->data.nix.reconnect) < 0)
            return -1;
        break;

    case VIR_DOMAIN_CHR_TYPE_NULL:
    case VIR_DOMAIN_CHR_TYPE_VC:
    case VIR_DOMAIN_CHR_TYPE_PTY:
    case VIR_DOMAIN_CHR_TYPE_DEV:
    case VIR_DOMAIN_CHR_TYPE_FILE:
    case VIR_DOMAIN_CHR_TYPE_PIPE:
    case VIR_DOMAIN_CHR_TYPE_STDIO:
    case VIR_DOMAIN_CHR_TYPE_UDP:
    case VIR_DOMAIN_CHR_TYPE_SPICEVMC:
    case VIR_DOMAIN_CHR_TYPE_SPICEPORT:
    case VIR_DOMAIN_CHR_TYPE_NMDM:
    case VIR_DOMAIN_CHR_TYPE_LAST:
        break;
    }

    return 0;
}


5122 5123 5124
static int
qemuDomainChrSerialTargetTypeToAddressType(int targetType)
{
5125
    switch ((virDomainChrSerialTargetType)targetType) {
5126 5127 5128 5129 5130 5131
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA:
        return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_ISA;
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB:
        return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_USB;
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI:
        return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
5132 5133
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
        return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO;
5134
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
5135
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
5136 5137 5138 5139 5140 5141 5142 5143 5144
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        break;
    }

    return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE;
}


5145 5146 5147 5148 5149 5150 5151 5152 5153 5154
static int
qemuDomainChrSerialTargetModelToTargetType(int targetModel)
{
    switch ((virDomainChrSerialTargetModel) targetModel) {
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_ISA_SERIAL:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA;
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_USB_SERIAL:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB;
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PCI_SERIAL:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI;
5155 5156
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
5157
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
5158
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
5159
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
5160 5161 5162
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
5163 5164 5165 5166 5167 5168 5169 5170 5171
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_NONE:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_LAST:
        break;
    }

    return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE;
}


5172
static int
5173
qemuDomainChrTargetDefValidate(const virDomainChrDef *chr)
5174
{
5175 5176
    int expected;

5177
    switch ((virDomainChrDeviceType)chr->deviceType) {
5178 5179 5180
    case VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL:

        /* Validate target type */
5181
        switch ((virDomainChrSerialTargetType)chr->targetType) {
5182
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA:
5183 5184
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI:
5185
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
5186

5187
            expected = qemuDomainChrSerialTargetTypeToAddressType(chr->targetType);
5188 5189

            if (chr->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
5190 5191 5192 5193 5194
                chr->info.type != expected) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Target type '%s' requires address type '%s'"),
                               virDomainChrSerialTargetTypeToString(chr->targetType),
                               virDomainDeviceAddressTypeToString(expected));
5195 5196 5197 5198
                return -1;
            }
            break;

5199
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
5200
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
5201 5202 5203 5204 5205 5206 5207 5208 5209
            if (chr->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Target type '%s' cannot have an "
                                 "associated address"),
                               virDomainChrSerialTargetTypeToString(chr->targetType));
                return -1;
            }
            break;

5210 5211 5212 5213
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            break;
        }
5214 5215 5216 5217 5218 5219

        /* Validate target model */
        switch ((virDomainChrSerialTargetModel) chr->targetModel) {
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_ISA_SERIAL:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_USB_SERIAL:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PCI_SERIAL:
5220
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
5221
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
5222 5223
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
5224
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240

            expected = qemuDomainChrSerialTargetModelToTargetType(chr->targetModel);

            if (chr->targetType != expected) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Target model '%s' requires target type '%s'"),
                               virDomainChrSerialTargetModelTypeToString(chr->targetModel),
                               virDomainChrSerialTargetTypeToString(expected));
                return -1;
            }
            break;

        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_LAST:
            break;
        }
5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254
        break;

    case VIR_DOMAIN_CHR_DEVICE_TYPE_CONSOLE:
    case VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL:
    case VIR_DOMAIN_CHR_DEVICE_TYPE_CHANNEL:
    case VIR_DOMAIN_CHR_DEVICE_TYPE_LAST:
        /* Nothing to do */
        break;
    }

    return 0;
}


5255
static int
5256
qemuDomainChrDefValidate(const virDomainChrDef *dev,
5257
                         const virDomainDef *def)
5258
{
5259
    if (qemuDomainChrSourceDefValidate(dev->source) < 0)
5260 5261
        return -1;

5262
    if (qemuDomainChrTargetDefValidate(dev) < 0)
5263 5264
        return -1;

5265
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL &&
5266
        (ARCH_IS_S390(def->os.arch) || qemuDomainIsPSeries(def))) {
5267 5268 5269 5270 5271
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("parallel ports are not supported"));
            return -1;
    }

5272 5273 5274
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL) {
        bool isCompatible = true;

5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285
        if (dev->targetType == VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM) {
            if (dev->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011 &&
                !qemuDomainIsARMVirt(def)) {
                isCompatible = false;
            }
            if (dev->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A &&
                !qemuDomainIsRISCVVirt(def)) {
                isCompatible = false;
            }
        }

5286 5287 5288 5289 5290 5291
        if (!qemuDomainIsPSeries(def) &&
            (dev->targetType == VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO ||
             dev->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY)) {
            isCompatible = false;
        }

5292 5293 5294 5295 5296 5297 5298
        if (!ARCH_IS_S390(def->os.arch) &&
            (dev->targetType == VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP ||
             dev->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE ||
             dev->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE)) {
            isCompatible = false;
        }

5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309
        if (!isCompatible) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Serial device with target type '%s' and "
                             "target model '%s' not compatible with guest "
                             "architecture or machine type"),
                           virDomainChrSerialTargetTypeToString(dev->targetType),
                           virDomainChrSerialTargetModelTypeToString(dev->targetModel));
            return -1;
        }
    }

5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325
    return 0;
}


static int
qemuDomainSmartcardDefValidate(const virDomainSmartcardDef *def)
{
    if (def->type == VIR_DOMAIN_SMARTCARD_TYPE_PASSTHROUGH &&
        qemuDomainChrSourceDefValidate(def->data.passthru) < 0)
        return -1;

    return 0;
}


static int
5326
qemuDomainRNGDefValidate(const virDomainRNGDef *def,
J
Ján Tomko 已提交
5327
                         virQEMUCapsPtr qemuCaps G_GNUC_UNUSED)
5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346
{
    if (def->backend == VIR_DOMAIN_RNG_BACKEND_EGD &&
        qemuDomainChrSourceDefValidate(def->source.chardev) < 0)
        return -1;

    return 0;
}


static int
qemuDomainRedirdevDefValidate(const virDomainRedirdevDef *def)
{
    if (qemuDomainChrSourceDefValidate(def->source) < 0)
        return -1;

    return 0;
}


5347 5348 5349 5350 5351 5352 5353 5354 5355
static int
qemuDomainWatchdogDefValidate(const virDomainWatchdogDef *dev,
                              const virDomainDef *def)
{
    switch ((virDomainWatchdogModel) dev->model) {
    case VIR_DOMAIN_WATCHDOG_MODEL_I6300ESB:
        if (dev->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
            dev->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
Y
Yuri Chornoivan 已提交
5356
                           _("%s model of watchdog can go only on PCI bus"),
5357 5358 5359 5360 5361 5362 5363 5364 5365
                           virDomainWatchdogModelTypeToString(dev->model));
            return -1;
        }
        break;

    case VIR_DOMAIN_WATCHDOG_MODEL_IB700:
        if (dev->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
            dev->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_ISA) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
Y
Yuri Chornoivan 已提交
5366
                           _("%s model of watchdog can go only on ISA bus"),
5367 5368 5369 5370 5371 5372 5373 5374
                           virDomainWatchdogModelTypeToString(dev->model));
            return -1;
        }
        break;

    case VIR_DOMAIN_WATCHDOG_MODEL_DIAG288:
        if (dev->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
Y
Yuri Chornoivan 已提交
5375
                           _("%s model of watchdog is virtual and cannot go on any bus."),
5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394
                           virDomainWatchdogModelTypeToString(dev->model));
            return -1;
        }
        if (!(ARCH_IS_S390(def->os.arch))) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("%s model of watchdog is allowed for s390 and s390x only"),
                           virDomainWatchdogModelTypeToString(dev->model));
            return -1;
        }
        break;

    case VIR_DOMAIN_WATCHDOG_MODEL_LAST:
        break;
    }

    return 0;
}


5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474
int
qemuDomainValidateActualNetDef(const virDomainNetDef *net,
                               virQEMUCapsPtr qemuCaps)
{
    /*
     * Validations that can only be properly checked at runtime (after
     * an <interface type='network'> has been resolved to its actual
     * type.
     *
     * (In its current form this function can still be called before
     * the actual type has been resolved (e.g. at domain definition
     * time), but only if the validations would SUCCEED for
     * type='network'.)
     */
    virDomainNetType actualType = virDomainNetGetActualType(net);

    /* Only tap/macvtap devices support multiqueue. */
    if (net->driver.virtio.queues > 0) {

        if (!(actualType == VIR_DOMAIN_NET_TYPE_NETWORK ||
              actualType == VIR_DOMAIN_NET_TYPE_BRIDGE ||
              actualType == VIR_DOMAIN_NET_TYPE_DIRECT ||
              actualType == VIR_DOMAIN_NET_TYPE_ETHERNET ||
              actualType == VIR_DOMAIN_NET_TYPE_VHOSTUSER)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("multiqueue network is not supported for: %s"),
                           virDomainNetTypeToString(actualType));
            return -1;
        }

        if (net->driver.virtio.queues > 1 &&
            actualType == VIR_DOMAIN_NET_TYPE_VHOSTUSER &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_VHOSTUSER_MULTIQUEUE)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("multiqueue network is not supported for vhost-user "
                             "with this QEMU binary"));
            return -1;
        }
    }

    /*
     * Only standard tap devices support nwfilter rules, and even then only
     * when *not* connected to an OVS bridge or midonet (indicated by having
     * a <virtualport> element in the config)
     */
    if (net->filter) {
        virNetDevVPortProfilePtr vport = virDomainNetGetActualVirtPortProfile(net);
        if (!(actualType == VIR_DOMAIN_NET_TYPE_NETWORK ||
              actualType == VIR_DOMAIN_NET_TYPE_BRIDGE ||
              actualType == VIR_DOMAIN_NET_TYPE_ETHERNET)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("filterref is not supported for "
                             "network interfaces of type %s"),
                           virDomainNetTypeToString(actualType));
            return -1;
        }
        if (vport && vport->virtPortType != VIR_NETDEV_VPORT_PROFILE_NONE) {
            /* currently none of the defined virtualport types support iptables */
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("filterref is not supported for "
                             "network interfaces with virtualport type %s"),
                           virNetDevVPortTypeToString(vport->virtPortType));
            return -1;
        }
    }

    if (net->backend.tap &&
        !(actualType == VIR_DOMAIN_NET_TYPE_NETWORK ||
          actualType == VIR_DOMAIN_NET_TYPE_BRIDGE ||
          actualType == VIR_DOMAIN_NET_TYPE_ETHERNET)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Custom tap device path is not supported for: %s"),
                       virDomainNetTypeToString(actualType));
        return -1;
    }

    return 0;
}


5475
static int
5476
qemuDomainDeviceDefValidateNetwork(const virDomainNetDef *net)
5477
{
5478 5479
    bool hasIPv4 = false;
    bool hasIPv6 = false;
5480
    size_t i;
5481

5482 5483 5484 5485 5486 5487 5488 5489 5490 5491
    if (net->type == VIR_DOMAIN_NET_TYPE_USER) {
        if (net->guestIP.nroutes) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Invalid attempt to set network interface "
                             "guest-side IP route, not supported by QEMU"));
            return -1;
        }

        for (i = 0; i < net->guestIP.nips; i++) {
            const virNetDevIPAddr *ip = net->guestIP.ips[i];
5492

5493
            if (VIR_SOCKET_ADDR_VALID(&net->guestIP.ips[i]->peer)) {
5494
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
5495 5496
                               _("Invalid attempt to set peer IP for guest"));
                return -1;
5497
            }
5498

5499 5500
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ip->address, AF_INET)) {
                if (hasIPv4) {
5501
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
5502 5503 5504
                                   _("Only one IPv4 address per "
                                     "interface is allowed"));
                    return -1;
5505
                }
5506
                hasIPv4 = true;
5507

5508 5509
                if (ip->prefix > 0 &&
                    (ip->prefix < 4 || ip->prefix > 27)) {
5510
                    virReportError(VIR_ERR_XML_ERROR, "%s",
5511
                                   _("invalid prefix, must be in range of 4-27"));
5512
                    return -1;
5513
                }
5514
            }
5515

5516 5517 5518 5519 5520 5521
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ip->address, AF_INET6)) {
                if (hasIPv6) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                                   _("Only one IPv6 address per "
                                     "interface is allowed"));
                    return -1;
5522
                }
5523
                hasIPv6 = true;
5524

5525 5526 5527 5528 5529
                if (ip->prefix > 120) {
                    virReportError(VIR_ERR_XML_ERROR, "%s",
                                   _("prefix too long"));
                    return -1;
                }
5530
            }
5531
        }
5532 5533 5534 5535 5536 5537 5538
    } else if (net->guestIP.nroutes || net->guestIP.nips) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Invalid attempt to set network interface "
                         "guest-side IP route and/or address info, "
                         "not supported by QEMU"));
        return -1;
    }
5539

5540
    if (virDomainNetIsVirtioModel(net)) {
5541 5542 5543 5544 5545 5546 5547 5548 5549
        if (net->driver.virtio.rx_queue_size & (net->driver.virtio.rx_queue_size - 1)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("rx_queue_size has to be a power of two"));
            return -1;
        }
        if (net->driver.virtio.tx_queue_size & (net->driver.virtio.tx_queue_size - 1)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("tx_queue_size has to be a power of two"));
            return -1;
5550
        }
5551
    }
5552

5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571
    if (net->mtu &&
        !qemuDomainNetSupportsMTU(net->type)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("setting MTU on interface type %s is not supported yet"),
                       virDomainNetTypeToString(net->type));
        return -1;
    }

    if (net->coalesce && !qemuDomainNetSupportsCoalesce(net->type)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("coalesce settings on interface type %s are not supported"),
                       virDomainNetTypeToString(net->type));
        return -1;
    }

    return 0;
}


5572
static int
5573
qemuDomainMdevDefVFIOPCIValidate(const virDomainHostdevDef *hostdev,
5574 5575
                                 const virDomainDef *def,
                                 virQEMUCapsPtr qemuCaps)
5576
{
5577 5578
    const virDomainHostdevSubsysMediatedDev *dev;

5579 5580 5581 5582 5583 5584 5585 5586 5587
    /* VFIO-PCI does not support boot */
    if (hostdev->info->bootIndex) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("booting from assigned devices is not "
                         "supported by mediated devices of "
                         "model vfio-pci"));
        return -1;
    }

5588
    dev = &hostdev->source.subsys.u.mdev;
5589
    if (dev->display == VIR_TRISTATE_SWITCH_ABSENT)
5590 5591 5592 5593 5594 5595 5596 5597 5598
        return 0;

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VFIO_PCI_DISPLAY)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("display property of device vfio-pci is "
                         "not supported by this version of QEMU"));
        return -1;
    }

5599
    if (dev->model != VIR_MDEV_MODEL_TYPE_VFIO_PCI) {
5600 5601 5602 5603 5604 5605 5606
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("<hostdev> attribute 'display' is only supported"
                         " with model='vfio-pci'"));

        return -1;
    }

5607
    if (dev->display == VIR_TRISTATE_SWITCH_ON) {
5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619
        if (def->ngraphics == 0) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("graphics device is needed for attribute value "
                             "'display=on' in <hostdev>"));
            return -1;
        }
    }

    return 0;
}


5620
static int
5621
qemuDomainMdevDefVFIOAPValidate(const virDomainHostdevDef *hostdev,
5622
                                const virDomainDef *def)
5623 5624 5625 5626
{
    size_t i;
    bool vfioap_found = false;

5627 5628 5629 5630 5631 5632 5633 5634 5635
    /* VFIO-AP does not support boot */
    if (hostdev->info->bootIndex) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("booting from assigned devices is not "
                         "supported by mediated devices of "
                         "model vfio-ap"));
        return -1;
    }

5636 5637
    /* VFIO-AP is restricted to a single mediated device only */
    for (i = 0; i < def->nhostdevs; i++) {
5638
        virDomainHostdevDefPtr hdev = def->hostdevs[i];
5639

5640 5641
        if (virHostdevIsMdevDevice(hdev) &&
            hdev->source.subsys.u.mdev.model == VIR_MDEV_MODEL_TYPE_VFIO_AP) {
5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655
            if (vfioap_found) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("Only one hostdev of model vfio-ap is "
                                 "supported"));
                return -1;
            }
            vfioap_found = true;
        }
    }

    return 0;
}


5656
static int
5657
qemuDomainMdevDefValidate(const virDomainHostdevDef *hostdev,
5658 5659 5660
                          const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
{
5661 5662 5663
    const virDomainHostdevSubsysMediatedDev *mdevsrc;

    mdevsrc = &hostdev->source.subsys.u.mdev;
5664 5665
    switch ((virMediatedDeviceModelType) mdevsrc->model) {
    case VIR_MDEV_MODEL_TYPE_VFIO_PCI:
5666
        return qemuDomainMdevDefVFIOPCIValidate(hostdev, def, qemuCaps);
5667
    case VIR_MDEV_MODEL_TYPE_VFIO_AP:
5668
        return qemuDomainMdevDefVFIOAPValidate(hostdev, def);
5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681
    case VIR_MDEV_MODEL_TYPE_VFIO_CCW:
        break;
    case VIR_MDEV_MODEL_TYPE_LAST:
    default:
        virReportEnumRangeError(virMediatedDeviceModelType,
                                mdevsrc->model);
        return -1;
    }

    return 0;
}


5682 5683
static int
qemuDomainDeviceDefValidateHostdev(const virDomainHostdevDef *hostdev,
5684 5685
                                   const virDomainDef *def,
                                   virQEMUCapsPtr qemuCaps)
5686 5687 5688 5689 5690 5691 5692 5693 5694 5695
{
    /* forbid capabilities mode hostdev in this kind of hypervisor */
    if (hostdev->mode == VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("hostdev mode 'capabilities' is not "
                         "supported in %s"),
                       virDomainVirtTypeToString(def->virtType));
        return -1;
    }

5696 5697 5698 5699 5700
    if (hostdev->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
        switch ((virDomainHostdevSubsysType) hostdev->source.subsys.type) {
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI:
5701
            break;
5702
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI_HOST:
5703 5704
            if (hostdev->info->bootIndex) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
5705 5706
                               _("booting from assigned devices is not "
                                 "supported by vhost SCSI devices"));
5707 5708
                return -1;
            }
5709 5710
            break;
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
5711
            return qemuDomainMdevDefValidate(hostdev, def, qemuCaps);
5712 5713 5714 5715 5716 5717 5718 5719
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
        default:
            virReportEnumRangeError(virDomainHostdevSubsysType,
                                    hostdev->source.subsys.type);
            return -1;
        }
    }

5720 5721 5722 5723
    return 0;
}


5724 5725 5726
static int
qemuDomainDeviceDefValidateVideo(const virDomainVideoDef *video)
{
5727
    switch ((virDomainVideoType) video->type) {
5728 5729
    case VIR_DOMAIN_VIDEO_TYPE_NONE:
        return 0;
5730 5731 5732
    case VIR_DOMAIN_VIDEO_TYPE_XEN:
    case VIR_DOMAIN_VIDEO_TYPE_VBOX:
    case VIR_DOMAIN_VIDEO_TYPE_PARALLELS:
5733
    case VIR_DOMAIN_VIDEO_TYPE_GOP:
5734 5735 5736 5737 5738 5739 5740 5741 5742 5743
    case VIR_DOMAIN_VIDEO_TYPE_DEFAULT:
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("video type '%s' is not supported with QEMU"),
                       virDomainVideoTypeToString(video->type));
        return -1;
    case VIR_DOMAIN_VIDEO_TYPE_VGA:
    case VIR_DOMAIN_VIDEO_TYPE_CIRRUS:
    case VIR_DOMAIN_VIDEO_TYPE_VMVGA:
    case VIR_DOMAIN_VIDEO_TYPE_QXL:
    case VIR_DOMAIN_VIDEO_TYPE_VIRTIO:
5744
    case VIR_DOMAIN_VIDEO_TYPE_BOCHS:
5745
    case VIR_DOMAIN_VIDEO_TYPE_RAMFB:
5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776
    case VIR_DOMAIN_VIDEO_TYPE_LAST:
        break;
    }

    if (!video->primary &&
        video->type != VIR_DOMAIN_VIDEO_TYPE_QXL &&
        video->type != VIR_DOMAIN_VIDEO_TYPE_VIRTIO) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("video type '%s' is only valid as primary "
                         "video device"),
                       virDomainVideoTypeToString(video->type));
        return -1;
    }

    if (video->accel && video->accel->accel2d == VIR_TRISTATE_SWITCH_ON) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("qemu does not support the accel2d setting"));
        return -1;
    }

    if (video->type == VIR_DOMAIN_VIDEO_TYPE_QXL) {
        if (video->vram > (UINT_MAX / 1024)) {
            virReportError(VIR_ERR_OVERFLOW,
                           _("value for 'vram' must be less than '%u'"),
                           UINT_MAX / 1024);
            return -1;
        }
        if (video->ram > (UINT_MAX / 1024)) {
            virReportError(VIR_ERR_OVERFLOW,
                           _("value for 'ram' must be less than '%u'"),
                           UINT_MAX / 1024);
5777 5778
            return -1;
        }
5779 5780 5781 5782 5783 5784 5785
        if (video->vgamem) {
            if (video->vgamem < 1024) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("value for 'vgamem' must be at least 1 MiB "
                                 "(1024 KiB)"));
                return -1;
            }
5786

5787 5788 5789 5790 5791 5792 5793 5794
            if (video->vgamem != VIR_ROUND_UP_POWER_OF_TWO(video->vgamem)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("value for 'vgamem' must be power of two"));
                return -1;
            }
        }
    }

5795 5796 5797 5798 5799 5800 5801 5802 5803 5804 5805
    if (video->type != VIR_DOMAIN_VIDEO_TYPE_VGA &&
        video->type != VIR_DOMAIN_VIDEO_TYPE_QXL &&
        video->type != VIR_DOMAIN_VIDEO_TYPE_VIRTIO &&
        video->type != VIR_DOMAIN_VIDEO_TYPE_BOCHS) {
        if (video->res) {
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("model resolution is not supported"));
            return -1;
        }
    }

5806 5807 5808 5809 5810 5811
    if (video->type == VIR_DOMAIN_VIDEO_TYPE_VGA ||
        video->type == VIR_DOMAIN_VIDEO_TYPE_VMVGA) {
        if (video->vram && video->vram < 1024) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           "%s", _("value for 'vram' must be at least "
                                   "1 MiB (1024 KiB)"));
5812 5813 5814 5815 5816 5817 5818 5819
            return -1;
        }
    }

    return 0;
}


5820
int
5821 5822
qemuDomainValidateStorageSource(virStorageSourcePtr src,
                                virQEMUCapsPtr qemuCaps)
5823
{
5824 5825
    int actualType = virStorageSourceGetActualType(src);

5826 5827 5828 5829 5830 5831
    if (src->format == VIR_STORAGE_FILE_COW) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                      _("'cow' storage format is not supported"));
        return -1;
    }

5832 5833 5834 5835 5836 5837 5838
    if (src->format == VIR_STORAGE_FILE_DIR) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("'directory' storage format is not directly supported by QEMU, "
                         "use 'dir' disk type instead"));
        return -1;
    }

5839 5840 5841 5842 5843 5844 5845
    if (src->format == VIR_STORAGE_FILE_ISO) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("storage format 'iso' is not directly supported by QEMU, "
                         "use 'raw' instead"));
        return -1;
    }

5846 5847 5848 5849 5850 5851 5852 5853 5854 5855
    if ((src->format == VIR_STORAGE_FILE_QCOW ||
         src->format == VIR_STORAGE_FILE_QCOW2) &&
        src->encryption &&
        (src->encryption->format == VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT ||
         src->encryption->format == VIR_STORAGE_ENCRYPTION_FORMAT_QCOW)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("old qcow/qcow2 encryption is not supported"));
            return -1;
    }

5856 5857 5858 5859 5860
    if (src->format == VIR_STORAGE_FILE_QCOW2 &&
        src->encryption &&
        src->encryption->format == VIR_STORAGE_ENCRYPTION_FORMAT_LUKS &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_QCOW2_LUKS)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
5861
                       _("LUKS encrypted QCOW2 images are not supported by this QEMU"));
5862 5863 5864
        return -1;
    }

5865
    if (src->format == VIR_STORAGE_FILE_FAT &&
5866
        actualType != VIR_STORAGE_TYPE_VOLUME &&
5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888
        actualType != VIR_STORAGE_TYPE_DIR) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("storage format 'fat' is supported only with 'dir' "
                         "storage type"));
        return -1;
    }

    if (actualType == VIR_STORAGE_TYPE_DIR) {
        if (src->format > 0 &&
            src->format != VIR_STORAGE_FILE_FAT) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("storage type 'dir' requires use of storage format 'fat'"));
            return -1;
        }

        if (!src->readonly) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtual FAT storage can't be accessed in read-write mode"));
            return -1;
        }
    }

5889 5890 5891 5892 5893
    if (src->pr &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_PR_MANAGER_HELPER)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("reservations not supported with this QEMU binary"));
        return -1;
5894 5895
    }

5896 5897 5898 5899 5900 5901 5902 5903 5904
    /* Use QEMU_CAPS_ISCSI_PASSWORD_SECRET as witness that iscsi 'initiator-name'
     * option is available, it was introduced at the same time. */
    if (src->initiator.iqn &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("iSCSI initiator IQN not supported with this QEMU binary"));
        return -1;
    }

5905 5906 5907 5908
    return 0;
}


5909
int
5910 5911
qemuDomainDeviceDefValidateDisk(const virDomainDiskDef *disk,
                                virQEMUCapsPtr qemuCaps)
5912
{
5913
    const char *driverName = virDomainDiskGetDriver(disk);
5914
    virStorageSourcePtr n;
5915 5916
    int idx;
    int partition;
5917

5918 5919 5920 5921 5922 5923
    if (disk->src->shared && !disk->src->readonly &&
        !qemuBlockStorageSourceSupportsConcurrentAccess(disk->src)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("shared access for disk '%s' requires use of "
                         "supported storage format"), disk->dst);
        return -1;
5924 5925
    }

5926 5927 5928 5929 5930 5931 5932 5933 5934 5935 5936 5937 5938 5939 5940
    if (disk->copy_on_read == VIR_TRISTATE_SWITCH_ON) {
        if (disk->src->readonly) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("copy_on_read is not compatible with read-only disk '%s'"),
                           disk->dst);
            return -1;
        }

        if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM ||
            disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("copy_on_read is not supported with removable disk '%s'"),
                           disk->dst);
            return -1;
        }
5941 5942
    }

5943 5944 5945 5946 5947 5948 5949 5950 5951 5952 5953 5954 5955 5956 5957 5958 5959 5960 5961 5962 5963 5964 5965 5966 5967 5968 5969
    if (disk->geometry.cylinders > 0 &&
        disk->geometry.heads > 0 &&
        disk->geometry.sectors > 0) {
        if (disk->bus == VIR_DOMAIN_DISK_BUS_USB ||
            disk->bus == VIR_DOMAIN_DISK_BUS_SD) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("CHS geometry can not be set for '%s' bus"),
                           virDomainDiskBusTypeToString(disk->bus));
            return -1;
        }

        if (disk->geometry.trans != VIR_DOMAIN_DISK_TRANS_DEFAULT &&
            disk->bus != VIR_DOMAIN_DISK_BUS_IDE) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("CHS translation mode can only be set for 'ide' bus not '%s'"),
                           virDomainDiskBusTypeToString(disk->bus));
            return -1;
        }
    }

    if (disk->serial && disk->bus == VIR_DOMAIN_DISK_BUS_SD) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Serial property not supported for drive bus '%s'"),
                       virDomainDiskBusTypeToString(disk->bus));
        return -1;
    }

5970 5971 5972 5973 5974 5975 5976
    if (driverName && STRNEQ(driverName, "qemu")) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("unsupported driver name '%s' for disk '%s'"),
                       driverName, disk->dst);
        return -1;
    }

P
Peter Krempa 已提交
5977 5978 5979 5980 5981 5982 5983 5984
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
        disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("disk type 'virtio' of '%s' does not support ejectable media"),
                       disk->dst);
        return -1;
    }

5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997
    if (virDiskNameParse(disk->dst, &idx, &partition) < 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("invalid disk target '%s'"), disk->dst);
        return -1;
    }

    if (partition != 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("invalid disk target '%s', partitions can't appear in disk targets"),
                       disk->dst);
        return -1;
    }

5998 5999 6000 6001 6002
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if (qemuDomainValidateStorageSource(n, qemuCaps) < 0)
            return -1;
    }

6003 6004 6005 6006
    return 0;
}


6007 6008 6009 6010
static int
qemuDomainDeviceDefValidateControllerAttributes(const virDomainControllerDef *controller)
{
    if (!(controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
6011 6012 6013
          (controller->model == VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI ||
           controller->model == VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL ||
           controller->model == VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL))) {
6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027 6028 6029 6030 6031 6032 6033
        if (controller->queues) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'queues' is only supported by virtio-scsi controller"));
            return -1;
        }
        if (controller->cmd_per_lun) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'cmd_per_lun' is only supported by virtio-scsi controller"));
            return -1;
        }
        if (controller->max_sectors) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'max_sectors' is only supported by virtio-scsi controller"));
            return -1;
        }
        if (controller->ioeventfd) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'ioeventfd' is only supported by virtio-scsi controller"));
            return -1;
        }
6034 6035 6036 6037 6038
        if (controller->iothread) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'iothread' is only supported for virtio-scsi controller"));
            return -1;
        }
6039 6040 6041 6042 6043 6044
    }

    return 0;
}


6045 6046 6047 6048 6049 6050
/**
 * @qemuCaps: QEMU capabilities
 * @model: SCSI model to check
 *
 * Using the @qemuCaps, let's ensure the provided @model can be supported
 *
J
Ján Tomko 已提交
6051
 * Returns true if acceptable, false otherwise with error message set.
6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066
 */
static bool
qemuDomainCheckSCSIControllerModel(virQEMUCapsPtr qemuCaps,
                                   int model)
{
    switch ((virDomainControllerModelSCSI) model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSILOGIC:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_LSI)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support "
                             "the LSI 53C895A SCSI controller"));
            return false;
        }
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI:
6067 6068
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL:
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL:
6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089 6090 6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_SCSI)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support "
                             "virtio scsi controller"));
            return false;
        }
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_IBMVSCSI:
        /*TODO: need checking work here if necessary */
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSISAS1068:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_MPTSAS1068)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support "
                             "the LSI SAS1068 (MPT Fusion) controller"));
            return false;
        }
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSISAS1078:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_MEGASAS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support "
                             "the LSI SAS1078 (MegaRAID) controller"));
            return false;
        }
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_AUTO:
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_BUSLOGIC:
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VMPVSCSI:
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Unsupported controller model: %s"),
                       virDomainControllerModelSCSITypeToString(model));
        return false;
6102 6103 6104 6105 6106 6107
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LAST:
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unexpected SCSI controller model %d"),
                       model);
        return false;
6108 6109 6110 6111 6112 6113
    }

    return true;
}


6114 6115 6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133 6134 6135 6136 6137 6138 6139
static int
qemuDomainDeviceDefValidateControllerIDE(const virDomainControllerDef *controller,
                                         const virDomainDef *def)
{
    /* first IDE controller is implicit on various machines */
    if (controller->idx == 0 && qemuDomainHasBuiltinIDE(def))
        return 0;

    /* Since we currently only support the integrated IDE
     * controller on various boards, if we ever get to here, it's
     * because some other machinetype had an IDE controller
     * specified, or one with a single IDE controller had multiple
     * IDE controllers specified.
     */
    if (qemuDomainHasBuiltinIDE(def))
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Only a single IDE controller is supported "
                         "for this machine type"));
    else
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("IDE controllers are unsupported for "
                         "this QEMU binary or machine type"));
    return -1;
}


6140 6141 6142 6143 6144 6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156
/* qemuDomainCheckSCSIControllerIOThreads:
 * @controller: Pointer to controller def
 * @def: Pointer to domain def
 *
 * If this controller definition has iothreads set, let's make sure the
 * configuration is right before adding to the command line
 *
 * Returns true if either supported or there are no iothreads for controller;
 * otherwise, returns false if configuration is not quite right.
 */
static bool
qemuDomainCheckSCSIControllerIOThreads(const virDomainControllerDef *controller,
                                       const virDomainDef *def)
{
    if (!controller->iothread)
        return true;

6157 6158
    if (controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI &&
6159 6160
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
       virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
6161 6162
                       _("virtio-scsi IOThreads only available for virtio "
                         "pci and virtio ccw controllers"));
6163 6164 6165 6166 6167 6168 6169 6170 6171 6172 6173 6174 6175 6176 6177 6178 6179 6180 6181 6182 6183
       return false;
    }

    /* Can we find the controller iothread in the iothreadid list? */
    if (!virDomainIOThreadIDFind(def, controller->iothread)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("controller iothread '%u' not defined in iothreadid"),
                       controller->iothread);
        return false;
    }

    return true;
}


static int
qemuDomainDeviceDefValidateControllerSCSI(const virDomainControllerDef *controller,
                                          const virDomainDef *def)
{
    switch ((virDomainControllerModelSCSI) controller->model) {
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI:
6184 6185
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL:
6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196
            if (!qemuDomainCheckSCSIControllerIOThreads(controller, def))
                return -1;
            break;

        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_AUTO:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_BUSLOGIC:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSILOGIC:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSISAS1068:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VMPVSCSI:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_IBMVSCSI:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSISAS1078:
6197
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT:
6198 6199 6200 6201 6202 6203 6204 6205
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LAST:
            break;
    }

    return 0;
}


6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236 6237
/**
 * virDomainControllerPCIModelNameToQEMUCaps:
 * @modelName: model name
 *
 * Maps model names for PCI controllers (virDomainControllerPCIModelName)
 * to the QEMU capabilities required to use them (virQEMUCapsFlags).
 *
 * Returns: the QEMU capability itself (>0) on success; 0 if no QEMU
 *          capability is needed; <0 on error.
 */
static int
virDomainControllerPCIModelNameToQEMUCaps(int modelName)
{
    switch ((virDomainControllerPCIModelName) modelName) {
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCI_BRIDGE:
        return QEMU_CAPS_DEVICE_PCI_BRIDGE;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_I82801B11_BRIDGE:
        return QEMU_CAPS_DEVICE_DMI_TO_PCI_BRIDGE;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_IOH3420:
        return QEMU_CAPS_DEVICE_IOH3420;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_X3130_UPSTREAM:
        return QEMU_CAPS_DEVICE_X3130_UPSTREAM;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_XIO3130_DOWNSTREAM:
        return QEMU_CAPS_DEVICE_XIO3130_DOWNSTREAM;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PXB:
        return QEMU_CAPS_DEVICE_PXB;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PXB_PCIE:
        return QEMU_CAPS_DEVICE_PXB_PCIE;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCIE_ROOT_PORT:
        return QEMU_CAPS_DEVICE_PCIE_ROOT_PORT;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_SPAPR_PCI_HOST_BRIDGE:
        return QEMU_CAPS_DEVICE_SPAPR_PCI_HOST_BRIDGE;
6238 6239
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCIE_PCI_BRIDGE:
        return QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE;
6240 6241 6242 6243 6244
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE:
        return 0;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_LAST:
    default:
        return -1;
6245 6246
    }

6247
    return -1;
6248 6249 6250
}


6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267
#define virReportControllerMissingOption(cont, model, modelName, option) \
    virReportError(VIR_ERR_INTERNAL_ERROR, \
                   _("Required option '%s' is not set for PCI controller " \
                     "with index '%d', model '%s' and modelName '%s'"), \
                   (option), (cont->idx), (model), (modelName));
#define virReportControllerInvalidOption(cont, model, modelName, option) \
    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, \
                   _("Option '%s' is not valid for PCI controller " \
                     "with index '%d', model '%s' and modelName '%s'"), \
                   (option), (cont->idx), (model), (modelName));
#define virReportControllerInvalidValue(cont, model, modelName, option) \
    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, \
                   _("Option '%s' has invalid value for PCI controller " \
                     "with index '%d', model '%s' and modelName '%s'"), \
                   (option), (cont->idx), (model), (modelName));


6268 6269 6270 6271 6272 6273 6274 6275 6276
static int
qemuDomainDeviceDefValidateControllerPCI(const virDomainControllerDef *cont,
                                         const virDomainDef *def,
                                         virQEMUCapsPtr qemuCaps)

{
    const virDomainPCIControllerOpts *pciopts = &cont->opts.pciopts;
    const char *model = virDomainControllerModelPCITypeToString(cont->model);
    const char *modelName = virDomainControllerPCIModelNameTypeToString(pciopts->modelName);
6277
    int cap = virDomainControllerPCIModelNameToQEMUCaps(pciopts->modelName);
6278 6279 6280 6281 6282 6283 6284 6285 6286 6287

    if (!model) {
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }
    if (!modelName) {
        virReportEnumRangeError(virDomainControllerPCIModelName, pciopts->modelName);
        return -1;
    }

6288 6289 6290 6291 6292 6293 6294 6295 6296
    /* modelName */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
6297
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
6298 6299 6300 6301 6302 6303 6304 6305 6306 6307 6308 6309 6310 6311 6312 6313 6314 6315 6316 6317 6318 6319 6320 6321 6322 6323 6324 6325 6326 6327 6328 6329 6330 6331 6332 6333 6334 6335 6336 6337 6338 6339 6340 6341 6342 6343 6344 6345 6346 6347 6348 6349 6350 6351 6352 6353 6354 6355 6356 6357 6358 6359 6360 6361 6362 6363 6364 6365 6366 6367 6368 6369 6370 6371 6372 6373 6374 6375 6376 6377 6378 6379 6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397 6398 6399 6400 6401
        /* modelName should have been set automatically */
        if (pciopts->modelName == VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE) {
            virReportControllerMissingOption(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
        /* modelName must be set for pSeries guests, but it's an error
         * for it to be set for any other guest */
        if (qemuDomainIsPSeries(def)) {
            if (pciopts->modelName == VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE) {
                virReportControllerMissingOption(cont, model, modelName, "modelName");
                return -1;
            }
        } else {
            if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE) {
                virReportControllerInvalidOption(cont, model, modelName, "modelName");
                return -1;
            }
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE) {
            virReportControllerInvalidOption(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

    /* modelName (cont'd) */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE &&
            pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_SPAPR_PCI_HOST_BRIDGE) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCI_BRIDGE) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_I82801B11_BRIDGE) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_IOH3420 &&
            pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCIE_ROOT_PORT) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_X3130_UPSTREAM) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_XIO3130_DOWNSTREAM) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PXB) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PXB_PCIE) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

6402 6403 6404 6405 6406 6407 6408
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
        if (pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCIE_PCI_BRIDGE) {
            virReportControllerInvalidValue(cont, model, modelName, "modelName");
            return -1;
        }
        break;

6409 6410 6411 6412 6413 6414 6415
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

6416 6417 6418 6419 6420 6421 6422 6423 6424
    /* index */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
6425
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
6426 6427 6428 6429 6430 6431 6432 6433 6434 6435 6436 6437 6438 6439 6440 6441 6442 6443 6444 6445 6446 6447 6448 6449 6450 6451 6452 6453 6454 6455 6456 6457 6458 6459
        if (cont->idx == 0) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Index for '%s' controllers must be > 0"),
                           model);
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
        /* pSeries guests can have multiple PHBs, so it's expected that
         * the index will not be zero for some of them */
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT &&
            pciopts->modelName == VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_SPAPR_PCI_HOST_BRIDGE) {
            break;
        }

        /* For all other pci-root and pcie-root controllers, though,
         * the index must be zero*/
        if (cont->idx != 0) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Index for '%s' controllers must be 0"),
                           model);
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

6460 6461 6462 6463 6464 6465 6466 6467 6468 6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479 6480 6481 6482 6483 6484 6485 6486
    /* targetIndex */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
        /* PHBs for pSeries guests must have been assigned a targetIndex */
        if (pciopts->targetIndex == -1 &&
            pciopts->modelName == VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_SPAPR_PCI_HOST_BRIDGE) {
            virReportControllerMissingOption(cont, model, modelName, "targetIndex");
            return -1;
        }

        /* targetIndex only applies to PHBs, so for any other pci-root
         * controller it being present is an error */
        if (pciopts->targetIndex != -1 &&
            pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_SPAPR_PCI_HOST_BRIDGE) {
            virReportControllerInvalidOption(cont, model, modelName, "targetIndex");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
6487
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
6488 6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500
        if (pciopts->targetIndex != -1) {
            virReportControllerInvalidOption(cont, model, modelName, "targetIndex");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517 6518 6519 6520
    /* pcihole64 */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
        /* The pcihole64 option only applies to x86 guests */
        if ((pciopts->pcihole64 ||
             pciopts->pcihole64size != 0) &&
            !ARCH_IS_X86(def->os.arch)) {
            virReportControllerInvalidOption(cont, model, modelName, "pcihole64");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
6521
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
6522 6523 6524 6525 6526 6527 6528 6529 6530 6531 6532 6533 6534 6535
        if (pciopts->pcihole64 ||
            pciopts->pcihole64size != 0) {
            virReportControllerInvalidOption(cont, model, modelName, "pcihole64");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

6536 6537 6538 6539 6540 6541 6542 6543 6544 6545 6546 6547 6548 6549 6550 6551 6552
    /* busNr */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
        if (pciopts->busNr == -1) {
            virReportControllerMissingOption(cont, model, modelName, "busNr");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
6553
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
6554 6555 6556 6557 6558 6559 6560 6561 6562 6563 6564 6565 6566
        if (pciopts->busNr != -1) {
            virReportControllerInvalidOption(cont, model, modelName, "busNr");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

6567 6568 6569 6570 6571 6572 6573 6574 6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597
    /* numaNode */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
        /* numaNode can be used for these controllers, but it's not set
         * automatically so it can be missing */
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
        /* Only PHBs support numaNode */
        if (pciopts->numaNode != -1 &&
            pciopts->modelName != VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_SPAPR_PCI_HOST_BRIDGE) {
            virReportControllerInvalidOption(cont, model, modelName, "numaNode");
            return -1;
        }

        /* However, the default PHB doesn't support numaNode */
        if (pciopts->numaNode != -1 &&
            pciopts->modelName == VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_SPAPR_PCI_HOST_BRIDGE &&
            pciopts->targetIndex == 0) {
            virReportControllerInvalidOption(cont, model, modelName, "numaNode");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
6598
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
6599 6600 6601 6602 6603 6604 6605 6606 6607 6608 6609 6610 6611
        if (pciopts->numaNode != -1) {
            virReportControllerInvalidOption(cont, model, modelName, "numaNode");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

6612 6613 6614 6615 6616 6617 6618 6619 6620 6621 6622 6623 6624 6625 6626 6627 6628
    /* chassisNr */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
        if (pciopts->chassisNr == -1) {
            virReportControllerMissingOption(cont, model, modelName, "chassisNr");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
6629
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
6630 6631 6632 6633 6634 6635 6636 6637 6638 6639 6640 6641 6642
        if (pciopts->chassisNr != -1) {
            virReportControllerInvalidOption(cont, model, modelName, "chassisNr");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

6643 6644 6645 6646 6647 6648 6649 6650 6651 6652 6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663
    /* chassis and port */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
        if (pciopts->chassis == -1) {
            virReportControllerMissingOption(cont, model, modelName, "chassis");
            return -1;
        }
        if (pciopts->port == -1) {
            virReportControllerMissingOption(cont, model, modelName, "port");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
6664
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676 6677 6678 6679 6680
        if (pciopts->chassis != -1) {
            virReportControllerInvalidOption(cont, model, modelName, "chassis");
            return -1;
        }
        if (pciopts->port != -1) {
            virReportControllerInvalidOption(cont, model, modelName, "port");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
    }

6681 6682 6683 6684 6685 6686 6687 6688 6689 6690 6691 6692 6693 6694 6695 6696 6697 6698 6699 6700 6701 6702 6703 6704 6705 6706
    /* QEMU device availability */
    if (cap < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unknown QEMU device for '%s' controller"),
                       modelName);
        return -1;
    }
    if (cap > 0 && !virQEMUCapsGet(qemuCaps, cap)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("The '%s' device is not supported by this QEMU binary"),
                       modelName);
        return -1;
    }

    /* PHBs didn't support numaNode from the very beginning, so an extra
     * capability check is required */
    if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT &&
        pciopts->modelName == VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_SPAPR_PCI_HOST_BRIDGE &&
        pciopts->numaNode != -1 &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_SPAPR_PCI_HOST_BRIDGE_NUMA_NODE)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Option '%s' is not supported by '%s' device with this QEMU binary"),
                       "numaNode", modelName);
        return -1;
    }

6707
    return 0;
6708 6709 6710
}


6711 6712 6713 6714 6715
#undef virReportControllerInvalidValue
#undef virReportControllerInvalidOption
#undef virReportControllerMissingOption


6716 6717 6718 6719 6720 6721 6722 6723 6724 6725 6726 6727 6728 6729 6730 6731 6732 6733
static int
qemuDomainDeviceDefValidateControllerSATA(const virDomainControllerDef *controller,
                                          const virDomainDef *def,
                                          virQEMUCapsPtr qemuCaps)
{
    /* first SATA controller on Q35 machines is implicit */
    if (controller->idx == 0 && qemuDomainIsQ35(def))
        return 0;

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_AHCI)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("SATA is not supported with this QEMU binary"));
        return -1;
    }
    return 0;
}


6734
static int
6735 6736 6737
qemuDomainDeviceDefValidateController(const virDomainControllerDef *controller,
                                      const virDomainDef *def,
                                      virQEMUCapsPtr qemuCaps)
6738
{
6739 6740
    int ret = 0;

6741
    if (!qemuDomainCheckCCWS390AddressSupport(def, &controller->info, qemuCaps,
6742 6743 6744
                                              "controller"))
        return -1;

6745 6746 6747 6748
    if (controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
        !qemuDomainCheckSCSIControllerModel(qemuCaps, controller->model))
        return -1;

6749 6750 6751
    if (qemuDomainDeviceDefValidateControllerAttributes(controller) < 0)
        return -1;

6752
    switch ((virDomainControllerType)controller->type) {
6753
    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
6754 6755 6756
        ret = qemuDomainDeviceDefValidateControllerIDE(controller, def);
        break;

6757
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
6758 6759 6760
        ret = qemuDomainDeviceDefValidateControllerSCSI(controller, def);
        break;

6761
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
6762 6763
        ret = qemuDomainDeviceDefValidateControllerPCI(controller, def,
                                                       qemuCaps);
6764 6765
        break;

6766
    case VIR_DOMAIN_CONTROLLER_TYPE_SATA:
6767 6768 6769 6770 6771
        ret = qemuDomainDeviceDefValidateControllerSATA(controller, def,
                                                        qemuCaps);
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
6772 6773 6774
    case VIR_DOMAIN_CONTROLLER_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_CONTROLLER_TYPE_CCID:
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
6775
    case VIR_DOMAIN_CONTROLLER_TYPE_XENBUS:
6776 6777 6778 6779
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
    }

6780
    return ret;
6781 6782 6783
}


6784
static int
6785 6786
qemuDomainDeviceDefValidateVsock(const virDomainVsockDef *vsock,
                                 const virDomainDef *def,
6787 6788 6789 6790 6791 6792 6793 6794
                                 virQEMUCapsPtr qemuCaps)
{
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VHOST_VSOCK)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("vsock device is not supported "
                         "with this QEMU binary"));
        return -1;
    }
6795

6796
    if (!qemuDomainCheckCCWS390AddressSupport(def, &vsock->info, qemuCaps,
6797 6798 6799
                                              "vsock"))
        return -1;

6800 6801
    return 0;
}
6802

6803 6804 6805

static int
qemuDomainDeviceDefValidateTPM(virDomainTPMDef *tpm,
6806 6807
                               const virDomainDef *def,
                               virQEMUCapsPtr qemuCaps)
6808 6809 6810 6811 6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828
{
    /* TPM 1.2 and 2 are not compatible, so we choose a specific version here */
    if (tpm->version == VIR_DOMAIN_TPM_VERSION_DEFAULT)
        tpm->version = VIR_DOMAIN_TPM_VERSION_1_2;

    switch (tpm->version) {
    case VIR_DOMAIN_TPM_VERSION_1_2:
        /* only TIS available for emulator */
        if (tpm->type == VIR_DOMAIN_TPM_TYPE_EMULATOR &&
            tpm->model != VIR_DOMAIN_TPM_MODEL_TIS) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported interface %s for TPM 1.2"),
                           virDomainTPMModelTypeToString(tpm->model));
            return -1;
        }
        break;
    case VIR_DOMAIN_TPM_VERSION_2_0:
    case VIR_DOMAIN_TPM_VERSION_DEFAULT:
    case VIR_DOMAIN_TPM_VERSION_LAST:
        break;
    }
6829 6830 6831 6832 6833 6834 6835 6836 6837 6838 6839 6840 6841 6842 6843 6844

    switch (tpm->type) {
    case VIR_DOMAIN_TPM_TYPE_PASSTHROUGH:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_TPM_PASSTHROUGH))
            goto no_support;
        break;

    case VIR_DOMAIN_TPM_TYPE_EMULATOR:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_TPM_EMULATOR))
            goto no_support;

        break;
    case VIR_DOMAIN_TPM_TYPE_LAST:
        break;
    }

6845
    return 0;
6846 6847 6848 6849 6850 6851 6852 6853

 no_support:
    virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                   _("The QEMU executable %s does not support TPM "
                     "backend type %s"),
                   def->emulator,
                   virDomainTPMBackendTypeToString(tpm->type));
    return -1;
6854 6855 6856
}


6857 6858 6859 6860 6861 6862 6863 6864 6865
static int
qemuDomainDeviceDefValidateGraphics(const virDomainGraphicsDef *graphics,
                                    const virDomainDef *def,
                                    virQEMUCapsPtr qemuCaps)
{
    bool have_egl_headless = false;
    size_t i;

    for (i = 0; i < def->ngraphics; i++) {
6866
        if (def->graphics[i]->type == VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS) {
6867 6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882 6883 6884 6885 6886 6887 6888 6889 6890 6891 6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906 6907 6908
            have_egl_headless = true;
            break;
        }
    }

    /* Only VNC and SPICE can be paired with egl-headless, the other types
     * either don't make sense to pair with egl-headless or aren't even
     * supported by QEMU.
     */
    if (have_egl_headless) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_EGL_HEADLESS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("egl-headless display is not supported with this "
                             "QEMU binary"));
            return -1;
        }

        if (graphics->type != VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS &&
            graphics->type != VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
            graphics->type != VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("graphics type 'egl-headless' is only supported "
                             "with one of: 'vnc', 'spice' graphics types"));
            return -1;
        }

        /* '-spice gl=on' and '-display egl-headless' are mutually
         * exclusive
         */
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE &&
            graphics->data.spice.gl == VIR_TRISTATE_BOOL_YES) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("multiple OpenGL displays are not supported "
                             "by QEMU"));
            return -1;
        }
    }

    return 0;
}


6909 6910
static int
qemuDomainDeviceDefValidateInput(const virDomainInputDef *input,
J
Ján Tomko 已提交
6911
                                 const virDomainDef *def G_GNUC_UNUSED,
6912 6913
                                 virQEMUCapsPtr qemuCaps)
{
6914 6915 6916 6917
    const char *baseName;
    int cap;
    int ccwCap;

6918 6919 6920
    if (input->bus != VIR_DOMAIN_INPUT_BUS_VIRTIO)
        return 0;

6921 6922 6923 6924 6925 6926 6927 6928 6929 6930 6931 6932 6933 6934 6935 6936 6937 6938 6939 6940 6941 6942 6943 6944 6945 6946 6947 6948 6949 6950 6951 6952
    /* Only type=passthrough supports model=virtio-(non-)transitional */
    switch ((virDomainInputModel)input->model) {
    case VIR_DOMAIN_INPUT_MODEL_VIRTIO_TRANSITIONAL:
    case VIR_DOMAIN_INPUT_MODEL_VIRTIO_NON_TRANSITIONAL:
        switch ((virDomainInputType)input->type) {
        case VIR_DOMAIN_INPUT_TYPE_MOUSE:
        case VIR_DOMAIN_INPUT_TYPE_TABLET:
        case VIR_DOMAIN_INPUT_TYPE_KBD:
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("virtio (non-)transitional models are not "
                             "supported for input type=%s"),
                           virDomainInputTypeToString(input->type));
            return -1;
        case VIR_DOMAIN_INPUT_TYPE_PASSTHROUGH:
            break;
        case VIR_DOMAIN_INPUT_TYPE_LAST:
        default:
            virReportEnumRangeError(virDomainInputType,
                                    input->type);
            return -1;
        }
        break;
    case VIR_DOMAIN_INPUT_MODEL_VIRTIO:
    case VIR_DOMAIN_INPUT_MODEL_DEFAULT:
        break;
    case VIR_DOMAIN_INPUT_MODEL_LAST:
    default:
        virReportEnumRangeError(virDomainInputModel,
                                input->model);
        return -1;
    }

6953 6954
    switch ((virDomainInputType)input->type) {
    case VIR_DOMAIN_INPUT_TYPE_MOUSE:
6955 6956 6957
        baseName = "virtio-mouse";
        cap = QEMU_CAPS_VIRTIO_MOUSE;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_MOUSE_CCW;
6958 6959
        break;
    case VIR_DOMAIN_INPUT_TYPE_TABLET:
6960 6961 6962
        baseName = "virtio-tablet";
        cap = QEMU_CAPS_VIRTIO_TABLET;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_TABLET_CCW;
6963 6964
        break;
    case VIR_DOMAIN_INPUT_TYPE_KBD:
6965 6966 6967
        baseName = "virtio-keyboard";
        cap = QEMU_CAPS_VIRTIO_KEYBOARD;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_KEYBOARD_CCW;
6968 6969
        break;
    case VIR_DOMAIN_INPUT_TYPE_PASSTHROUGH:
6970 6971 6972
        baseName = "virtio-input-host";
        cap = QEMU_CAPS_VIRTIO_INPUT_HOST;
        ccwCap = QEMU_CAPS_LAST;
6973 6974 6975 6976 6977 6978 6979 6980
        break;
    case VIR_DOMAIN_INPUT_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainInputType,
                                input->type);
        return -1;
    }

6981 6982 6983 6984 6985 6986 6987 6988 6989
    if (!virQEMUCapsGet(qemuCaps, cap) ||
        (input->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW &&
         !virQEMUCapsGet(qemuCaps, ccwCap))) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("%s is not supported by this QEMU binary"),
                       baseName);
        return -1;
    }

6990 6991 6992 6993
    return 0;
}


6994 6995 6996 6997 6998 6999 7000 7001 7002
static int
qemuDomainDeviceDefValidateMemballoon(const virDomainMemballoonDef *memballoon,
                                      virQEMUCapsPtr qemuCaps)
{
    if (!memballoon ||
        memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_NONE) {
        return 0;
    }

7003 7004 7005
    if (memballoon->model != VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO &&
        memballoon->model != VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO_TRANSITIONAL &&
        memballoon->model != VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO_NON_TRANSITIONAL) {
7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020 7021 7022
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Memory balloon device type '%s' is not supported by this version of qemu"),
                       virDomainMemballoonModelTypeToString(memballoon->model));
        return -1;
    }

    if (memballoon->autodeflate != VIR_TRISTATE_SWITCH_ABSENT &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_BALLOON_AUTODEFLATE)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("deflate-on-oom is not supported by this QEMU binary"));
        return -1;
    }

    return 0;
}


7023 7024 7025 7026 7027 7028 7029 7030 7031 7032 7033 7034 7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046
static int
qemuDomainDeviceDefValidateIOMMU(const virDomainIOMMUDef *iommu,
                                 const virDomainDef *def,
                                 virQEMUCapsPtr qemuCaps)
{
    switch (iommu->model) {
    case VIR_DOMAIN_IOMMU_MODEL_INTEL:
        if (!qemuDomainIsQ35(def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("IOMMU device: '%s' is only supported with "
                             "Q35 machines"),
                           virDomainIOMMUModelTypeToString(iommu->model));
            return -1;
        }
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_INTEL_IOMMU) &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_IOMMU)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("IOMMU device: '%s' is not supported with "
                             "this QEMU binary"),
                           virDomainIOMMUModelTypeToString(iommu->model));
            return -1;
        }
        break;

7047
    case VIR_DOMAIN_IOMMU_MODEL_SMMUV3:
7048 7049 7050 7051 7052 7053 7054 7055 7056 7057 7058 7059 7060 7061
        if (!qemuDomainIsARMVirt(def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("IOMMU device: '%s' is only supported with "
                             "ARM Virt machines"),
                           virDomainIOMMUModelTypeToString(iommu->model));
            return -1;
        }
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_VIRT_IOMMU)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("IOMMU device: '%s' is not supported with "
                             "this QEMU binary"),
                           virDomainIOMMUModelTypeToString(iommu->model));
            return -1;
        }
7062 7063
        break;

7064 7065 7066
    case VIR_DOMAIN_IOMMU_MODEL_LAST:
    default:
        virReportEnumRangeError(virDomainIOMMUModel, iommu->model);
7067
        return -1;
7068 7069
    }

7070 7071 7072 7073 7074 7075 7076 7077 7078 7079 7080 7081 7082 7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093 7094 7095 7096 7097 7098 7099 7100 7101 7102 7103
    /* These capability checks ensure we're not trying to use features
     * of Intel IOMMU that the QEMU binary does not support, but they
     * also make sure we report an error when trying to use features
     * that are not implemented by SMMUv3 */

    if (iommu->intremap != VIR_TRISTATE_SWITCH_ABSENT &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_INTEL_IOMMU_INTREMAP)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("iommu: interrupt remapping is not supported "
                         "with this QEMU binary"));
        return -1;
    }
    if (iommu->caching_mode != VIR_TRISTATE_SWITCH_ABSENT &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_INTEL_IOMMU_CACHING_MODE))  {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("iommu: caching mode is not supported "
                         "with this QEMU binary"));
        return -1;
    }
    if (iommu->eim != VIR_TRISTATE_SWITCH_ABSENT &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_INTEL_IOMMU_EIM))  {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("iommu: eim is not supported "
                         "with this QEMU binary"));
        return -1;
    }
    if (iommu->iotlb != VIR_TRISTATE_SWITCH_ABSENT &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_INTEL_IOMMU_DEVICE_IOTLB)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("iommu: device IOTLB is not supported "
                         "with this QEMU binary"));
        return -1;
    }

7104 7105 7106 7107
    return 0;
}


7108 7109
static int
qemuDomainDeviceDefValidateFS(virDomainFSDefPtr fs,
J
Ján Tomko 已提交
7110 7111
                              const virDomainDef *def G_GNUC_UNUSED,
                              virQEMUCapsPtr qemuCaps G_GNUC_UNUSED)
7112 7113 7114 7115 7116 7117 7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131 7132 7133 7134 7135 7136 7137 7138 7139 7140 7141 7142 7143 7144 7145 7146 7147 7148
{
    if (fs->type != VIR_DOMAIN_FS_TYPE_MOUNT) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("only supports mount filesystem type"));
        return -1;
    }

    switch ((virDomainFSDriverType) fs->fsdriver) {
    case VIR_DOMAIN_FS_DRIVER_TYPE_DEFAULT:
    case VIR_DOMAIN_FS_DRIVER_TYPE_PATH:
        break;

    case VIR_DOMAIN_FS_DRIVER_TYPE_HANDLE:
        if (fs->accessmode != VIR_DOMAIN_FS_ACCESSMODE_PASSTHROUGH) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("only supports passthrough accessmode"));
            return -1;
        }
        break;

    case VIR_DOMAIN_FS_DRIVER_TYPE_LOOP:
    case VIR_DOMAIN_FS_DRIVER_TYPE_NBD:
    case VIR_DOMAIN_FS_DRIVER_TYPE_PLOOP:
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Filesystem driver type not supported"));
        return -1;

    case VIR_DOMAIN_FS_DRIVER_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainFSDriverType, fs->fsdriver);
        return -1;
    }

    return 0;
}


7149 7150 7151 7152 7153 7154 7155 7156 7157 7158 7159 7160 7161 7162 7163 7164 7165 7166 7167 7168 7169 7170 7171 7172 7173
static int
qemuDomainDeviceDefValidateZPCIAddress(virDomainDeviceInfoPtr info,
                                       virQEMUCapsPtr qemuCaps)
{
    if (!virZPCIDeviceAddressIsEmpty(&info->addr.pci.zpci) &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_ZPCI)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s",
                       _("This QEMU binary doesn't support zPCI"));
        return -1;
    }

    return 0;
}


static int
qemuDomainDeviceDefValidateAddress(const virDomainDeviceDef *dev,
                                   virQEMUCapsPtr qemuCaps)
{
    virDomainDeviceInfoPtr info;

    if (!(info = virDomainDeviceGetInfo((virDomainDeviceDef *)dev)))
        return 0;

7174 7175
    switch ((virDomainDeviceAddressType) info->type) {
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI:
7176 7177
        return qemuDomainDeviceDefValidateZPCIAddress(info, qemuCaps);

7178 7179 7180 7181 7182 7183
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE:
        /* Address validation might happen before we have had a chance to
         * automatically assign addresses to devices for which the user
         * didn't specify one themselves */
        break;

7184 7185 7186 7187 7188 7189 7190 7191 7192 7193 7194 7195 7196 7197
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO: {
        virDomainDeviceSpaprVioAddressPtr addr = &(info->addr.spaprvio);

        if (addr->has_reg && addr->reg > 0xffffffff) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("spapr-vio reg='0x%llx' exceeds maximum "
                             "possible value (0xffffffff)"),
                           addr->reg);
            return -1;
        }

        break;
        }

7198 7199 7200 7201 7202 7203 7204 7205 7206 7207 7208 7209 7210 7211 7212 7213 7214 7215
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCID:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_USB:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_MMIO:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_ISA:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DIMM:
        /* No validation for these address types yet */
        break;

    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainDeviceAddressType, info->type);
        return -1;
    }

7216 7217 7218 7219
    return 0;
}


7220 7221 7222
static int
qemuDomainDeviceDefValidate(const virDomainDeviceDef *dev,
                            const virDomainDef *def,
7223
                            void *opaque)
7224
{
7225
    int ret = 0;
7226 7227
    virQEMUDriverPtr driver = opaque;
    virQEMUCapsPtr qemuCaps = NULL;
7228
    virDomainCapsPtr domCaps = NULL;
7229 7230 7231 7232

    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
                                            def->emulator)))
        return -1;
7233

7234 7235 7236 7237 7238 7239
    if (!(domCaps = virQEMUDriverGetDomainCapabilities(driver, qemuCaps,
                                                       def->os.machine,
                                                       def->os.arch,
                                                       def->virtType)))
        goto cleanup;

7240 7241 7242
    if ((ret = qemuDomainDeviceDefValidateAddress(dev, qemuCaps)) < 0)
        goto cleanup;

7243 7244 7245
    if (virDomainCapsDeviceDefValidate(domCaps, dev, def) < 0)
        goto cleanup;

7246
    switch ((virDomainDeviceType)dev->type) {
7247 7248 7249 7250 7251 7252 7253 7254 7255 7256 7257 7258 7259
    case VIR_DOMAIN_DEVICE_NET:
        ret = qemuDomainDeviceDefValidateNetwork(dev->data.net);
        break;

    case VIR_DOMAIN_DEVICE_CHR:
        ret = qemuDomainChrDefValidate(dev->data.chr, def);
        break;

    case VIR_DOMAIN_DEVICE_SMARTCARD:
        ret = qemuDomainSmartcardDefValidate(dev->data.smartcard);
        break;

    case VIR_DOMAIN_DEVICE_RNG:
7260
        ret = qemuDomainRNGDefValidate(dev->data.rng, qemuCaps);
7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271
        break;

    case VIR_DOMAIN_DEVICE_REDIRDEV:
        ret = qemuDomainRedirdevDefValidate(dev->data.redirdev);
        break;

    case VIR_DOMAIN_DEVICE_WATCHDOG:
        ret = qemuDomainWatchdogDefValidate(dev->data.watchdog, def);
        break;

    case VIR_DOMAIN_DEVICE_HOSTDEV:
7272 7273
        ret = qemuDomainDeviceDefValidateHostdev(dev->data.hostdev, def,
                                                 qemuCaps);
7274 7275 7276 7277 7278 7279 7280
        break;

    case VIR_DOMAIN_DEVICE_VIDEO:
        ret = qemuDomainDeviceDefValidateVideo(dev->data.video);
        break;

    case VIR_DOMAIN_DEVICE_DISK:
7281
        ret = qemuDomainDeviceDefValidateDisk(dev->data.disk, qemuCaps);
7282 7283
        break;

7284
    case VIR_DOMAIN_DEVICE_CONTROLLER:
7285 7286
        ret = qemuDomainDeviceDefValidateController(dev->data.controller, def,
                                                    qemuCaps);
7287 7288
        break;

7289
    case VIR_DOMAIN_DEVICE_VSOCK:
7290
        ret = qemuDomainDeviceDefValidateVsock(dev->data.vsock, def, qemuCaps);
7291 7292
        break;

7293
    case VIR_DOMAIN_DEVICE_TPM:
7294
        ret = qemuDomainDeviceDefValidateTPM(dev->data.tpm, def, qemuCaps);
7295 7296
        break;

7297 7298 7299 7300 7301
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainDeviceDefValidateGraphics(dev->data.graphics, def,
                                                  qemuCaps);
        break;

7302 7303 7304 7305
    case VIR_DOMAIN_DEVICE_INPUT:
        ret = qemuDomainDeviceDefValidateInput(dev->data.input, def, qemuCaps);
        break;

7306 7307 7308 7309
    case VIR_DOMAIN_DEVICE_MEMBALLOON:
        ret = qemuDomainDeviceDefValidateMemballoon(dev->data.memballoon, qemuCaps);
        break;

7310 7311 7312 7313
    case VIR_DOMAIN_DEVICE_IOMMU:
        ret = qemuDomainDeviceDefValidateIOMMU(dev->data.iommu, def, qemuCaps);
        break;

7314
    case VIR_DOMAIN_DEVICE_FS:
7315 7316 7317 7318
        ret = qemuDomainDeviceDefValidateFS(dev->data.fs, def, qemuCaps);
        break;

    case VIR_DOMAIN_DEVICE_LEASE:
7319 7320 7321 7322
    case VIR_DOMAIN_DEVICE_SOUND:
    case VIR_DOMAIN_DEVICE_HUB:
    case VIR_DOMAIN_DEVICE_NVRAM:
    case VIR_DOMAIN_DEVICE_SHMEM:
7323
    case VIR_DOMAIN_DEVICE_MEMORY:
7324 7325 7326 7327
    case VIR_DOMAIN_DEVICE_PANIC:
    case VIR_DOMAIN_DEVICE_NONE:
    case VIR_DOMAIN_DEVICE_LAST:
        break;
7328 7329
    }

7330
 cleanup:
7331
    virObjectUnref(qemuCaps);
7332
    virObjectUnref(domCaps);
7333 7334 7335 7336
    return ret;
}


7337 7338 7339 7340 7341 7342 7343 7344 7345
/**
 * qemuDomainDefaultNetModel:
 * @def: domain definition
 * @qemuCaps: qemu capabilities
 *
 * Returns the default network model for a given domain. Note that if @qemuCaps
 * is NULL this function may return NULL if the default model depends on the
 * capabilities.
 */
7346
static int
7347 7348
qemuDomainDefaultNetModel(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
7349
{
7350
    if (ARCH_IS_S390(def->os.arch))
7351
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
7352

S
Stefan Schallenberg 已提交
7353 7354
    if (def->os.arch == VIR_ARCH_ARMV6L ||
        def->os.arch == VIR_ARCH_ARMV7L ||
7355
        def->os.arch == VIR_ARCH_AARCH64) {
7356
        if (STREQ(def->os.machine, "versatilepb"))
7357
            return VIR_DOMAIN_NET_MODEL_SMC91C111;
7358

7359
        if (qemuDomainIsARMVirt(def))
7360
            return VIR_DOMAIN_NET_MODEL_VIRTIO;
7361

7362 7363
        /* Incomplete. vexpress (and a few others) use this, but not all
         * arm boards */
7364
        return VIR_DOMAIN_NET_MODEL_LAN9118;
7365 7366
    }

7367 7368
    /* virtio is a sensible default for RISC-V virt guests */
    if (qemuDomainIsRISCVVirt(def))
7369
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
7370

7371 7372 7373
    /* In all other cases the model depends on the capabilities. If they were
     * not provided don't report any default. */
    if (!qemuCaps)
7374
        return VIR_DOMAIN_NET_MODEL_UNKNOWN;
7375

7376 7377 7378 7379
    /* Try several network devices in turn; each of these devices is
     * less likely be supported out-of-the-box by the guest operating
     * system than the previous one */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_RTL8139))
7380
        return VIR_DOMAIN_NET_MODEL_RTL8139;
7381
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_E1000))
7382
        return VIR_DOMAIN_NET_MODEL_E1000;
7383
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_NET))
7384
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
7385 7386 7387

    /* We've had no luck detecting support for any network device,
     * but we have to return something: might as well be rtl8139 */
7388
    return VIR_DOMAIN_NET_MODEL_RTL8139;
7389
}
7390

7391 7392

/*
7393 7394 7395 7396 7397 7398 7399 7400 7401 7402 7403 7404 7405 7406 7407 7408
 * Clear auto generated unix socket paths:
 *
 * libvirt 1.2.18 and older:
 *     {cfg->channelTargetDir}/{dom-name}.{target-name}
 *
 * libvirt 1.2.19 - 1.3.2:
 *     {cfg->channelTargetDir}/domain-{dom-name}/{target-name}
 *
 * libvirt 1.3.3 and newer:
 *     {cfg->channelTargetDir}/domain-{dom-id}-{short-dom-name}/{target-name}
 *
 * The unix socket path was stored in config XML until libvirt 1.3.0.
 * If someone specifies the same path as we generate, they shouldn't do it.
 *
 * This function clears the path for migration as well, so we need to clear
 * the path even if we are not storing it in the XML.
7409
 */
7410
static int
7411 7412 7413
qemuDomainChrDefDropDefaultPath(virDomainChrDefPtr chr,
                                virQEMUDriverPtr driver)
{
7414 7415 7416 7417
    virQEMUDriverConfigPtr cfg;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *regexp = NULL;
    int ret = -1;
7418

7419 7420 7421 7422 7423
    if (chr->deviceType != VIR_DOMAIN_CHR_DEVICE_TYPE_CHANNEL ||
        chr->targetType != VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO ||
        chr->source->type != VIR_DOMAIN_CHR_TYPE_UNIX ||
        !chr->source->data.nix.path) {
        return 0;
7424 7425
    }

7426 7427 7428 7429 7430 7431 7432 7433 7434 7435 7436 7437 7438
    cfg = virQEMUDriverGetConfig(driver);

    virBufferEscapeRegex(&buf, "^%s", cfg->channelTargetDir);
    virBufferAddLit(&buf, "/([^/]+\\.)|(domain-[^/]+/)");
    virBufferEscapeRegex(&buf, "%s$", chr->target.name);

    regexp = virBufferContentAndReset(&buf);

    if (virStringMatch(chr->source->data.nix.path, regexp))
        VIR_FREE(chr->source->data.nix.path);

    ret = 0;
    VIR_FREE(regexp);
7439
    virObjectUnref(cfg);
7440
    return ret;
7441 7442 7443
}


7444 7445 7446 7447 7448 7449 7450 7451 7452 7453 7454 7455 7456 7457 7458 7459 7460 7461 7462 7463 7464 7465 7466 7467 7468 7469 7470 7471 7472 7473 7474 7475 7476 7477 7478 7479 7480 7481 7482 7483 7484 7485 7486 7487 7488 7489 7490 7491 7492 7493
static int
qemuDomainShmemDefPostParse(virDomainShmemDefPtr shm)
{
    /* This was the default since the introduction of this device. */
    if (shm->model != VIR_DOMAIN_SHMEM_MODEL_IVSHMEM_DOORBELL && !shm->size)
        shm->size = 4 << 20;

    /* Nothing more to check/change for IVSHMEM */
    if (shm->model == VIR_DOMAIN_SHMEM_MODEL_IVSHMEM)
        return 0;

    if (!shm->server.enabled) {
        if (shm->model == VIR_DOMAIN_SHMEM_MODEL_IVSHMEM_DOORBELL) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("shmem model '%s' is supported "
                             "only with server option enabled"),
                           virDomainShmemModelTypeToString(shm->model));
            return -1;
        }

        if (shm->msi.enabled) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("shmem model '%s' doesn't support "
                             "msi"),
                           virDomainShmemModelTypeToString(shm->model));
        }
    } else {
        if (shm->model == VIR_DOMAIN_SHMEM_MODEL_IVSHMEM_PLAIN) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("shmem model '%s' is supported "
                             "only with server option disabled"),
                           virDomainShmemModelTypeToString(shm->model));
            return -1;
        }

        if (shm->size) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("shmem model '%s' does not support size setting"),
                           virDomainShmemModelTypeToString(shm->model));
            return -1;
        }
        shm->msi.enabled = true;
        if (!shm->msi.ioeventfd)
            shm->msi.ioeventfd = VIR_TRISTATE_SWITCH_ON;
    }

    return 0;
}


7494
#define QEMU_USB_XHCI_MAXPORTS 15
7495 7496


7497 7498 7499
static int
qemuDomainControllerDefPostParse(virDomainControllerDefPtr cont,
                                 const virDomainDef *def,
7500 7501
                                 virQEMUCapsPtr qemuCaps,
                                 unsigned int parseFlags)
7502
{
7503 7504
    switch ((virDomainControllerType)cont->type) {
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
7505 7506 7507
        /* Set the default SCSI controller model if not already set */
        if (qemuDomainSetSCSIControllerModel(def, cont, qemuCaps) < 0)
            return -1;
7508
        break;
7509

7510
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
7511
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT && qemuCaps) {
7512
            /* Pick a suitable default model for the USB controller if none
7513 7514
             * has been selected by the user and we have the qemuCaps for
             * figuring out which contollers are supported.
7515 7516 7517 7518 7519 7520 7521
             *
             * We rely on device availability instead of setting the model
             * unconditionally because, for some machine types, there's a
             * chance we will get away with using the legacy USB controller
             * when the relevant device is not available.
             *
             * See qemuBuildControllerDevCommandLine() */
7522 7523 7524 7525 7526 7527 7528 7529 7530 7531 7532

            /* Default USB controller is piix3-uhci if available. */
            if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PIIX3_USB_UHCI))
                cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_PIIX3_UHCI;

            if (ARCH_IS_S390(def->os.arch)) {
                if (cont->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE) {
                    /* set the default USB model to none for s390 unless an
                     * address is found */
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NONE;
                }
7533
            } else if (ARCH_IS_PPC64(def->os.arch)) {
7534 7535
                /* To not break migration we need to set default USB controller
                 * for ppc64 to pci-ohci if we cannot change ABI of the VM.
7536 7537
                 * The nec-usb-xhci or qemu-xhci controller is used as default
                 * only for newly defined domains or devices. */
7538
                if ((parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE) &&
7539 7540 7541
                    virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_QEMU_XHCI)) {
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_QEMU_XHCI;
                } else if ((parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE) &&
7542 7543 7544
                    virQEMUCapsGet(qemuCaps, QEMU_CAPS_NEC_USB_XHCI)) {
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
                } else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_OHCI)) {
7545
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_PCI_OHCI;
7546 7547 7548
                } else {
                    /* Explicitly fallback to legacy USB controller for PPC64. */
                    cont->model = -1;
7549
                }
7550
            } else if (def->os.arch == VIR_ARCH_AARCH64) {
7551 7552 7553
                if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_QEMU_XHCI))
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_QEMU_XHCI;
                else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_NEC_USB_XHCI))
7554
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
7555 7556 7557 7558 7559 7560 7561 7562 7563 7564 7565
            }
        }
        /* forbid usb model 'qusb1' and 'qusb2' in this kind of hyperviosr */
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_QUSB1 ||
            cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_QUSB2) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("USB controller model type 'qusb1' or 'qusb2' "
                             "is not supported in %s"),
                           virDomainVirtTypeToString(def->virtType));
            return -1;
        }
7566 7567 7568
        if ((cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI ||
             cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_QEMU_XHCI) &&
            cont->opts.usbopts.ports > QEMU_USB_XHCI_MAXPORTS) {
7569
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
7570 7571 7572
                           _("'%s' controller only supports up to '%u' ports"),
                           virDomainControllerModelUSBTypeToString(cont->model),
                           QEMU_USB_XHCI_MAXPORTS);
7573 7574 7575
            return -1;
        }
        break;
7576

7577
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
7578 7579 7580 7581 7582 7583 7584

        /* pSeries guests can have multiple pci-root controllers,
         * but other machine types only support a single one */
        if (!qemuDomainIsPSeries(def) &&
            (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT ||
             cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT) &&
            cont->idx != 0) {
7585 7586 7587 7588 7589 7590
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("pci-root and pcie-root controllers "
                             "should have index 0"));
            return -1;
        }

7591
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS &&
7592
            !qemuDomainIsI440FX(def)) {
7593 7594 7595 7596 7597 7598
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("pci-expander-bus controllers are only supported "
                             "on 440fx-based machinetypes"));
            return -1;
        }
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS &&
7599
            !qemuDomainIsQ35(def)) {
7600 7601 7602 7603 7604 7605
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("pcie-expander-bus controllers are only supported "
                             "on q35-based machinetypes"));
            return -1;
        }

7606 7607 7608
        /* if a PCI expander bus or pci-root on Pseries has a NUMA node
         * set, make sure that NUMA node is configured in the guest
         * <cpu><numa> array. NUMA cell id's in this array are numbered
7609 7610
         * from 0 .. size-1.
         */
7611 7612
        if (cont->opts.pciopts.numaNode >= 0 &&
            cont->opts.pciopts.numaNode >=
7613
            (int)virDomainNumaGetNodeCount(def->numa)) {
7614 7615 7616 7617 7618 7619 7620 7621 7622 7623
            virReportError(VIR_ERR_XML_ERROR,
                           _("%s with index %d is "
                             "configured for a NUMA node (%d) "
                             "not present in the domain's "
                             "<cpu><numa> array (%zu)"),
                           virDomainControllerModelPCITypeToString(cont->model),
                           cont->idx, cont->opts.pciopts.numaNode,
                           virDomainNumaGetNodeCount(def->numa));
            return -1;
        }
7624 7625 7626 7627 7628 7629 7630
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_SATA:
    case VIR_DOMAIN_CONTROLLER_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_CONTROLLER_TYPE_CCID:
    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
7631
    case VIR_DOMAIN_CONTROLLER_TYPE_XENBUS:
7632 7633
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
7634 7635 7636 7637 7638
    }

    return 0;
}

7639 7640 7641 7642 7643 7644
static int
qemuDomainChrDefPostParse(virDomainChrDefPtr chr,
                          const virDomainDef *def,
                          virQEMUDriverPtr driver,
                          unsigned int parseFlags)
{
7645 7646 7647 7648 7649 7650 7651 7652 7653 7654 7655 7656 7657 7658
    /* Historically, isa-serial and the default matched, so in order to
     * maintain backwards compatibility we map them here. The actual default
     * will be picked below based on the architecture and machine type. */
    if (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL &&
        chr->targetType == VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA) {
        chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE;
    }

    /* Set the default serial type */
    if (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL &&
        chr->targetType == VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE) {
        if (ARCH_IS_X86(def->os.arch)) {
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA;
        } else if (qemuDomainIsPSeries(def)) {
7659
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
7660
        } else if (qemuDomainIsARMVirt(def) || qemuDomainIsRISCVVirt(def)) {
7661
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
7662 7663
        } else if (ARCH_IS_S390(def->os.arch)) {
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
7664 7665 7666
        }
    }

7667 7668 7669
    /* Set the default target model */
    if (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL &&
        chr->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_NONE) {
7670
        switch ((virDomainChrSerialTargetType)chr->targetType) {
7671 7672 7673 7674 7675 7676 7677 7678 7679
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_ISA_SERIAL;
            break;
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_USB_SERIAL;
            break;
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PCI_SERIAL;
            break;
7680 7681 7682
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY;
            break;
7683
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
7684 7685 7686 7687 7688
            if (qemuDomainIsARMVirt(def)) {
                chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011;
            } else if (qemuDomainIsRISCVVirt(def)) {
                chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A;
            }
7689
            break;
7690 7691 7692
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE;
            break;
7693 7694 7695 7696 7697 7698 7699
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            /* Nothing to do */
            break;
        }
    }

7700 7701 7702 7703 7704 7705 7706 7707 7708 7709 7710 7711 7712 7713 7714 7715
    /* clear auto generated unix socket path for inactive definitions */
    if (parseFlags & VIR_DOMAIN_DEF_PARSE_INACTIVE) {
        if (qemuDomainChrDefDropDefaultPath(chr, driver) < 0)
            return -1;

        /* For UNIX chardev if no path is provided we generate one.
         * This also implies that the mode is 'bind'. */
        if (chr->source &&
            chr->source->type == VIR_DOMAIN_CHR_TYPE_UNIX &&
            !chr->source->data.nix.path) {
            chr->source->data.nix.listen = true;
        }
    }

    return 0;
}
7716

7717

7718 7719 7720 7721 7722 7723 7724 7725 7726 7727 7728 7729 7730 7731 7732 7733 7734 7735 7736 7737 7738 7739 7740 7741 7742 7743 7744 7745 7746 7747 7748 7749 7750 7751 7752 7753 7754 7755 7756 7757 7758 7759 7760 7761 7762 7763 7764 7765 7766 7767 7768 7769 7770 7771 7772 7773 7774 7775 7776 7777 7778 7779 7780 7781 7782 7783 7784 7785 7786 7787 7788 7789 7790 7791 7792 7793 7794 7795 7796 7797 7798
/**
 * qemuDomainDeviceDiskDefPostParseRestoreSecAlias:
 *
 * Re-generate aliases for objects related to the storage source if they
 * were not stored in the status XML by an older libvirt.
 *
 * Note that qemuCaps should be always present for a status XML.
 */
static int
qemuDomainDeviceDiskDefPostParseRestoreSecAlias(virDomainDiskDefPtr disk,
                                                virQEMUCapsPtr qemuCaps,
                                                unsigned int parseFlags)
{
    qemuDomainStorageSourcePrivatePtr priv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(disk->src);
    bool restoreAuthSecret = false;
    bool restoreEncSecret = false;
    char *authalias = NULL;
    char *encalias = NULL;
    int ret = -1;

    if (!(parseFlags & VIR_DOMAIN_DEF_PARSE_STATUS) ||
        !qemuCaps ||
        virStorageSourceIsEmpty(disk->src) ||
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

    /* network storage authentication secret */
    if (disk->src->auth &&
        (!priv || !priv->secinfo)) {

        /* only RBD and iSCSI (with capability) were supporting authentication
         * using secret object at the time we did not format the alias into the
         * status XML */
        if (virStorageSourceGetActualType(disk->src) == VIR_STORAGE_TYPE_NETWORK &&
            (disk->src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD ||
             (disk->src->protocol == VIR_STORAGE_NET_PROTOCOL_ISCSI &&
              virQEMUCapsGet(qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET))))
            restoreAuthSecret = true;
    }

    /* disk encryption secret */
    if (disk->src->encryption &&
        disk->src->encryption->format == VIR_STORAGE_ENCRYPTION_FORMAT_LUKS &&
        (!priv || !priv->encinfo))
        restoreEncSecret = true;

    if (!restoreAuthSecret && !restoreEncSecret)
        return 0;

    if (!priv) {
        if (!(disk->src->privateData = qemuDomainStorageSourcePrivateNew()))
            return -1;

        priv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(disk->src);
    }

    if (restoreAuthSecret) {
        if (!(authalias = qemuDomainGetSecretAESAlias(disk->info.alias, false)))
            goto cleanup;

        if (qemuStorageSourcePrivateDataAssignSecinfo(&priv->secinfo, &authalias) < 0)
            goto cleanup;
    }

    if (restoreEncSecret) {
        if (!(encalias = qemuDomainGetSecretAESAlias(disk->info.alias, true)))
            goto cleanup;

        if (qemuStorageSourcePrivateDataAssignSecinfo(&priv->encinfo, &encalias) < 0)
            goto cleanup;
    }

    ret = 0;

 cleanup:
    VIR_FREE(authalias);
    VIR_FREE(encalias);
    return ret;
}


7799 7800
static int
qemuDomainDeviceDiskDefPostParse(virDomainDiskDefPtr disk,
7801
                                 virQEMUCapsPtr qemuCaps,
7802
                                 unsigned int parseFlags)
7803 7804
{
    /* set default disk types and drivers */
7805 7806 7807
    if (!virDomainDiskGetDriver(disk) &&
        virDomainDiskSetDriver(disk, "qemu") < 0)
        return -1;
7808

7809 7810
    /* default disk format for drives */
    if (virDomainDiskGetFormat(disk) == VIR_STORAGE_FILE_NONE &&
7811
        virDomainDiskGetType(disk) != VIR_STORAGE_TYPE_VOLUME)
7812
        virDomainDiskSetFormat(disk, VIR_STORAGE_FILE_RAW);
7813

7814 7815 7816 7817
    /* default disk format for mirrored drive */
    if (disk->mirror &&
        disk->mirror->format == VIR_STORAGE_FILE_NONE)
        disk->mirror->format = VIR_STORAGE_FILE_RAW;
7818

7819 7820 7821 7822
    if (qemuDomainDeviceDiskDefPostParseRestoreSecAlias(disk, qemuCaps,
                                                        parseFlags) < 0)
        return -1;

7823 7824 7825 7826 7827 7828 7829
    /* regenerate TLS alias for old status XMLs */
    if (parseFlags & VIR_DOMAIN_DEF_PARSE_STATUS &&
        disk->src->haveTLS == VIR_TRISTATE_BOOL_YES &&
        !disk->src->tlsAlias &&
        !(disk->src->tlsAlias = qemuAliasTLSObjFromSrcAlias(disk->info.alias)))
        return -1;

7830 7831 7832 7833
    return 0;
}


7834 7835 7836 7837 7838 7839
static int
qemuDomainDeviceNetDefPostParse(virDomainNetDefPtr net,
                                const virDomainDef *def,
                                virQEMUCapsPtr qemuCaps)
{
    if (net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
7840 7841
        !virDomainNetGetModelString(net))
        net->model = qemuDomainDefaultNetModel(def, qemuCaps);
7842 7843 7844 7845 7846

    return 0;
}


7847 7848 7849 7850 7851 7852 7853
static int
qemuDomainDeviceVideoDefPostParse(virDomainVideoDefPtr video,
                                  const virDomainDef *def)
{
    if (video->type == VIR_DOMAIN_VIDEO_TYPE_DEFAULT) {
        if (ARCH_IS_PPC64(def->os.arch))
            video->type = VIR_DOMAIN_VIDEO_TYPE_VGA;
7854 7855 7856
        else if (qemuDomainIsARMVirt(def) ||
                 qemuDomainIsRISCVVirt(def) ||
                 ARCH_IS_S390(def->os.arch))
7857 7858 7859 7860 7861 7862 7863 7864 7865 7866 7867 7868 7869 7870
            video->type = VIR_DOMAIN_VIDEO_TYPE_VIRTIO;
        else
            video->type = VIR_DOMAIN_VIDEO_TYPE_CIRRUS;
    }

    if (video->type == VIR_DOMAIN_VIDEO_TYPE_QXL &&
        !video->vgamem) {
        video->vgamem = QEMU_QXL_VGAMEM_DEFAULT;
    }

    return 0;
}


7871 7872 7873 7874 7875 7876 7877 7878 7879 7880 7881 7882 7883 7884 7885 7886 7887
static int
qemuDomainDevicePanicDefPostParse(virDomainPanicDefPtr panic,
                                  const virDomainDef *def)
{
    if (panic->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT) {
        if (qemuDomainIsPSeries(def))
            panic->model = VIR_DOMAIN_PANIC_MODEL_PSERIES;
        else if (ARCH_IS_S390(def->os.arch))
            panic->model = VIR_DOMAIN_PANIC_MODEL_S390;
        else
            panic->model = VIR_DOMAIN_PANIC_MODEL_ISA;
    }

    return 0;
}


J
Ján Tomko 已提交
7888 7889 7890 7891 7892 7893 7894 7895 7896 7897
static int
qemuDomainVsockDefPostParse(virDomainVsockDefPtr vsock)
{
    if (vsock->model == VIR_DOMAIN_VSOCK_MODEL_DEFAULT)
        vsock->model = VIR_DOMAIN_VSOCK_MODEL_VIRTIO;

    return 0;
}


7898 7899 7900 7901 7902 7903 7904
static int
qemuDomainHostdevDefMdevPostParse(virDomainHostdevSubsysMediatedDevPtr mdevsrc,
                                  virQEMUCapsPtr qemuCaps)
{
    /* QEMU 2.12 added support for vfio-pci display type, we default to
     * 'display=off' to stay safe from future changes */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_VFIO_PCI_DISPLAY) &&
7905
        mdevsrc->model == VIR_MDEV_MODEL_TYPE_VFIO_PCI &&
7906 7907 7908 7909 7910 7911 7912 7913 7914 7915 7916 7917 7918 7919 7920 7921 7922 7923 7924 7925 7926 7927
        mdevsrc->display == VIR_TRISTATE_SWITCH_ABSENT)
        mdevsrc->display = VIR_TRISTATE_SWITCH_OFF;

    return 0;
}


static int
qemuDomainHostdevDefPostParse(virDomainHostdevDefPtr hostdev,
                              virQEMUCapsPtr qemuCaps)
{
    virDomainHostdevSubsysPtr subsys = &hostdev->source.subsys;

    if (hostdev->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS &&
        hostdev->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV &&
        qemuDomainHostdevDefMdevPostParse(&subsys->u.mdev, qemuCaps) < 0)
        return -1;

    return 0;
}


7928 7929
static int
qemuDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
7930
                             const virDomainDef *def,
J
Ján Tomko 已提交
7931
                             virCapsPtr caps G_GNUC_UNUSED,
7932
                             unsigned int parseFlags,
7933
                             void *opaque,
7934
                             void *parseOpaque)
7935
{
7936
    virQEMUDriverPtr driver = opaque;
7937 7938 7939
    /* Note that qemuCaps may be NULL when this function is called. This
     * function shall not fail in that case. It will be re-run on VM startup
     * with the capabilities populated. */
7940
    virQEMUCapsPtr qemuCaps = parseOpaque;
7941
    int ret = -1;
7942

7943 7944 7945 7946
    switch ((virDomainDeviceType) dev->type) {
    case VIR_DOMAIN_DEVICE_NET:
        ret = qemuDomainDeviceNetDefPostParse(dev->data.net, def, qemuCaps);
        break;
7947

7948
    case VIR_DOMAIN_DEVICE_DISK:
7949
        ret = qemuDomainDeviceDiskDefPostParse(dev->data.disk, qemuCaps,
7950
                                               parseFlags);
7951
        break;
7952

7953 7954 7955
    case VIR_DOMAIN_DEVICE_VIDEO:
        ret = qemuDomainDeviceVideoDefPostParse(dev->data.video, def);
        break;
7956

7957 7958 7959
    case VIR_DOMAIN_DEVICE_PANIC:
        ret = qemuDomainDevicePanicDefPostParse(dev->data.panic, def);
        break;
7960

7961 7962 7963 7964
    case VIR_DOMAIN_DEVICE_CONTROLLER:
        ret = qemuDomainControllerDefPostParse(dev->data.controller, def,
                                               qemuCaps, parseFlags);
        break;
7965

7966 7967 7968
    case VIR_DOMAIN_DEVICE_SHMEM:
        ret = qemuDomainShmemDefPostParse(dev->data.shmem);
        break;
7969

7970 7971 7972
    case VIR_DOMAIN_DEVICE_CHR:
        ret = qemuDomainChrDefPostParse(dev->data.chr, def, driver, parseFlags);
        break;
7973

J
Ján Tomko 已提交
7974 7975 7976 7977
    case VIR_DOMAIN_DEVICE_VSOCK:
        ret = qemuDomainVsockDefPostParse(dev->data.vsock);
        break;

7978 7979 7980 7981
    case VIR_DOMAIN_DEVICE_HOSTDEV:
        ret = qemuDomainHostdevDefPostParse(dev->data.hostdev, qemuCaps);
        break;

7982 7983 7984 7985 7986 7987 7988 7989 7990 7991 7992 7993 7994 7995 7996 7997 7998 7999 8000 8001 8002 8003 8004 8005 8006 8007 8008 8009
    case VIR_DOMAIN_DEVICE_LEASE:
    case VIR_DOMAIN_DEVICE_FS:
    case VIR_DOMAIN_DEVICE_INPUT:
    case VIR_DOMAIN_DEVICE_SOUND:
    case VIR_DOMAIN_DEVICE_WATCHDOG:
    case VIR_DOMAIN_DEVICE_GRAPHICS:
    case VIR_DOMAIN_DEVICE_HUB:
    case VIR_DOMAIN_DEVICE_REDIRDEV:
    case VIR_DOMAIN_DEVICE_SMARTCARD:
    case VIR_DOMAIN_DEVICE_MEMBALLOON:
    case VIR_DOMAIN_DEVICE_NVRAM:
    case VIR_DOMAIN_DEVICE_RNG:
    case VIR_DOMAIN_DEVICE_TPM:
    case VIR_DOMAIN_DEVICE_MEMORY:
    case VIR_DOMAIN_DEVICE_IOMMU:
        ret = 0;
        break;

    case VIR_DOMAIN_DEVICE_NONE:
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("unexpected VIR_DOMAIN_DEVICE_NONE"));
        break;

    case VIR_DOMAIN_DEVICE_LAST:
    default:
        virReportEnumRangeError(virDomainDeviceType, dev->type);
        break;
    }
8010 8011

    return ret;
8012 8013 8014
}


8015 8016
static int
qemuDomainDefAssignAddresses(virDomainDef *def,
J
Ján Tomko 已提交
8017 8018
                             virCapsPtr caps G_GNUC_UNUSED,
                             unsigned int parseFlags G_GNUC_UNUSED,
8019
                             void *opaque,
8020
                             void *parseOpaque)
8021 8022
{
    virQEMUDriverPtr driver = opaque;
8023 8024 8025
    /* Note that qemuCaps may be NULL when this function is called. This
     * function shall not fail in that case. It will be re-run on VM startup
     * with the capabilities populated. */
8026
    virQEMUCapsPtr qemuCaps = parseOpaque;
8027
    bool newDomain = parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE;
8028

8029 8030 8031 8032 8033 8034 8035
    /* Skip address assignment if @qemuCaps is not present. In such case devices
     * which are automatically added may be missing. Additionally @qemuCaps should
     * only be missing when reloading configs, thus addresses were already
     * assigned. */
    if (!qemuCaps)
        return 1;

8036 8037 8038 8039 8040 8041
    return qemuDomainAssignAddresses(def, qemuCaps, driver, NULL, newDomain);
}


static int
qemuDomainPostParseDataAlloc(const virDomainDef *def,
J
Ján Tomko 已提交
8042 8043
                             virCapsPtr caps G_GNUC_UNUSED,
                             unsigned int parseFlags G_GNUC_UNUSED,
8044 8045 8046 8047 8048 8049
                             void *opaque,
                             void **parseOpaque)
{
    virQEMUDriverPtr driver = opaque;

    if (!(*parseOpaque = virQEMUCapsCacheLookup(driver->qemuCapsCache,
8050
                                                def->emulator)))
8051
        return 1;
8052

8053 8054 8055 8056 8057 8058 8059 8060
    return 0;
}


static void
qemuDomainPostParseDataFree(void *parseOpaque)
{
    virQEMUCapsPtr qemuCaps = parseOpaque;
8061 8062 8063 8064 8065

    virObjectUnref(qemuCaps);
}


8066
virDomainDefParserConfig virQEMUDriverDomainDefParserConfig = {
8067
    .domainPostParseBasicCallback = qemuDomainDefPostParseBasic,
8068 8069
    .domainPostParseDataAlloc = qemuDomainPostParseDataAlloc,
    .domainPostParseDataFree = qemuDomainPostParseDataFree,
8070
    .devicesPostParseCallback = qemuDomainDeviceDefPostParse,
8071
    .domainPostParseCallback = qemuDomainDefPostParse,
8072
    .assignAddressesCallback = qemuDomainDefAssignAddresses,
8073
    .domainValidateCallback = qemuDomainDefValidate,
8074 8075
    .deviceValidateCallback = qemuDomainDeviceDefValidate,

8076
    .features = VIR_DOMAIN_DEF_FEATURE_MEMORY_HOTPLUG |
8077
                VIR_DOMAIN_DEF_FEATURE_OFFLINE_VCPUPIN |
8078
                VIR_DOMAIN_DEF_FEATURE_INDIVIDUAL_VCPUS |
8079
                VIR_DOMAIN_DEF_FEATURE_USER_ALIAS |
8080 8081
                VIR_DOMAIN_DEF_FEATURE_FW_AUTOSELECT |
                VIR_DOMAIN_DEF_FEATURE_NET_MODEL_STRING,
8082 8083 8084
};


8085
static void
8086 8087
qemuDomainObjSaveStatus(virQEMUDriverPtr driver,
                        virDomainObjPtr obj)
8088
{
8089 8090 8091
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    if (virDomainObjIsActive(obj)) {
8092
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj, driver->caps) < 0)
8093
            VIR_WARN("Failed to save status on vm %s", obj->def->name);
8094
    }
8095

8096
    virObjectUnref(cfg);
8097 8098
}

8099 8100 8101 8102 8103 8104 8105 8106

void
qemuDomainSaveStatus(virDomainObjPtr obj)
{
    qemuDomainObjSaveStatus(QEMU_DOMAIN_PRIVATE(obj)->driver, obj);
}


8107 8108 8109 8110
void
qemuDomainSaveConfig(virDomainObjPtr obj)
{
    virQEMUDriverPtr driver = QEMU_DOMAIN_PRIVATE(obj)->driver;
8111
    g_autoptr(virQEMUDriverConfig) cfg = NULL;
8112 8113 8114 8115 8116 8117 8118 8119 8120 8121 8122 8123 8124 8125 8126 8127 8128
    virDomainDefPtr def = NULL;

    if (virDomainObjIsActive(obj))
        def = obj->newDef;
    else
        def = obj->def;

    if (!def)
        return;

    cfg = virQEMUDriverGetConfig(driver);

    if (virDomainSaveConfig(cfg->configDir, driver->caps, def) < 0)
        VIR_WARN("Failed to save config of vm %s", obj->def->name);
}


J
Jiri Denemark 已提交
8129
void
8130
qemuDomainObjSetJobPhase(virQEMUDriverPtr driver,
J
Jiri Denemark 已提交
8131 8132 8133 8134
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
8135
    unsigned long long me = virThreadSelfID();
J
Jiri Denemark 已提交
8136 8137 8138 8139

    if (!priv->job.asyncJob)
        return;

8140 8141 8142 8143 8144
    VIR_DEBUG("Setting '%s' phase to '%s'",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob, phase));

    if (priv->job.asyncOwner && me != priv->job.asyncOwner) {
8145
        VIR_WARN("'%s' async job is owned by thread %llu",
8146 8147 8148 8149
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }

J
Jiri Denemark 已提交
8150
    priv->job.phase = phase;
8151
    priv->job.asyncOwner = me;
8152
    qemuDomainObjSaveStatus(driver, obj);
J
Jiri Denemark 已提交
8153 8154
}

8155
void
8156 8157
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
8158 8159 8160
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

8161 8162 8163 8164 8165 8166 8167
    if (!priv->job.asyncJob)
        return;

    priv->job.mask = allowedJobs | JOB_MASK(QEMU_JOB_DESTROY);
}

void
8168
qemuDomainObjDiscardAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
8169 8170 8171 8172 8173 8174
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
8175
    qemuDomainObjSaveStatus(driver, obj);
8176 8177
}

8178 8179 8180 8181 8182 8183 8184 8185 8186
void
qemuDomainObjReleaseAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    VIR_DEBUG("Releasing ownership of '%s' async job",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));

    if (priv->job.asyncOwner != virThreadSelfID()) {
8187
        VIR_WARN("'%s' async job is owned by thread %llu",
8188 8189 8190 8191 8192 8193
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }
    priv->job.asyncOwner = 0;
}

8194
static bool
8195
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
8196
{
8197 8198 8199
    return !priv->job.asyncJob ||
           job == QEMU_JOB_NONE ||
           (priv->job.mask & JOB_MASK(job)) != 0;
8200 8201
}

8202
bool
8203
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
8204 8205 8206 8207
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

8208 8209 8210 8211 8212 8213 8214 8215 8216 8217 8218
static bool
qemuDomainObjCanSetJob(qemuDomainObjPrivatePtr priv,
                       qemuDomainJob job,
                       qemuDomainAgentJob agentJob)
{
    return ((job == QEMU_JOB_NONE ||
             priv->job.active == QEMU_JOB_NONE) &&
            (agentJob == QEMU_AGENT_JOB_NONE ||
             priv->job.agentActive == QEMU_AGENT_JOB_NONE));
}

8219 8220 8221
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

8222 8223 8224 8225 8226 8227
/**
 * qemuDomainObjBeginJobInternal:
 * @driver: qemu driver
 * @obj: domain object
 * @job: qemuDomainJob to start
 * @asyncJob: qemuDomainAsyncJob to start
8228
 * @nowait: don't wait trying to acquire @job
8229 8230 8231 8232
 *
 * Acquires job for a domain object which must be locked before
 * calling. If there's already a job running waits up to
 * QEMU_JOB_WAIT_TIME after which the functions fails reporting
8233 8234 8235 8236 8237
 * an error unless @nowait is set.
 *
 * If @nowait is true this function tries to acquire job and if
 * it fails, then it returns immediately without waiting. No
 * error is reported in this case.
8238 8239 8240 8241 8242
 *
 * Returns: 0 on success,
 *         -2 if unable to start job because of timeout or
 *            maxQueuedJobs limit,
 *         -1 otherwise.
8243
 */
8244
static int ATTRIBUTE_NONNULL(1)
8245
qemuDomainObjBeginJobInternal(virQEMUDriverPtr driver,
8246
                              virDomainObjPtr obj,
8247
                              qemuDomainJob job,
8248
                              qemuDomainAgentJob agentJob,
8249 8250
                              qemuDomainAsyncJob asyncJob,
                              bool nowait)
8251 8252
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
8253
    unsigned long long now;
8254
    unsigned long long then;
8255
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
J
Jiri Denemark 已提交
8256
    bool async = job == QEMU_JOB_ASYNC;
8257
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8258
    const char *blocker = NULL;
8259
    const char *agentBlocker = NULL;
J
Jiri Denemark 已提交
8260
    int ret = -1;
J
Jiri Denemark 已提交
8261
    unsigned long long duration = 0;
8262
    unsigned long long agentDuration = 0;
J
Jiri Denemark 已提交
8263
    unsigned long long asyncDuration = 0;
8264

8265 8266 8267 8268 8269 8270
    VIR_DEBUG("Starting job: job=%s agentJob=%s asyncJob=%s "
              "(vm=%p name=%s, current job=%s agentJob=%s async=%s)",
              qemuDomainJobTypeToString(job),
              qemuDomainAgentJobTypeToString(agentJob),
              qemuDomainAsyncJobTypeToString(asyncJob),
              obj, obj->def->name,
J
Jiri Denemark 已提交
8271
              qemuDomainJobTypeToString(priv->job.active),
8272
              qemuDomainAgentJobTypeToString(priv->job.agentActive),
J
Jiri Denemark 已提交
8273
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
8274

8275 8276
    if (virTimeMillisNow(&now) < 0) {
        virObjectUnref(cfg);
8277
        return -1;
8278 8279
    }

8280
    priv->jobs_queued++;
J
Jiri Denemark 已提交
8281
    then = now + QEMU_JOB_WAIT_TIME;
8282

8283
 retry:
8284 8285
    if ((!async && job != QEMU_JOB_DESTROY) &&
        cfg->maxQueuedJobs &&
8286
        priv->jobs_queued > cfg->maxQueuedJobs) {
8287 8288 8289
        goto error;
    }

8290
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
8291 8292 8293
        if (nowait)
            goto cleanup;

8294
        VIR_DEBUG("Waiting for async job (vm=%p name=%s)", obj, obj->def->name);
8295
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->parent.lock, then) < 0)
8296 8297 8298
            goto error;
    }

8299
    while (!qemuDomainObjCanSetJob(priv, job, agentJob)) {
8300 8301 8302
        if (nowait)
            goto cleanup;

8303
        VIR_DEBUG("Waiting for job (vm=%p name=%s)", obj, obj->def->name);
8304
        if (virCondWaitUntil(&priv->job.cond, &obj->parent.lock, then) < 0)
8305
            goto error;
8306
    }
8307 8308 8309

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
8310
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
8311 8312
        goto retry;

J
Jiri Denemark 已提交
8313 8314
    ignore_value(virTimeMillisNow(&now));

8315 8316 8317 8318 8319 8320 8321 8322 8323 8324 8325 8326 8327 8328 8329 8330 8331 8332 8333 8334 8335 8336 8337 8338 8339 8340 8341 8342 8343 8344 8345 8346 8347 8348 8349 8350 8351 8352 8353 8354
    if (job) {
        qemuDomainObjResetJob(priv);

        if (job != QEMU_JOB_ASYNC) {
            VIR_DEBUG("Started job: %s (async=%s vm=%p name=%s)",
                      qemuDomainJobTypeToString(job),
                      qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                      obj, obj->def->name);
            priv->job.active = job;
            priv->job.owner = virThreadSelfID();
            priv->job.ownerAPI = virThreadJobGet();
            priv->job.started = now;
        } else {
            VIR_DEBUG("Started async job: %s (vm=%p name=%s)",
                      qemuDomainAsyncJobTypeToString(asyncJob),
                      obj, obj->def->name);
            qemuDomainObjResetAsyncJob(priv);
            if (VIR_ALLOC(priv->job.current) < 0)
                goto cleanup;
            priv->job.current->status = QEMU_DOMAIN_JOB_STATUS_ACTIVE;
            priv->job.asyncJob = asyncJob;
            priv->job.asyncOwner = virThreadSelfID();
            priv->job.asyncOwnerAPI = virThreadJobGet();
            priv->job.asyncStarted = now;
            priv->job.current->started = now;
        }
    }

    if (agentJob) {
        qemuDomainObjResetAgentJob(priv);

        VIR_DEBUG("Started agent job: %s (vm=%p name=%s job=%s async=%s)",
                  qemuDomainAgentJobTypeToString(agentJob),
                  obj, obj->def->name,
                  qemuDomainJobTypeToString(priv->job.active),
                  qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
        priv->job.agentActive = agentJob;
        priv->job.agentOwner = virThreadSelfID();
        priv->job.agentOwnerAPI = virThreadJobGet();
        priv->job.agentStarted = now;
8355
    }
8356

8357
    if (qemuDomainTrackJob(job))
8358
        qemuDomainObjSaveStatus(driver, obj);
8359

8360
    virObjectUnref(cfg);
8361
    return 0;
8362

8363
 error:
J
Jiri Denemark 已提交
8364 8365 8366
    ignore_value(virTimeMillisNow(&now));
    if (priv->job.active && priv->job.started)
        duration = now - priv->job.started;
8367 8368
    if (priv->job.agentActive && priv->job.agentStarted)
        agentDuration = now - priv->job.agentStarted;
J
Jiri Denemark 已提交
8369 8370 8371
    if (priv->job.asyncJob && priv->job.asyncStarted)
        asyncDuration = now - priv->job.asyncStarted;

8372 8373 8374 8375
    VIR_WARN("Cannot start job (%s, %s, %s) for domain %s; "
             "current job is (%s, %s, %s) "
             "owned by (%llu %s, %llu %s, %llu %s (flags=0x%lx)) "
             "for (%llus, %llus, %llus)",
8376
             qemuDomainJobTypeToString(job),
8377
             qemuDomainAgentJobTypeToString(agentJob),
8378 8379 8380
             qemuDomainAsyncJobTypeToString(asyncJob),
             obj->def->name,
             qemuDomainJobTypeToString(priv->job.active),
8381
             qemuDomainAgentJobTypeToString(priv->job.agentActive),
8382
             qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
8383
             priv->job.owner, NULLSTR(priv->job.ownerAPI),
8384
             priv->job.agentOwner, NULLSTR(priv->job.agentOwnerAPI),
J
Jiri Denemark 已提交
8385
             priv->job.asyncOwner, NULLSTR(priv->job.asyncOwnerAPI),
8386
             priv->job.apiFlags,
8387
             duration / 1000, agentDuration / 1000, asyncDuration / 1000);
8388

8389 8390 8391 8392 8393 8394 8395 8396 8397
    if (job) {
        if (nested || qemuDomainNestedJobAllowed(priv, job))
            blocker = priv->job.ownerAPI;
        else
            blocker = priv->job.asyncOwnerAPI;
    }

    if (agentJob)
        agentBlocker = priv->job.agentOwnerAPI;
8398

8399
    if (errno == ETIMEDOUT) {
8400 8401 8402 8403 8404 8405
        if (blocker && agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by monitor=%s agent=%s)"),
                           blocker, agentBlocker);
        } else if (blocker) {
8406
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
8407 8408
                           _("cannot acquire state change "
                             "lock (held by monitor=%s)"),
8409
                           blocker);
8410 8411 8412 8413 8414
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by agent=%s)"),
                           agentBlocker);
8415 8416 8417 8418
        } else {
            virReportError(VIR_ERR_OPERATION_TIMEOUT, "%s",
                           _("cannot acquire state change lock"));
        }
8419 8420 8421
        ret = -2;
    } else if (cfg->maxQueuedJobs &&
               priv->jobs_queued > cfg->maxQueuedJobs) {
8422 8423 8424 8425 8426 8427 8428
        if (blocker && agentBlocker) {
            virReportError(VIR_ERR_OPERATION_FAILED,
                           _("cannot acquire state change "
                             "lock (held by monitor=%s agent=%s) "
                             "due to max_queued limit"),
                           blocker, agentBlocker);
        } else if (blocker) {
8429
            virReportError(VIR_ERR_OPERATION_FAILED,
8430 8431
                           _("cannot acquire state change "
                             "lock (held by monitor=%s) "
8432 8433
                             "due to max_queued limit"),
                           blocker);
8434 8435 8436 8437 8438 8439
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_FAILED,
                           _("cannot acquire state change "
                             "lock (held by agent=%s) "
                             "due to max_queued limit"),
                           agentBlocker);
8440 8441 8442 8443 8444
        } else {
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("cannot acquire state change lock "
                             "due to max_queued limit"));
        }
8445 8446
        ret = -2;
    } else {
8447
        virReportSystemError(errno, "%s", _("cannot acquire job mutex"));
8448
    }
J
Jiri Denemark 已提交
8449 8450

 cleanup:
8451
    priv->jobs_queued--;
8452
    virObjectUnref(cfg);
8453
    return ret;
8454 8455 8456
}

/*
8457
 * obj must be locked before calling
8458 8459 8460 8461
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
8462
 * Successful calls must be followed by EndJob eventually
8463
 */
8464
int qemuDomainObjBeginJob(virQEMUDriverPtr driver,
8465
                          virDomainObjPtr obj,
8466
                          qemuDomainJob job)
8467
{
8468
    if (qemuDomainObjBeginJobInternal(driver, obj, job,
8469
                                      QEMU_AGENT_JOB_NONE,
8470
                                      QEMU_ASYNC_JOB_NONE, false) < 0)
8471 8472 8473
        return -1;
    else
        return 0;
8474 8475
}

8476 8477 8478 8479 8480 8481 8482 8483 8484 8485 8486 8487 8488 8489 8490 8491 8492 8493 8494 8495 8496 8497 8498 8499 8500 8501 8502 8503 8504 8505 8506 8507 8508 8509 8510 8511 8512
/**
 * qemuDomainObjBeginAgentJob:
 *
 * Grabs agent type of job. Use if caller talks to guest agent only.
 *
 * To end job call qemuDomainObjEndAgentJob.
 */
int
qemuDomainObjBeginAgentJob(virQEMUDriverPtr driver,
                           virDomainObjPtr obj,
                           qemuDomainAgentJob agentJob)
{
    return qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_NONE,
                                         agentJob,
                                         QEMU_ASYNC_JOB_NONE, false);
}

/**
 * qemuDomainObjBeginJobWithAgent:
 *
 * Grabs both monitor and agent types of job. Use if caller talks to
 * both monitor and guest agent. However, if @job (or @agentJob) is
 * QEMU_JOB_NONE (or QEMU_AGENT_JOB_NONE) only agent job is acquired (or
 * monitor job).
 *
 * To end job call qemuDomainObjEndJobWithAgent.
 */
int
qemuDomainObjBeginJobWithAgent(virQEMUDriverPtr driver,
                               virDomainObjPtr obj,
                               qemuDomainJob job,
                               qemuDomainAgentJob agentJob)
{
    return qemuDomainObjBeginJobInternal(driver, obj, job, agentJob,
                                         QEMU_ASYNC_JOB_NONE, false);
}

8513
int qemuDomainObjBeginAsyncJob(virQEMUDriverPtr driver,
8514
                               virDomainObjPtr obj,
8515
                               qemuDomainAsyncJob asyncJob,
8516 8517
                               virDomainJobOperation operation,
                               unsigned long apiFlags)
8518
{
8519 8520
    qemuDomainObjPrivatePtr priv;

8521
    if (qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_ASYNC,
8522
                                      QEMU_AGENT_JOB_NONE,
8523
                                      asyncJob, false) < 0)
8524
        return -1;
8525 8526 8527

    priv = obj->privateData;
    priv->job.current->operation = operation;
8528
    priv->job.apiFlags = apiFlags;
8529
    return 0;
8530 8531
}

8532
int
8533 8534
qemuDomainObjBeginNestedJob(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
8535
                            qemuDomainAsyncJob asyncJob)
8536 8537 8538 8539 8540
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (asyncJob != priv->job.asyncJob) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
8541 8542
                       _("unexpected async job %d type expected %d"),
                       asyncJob, priv->job.asyncJob);
8543 8544 8545 8546
        return -1;
    }

    if (priv->job.asyncOwner != virThreadSelfID()) {
8547
        VIR_WARN("This thread doesn't seem to be the async job owner: %llu",
8548 8549 8550 8551 8552
                 priv->job.asyncOwner);
    }

    return qemuDomainObjBeginJobInternal(driver, obj,
                                         QEMU_JOB_ASYNC_NESTED,
8553
                                         QEMU_AGENT_JOB_NONE,
8554 8555
                                         QEMU_ASYNC_JOB_NONE,
                                         false);
8556 8557
}

8558 8559 8560 8561 8562 8563 8564 8565 8566 8567 8568 8569 8570 8571 8572 8573 8574 8575 8576
/**
 * qemuDomainObjBeginJobNowait:
 *
 * @driver: qemu driver
 * @obj: domain object
 * @job: qemuDomainJob to start
 *
 * Acquires job for a domain object which must be locked before
 * calling. If there's already a job running it returns
 * immediately without any error reported.
 *
 * Returns: see qemuDomainObjBeginJobInternal
 */
int
qemuDomainObjBeginJobNowait(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
                            qemuDomainJob job)
{
    return qemuDomainObjBeginJobInternal(driver, obj, job,
8577
                                         QEMU_AGENT_JOB_NONE,
8578 8579
                                         QEMU_ASYNC_JOB_NONE, true);
}
8580

8581
/*
8582
 * obj must be locked and have a reference before calling
8583 8584 8585 8586
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 */
8587 8588
void
qemuDomainObjEndJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
8589 8590
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
8591
    qemuDomainJob job = priv->job.active;
8592

8593 8594
    priv->jobs_queued--;

8595
    VIR_DEBUG("Stopping job: %s (async=%s vm=%p name=%s)",
8596
              qemuDomainJobTypeToString(job),
8597 8598
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
8599

8600
    qemuDomainObjResetJob(priv);
8601
    if (qemuDomainTrackJob(job))
8602
        qemuDomainObjSaveStatus(driver, obj);
8603 8604 8605 8606 8607 8608 8609 8610 8611 8612 8613 8614 8615 8616 8617 8618 8619 8620 8621 8622 8623 8624 8625 8626 8627 8628 8629 8630 8631 8632 8633 8634 8635 8636 8637 8638 8639 8640 8641 8642 8643 8644 8645
    /* We indeed need to wake up ALL threads waiting because
     * grabbing a job requires checking more variables. */
    virCondBroadcast(&priv->job.cond);
}

void
qemuDomainObjEndAgentJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
    qemuDomainAgentJob agentJob = priv->job.agentActive;

    priv->jobs_queued--;

    VIR_DEBUG("Stopping agent job: %s (async=%s vm=%p name=%s)",
              qemuDomainAgentJobTypeToString(agentJob),
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);

    qemuDomainObjResetAgentJob(priv);
    /* We indeed need to wake up ALL threads waiting because
     * grabbing a job requires checking more variables. */
    virCondBroadcast(&priv->job.cond);
}

void
qemuDomainObjEndJobWithAgent(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
    qemuDomainJob job = priv->job.active;
    qemuDomainAgentJob agentJob = priv->job.agentActive;

    priv->jobs_queued--;

    VIR_DEBUG("Stopping both jobs: %s %s (async=%s vm=%p name=%s)",
              qemuDomainJobTypeToString(job),
              qemuDomainAgentJobTypeToString(agentJob),
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAgentJob(priv);
    if (qemuDomainTrackJob(job))
8646
        qemuDomainObjSaveStatus(driver, obj);
8647 8648 8649
    /* We indeed need to wake up ALL threads waiting because
     * grabbing a job requires checking more variables. */
    virCondBroadcast(&priv->job.cond);
8650 8651
}

8652
void
8653
qemuDomainObjEndAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
8654 8655
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
8656

8657 8658
    priv->jobs_queued--;

8659 8660 8661
    VIR_DEBUG("Stopping async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
8662

8663
    qemuDomainObjResetAsyncJob(priv);
8664
    qemuDomainObjSaveStatus(driver, obj);
8665 8666 8667
    virCondBroadcast(&priv->job.asyncCond);
}

8668 8669 8670 8671 8672
void
qemuDomainObjAbortAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

8673 8674 8675
    VIR_DEBUG("Requesting abort of async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
8676

8677 8678
    priv->job.abortJob = true;
    virDomainObjBroadcast(obj);
8679 8680
}

8681 8682 8683 8684
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU monitor API call
8685 8686 8687
 * Must have already either called qemuDomainObjBeginJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active; may not be used for nested async jobs.
8688 8689 8690
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
8691
static int
8692
qemuDomainObjEnterMonitorInternal(virQEMUDriverPtr driver,
8693
                                  virDomainObjPtr obj,
8694
                                  qemuDomainAsyncJob asyncJob)
8695 8696 8697
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

8698
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
8699 8700 8701
        int ret;
        if ((ret = qemuDomainObjBeginNestedJob(driver, obj, asyncJob)) < 0)
            return ret;
8702
        if (!virDomainObjIsActive(obj)) {
8703 8704
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
8705
            qemuDomainObjEndJob(driver, obj);
8706 8707
            return -1;
        }
8708 8709 8710
    } else if (priv->job.asyncOwner == virThreadSelfID()) {
        VIR_WARN("This thread seems to be the async job owner; entering"
                 " monitor without asking for a nested job is dangerous");
8711 8712 8713 8714 8715
    } else if (priv->job.owner != virThreadSelfID()) {
        VIR_WARN("Entering a monitor without owning a job. "
                 "Job %s owner %s (%llu)",
                 qemuDomainJobTypeToString(priv->job.active),
                 priv->job.ownerAPI, priv->job.owner);
8716 8717
    }

8718 8719
    VIR_DEBUG("Entering monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
8720
    virObjectLock(priv->mon);
8721
    virObjectRef(priv->mon);
8722
    ignore_value(virTimeMillisNow(&priv->monStart));
8723
    virObjectUnlock(obj);
8724 8725

    return 0;
8726 8727
}

8728
static void ATTRIBUTE_NONNULL(1)
8729
qemuDomainObjExitMonitorInternal(virQEMUDriverPtr driver,
8730
                                 virDomainObjPtr obj)
8731 8732
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
8733
    bool hasRefs;
8734

8735
    hasRefs = virObjectUnref(priv->mon);
8736

8737
    if (hasRefs)
8738
        virObjectUnlock(priv->mon);
8739

8740
    virObjectLock(obj);
8741 8742
    VIR_DEBUG("Exited monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
8743

8744
    priv->monStart = 0;
8745
    if (!hasRefs)
8746
        priv->mon = NULL;
8747

J
Jiri Denemark 已提交
8748 8749
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjEndJob(driver, obj);
8750 8751
}

8752
void qemuDomainObjEnterMonitor(virQEMUDriverPtr driver,
8753
                               virDomainObjPtr obj)
8754
{
8755
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, obj,
8756
                                                   QEMU_ASYNC_JOB_NONE));
8757 8758
}

8759
/* obj must NOT be locked before calling
8760 8761
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
8762 8763 8764 8765 8766 8767
 *
 * Returns -1 if the domain is no longer alive after exiting the monitor.
 * In that case, the caller should be careful when using obj's data,
 * e.g. the live definition in vm->def has been freed by qemuProcessStop
 * and replaced by the persistent definition, so pointers stolen
 * from the live definition could no longer be valid.
8768
 */
8769 8770
int qemuDomainObjExitMonitor(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
8771
{
8772
    qemuDomainObjExitMonitorInternal(driver, obj);
8773
    if (!virDomainObjIsActive(obj)) {
8774
        if (virGetLastErrorCode() == VIR_ERR_OK)
8775 8776
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
8777 8778 8779
        return -1;
    }
    return 0;
8780
}
8781 8782

/*
8783
 * obj must be locked before calling
8784 8785
 *
 * To be called immediately before any QEMU monitor API call.
8786
 * Must have already either called qemuDomainObjBeginJob()
8787 8788 8789 8790 8791
 * and checked that the VM is still active, with asyncJob of
 * QEMU_ASYNC_JOB_NONE; or already called qemuDomainObjBeginAsyncJob,
 * with the same asyncJob.
 *
 * Returns 0 if job was started, in which case this must be followed with
8792 8793 8794
 * qemuDomainObjExitMonitor(); -2 if waiting for the nested job times out;
 * or -1 if the job could not be started (probably because the vm exited
 * in the meantime).
8795 8796
 */
int
8797
qemuDomainObjEnterMonitorAsync(virQEMUDriverPtr driver,
8798
                               virDomainObjPtr obj,
8799
                               qemuDomainAsyncJob asyncJob)
8800
{
8801
    return qemuDomainObjEnterMonitorInternal(driver, obj, asyncJob);
8802 8803
}

D
Daniel P. Berrange 已提交
8804

8805 8806 8807 8808
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU agent API call.
8809 8810 8811
 * Must have already called qemuDomainObjBeginAgentJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active.
8812 8813 8814
 *
 * To be followed with qemuDomainObjExitAgent() once complete
 */
8815
qemuAgentPtr
8816
qemuDomainObjEnterAgent(virDomainObjPtr obj)
D
Daniel P. Berrange 已提交
8817 8818
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
8819
    qemuAgentPtr agent = priv->agent;
D
Daniel P. Berrange 已提交
8820

8821 8822
    VIR_DEBUG("Entering agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
8823 8824 8825

    virObjectLock(agent);
    virObjectRef(agent);
8826
    virObjectUnlock(obj);
8827 8828

    return agent;
D
Daniel P. Berrange 已提交
8829 8830
}

8831 8832 8833 8834 8835 8836

/* obj must NOT be locked before calling
 *
 * Should be paired with an earlier qemuDomainObjEnterAgent() call
 */
void
8837
qemuDomainObjExitAgent(virDomainObjPtr obj, qemuAgentPtr agent)
D
Daniel P. Berrange 已提交
8838
{
8839 8840
    virObjectUnlock(agent);
    virObjectUnref(agent);
8841
    virObjectLock(obj);
D
Daniel P. Berrange 已提交
8842

8843 8844
    VIR_DEBUG("Exited agent (agent=%p vm=%p name=%s)",
              agent, obj, obj->def->name);
D
Daniel P. Berrange 已提交
8845 8846
}

8847
void qemuDomainObjEnterRemote(virDomainObjPtr obj)
8848
{
8849 8850
    VIR_DEBUG("Entering remote (vm=%p name=%s)",
              obj, obj->def->name);
8851
    virObjectUnlock(obj);
8852 8853
}

8854 8855 8856 8857

int
qemuDomainObjExitRemote(virDomainObjPtr obj,
                        bool checkActive)
8858
{
8859
    virObjectLock(obj);
8860 8861
    VIR_DEBUG("Exited remote (vm=%p name=%s)",
              obj, obj->def->name);
8862 8863 8864 8865 8866 8867 8868 8869 8870

    if (checkActive && !virDomainObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("domain '%s' is not running"),
                       obj->def->name);
        return -1;
    }

    return 0;
8871
}
8872 8873


8874 8875
static virDomainDefPtr
qemuDomainDefFromXML(virQEMUDriverPtr driver,
8876
                     virQEMUCapsPtr qemuCaps,
8877 8878 8879 8880 8881 8882 8883 8884
                     const char *xml)
{
    virCapsPtr caps;
    virDomainDefPtr def;

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        return NULL;

8885
    def = virDomainDefParseString(xml, caps, driver->xmlopt, qemuCaps,
8886 8887 8888 8889 8890 8891 8892 8893
                                  VIR_DOMAIN_DEF_PARSE_INACTIVE |
                                  VIR_DOMAIN_DEF_PARSE_SKIP_VALIDATE);

    virObjectUnref(caps);
    return def;
}


8894 8895
virDomainDefPtr
qemuDomainDefCopy(virQEMUDriverPtr driver,
8896
                  virQEMUCapsPtr qemuCaps,
8897 8898 8899 8900
                  virDomainDefPtr src,
                  unsigned int flags)
{
    virDomainDefPtr ret = NULL;
8901
    char *xml;
8902

8903
    if (!(xml = qemuDomainDefFormatXML(driver, qemuCaps, src, flags)))
8904
        return NULL;
8905

8906
    ret = qemuDomainDefFromXML(driver, qemuCaps, xml);
8907 8908 8909 8910 8911

    VIR_FREE(xml);
    return ret;
}

8912 8913 8914

static int
qemuDomainDefFormatBufInternal(virQEMUDriverPtr driver,
8915
                               virQEMUCapsPtr qemuCaps,
8916 8917 8918 8919
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags,
                               virBuffer *buf)
8920
{
8921
    int ret = -1;
8922
    virDomainDefPtr copy = NULL;
8923 8924
    virCapsPtr caps = NULL;

8925 8926
    virCheckFlags(VIR_DOMAIN_XML_COMMON_FLAGS | VIR_DOMAIN_XML_UPDATE_CPU, -1);

8927 8928
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;
8929

8930 8931 8932
    if (!(flags & (VIR_DOMAIN_XML_UPDATE_CPU | VIR_DOMAIN_XML_MIGRATABLE)))
        goto format;

8933
    if (!(copy = virDomainDefCopy(def, caps, driver->xmlopt, qemuCaps,
8934 8935 8936 8937 8938
                                  flags & VIR_DOMAIN_XML_MIGRATABLE)))
        goto cleanup;

    def = copy;

8939
    /* Update guest CPU requirements according to host CPU */
8940
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) &&
8941
        def->cpu &&
8942 8943
        (def->cpu->mode != VIR_CPU_MODE_CUSTOM ||
         def->cpu->model)) {
8944
        g_autoptr(virQEMUCaps) qCaps = NULL;
8945 8946 8947 8948 8949 8950 8951 8952 8953

        if (qemuCaps) {
            qCaps = virObjectRef(qemuCaps);
        } else {
            if (!(qCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
                                                     def->emulator,
                                                     def->os.machine)))
                goto cleanup;
        }
8954 8955

        if (virCPUUpdate(def->os.arch, def->cpu,
8956
                         virQEMUCapsGetHostModel(qCaps, def->virtType,
8957
                                                 VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE)) < 0)
8958 8959 8960
            goto cleanup;
    }

8961
    if ((flags & VIR_DOMAIN_XML_MIGRATABLE)) {
8962
        size_t i;
8963
        int toremove = 0;
8964
        virDomainControllerDefPtr usb = NULL, pci = NULL;
8965 8966 8967 8968 8969 8970 8971 8972 8973 8974 8975 8976 8977 8978 8979

        /* If only the default USB controller is present, we can remove it
         * and make the XML compatible with older versions of libvirt which
         * didn't support USB controllers in the XML but always added the
         * default one to qemu anyway.
         */
        for (i = 0; i < def->ncontrollers; i++) {
            if (def->controllers[i]->type == VIR_DOMAIN_CONTROLLER_TYPE_USB) {
                if (usb) {
                    usb = NULL;
                    break;
                }
                usb = def->controllers[i];
            }
        }
8980 8981 8982 8983 8984 8985 8986 8987 8988

        /* In order to maintain compatibility with version of libvirt that
         * didn't support <controller type='usb'/> (<= 0.9.4), we need to
         * drop the default USB controller, ie. a USB controller at index
         * zero with no model or with the default piix3-ohci model.
         *
         * However, we only need to do so for x86 i440fx machine types,
         * because other architectures and machine types were introduced
         * when libvirt already supported <controller type='usb'/>.
8989
         */
8990
        if (qemuDomainIsI440FX(def) &&
8991
            usb && usb->idx == 0 &&
8992
            (usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT ||
8993 8994
             usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_PIIX3_UHCI) &&
            !virDomainDeviceAliasIsUserAlias(usb->info.alias)) {
8995 8996
            VIR_DEBUG("Removing default USB controller from domain '%s'"
                      " for migration compatibility", def->name);
8997
            toremove++;
8998 8999 9000 9001 9002 9003 9004 9005 9006 9007 9008 9009 9010 9011 9012 9013 9014
        } else {
            usb = NULL;
        }

        /* Remove the default PCI controller if there is only one present
         * and its model is pci-root */
        for (i = 0; i < def->ncontrollers; i++) {
            if (def->controllers[i]->type == VIR_DOMAIN_CONTROLLER_TYPE_PCI) {
                if (pci) {
                    pci = NULL;
                    break;
                }
                pci = def->controllers[i];
            }
        }

        if (pci && pci->idx == 0 &&
9015 9016 9017
            pci->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT &&
            !virDomainDeviceAliasIsUserAlias(pci->info.alias) &&
            !pci->opts.pciopts.pcihole64) {
L
Laine Stump 已提交
9018
            VIR_DEBUG("Removing default pci-root from domain '%s'"
9019
                      " for migration compatibility", def->name);
9020
            toremove++;
9021 9022 9023 9024
        } else {
            pci = NULL;
        }

9025
        if (toremove) {
9026 9027 9028
            virDomainControllerDefPtr *controllers = def->controllers;
            int ncontrollers = def->ncontrollers;

9029
            if (VIR_ALLOC_N(def->controllers, ncontrollers - toremove) < 0) {
9030
                def->controllers = controllers;
9031 9032 9033 9034 9035
                goto cleanup;
            }

            def->ncontrollers = 0;
            for (i = 0; i < ncontrollers; i++) {
9036
                if (controllers[i] != usb && controllers[i] != pci)
9037 9038
                    def->controllers[def->ncontrollers++] = controllers[i];
            }
9039 9040 9041 9042

            VIR_FREE(controllers);
            virDomainControllerDefFree(pci);
            virDomainControllerDefFree(usb);
9043
        }
9044

9045 9046 9047 9048 9049 9050 9051 9052 9053
        /* Remove the panic device for selected models if present */
        for (i = 0; i < def->npanics; i++) {
            if (def->panics[i]->model == VIR_DOMAIN_PANIC_MODEL_S390 ||
                def->panics[i]->model == VIR_DOMAIN_PANIC_MODEL_PSERIES) {
                VIR_DELETE_ELEMENT(def->panics, i, def->npanics);
                break;
            }
        }

9054 9055 9056 9057
        for (i = 0; i < def->nchannels; i++) {
            if (qemuDomainChrDefDropDefaultPath(def->channels[i], driver) < 0)
                goto cleanup;
        }
9058

9059 9060 9061 9062 9063 9064 9065 9066 9067
        for (i = 0; i < def->nserials; i++) {
            virDomainChrDefPtr serial = def->serials[i];

            /* Historically, the native console type for some machine types
             * was not set at all, which means it defaulted to ISA even
             * though that was not even remotely accurate. To ensure migration
             * towards older libvirt versions works for such guests, we switch
             * it back to the default here */
            if (flags & VIR_DOMAIN_XML_MIGRATABLE) {
9068
                switch ((virDomainChrSerialTargetType)serial->targetType) {
9069
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
9070
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
9071 9072 9073 9074 9075 9076
                    serial->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE;
                    serial->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_NONE;
                    break;
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA:
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI:
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB:
9077
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
9078 9079 9080 9081 9082 9083 9084 9085
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
                    /* Nothing to do */
                    break;
                }
            }
        }

9086 9087 9088 9089 9090 9091 9092 9093 9094
        /* Replace the CPU definition updated according to QEMU with the one
         * used for starting the domain. The updated def will be sent
         * separately for backward compatibility.
         */
        if (origCPU) {
            virCPUDefFree(def->cpu);
            if (!(def->cpu = virCPUDefCopy(origCPU)))
                goto cleanup;
        }
9095 9096
    }

9097 9098
 format:
    ret = virDomainDefFormatInternal(def, caps,
9099
                                     virDomainDefFormatConvertXMLFlags(flags),
9100
                                     buf, driver->xmlopt);
9101

9102
 cleanup:
9103
    virDomainDefFree(copy);
9104
    virObjectUnref(caps);
9105 9106
    return ret;
}
9107

9108 9109 9110

int
qemuDomainDefFormatBuf(virQEMUDriverPtr driver,
9111
                       virQEMUCapsPtr qemuCaps,
9112 9113 9114 9115
                       virDomainDefPtr def,
                       unsigned int flags,
                       virBufferPtr buf)
{
9116
    return qemuDomainDefFormatBufInternal(driver, qemuCaps, def, NULL, flags, buf);
9117 9118 9119 9120 9121
}


static char *
qemuDomainDefFormatXMLInternal(virQEMUDriverPtr driver,
9122
                               virQEMUCapsPtr qemuCaps,
9123 9124 9125
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags)
9126 9127 9128
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

9129
    if (qemuDomainDefFormatBufInternal(driver, qemuCaps, def, origCPU, flags, &buf) < 0)
9130 9131 9132 9133 9134
        return NULL;

    return virBufferContentAndReset(&buf);
}

9135 9136 9137

char *
qemuDomainDefFormatXML(virQEMUDriverPtr driver,
9138
                       virQEMUCapsPtr qemuCaps,
9139 9140 9141
                       virDomainDefPtr def,
                       unsigned int flags)
{
9142
    return qemuDomainDefFormatXMLInternal(driver, qemuCaps, def, NULL, flags);
9143 9144 9145
}


9146
char *qemuDomainFormatXML(virQEMUDriverPtr driver,
9147
                          virDomainObjPtr vm,
9148
                          unsigned int flags)
9149 9150
{
    virDomainDefPtr def;
9151 9152
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCPUDefPtr origCPU = NULL;
9153

9154
    if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef) {
9155
        def = vm->newDef;
9156
    } else {
9157
        def = vm->def;
9158
        origCPU = priv->origCPU;
9159
    }
9160

9161
    return qemuDomainDefFormatXMLInternal(driver, priv->qemuCaps, def, origCPU, flags);
9162 9163
}

9164
char *
9165
qemuDomainDefFormatLive(virQEMUDriverPtr driver,
9166
                        virQEMUCapsPtr qemuCaps,
9167
                        virDomainDefPtr def,
9168
                        virCPUDefPtr origCPU,
9169 9170
                        bool inactive,
                        bool compatible)
9171 9172 9173 9174 9175
{
    unsigned int flags = QEMU_DOMAIN_FORMAT_LIVE_FLAGS;

    if (inactive)
        flags |= VIR_DOMAIN_XML_INACTIVE;
9176 9177
    if (compatible)
        flags |= VIR_DOMAIN_XML_MIGRATABLE;
9178

9179
    return qemuDomainDefFormatXMLInternal(driver, qemuCaps, def, origCPU, flags);
9180 9181
}

9182

9183
void qemuDomainObjTaint(virQEMUDriverPtr driver,
9184
                        virDomainObjPtr obj,
9185
                        virDomainTaintFlags taint,
9186
                        qemuDomainLogContextPtr logCtxt)
9187
{
9188
    virErrorPtr orig_err = NULL;
9189 9190
    char *timestamp = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
9191
    int rc;
9192

9193 9194 9195 9196 9197 9198 9199 9200 9201 9202 9203 9204 9205 9206
    if (!virDomainObjTaint(obj, taint))
        return;

    virUUIDFormat(obj->def->uuid, uuidstr);

    VIR_WARN("Domain id=%d name='%s' uuid=%s is tainted: %s",
             obj->def->id,
             obj->def->name,
             uuidstr,
             virDomainTaintTypeToString(taint));

    /* We don't care about errors logging taint info, so
     * preserve original error, and clear any error that
     * is raised */
9207
    virErrorPreserveLast(&orig_err);
9208

9209 9210 9211
    if (!(timestamp = virTimeStringNow()))
        goto cleanup;

9212 9213 9214 9215 9216 9217 9218 9219 9220 9221 9222 9223
    if (logCtxt) {
        rc = qemuDomainLogContextWrite(logCtxt,
                                       "%s: Domain id=%d is tainted: %s\n",
                                       timestamp,
                                       obj->def->id,
                                       virDomainTaintTypeToString(taint));
    } else {
        rc = qemuDomainLogAppendMessage(driver, obj,
                                        "%s: Domain id=%d is tainted: %s\n",
                                        timestamp,
                                        obj->def->id,
                                        virDomainTaintTypeToString(taint));
9224 9225
    }

9226
    if (rc < 0)
9227 9228 9229 9230
        virResetLastError();

 cleanup:
    VIR_FREE(timestamp);
9231
    virErrorRestore(&orig_err);
9232 9233 9234
}


9235
void qemuDomainObjCheckTaint(virQEMUDriverPtr driver,
9236
                             virDomainObjPtr obj,
9237
                             qemuDomainLogContextPtr logCtxt)
9238
{
9239
    size_t i;
9240
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
9241
    qemuDomainObjPrivatePtr priv = obj->privateData;
9242
    bool custom_hypervisor_feat = false;
9243

9244
    if (virQEMUDriverIsPrivileged(driver) &&
9245 9246 9247
        (!cfg->clearEmulatorCapabilities ||
         cfg->user == 0 ||
         cfg->group == 0))
9248
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
9249

9250
    if (priv->hookRun)
9251
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOOK, logCtxt);
9252

9253
    if (obj->def->namespaceData) {
9254 9255
        qemuDomainXmlNsDefPtr qemuxmlns = obj->def->namespaceData;
        if (qemuxmlns->num_args || qemuxmlns->num_env)
9256
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logCtxt);
9257 9258 9259 9260
        if (qemuxmlns->ncapsadd > 0 || qemuxmlns->ncapsdel > 0)
            custom_hypervisor_feat = true;
    }

9261 9262
    if (custom_hypervisor_feat ||
        (cfg->capabilityfilters && *cfg->capabilityfilters)) {
9263 9264
        qemuDomainObjTaint(driver, obj,
                           VIR_DOMAIN_TAINT_CUSTOM_HYPERVISOR_FEATURE, logCtxt);
9265 9266
    }

9267
    if (obj->def->cpu && obj->def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
9268
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOST_CPU, logCtxt);
9269

9270
    for (i = 0; i < obj->def->ndisks; i++)
9271
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logCtxt);
9272

9273 9274
    for (i = 0; i < obj->def->nhostdevs; i++)
        qemuDomainObjCheckHostdevTaint(driver, obj, obj->def->hostdevs[i],
9275
                                       logCtxt);
9276

9277
    for (i = 0; i < obj->def->nnets; i++)
9278
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logCtxt);
9279

9280
    if (obj->def->os.dtb)
9281
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_DTB, logCtxt);
9282

9283
    virObjectUnref(cfg);
9284 9285 9286
}


9287
void qemuDomainObjCheckDiskTaint(virQEMUDriverPtr driver,
9288
                                 virDomainObjPtr obj,
9289
                                 virDomainDiskDefPtr disk,
9290
                                 qemuDomainLogContextPtr logCtxt)
9291
{
9292
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
9293

9294 9295
    if (disk->rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
9296
                           logCtxt);
9297

9298 9299
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
        virStorageSourceGetActualType(disk->src) == VIR_STORAGE_TYPE_BLOCK &&
9300
        disk->src->path && virFileIsCDROM(disk->src->path) == 1)
9301
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CDROM_PASSTHROUGH,
9302
                           logCtxt);
9303

9304
    virObjectUnref(cfg);
9305 9306 9307
}


9308 9309 9310
void qemuDomainObjCheckHostdevTaint(virQEMUDriverPtr driver,
                                    virDomainObjPtr obj,
                                    virDomainHostdevDefPtr hostdev,
9311
                                    qemuDomainLogContextPtr logCtxt)
9312
{
9313
    if (!virHostdevIsSCSIDevice(hostdev))
9314
        return;
9315

9316 9317
    if (hostdev->source.subsys.u.scsi.rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
9318 9319 9320
}


9321
void qemuDomainObjCheckNetTaint(virQEMUDriverPtr driver,
9322
                                virDomainObjPtr obj,
9323
                                virDomainNetDefPtr net,
9324
                                qemuDomainLogContextPtr logCtxt)
9325
{
9326 9327 9328 9329 9330 9331
    /* script is only useful for NET_TYPE_ETHERNET (qemu) and
     * NET_TYPE_BRIDGE (xen), but could be (incorrectly) specified for
     * any interface type. In any case, it's adding user sauce into
     * the soup, so it should taint the domain.
     */
    if (net->script != NULL)
9332
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logCtxt);
9333
}
9334 9335


9336 9337 9338 9339 9340 9341 9342
qemuDomainLogContextPtr qemuDomainLogContextNew(virQEMUDriverPtr driver,
                                                virDomainObjPtr vm,
                                                qemuDomainLogContextMode mode)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainLogContextPtr ctxt = NULL;

9343
    if (qemuDomainInitialize() < 0)
9344 9345 9346 9347
        goto cleanup;

    if (!(ctxt = virObjectNew(qemuDomainLogContextClass)))
        goto cleanup;
9348

9349
    VIR_DEBUG("Context new %p stdioLogD=%d", ctxt, cfg->stdioLogD);
9350 9351 9352
    ctxt->writefd = -1;
    ctxt->readfd = -1;

9353
    ctxt->path = g_strdup_printf("%s/%s.log", cfg->logDir, vm->def->name);
9354

9355 9356 9357 9358
    if (cfg->stdioLogD) {
        ctxt->manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver));
        if (!ctxt->manager)
            goto error;
9359

9360 9361 9362 9363
        ctxt->writefd = virLogManagerDomainOpenLogFile(ctxt->manager,
                                                       "qemu",
                                                       vm->def->uuid,
                                                       vm->def->name,
9364
                                                       ctxt->path,
9365 9366 9367 9368 9369 9370
                                                       0,
                                                       &ctxt->inode,
                                                       &ctxt->pos);
        if (ctxt->writefd < 0)
            goto error;
    } else {
9371
        if ((ctxt->writefd = open(ctxt->path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
9372
            virReportSystemError(errno, _("failed to create logfile %s"),
9373
                                 ctxt->path);
9374 9375
            goto error;
        }
9376
        if (virSetCloseExec(ctxt->writefd) < 0) {
9377
            virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
9378
                                 ctxt->path);
9379 9380 9381
            goto error;
        }

9382 9383 9384 9385 9386 9387 9388
        /* For unprivileged startup we must truncate the file since
         * we can't rely on logrotate. We don't use O_TRUNC since
         * it is better for SELinux policy if we truncate afterwards */
        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START &&
            !virQEMUDriverIsPrivileged(driver) &&
            ftruncate(ctxt->writefd, 0) < 0) {
            virReportSystemError(errno, _("failed to truncate %s"),
9389
                                 ctxt->path);
9390 9391 9392 9393
            goto error;
        }

        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START) {
9394
            if ((ctxt->readfd = open(ctxt->path, O_RDONLY, S_IRUSR | S_IWUSR)) < 0) {
9395
                virReportSystemError(errno, _("failed to open logfile %s"),
9396
                                     ctxt->path);
9397 9398 9399 9400
                goto error;
            }
            if (virSetCloseExec(ctxt->readfd) < 0) {
                virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
9401
                                     ctxt->path);
9402 9403 9404 9405
                goto error;
            }
        }

9406 9407
        if ((ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END)) < 0) {
            virReportSystemError(errno, _("failed to seek in log file %s"),
9408
                                 ctxt->path);
9409 9410 9411 9412
            goto error;
        }
    }

9413
 cleanup:
9414 9415 9416 9417
    virObjectUnref(cfg);
    return ctxt;

 error:
9418
    virObjectUnref(ctxt);
9419 9420
    ctxt = NULL;
    goto cleanup;
9421 9422 9423 9424 9425 9426 9427 9428 9429 9430 9431 9432 9433 9434
}


int qemuDomainLogContextWrite(qemuDomainLogContextPtr ctxt,
                              const char *fmt, ...)
{
    va_list argptr;
    char *message = NULL;
    int ret = -1;

    va_start(argptr, fmt);

    if (virVasprintf(&message, fmt, argptr) < 0)
        goto cleanup;
9435 9436
    if (!ctxt->manager &&
        lseek(ctxt->writefd, 0, SEEK_END) < 0) {
9437
        virReportSystemError(errno, "%s",
9438
                             _("Unable to seek to end of domain logfile"));
9439 9440 9441 9442 9443 9444 9445 9446 9447 9448 9449 9450 9451 9452 9453 9454 9455 9456 9457 9458
        goto cleanup;
    }
    if (safewrite(ctxt->writefd, message, strlen(message)) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to write to domain logfile"));
        goto cleanup;
    }

    ret = 0;

 cleanup:
    va_end(argptr);
    VIR_FREE(message);
    return ret;
}


ssize_t qemuDomainLogContextRead(qemuDomainLogContextPtr ctxt,
                                 char **msg)
{
9459 9460 9461 9462
    VIR_DEBUG("Context read %p manager=%p inode=%llu pos=%llu",
              ctxt, ctxt->manager,
              (unsigned long long)ctxt->inode,
              (unsigned long long)ctxt->pos);
9463
    char *buf;
9464 9465 9466
    size_t buflen;
    if (ctxt->manager) {
        buf = virLogManagerDomainReadLogFile(ctxt->manager,
9467
                                             ctxt->path,
9468 9469 9470 9471 9472 9473 9474 9475 9476
                                             ctxt->inode,
                                             ctxt->pos,
                                             1024 * 128,
                                             0);
        if (!buf)
            return -1;
        buflen = strlen(buf);
    } else {
        ssize_t got;
9477

9478
        buflen = 1024 * 128;
9479

9480 9481
        /* Best effort jump to start of messages */
        ignore_value(lseek(ctxt->readfd, ctxt->pos, SEEK_SET));
9482

9483 9484
        if (VIR_ALLOC_N(buf, buflen) < 0)
            return -1;
9485

9486 9487 9488 9489 9490 9491 9492 9493 9494 9495 9496 9497 9498
        got = saferead(ctxt->readfd, buf, buflen - 1);
        if (got < 0) {
            VIR_FREE(buf);
            virReportSystemError(errno, "%s",
                                 _("Unable to read from log file"));
            return -1;
        }

        buf[got] = '\0';

        ignore_value(VIR_REALLOC_N_QUIET(buf, got + 1));
        buflen = got;
    }
9499 9500 9501

    *msg = buf;

9502
    return buflen;
9503 9504 9505
}


9506 9507 9508 9509 9510 9511 9512 9513 9514 9515 9516 9517 9518 9519 9520 9521 9522 9523 9524 9525 9526 9527 9528 9529 9530 9531 9532
/**
 * qemuDomainLogAppendMessage:
 *
 * This is a best-effort attempt to add a log message to the qemu log file
 * either by using virtlogd or the legacy approach */
int
qemuDomainLogAppendMessage(virQEMUDriverPtr driver,
                           virDomainObjPtr vm,
                           const char *fmt,
                           ...)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    virLogManagerPtr manager = NULL;
    va_list ap;
    char *path = NULL;
    int writefd = -1;
    char *message = NULL;
    int ret = -1;

    va_start(ap, fmt);

    if (virVasprintf(&message, fmt, ap) < 0)
        goto cleanup;

    VIR_DEBUG("Append log message (vm='%s' message='%s) stdioLogD=%d",
              vm->def->name, message, cfg->stdioLogD);

9533
    path = g_strdup_printf("%s/%s.log", cfg->logDir, vm->def->name);
9534 9535 9536 9537 9538 9539 9540 9541 9542 9543 9544 9545 9546 9547 9548 9549 9550 9551 9552 9553 9554 9555 9556 9557 9558 9559 9560 9561 9562 9563 9564 9565 9566

    if (cfg->stdioLogD) {
        if (!(manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver))))
            goto cleanup;

        if (virLogManagerDomainAppendMessage(manager, "qemu", vm->def->uuid,
                                             vm->def->name, path, message, 0) < 0)
            goto cleanup;
    } else {
        if ((writefd = open(path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
            virReportSystemError(errno, _("failed to create logfile %s"),
                                 path);
            goto cleanup;
        }

        if (safewrite(writefd, message, strlen(message)) < 0)
            goto cleanup;
    }

    ret = 0;

 cleanup:
    va_end(ap);
    VIR_FREE(message);
    VIR_FORCE_CLOSE(writefd);
    virLogManagerFree(manager);
    virObjectUnref(cfg);
    VIR_FREE(path);

    return ret;
}


9567 9568 9569 9570 9571 9572 9573 9574
int qemuDomainLogContextGetWriteFD(qemuDomainLogContextPtr ctxt)
{
    return ctxt->writefd;
}


void qemuDomainLogContextMarkPosition(qemuDomainLogContextPtr ctxt)
{
9575 9576
    if (ctxt->manager)
        virLogManagerDomainGetLogFilePosition(ctxt->manager,
9577
                                              ctxt->path,
9578 9579 9580 9581 9582
                                              0,
                                              &ctxt->inode,
                                              &ctxt->pos);
    else
        ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END);
9583 9584 9585
}


9586 9587 9588 9589 9590 9591
virLogManagerPtr qemuDomainLogContextGetManager(qemuDomainLogContextPtr ctxt)
{
    return ctxt->manager;
}


9592 9593
/* Locate an appropriate 'qemu-img' binary.  */
const char *
9594
qemuFindQemuImgBinary(virQEMUDriverPtr driver)
9595
{
9596 9597
    if (!driver->qemuImgBinary)
        virReportError(VIR_ERR_INTERNAL_ERROR,
9598
                       "%s", _("unable to find qemu-img"));
9599 9600 9601 9602 9603 9604

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
9605
                                virDomainMomentObjPtr snapshot,
9606
                                virCapsPtr caps,
9607
                                virDomainXMLOptionPtr xmlopt,
9608
                                const char *snapshotDir)
9609 9610 9611 9612 9613 9614
{
    char *newxml = NULL;
    int ret = -1;
    char *snapDir = NULL;
    char *snapFile = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
9615 9616
    unsigned int flags = VIR_DOMAIN_SNAPSHOT_FORMAT_SECURE |
        VIR_DOMAIN_SNAPSHOT_FORMAT_INTERNAL;
9617
    virDomainSnapshotDefPtr def = virDomainSnapshotObjGetDef(snapshot);
9618

9619
    if (virDomainSnapshotGetCurrent(vm->snapshots) == snapshot)
9620
        flags |= VIR_DOMAIN_SNAPSHOT_FORMAT_CURRENT;
9621
    virUUIDFormat(vm->def->uuid, uuidstr);
9622
    newxml = virDomainSnapshotDefFormat(uuidstr, def, caps, xmlopt, flags);
9623
    if (newxml == NULL)
9624 9625
        return -1;

9626
    snapDir = g_strdup_printf("%s/%s", snapshotDir, vm->def->name);
9627 9628 9629 9630 9631 9632
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
        goto cleanup;
    }

9633
    snapFile = g_strdup_printf("%s/%s.xml", snapDir, def->parent.name);
9634

J
Ján Tomko 已提交
9635
    ret = virXMLSaveFile(snapFile, NULL, "snapshot-edit", newxml);
9636

9637
 cleanup:
9638 9639 9640 9641 9642 9643
    VIR_FREE(snapFile);
    VIR_FREE(snapDir);
    VIR_FREE(newxml);
    return ret;
}

9644

9645 9646
/* The domain is expected to be locked and inactive. Return -1 on normal
 * failure, 1 if we skipped a disk due to try_all.  */
9647
static int
9648
qemuDomainSnapshotForEachQcow2Raw(virQEMUDriverPtr driver,
9649 9650 9651 9652 9653
                                  virDomainDefPtr def,
                                  const char *name,
                                  const char *op,
                                  bool try_all,
                                  int ndisks)
9654 9655
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
9656
    size_t i;
9657 9658 9659 9660 9661 9662 9663 9664 9665
    bool skipped = false;

    qemuimgarg[0] = qemuFindQemuImgBinary(driver);
    if (qemuimgarg[0] == NULL) {
        /* qemuFindQemuImgBinary set the error */
        return -1;
    }

    qemuimgarg[2] = op;
9666
    qemuimgarg[3] = name;
9667

9668
    for (i = 0; i < ndisks; i++) {
9669
        /* FIXME: we also need to handle LVM here */
9670
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
9671 9672 9673
            int format = virDomainDiskGetFormat(def->disks[i]);

            if (format > 0 && format != VIR_STORAGE_FILE_QCOW2) {
9674 9675 9676 9677 9678
                if (try_all) {
                    /* Continue on even in the face of error, since other
                     * disks in this VM may have the same snapshot name.
                     */
                    VIR_WARN("skipping snapshot action on %s",
9679
                             def->disks[i]->dst);
9680 9681
                    skipped = true;
                    continue;
9682 9683 9684 9685 9686
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
9687
                }
9688 9689 9690 9691
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("Disk device '%s' does not support"
                                 " snapshotting"),
                               def->disks[i]->dst);
9692 9693 9694
                return -1;
            }

9695
            qemuimgarg[4] = virDomainDiskGetSource(def->disks[i]);
9696 9697 9698 9699

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
9700
                             def->disks[i]->dst);
9701 9702
                    skipped = true;
                    continue;
9703 9704 9705 9706 9707
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
9708 9709 9710 9711 9712 9713 9714 9715 9716
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

9717 9718 9719
/* The domain is expected to be locked and inactive. Return -1 on normal
 * failure, 1 if we skipped a disk due to try_all.  */
int
9720
qemuDomainSnapshotForEachQcow2(virQEMUDriverPtr driver,
9721
                               virDomainObjPtr vm,
9722
                               virDomainMomentObjPtr snap,
9723 9724 9725 9726 9727 9728
                               const char *op,
                               bool try_all)
{
    /* Prefer action on the disks in use at the time the snapshot was
     * created; but fall back to current definition if dealing with a
     * snapshot created prior to libvirt 0.9.5.  */
9729
    virDomainDefPtr def = snap->def->dom;
9730 9731 9732

    if (!def)
        def = vm->def;
9733
    return qemuDomainSnapshotForEachQcow2Raw(driver, def, snap->def->name,
9734 9735 9736
                                             op, try_all, def->ndisks);
}

9737 9738
/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
9739
qemuDomainSnapshotDiscard(virQEMUDriverPtr driver,
9740
                          virDomainObjPtr vm,
9741
                          virDomainMomentObjPtr snap,
9742
                          bool update_parent,
9743 9744 9745 9746 9747
                          bool metadata_only)
{
    char *snapFile = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;
9748
    virDomainMomentObjPtr parentsnap = NULL;
9749
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
9750 9751 9752 9753 9754 9755 9756 9757 9758

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
                goto cleanup;
        } else {
            priv = vm->privateData;
9759
            qemuDomainObjEnterMonitor(driver, vm);
9760
            /* we continue on even in the face of error */
9761
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
9762
            ignore_value(qemuDomainObjExitMonitor(driver, vm));
9763 9764 9765
        }
    }

9766 9767
    snapFile = g_strdup_printf("%s/%s/%s.xml", cfg->snapshotDir, vm->def->name,
                               snap->def->name);
9768

9769 9770
    if (snap == virDomainSnapshotGetCurrent(vm->snapshots)) {
        virDomainSnapshotSetCurrent(vm->snapshots, NULL);
9771
        if (update_parent && snap->def->parent_name) {
9772
            parentsnap = virDomainSnapshotFindByName(vm->snapshots,
9773
                                                     snap->def->parent_name);
9774 9775
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
9776
                         snap->def->parent_name);
9777
            } else {
9778
                virDomainSnapshotSetCurrent(vm->snapshots, parentsnap);
9779
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap, driver->caps,
9780
                                                    driver->xmlopt,
9781
                                                    cfg->snapshotDir) < 0) {
9782
                    VIR_WARN("failed to set parent snapshot '%s' as current",
9783
                             snap->def->parent_name);
9784
                    virDomainSnapshotSetCurrent(vm->snapshots, NULL);
9785 9786 9787 9788 9789 9790 9791
                }
            }
        }
    }

    if (unlink(snapFile) < 0)
        VIR_WARN("Failed to unlink %s", snapFile);
9792
    if (update_parent)
9793
        virDomainMomentDropParent(snap);
9794
    virDomainSnapshotObjListRemove(vm->snapshots, snap);
9795 9796 9797

    ret = 0;

9798
 cleanup:
9799
    VIR_FREE(snapFile);
9800
    virObjectUnref(cfg);
9801 9802 9803 9804
    return ret;
}

/* Hash iterator callback to discard multiple snapshots.  */
9805
int qemuDomainMomentDiscardAll(void *payload,
J
Ján Tomko 已提交
9806
                               const void *name G_GNUC_UNUSED,
9807
                               void *data)
9808
{
9809 9810
    virDomainMomentObjPtr moment = payload;
    virQEMUMomentRemovePtr curr = data;
9811 9812
    int err;

9813 9814 9815 9816
    if (!curr->found && curr->current == moment)
        curr->found = true;
    err = curr->momentDiscard(curr->driver, curr->vm, moment, false,
                              curr->metadata_only);
9817 9818
    if (err && !curr->err)
        curr->err = err;
9819
    return 0;
9820 9821 9822
}

int
9823
qemuDomainSnapshotDiscardAllMetadata(virQEMUDriverPtr driver,
9824 9825
                                     virDomainObjPtr vm)
{
9826 9827 9828
    virQEMUMomentRemove rem = {
        .driver = driver,
        .vm = vm,
9829 9830
        .metadata_only = true,
        .momentDiscard = qemuDomainSnapshotDiscard,
9831
    };
9832

9833
    virDomainSnapshotForEach(vm->snapshots, qemuDomainMomentDiscardAll, &rem);
9834
    virDomainSnapshotObjListRemoveAll(vm->snapshots);
9835 9836 9837 9838

    return rem.err;
}

9839

9840 9841 9842
static void
qemuDomainRemoveInactiveCommon(virQEMUDriverPtr driver,
                               virDomainObjPtr vm)
9843
{
9844
    virQEMUDriverConfigPtr cfg;
9845 9846
    g_autofree char *snapDir = NULL;
    g_autofree char *chkDir = NULL;
9847 9848

    cfg = virQEMUDriverGetConfig(driver);
9849

9850 9851 9852 9853
    /* Remove any snapshot metadata prior to removing the domain */
    if (qemuDomainSnapshotDiscardAllMetadata(driver, vm) < 0) {
        VIR_WARN("unable to remove all snapshots for domain %s",
                 vm->def->name);
9854 9855 9856 9857 9858
    } else {
        snapDir = g_strdup_printf("%s/%s", cfg->snapshotDir, vm->def->name);

        if (rmdir(snapDir) < 0 && errno != ENOENT)
            VIR_WARN("unable to remove snapshot directory %s", snapDir);
9859
    }
9860
    /* Remove any checkpoint metadata prior to removing the domain */
9861
    if (qemuCheckpointDiscardAllMetadata(driver, vm) < 0) {
9862 9863
        VIR_WARN("unable to remove all checkpoints for domain %s",
                 vm->def->name);
9864 9865 9866 9867 9868
    } else {
        chkDir = g_strdup_printf("%s/%s", cfg->checkpointDir,
                                 vm->def->name);
        if (rmdir(chkDir) < 0 && errno != ENOENT)
            VIR_WARN("unable to remove checkpoint directory %s", chkDir);
9869
    }
9870
    qemuExtDevicesCleanupHost(driver, vm->def);
9871

9872
    virObjectUnref(cfg);
9873 9874 9875
}


9876 9877 9878 9879 9880 9881 9882 9883 9884 9885 9886 9887 9888 9889 9890 9891 9892 9893 9894 9895
/**
 * qemuDomainRemoveInactive:
 *
 * The caller must hold a lock to the vm.
 */
void
qemuDomainRemoveInactive(virQEMUDriverPtr driver,
                         virDomainObjPtr vm)
{
    if (vm->persistent) {
        /* Short-circuit, we don't want to remove a persistent domain */
        return;
    }

    qemuDomainRemoveInactiveCommon(driver, vm);

    virDomainObjListRemove(driver->domains, vm);
}


9896 9897 9898 9899 9900 9901 9902 9903 9904 9905 9906 9907 9908 9909 9910 9911 9912 9913 9914 9915 9916 9917
/**
 * qemuDomainRemoveInactiveLocked:
 *
 * The caller must hold a lock to the vm and must hold the
 * lock on driver->domains in order to call the remove obj
 * from locked list method.
 */
static void
qemuDomainRemoveInactiveLocked(virQEMUDriverPtr driver,
                               virDomainObjPtr vm)
{
    if (vm->persistent) {
        /* Short-circuit, we don't want to remove a persistent domain */
        return;
    }

    qemuDomainRemoveInactiveCommon(driver, vm);

    virDomainObjListRemoveLocked(driver->domains, vm);
}


9918 9919 9920 9921 9922 9923 9924 9925 9926 9927 9928 9929 9930 9931 9932 9933 9934
/**
 * qemuDomainRemoveInactiveJob:
 *
 * Just like qemuDomainRemoveInactive but it tries to grab a
 * QEMU_JOB_MODIFY first. Even though it doesn't succeed in
 * grabbing the job the control carries with
 * qemuDomainRemoveInactive call.
 */
void
qemuDomainRemoveInactiveJob(virQEMUDriverPtr driver,
                            virDomainObjPtr vm)
{
    bool haveJob;

    haveJob = qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) >= 0;

    qemuDomainRemoveInactive(driver, vm);
9935 9936

    if (haveJob)
9937
        qemuDomainObjEndJob(driver, vm);
9938
}
9939

9940

9941 9942 9943 9944 9945 9946 9947 9948 9949 9950 9951 9952 9953 9954 9955 9956 9957 9958 9959 9960 9961
/**
 * qemuDomainRemoveInactiveJobLocked:
 *
 * Similar to qemuDomainRemoveInactiveJob, except that the caller must
 * also hold the lock @driver->domains
 */
void
qemuDomainRemoveInactiveJobLocked(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm)
{
    bool haveJob;

    haveJob = qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) >= 0;

    qemuDomainRemoveInactiveLocked(driver, vm);

    if (haveJob)
        qemuDomainObjEndJob(driver, vm);
}


9962
void
9963
qemuDomainSetFakeReboot(virQEMUDriverPtr driver,
9964 9965 9966 9967
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
9968
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
9969 9970

    if (priv->fakeReboot == value)
9971
        goto cleanup;
9972 9973 9974

    priv->fakeReboot = value;

9975
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
9976
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
9977

9978
 cleanup:
9979
    virObjectUnref(cfg);
9980
}
M
Michal Privoznik 已提交
9981

9982
static void
9983 9984
qemuDomainCheckRemoveOptionalDisk(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
9985
                                  size_t diskIndex)
9986 9987
{
    char uuid[VIR_UUID_STRING_BUFLEN];
9988
    virObjectEventPtr event = NULL;
9989
    virDomainDiskDefPtr disk = vm->def->disks[diskIndex];
9990
    const char *src = virDomainDiskGetSource(disk);
9991 9992 9993 9994 9995

    virUUIDFormat(vm->def->uuid, uuid);

    VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
              "due to inaccessible source '%s'",
9996
              disk->dst, vm->def->name, uuid, src);
9997 9998 9999 10000

    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM ||
        disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY) {

10001
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
10002 10003
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
10004
        virDomainDiskEmptySource(disk);
10005 10006
        /* keeping the old startup policy would be invalid for new images */
        disk->startupPolicy = VIR_DOMAIN_STARTUP_POLICY_DEFAULT;
10007
    } else {
10008
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
10009 10010
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_DROP_MISSING_ON_START);
10011 10012
        virDomainDiskRemove(vm->def, diskIndex);
        virDomainDiskDefFree(disk);
10013 10014
    }

10015
    virObjectEventStateQueue(driver->domainEventState, event);
10016 10017
}

10018 10019 10020 10021 10022 10023 10024 10025 10026 10027 10028 10029 10030 10031 10032 10033 10034

/**
 * qemuDomainCheckDiskStartupPolicy:
 * @driver: qemu driver object
 * @vm: domain object
 * @disk: index of disk to check
 * @cold_boot: true if a new VM is being started
 *
 * This function should be called when the source storage for a disk device is
 * missing. The function checks whether the startup policy for the disk allows
 * removal of the source (or disk) according to the state of the VM.
 *
 * The function returns 0 if the source or disk was dropped and -1 if the state
 * of the VM does not allow this. This function does not report errors, but
 * clears any reported error if 0 is returned.
 */
int
10035 10036
qemuDomainCheckDiskStartupPolicy(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
10037
                                 size_t diskIndex,
10038 10039
                                 bool cold_boot)
{
10040
    int startupPolicy = vm->def->disks[diskIndex]->startupPolicy;
10041
    int device = vm->def->disks[diskIndex]->device;
10042

10043
    switch ((virDomainStartupPolicy) startupPolicy) {
10044
        case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
10045 10046 10047 10048 10049 10050
            /* Once started with an optional disk, qemu saves its section
             * in the migration stream, so later, when restoring from it
             * we must make sure the sections match. */
            if (!cold_boot &&
                device != VIR_DOMAIN_DISK_DEVICE_FLOPPY &&
                device != VIR_DOMAIN_DISK_DEVICE_CDROM)
10051
                return -1;
10052 10053
            break;

10054
        case VIR_DOMAIN_STARTUP_POLICY_DEFAULT:
10055
        case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
10056
            return -1;
10057 10058

        case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
10059
            if (cold_boot)
10060
                return -1;
10061 10062 10063 10064 10065 10066 10067
            break;

        case VIR_DOMAIN_STARTUP_POLICY_LAST:
            /* this should never happen */
            break;
    }

10068
    qemuDomainCheckRemoveOptionalDisk(driver, vm, diskIndex);
10069
    virResetLastError();
10070 10071 10072
    return 0;
}

10073

10074 10075 10076 10077 10078 10079 10080 10081 10082 10083

/*
 * The vm must be locked when any of the following cleanup functions is
 * called.
 */
int
qemuDomainCleanupAdd(virDomainObjPtr vm,
                     qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
10084
    size_t i;
10085 10086 10087 10088 10089 10090 10091 10092 10093 10094

    VIR_DEBUG("vm=%s, cb=%p", vm->def->name, cb);

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[i] == cb)
            return 0;
    }

    if (VIR_RESIZE_N(priv->cleanupCallbacks,
                     priv->ncleanupCallbacks_max,
10095
                     priv->ncleanupCallbacks, 1) < 0)
10096 10097 10098 10099 10100 10101 10102 10103 10104 10105 10106
        return -1;

    priv->cleanupCallbacks[priv->ncleanupCallbacks++] = cb;
    return 0;
}

void
qemuDomainCleanupRemove(virDomainObjPtr vm,
                        qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
10107
    size_t i;
10108 10109 10110 10111

    VIR_DEBUG("vm=%s, cb=%p", vm->def->name, cb);

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
10112 10113 10114
        if (priv->cleanupCallbacks[i] == cb)
            VIR_DELETE_ELEMENT_INPLACE(priv->cleanupCallbacks,
                                       i, priv->ncleanupCallbacks);
10115 10116 10117 10118 10119 10120 10121 10122
    }

    VIR_SHRINK_N(priv->cleanupCallbacks,
                 priv->ncleanupCallbacks_max,
                 priv->ncleanupCallbacks_max - priv->ncleanupCallbacks);
}

void
10123
qemuDomainCleanupRun(virQEMUDriverPtr driver,
10124 10125 10126
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
10127
    size_t i;
10128 10129 10130 10131

    VIR_DEBUG("driver=%p, vm=%s", driver, vm->def->name);

    /* run cleanup callbacks in reverse order */
10132 10133
    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[priv->ncleanupCallbacks - (i + 1)])
10134 10135 10136 10137 10138 10139 10140
            priv->cleanupCallbacks[i](driver, vm);
    }

    VIR_FREE(priv->cleanupCallbacks);
    priv->ncleanupCallbacks = 0;
    priv->ncleanupCallbacks_max = 0;
}
10141

10142 10143 10144
static void
qemuDomainGetImageIds(virQEMUDriverConfigPtr cfg,
                      virDomainObjPtr vm,
10145
                      virStorageSourcePtr src,
10146
                      virStorageSourcePtr parentSrc,
10147 10148 10149 10150 10151 10152 10153 10154 10155 10156 10157 10158 10159 10160 10161 10162 10163 10164
                      uid_t *uid, gid_t *gid)
{
    virSecurityLabelDefPtr vmlabel;
    virSecurityDeviceLabelDefPtr disklabel;

    if (uid)
        *uid = -1;
    if (gid)
        *gid = -1;

    if (cfg) {
        if (uid)
            *uid = cfg->user;

        if (gid)
            *gid = cfg->group;
    }

10165 10166
    if (vm && (vmlabel = virDomainDefGetSecurityLabelDef(vm->def, "dac")) &&
        vmlabel->label)
10167 10168
        virParseOwnershipIds(vmlabel->label, uid, gid);

10169 10170 10171 10172 10173
    if (parentSrc &&
        (disklabel = virStorageSourceGetSecurityLabelDef(parentSrc, "dac")) &&
        disklabel->label)
        virParseOwnershipIds(disklabel->label, uid, gid);

10174
    if ((disklabel = virStorageSourceGetSecurityLabelDef(src, "dac")) &&
10175
        disklabel->label)
10176 10177 10178 10179
        virParseOwnershipIds(disklabel->label, uid, gid);
}


10180 10181 10182
int
qemuDomainStorageFileInit(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
10183 10184
                          virStorageSourcePtr src,
                          virStorageSourcePtr parent)
10185 10186 10187 10188 10189 10190
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    uid_t uid;
    gid_t gid;
    int ret = -1;

10191
    qemuDomainGetImageIds(cfg, vm, src, parent, &uid, &gid);
10192 10193 10194 10195 10196 10197 10198 10199 10200 10201 10202 10203

    if (virStorageFileInitAs(src, uid, gid) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
}


10204 10205 10206 10207 10208
char *
qemuDomainStorageAlias(const char *device, int depth)
{
    char *alias;

10209
    device = qemuAliasDiskDriveSkipPrefix(device);
10210 10211

    if (!depth)
10212
        alias = g_strdup(device);
10213
    else
10214
        alias = g_strdup_printf("%s.%d", device, depth);
10215 10216 10217 10218
    return alias;
}


10219 10220 10221 10222 10223 10224 10225 10226 10227 10228 10229 10230 10231 10232 10233 10234 10235 10236 10237 10238 10239 10240 10241 10242 10243 10244 10245 10246 10247 10248 10249 10250 10251 10252 10253 10254 10255 10256 10257 10258 10259 10260 10261 10262 10263 10264 10265
/**
 * qemuDomainStorageSourceValidateDepth:
 * @src: storage source chain to validate
 * @add: offsets the calculated number of images
 * @diskdst: optional disk target to use in error message
 *
 * The XML parser limits the maximum element nesting to 256 layers. As libvirt
 * reports the chain into the status and in some cases the config XML we must
 * validate that any user-provided chains will not exceed the XML nesting limit
 * when formatted to the XML.
 *
 * This function validates that the storage source chain starting @src is at
 * most 200 layers deep. @add modifies the calculated value to offset the number
 * to allow checking cases when new layers are going to be added to the chain.
 *
 * Returns 0 on success and -1 if the chain is too deep. Error is reported.
 */
int
qemuDomainStorageSourceValidateDepth(virStorageSourcePtr src,
                                     int add,
                                     const char *diskdst)
{
    virStorageSourcePtr n;
    size_t nlayers = 0;

    for (n = src; virStorageSourceIsBacking(n); n = n->backingStore)
        nlayers++;

    nlayers += add;

    if (nlayers > 200) {
        if (diskdst)
            virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                           _("backing chains more than 200 layers deep are not "
                             "supported for disk '%s'"), diskdst);
        else
            virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s",
                           _("backing chains more than 200 layers deep are not "
                             "supported"));

        return -1;
    }

    return 0;
}


10266 10267 10268 10269 10270 10271 10272 10273 10274 10275 10276 10277 10278 10279 10280 10281 10282 10283 10284 10285 10286 10287 10288 10289 10290 10291
/**
 * qemuDomainPrepareStorageSourceConfig:
 * @src: storage source to configure
 * @cfg: qemu driver config object
 * @qemuCaps: capabilities of qemu
 *
 * Set properties of @src based on the qemu driver config @cfg.
 *
 */
static void
qemuDomainPrepareStorageSourceConfig(virStorageSourcePtr src,
                                     virQEMUDriverConfigPtr cfg,
                                     virQEMUCapsPtr qemuCaps)
{
    if (!cfg)
        return;

    if (src->type == VIR_STORAGE_TYPE_NETWORK &&
        src->protocol == VIR_STORAGE_NET_PROTOCOL_GLUSTER &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_GLUSTER_DEBUG_LEVEL)) {
        src->debug = true;
        src->debugLevel = cfg->glusterDebugLevel;
    }
}


10292 10293 10294 10295 10296 10297 10298 10299 10300 10301 10302 10303 10304
/**
 * qemuDomainDetermineDiskChain:
 * @driver: qemu driver object
 * @vm: domain object
 * @disk: disk definition
 * @disksrc: source to determine the chain for, may be NULL
 * @report_broken: report broken chain verbosely
 *
 * Prepares and initializes the backing chain of disk @disk. In cases where
 * a new source is to be associated with @disk the @disksrc parameter can be
 * used to override the source. If @report_broken is true missing images
 * in the backing chain are reported.
 */
10305
int
10306
qemuDomainDetermineDiskChain(virQEMUDriverPtr driver,
10307
                             virDomainObjPtr vm,
10308
                             virDomainDiskDefPtr disk,
10309
                             virStorageSourcePtr disksrc,
10310
                             bool report_broken)
10311
{
10312
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
10313 10314
    virStorageSourcePtr src; /* iterator for the backing chain declared in XML */
    virStorageSourcePtr n; /* iterator for the backing chain detected from disk */
10315
    qemuDomainObjPrivatePtr priv = vm->privateData;
10316 10317
    uid_t uid;
    gid_t gid;
10318

10319 10320 10321
    if (!disksrc)
        disksrc = disk->src;

10322 10323
    if (virStorageSourceIsEmpty(disksrc))
        return 0;
10324

10325 10326
    /* There is no need to check the backing chain for disks without backing
     * support */
10327 10328 10329
    if (virStorageSourceIsLocalStorage(disksrc) &&
        disksrc->format > VIR_STORAGE_FILE_NONE &&
        disksrc->format < VIR_STORAGE_FILE_BACKING) {
10330

10331
        if (!virFileExists(disksrc->path)) {
10332
            if (report_broken)
10333
                virStorageFileReportBrokenChain(errno, disksrc, disksrc);
10334

10335
            return -1;
10336 10337
        }

10338
        /* terminate the chain for such images as the code below would do */
10339
        if (!disksrc->backingStore &&
10340
            !(disksrc->backingStore = virStorageSourceNew()))
10341
            return -1;
10342

10343 10344 10345
        /* host cdrom requires special treatment in qemu, so we need to check
         * whether a block device is a cdrom */
        if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
10346 10347 10348 10349
            disksrc->format == VIR_STORAGE_FILE_RAW &&
            virStorageSourceIsBlockLocal(disksrc) &&
            virFileIsCDROM(disksrc->path) == 1)
            disksrc->hostcdrom = true;
10350

10351
        return 0;
10352 10353
    }

10354
    src = disksrc;
10355 10356
    /* skip to the end of the chain if there is any */
    while (virStorageSourceHasBacking(src)) {
10357 10358
        if (report_broken) {
            int rv = virStorageFileSupportsAccess(src);
10359

10360
            if (rv < 0)
10361
                return -1;
10362

10363
            if (rv > 0) {
10364
                if (qemuDomainStorageFileInit(driver, vm, src, disksrc) < 0)
10365
                    return -1;
10366 10367

                if (virStorageFileAccess(src, F_OK) < 0) {
10368
                    virStorageFileReportBrokenChain(errno, src, disksrc);
10369
                    virStorageFileDeinit(src);
10370
                    return -1;
10371 10372
                }

10373
                virStorageFileDeinit(src);
10374 10375
            }
        }
10376
        src = src->backingStore;
10377 10378 10379 10380
    }

    /* We skipped to the end of the chain. Skip detection if there's the
     * terminator. (An allocated but empty backingStore) */
10381 10382 10383 10384
    if (src->backingStore) {
        if (qemuDomainStorageSourceValidateDepth(disksrc, 0, disk->dst) < 0)
            return -1;

10385
        return 0;
10386
    }
10387

10388
    qemuDomainGetImageIds(cfg, vm, src, disksrc, &uid, &gid);
10389

10390
    if (virStorageFileGetMetadata(src, uid, gid, report_broken) < 0)
10391
        return -1;
10392

10393
    for (n = src->backingStore; virStorageSourceIsBacking(n); n = n->backingStore) {
10394
        if (qemuDomainValidateStorageSource(n, priv->qemuCaps) < 0)
10395
            return -1;
10396

10397
        qemuDomainPrepareStorageSourceConfig(n, cfg, priv->qemuCaps);
10398
        qemuDomainPrepareDiskSourceData(disk, n);
10399 10400 10401

        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV) &&
            qemuDomainPrepareStorageSourceBlockdev(disk, n, priv, cfg) < 0)
10402
            return -1;
10403
    }
10404

10405 10406 10407
    if (qemuDomainStorageSourceValidateDepth(disksrc, 0, disk->dst) < 0)
        return -1;

10408
    return 0;
10409
}
10410

10411

10412 10413 10414 10415 10416 10417 10418 10419 10420 10421 10422 10423 10424 10425 10426 10427
/**
 * qemuDomainDiskGetBackendAlias:
 * @disk: disk definition
 * @qemuCaps: emulator capabilities
 * @backendAlias: filled with the alias of the disk storage backend
 *
 * Returns the correct alias for the disk backend. This may be the alias of
 * -drive for legacy setup or the correct node name for -blockdev setups.
 *
 * @backendAlias may be NULL on success if the backend does not exist
 * (disk is empty). Caller is responsible for freeing @backendAlias.
 *
 * Returns 0 on success, -1 on error with libvirt error reported.
 */
int
qemuDomainDiskGetBackendAlias(virDomainDiskDefPtr disk,
10428
                              virQEMUCapsPtr qemuCaps,
10429 10430
                              char **backendAlias)
{
10431 10432
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);
    const char *nodename = NULL;
10433 10434
    *backendAlias = NULL;

10435 10436 10437 10438 10439 10440 10441 10442 10443 10444 10445 10446 10447 10448 10449
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if (!(*backendAlias = qemuAliasDiskDriveFromDisk(disk)))
            return -1;

        return 0;
    }

    if (virStorageSourceIsEmpty(disk->src))
        return 0;

    if (disk->copy_on_read == VIR_TRISTATE_SWITCH_ON)
        nodename = priv->nodeCopyOnRead;
    else
        nodename = disk->src->nodeformat;

10450
    *backendAlias = g_strdup(nodename);
10451 10452 10453 10454 10455

    return 0;
}


10456 10457 10458
typedef enum {
    /* revoke access to the image instead of allowing it */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE = 1 << 0,
10459 10460
    /* operate on full backing chain rather than single image */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN = 1 << 1,
10461 10462 10463 10464
    /* force permissions to read-only/read-write when allowing */
    /* currently does not properly work with QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_ONLY = 1 << 2,
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_WRITE = 1 << 3,
10465
    /* don't revoke permissions when modification has failed */
10466
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE = 1 << 4,
10467
    /* VM already has access to the source and we are just modifying it */
10468
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS = 1 << 5,
10469 10470 10471
} qemuDomainStorageSourceAccessFlags;


10472
/**
10473
 * qemuDomainStorageSourceAccessModify:
10474 10475 10476
 * @driver: qemu driver struct
 * @vm: domain object
 * @src: Source to prepare
10477
 * @flags: bitwise or of qemuDomainStorageSourceAccessFlags
10478 10479
 *
 * Setup the locks, cgroups and security permissions on a disk source and its
10480
 * backing chain.
10481 10482 10483
 *
 * Returns 0 on success and -1 on error. Reports libvirt error.
 */
10484
static int
10485 10486 10487
qemuDomainStorageSourceAccessModify(virQEMUDriverPtr driver,
                                    virDomainObjPtr vm,
                                    virStorageSourcePtr src,
10488
                                    qemuDomainStorageSourceAccessFlags flags)
10489
{
10490
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
10491 10492 10493
    const char *srcstr = NULLSTR(src->path);
    int ret = -1;
    virErrorPtr orig_err = NULL;
10494
    bool chain = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
10495 10496 10497
    bool force_ro = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_ONLY;
    bool force_rw = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_WRITE;
    bool revoke = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE;
10498
    int rc;
10499
    bool was_readonly = src->readonly;
10500 10501 10502 10503
    bool revoke_cgroup = false;
    bool revoke_label = false;
    bool revoke_namespace = false;
    bool revoke_lockspace = false;
10504

10505 10506 10507 10508
    VIR_DEBUG("src='%s' readonly=%d force_ro=%d force_rw=%d revoke=%d chain=%d",
              NULLSTR(src->path), src->readonly, force_ro, force_rw, revoke, chain);

    if (force_ro)
10509
        src->readonly = true;
10510

10511
    if (force_rw)
10512 10513
        src->readonly = false;

10514
    /* just tear down the disk access */
10515
    if (revoke) {
10516
        virErrorPreserveLast(&orig_err);
10517 10518 10519 10520
        revoke_cgroup = true;
        revoke_label = true;
        revoke_namespace = true;
        revoke_lockspace = true;
10521
        ret = 0;
10522
        goto revoke;
10523 10524 10525
    }

    if (virDomainLockImageAttach(driver->lockManager, cfg->uri, vm, src) < 0)
10526 10527 10528
        goto revoke;

    revoke_lockspace = true;
10529

10530 10531 10532 10533
    /* When modifying access of existing @src namespace does not need update */
    if (!(flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS)) {
        if (qemuDomainNamespaceSetupDisk(vm, src) < 0)
            goto revoke;
10534

10535 10536
        revoke_namespace = true;
    }
10537

10538
    if (qemuSecuritySetImageLabel(driver, vm, src, chain) < 0)
10539 10540 10541
        goto revoke;

    revoke_label = true;
10542

10543 10544 10545 10546 10547 10548
    if (chain)
        rc = qemuSetupImageChainCgroup(vm, src);
    else
        rc = qemuSetupImageCgroup(vm, src);

    if (rc < 0)
10549 10550 10551
        goto revoke;

    revoke_cgroup = true;
10552 10553 10554 10555

    ret = 0;
    goto cleanup;

10556
 revoke:
10557 10558 10559
    if (flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE)
        goto cleanup;

10560 10561 10562 10563 10564
    if (revoke_cgroup) {
        if (chain)
            rc = qemuTeardownImageChainCgroup(vm, src);
        else
            rc = qemuTeardownImageCgroup(vm, src);
10565

10566 10567 10568 10569 10570 10571 10572 10573 10574 10575 10576 10577 10578 10579 10580 10581 10582 10583
        if (rc < 0)
            VIR_WARN("Unable to tear down cgroup access on %s", srcstr);
    }

    if (revoke_label) {
        if (qemuSecurityRestoreImageLabel(driver, vm, src, chain) < 0)
            VIR_WARN("Unable to restore security label on %s", srcstr);
    }

    if (revoke_namespace) {
        if (qemuDomainNamespaceTeardownDisk(vm, src) < 0)
            VIR_WARN("Unable to remove /dev entry for %s", srcstr);
    }

    if (revoke_lockspace) {
        if (virDomainLockImageDetach(driver->lockManager, vm, src) < 0)
            VIR_WARN("Unable to release lock on %s", srcstr);
    }
10584 10585

 cleanup:
10586
    src->readonly = was_readonly;
10587 10588 10589 10590 10591 10592
    virErrorRestore(&orig_err);

    return ret;
}


10593 10594 10595 10596 10597
int
qemuDomainStorageSourceChainAccessAllow(virQEMUDriverPtr driver,
                                        virDomainObjPtr vm,
                                        virStorageSourcePtr src)
{
10598
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
10599 10600

    return qemuDomainStorageSourceAccessModify(driver, vm, src, flags);
10601 10602 10603 10604 10605 10606 10607 10608
}


int
qemuDomainStorageSourceChainAccessRevoke(virQEMUDriverPtr driver,
                                         virDomainObjPtr vm,
                                         virStorageSourcePtr src)
{
10609 10610
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE |
                                               QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
10611 10612

    return qemuDomainStorageSourceAccessModify(driver, vm, src, flags);
10613 10614 10615
}


10616
/**
10617
 * qemuDomainStorageSourceAccessRevoke:
10618 10619 10620 10621 10622
 *
 * Revoke access to a single backing chain element. This restores the labels,
 * removes cgroup ACLs for devices and removes locks.
 */
void
10623 10624 10625
qemuDomainStorageSourceAccessRevoke(virQEMUDriverPtr driver,
                                    virDomainObjPtr vm,
                                    virStorageSourcePtr elem)
10626
{
10627
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE;
10628

10629
    ignore_value(qemuDomainStorageSourceAccessModify(driver, vm, elem, flags));
10630 10631 10632 10633
}


/**
10634
 * qemuDomainStorageSourceAccessAllow:
10635 10636 10637 10638 10639
 * @driver: qemu driver data
 * @vm: domain object
 * @elem: source structure to set access for
 * @readonly: setup read-only access if true
 * @newSource: @elem describes a storage source which @vm can't access yet
10640 10641 10642
 *
 * Allow a VM access to a single element of a disk backing chain; this helper
 * ensures that the lock manager, cgroup device controller, and security manager
10643 10644 10645 10646 10647
 * labelling are all aware of each new file before it is added to a chain.
 *
 * When modifying permissions of @elem which @vm can already access (is in the
 * backing chain) @newSource needs to be set to false.
 */
10648
int
10649 10650 10651 10652 10653
qemuDomainStorageSourceAccessAllow(virQEMUDriverPtr driver,
                                   virDomainObjPtr vm,
                                   virStorageSourcePtr elem,
                                   bool readonly,
                                   bool newSource)
10654
{
10655
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE;
10656

10657
    if (readonly)
10658 10659 10660
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_ONLY;
    else
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_WRITE;
10661

10662
    if (!newSource)
10663
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS;
10664

10665
    return qemuDomainStorageSourceAccessModify(driver, vm, elem, flags);
10666 10667 10668
}


10669 10670 10671 10672 10673 10674 10675 10676 10677 10678
/*
 * Makes sure the @disk differs from @orig_disk only by the source
 * path and nothing else.  Fields that are being checked and the
 * information whether they are nullable (may not be specified) or is
 * taken from the virDomainDiskDefFormat() code.
 */
bool
qemuDomainDiskChangeSupported(virDomainDiskDefPtr disk,
                              virDomainDiskDefPtr orig_disk)
{
10679 10680 10681 10682 10683 10684 10685 10686 10687 10688
#define CHECK_EQ(field, field_name, nullable) \
    do { \
        if (nullable && !disk->field) \
            break; \
        if (disk->field != orig_disk->field) { \
            virReportError(VIR_ERR_OPERATION_UNSUPPORTED, \
                           _("cannot modify field '%s' of the disk"), \
                           field_name); \
            return false; \
        } \
10689 10690
    } while (0)

10691 10692 10693 10694 10695 10696 10697 10698 10699 10700 10701 10702
#define CHECK_STREQ_NULLABLE(field, field_name) \
    do { \
        if (!disk->field) \
            break; \
        if (STRNEQ_NULLABLE(disk->field, orig_disk->field)) { \
            virReportError(VIR_ERR_OPERATION_UNSUPPORTED, \
                           _("cannot modify field '%s' of the disk"), \
                           field_name); \
            return false; \
        } \
    } while (0)

10703
    CHECK_EQ(device, "device", false);
10704 10705 10706 10707
    CHECK_EQ(bus, "bus", false);
    if (STRNEQ(disk->dst, orig_disk->dst)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
10708
                       "target");
10709 10710 10711 10712
        return false;
    }
    CHECK_EQ(tray_status, "tray", true);
    CHECK_EQ(removable, "removable", true);
10713 10714 10715 10716 10717 10718 10719 10720 10721 10722 10723 10724 10725 10726 10727 10728 10729 10730 10731 10732 10733 10734 10735 10736 10737 10738 10739 10740 10741 10742 10743 10744 10745 10746 10747 10748 10749 10750 10751 10752 10753 10754 10755 10756 10757 10758 10759 10760 10761 10762 10763 10764 10765 10766

    if (disk->geometry.cylinders &&
        disk->geometry.heads &&
        disk->geometry.sectors) {
        CHECK_EQ(geometry.cylinders, "geometry cylinders", false);
        CHECK_EQ(geometry.heads, "geometry heads", false);
        CHECK_EQ(geometry.sectors, "geometry sectors", false);
        CHECK_EQ(geometry.trans, "BIOS-translation-modus", true);
    }

    CHECK_EQ(blockio.logical_block_size,
             "blockio logical_block_size", false);
    CHECK_EQ(blockio.physical_block_size,
             "blockio physical_block_size", false);

    CHECK_EQ(blkdeviotune.total_bytes_sec,
             "blkdeviotune total_bytes_sec",
             true);
    CHECK_EQ(blkdeviotune.read_bytes_sec,
             "blkdeviotune read_bytes_sec",
             true);
    CHECK_EQ(blkdeviotune.write_bytes_sec,
             "blkdeviotune write_bytes_sec",
             true);
    CHECK_EQ(blkdeviotune.total_iops_sec,
             "blkdeviotune total_iops_sec",
             true);
    CHECK_EQ(blkdeviotune.read_iops_sec,
             "blkdeviotune read_iops_sec",
             true);
    CHECK_EQ(blkdeviotune.write_iops_sec,
             "blkdeviotune write_iops_sec",
             true);
    CHECK_EQ(blkdeviotune.total_bytes_sec_max,
             "blkdeviotune total_bytes_sec_max",
             true);
    CHECK_EQ(blkdeviotune.read_bytes_sec_max,
             "blkdeviotune read_bytes_sec_max",
             true);
    CHECK_EQ(blkdeviotune.write_bytes_sec_max,
             "blkdeviotune write_bytes_sec_max",
             true);
    CHECK_EQ(blkdeviotune.total_iops_sec_max,
             "blkdeviotune total_iops_sec_max",
             true);
    CHECK_EQ(blkdeviotune.read_iops_sec_max,
             "blkdeviotune read_iops_sec_max",
             true);
    CHECK_EQ(blkdeviotune.write_iops_sec_max,
             "blkdeviotune write_iops_sec_max",
             true);
    CHECK_EQ(blkdeviotune.size_iops_sec,
             "blkdeviotune size_iops_sec",
             true);
10767 10768
    CHECK_STREQ_NULLABLE(blkdeviotune.group_name,
                         "blkdeviotune group name");
10769

10770 10771 10772 10773 10774 10775 10776 10777
    CHECK_STREQ_NULLABLE(serial,
                         "serial");
    CHECK_STREQ_NULLABLE(wwn,
                         "wwn");
    CHECK_STREQ_NULLABLE(vendor,
                         "vendor");
    CHECK_STREQ_NULLABLE(product,
                         "product");
10778

10779 10780 10781 10782 10783 10784 10785
    CHECK_EQ(cachemode, "cache", true);
    CHECK_EQ(error_policy, "error_policy", true);
    CHECK_EQ(rerror_policy, "rerror_policy", true);
    CHECK_EQ(iomode, "io", true);
    CHECK_EQ(ioeventfd, "ioeventfd", true);
    CHECK_EQ(event_idx, "event_idx", true);
    CHECK_EQ(copy_on_read, "copy_on_read", true);
10786
    /* "snapshot" is a libvirt internal field and thus can be changed */
10787
    /* startupPolicy is allowed to be updated. Therefore not checked here. */
10788
    CHECK_EQ(transient, "transient", true);
10789 10790 10791 10792 10793 10794 10795 10796 10797 10798 10799 10800 10801

    /* Note: For some address types the address auto generation for
     * @disk has still not happened at this point (e.g. driver
     * specific addresses) therefore we can't catch these possible
     * address modifications here. */
    if (disk->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
        !virDomainDeviceInfoAddressIsEqual(&disk->info, &orig_disk->info)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "address");
        return false;
    }

10802
    /* device alias is checked already in virDomainDefCompatibleDevice */
10803

10804
    CHECK_EQ(info.bootIndex, "boot order", true);
10805 10806 10807 10808 10809
    CHECK_EQ(rawio, "rawio", true);
    CHECK_EQ(sgio, "sgio", true);
    CHECK_EQ(discard, "discard", true);
    CHECK_EQ(iothread, "iothread", true);

10810 10811
    CHECK_STREQ_NULLABLE(domain_name,
                         "backenddomain");
10812

10813 10814 10815 10816 10817 10818
    /* checks for fields stored in disk->src */
    /* unfortunately 'readonly' and 'shared' can't be converted to tristate
     * values thus we need to ignore the check if the new value is 'false' */
    CHECK_EQ(src->readonly, "readonly", true);
    CHECK_EQ(src->shared, "shared", true);

10819 10820 10821 10822 10823 10824 10825 10826
    if (!virStoragePRDefIsEqual(disk->src->pr,
                                orig_disk->src->pr)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "reservations");
        return false;
    }

10827
#undef CHECK_EQ
10828
#undef CHECK_STREQ_NULLABLE
10829 10830 10831 10832

    return true;
}

10833 10834 10835 10836 10837 10838 10839 10840 10841 10842 10843 10844 10845 10846 10847

/* Return the format node name for a given disk of an online guest */
const char *
qemuDomainDiskNodeFormatLookup(virDomainObjPtr vm,
                               const char *disk)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++) {
        if (STREQ(vm->def->disks[i]->dst, disk))
            return vm->def->disks[i]->src->nodeformat;
    }
    return NULL;
}

10848 10849 10850
bool
qemuDomainDiskBlockJobIsActive(virDomainDiskDefPtr disk)
{
10851 10852
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

10853 10854 10855 10856 10857 10858 10859 10860
    if (disk->mirror) {
        virReportError(VIR_ERR_BLOCK_COPY_ACTIVE,
                       _("disk '%s' already in active block job"),
                       disk->dst);

        return true;
    }

10861
    if (diskPriv->blockjob &&
10862
        qemuBlockJobIsRunning(diskPriv->blockjob)) {
10863 10864 10865 10866 10867 10868 10869 10870 10871 10872
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("disk '%s' already in active block job"),
                       disk->dst);
        return true;
    }

    return false;
}


10873 10874 10875 10876 10877 10878 10879 10880 10881 10882 10883 10884 10885 10886 10887
/**
 * qemuDomainHasBlockjob:
 * @vm: domain object
 * @copy_only: Reject only block copy job
 *
 * Return true if @vm has at least one disk involved in a current block
 * copy/commit/pull job. If @copy_only is true this returns true only if the
 * disk is involved in a block copy.
 * */
bool
qemuDomainHasBlockjob(virDomainObjPtr vm,
                      bool copy_only)
{
    size_t i;
    for (i = 0; i < vm->def->ndisks; i++) {
10888 10889 10890
        virDomainDiskDefPtr disk = vm->def->disks[i];
        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

10891 10892
        if (!copy_only && diskPriv->blockjob &&
            qemuBlockJobIsRunning(diskPriv->blockjob))
10893 10894
            return true;

10895
        if (disk->mirror && disk->mirrorJob == VIR_DOMAIN_BLOCK_JOB_TYPE_COPY)
10896 10897 10898 10899 10900 10901 10902
            return true;
    }

    return false;
}


10903 10904
int
qemuDomainUpdateDeviceList(virQEMUDriverPtr driver,
10905 10906
                           virDomainObjPtr vm,
                           int asyncJob)
10907 10908 10909
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char **aliases;
10910
    int rc;
10911

10912 10913
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
10914 10915 10916 10917
    rc = qemuMonitorGetDeviceAliases(priv->mon, &aliases);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
    if (rc < 0)
10918 10919
        return -1;

10920
    virStringListFree(priv->qemuDevices);
10921 10922 10923
    priv->qemuDevices = aliases;
    return 0;
}
10924

10925 10926 10927 10928 10929 10930 10931 10932 10933 10934 10935 10936 10937 10938 10939 10940 10941 10942 10943

int
qemuDomainUpdateMemoryDeviceInfo(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
                                 int asyncJob)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virHashTablePtr meminfo = NULL;
    int rc;
    size_t i;

    if (vm->def->nmems == 0)
        return 0;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;

    rc = qemuMonitorGetMemoryDeviceInfo(priv->mon, &meminfo);

10944 10945
    if (qemuDomainObjExitMonitor(driver, vm) < 0) {
        virHashFree(meminfo);
10946
        return -1;
10947
    }
10948 10949 10950 10951 10952 10953 10954 10955 10956 10957 10958 10959 10960 10961 10962 10963 10964 10965 10966 10967 10968 10969 10970 10971 10972 10973 10974 10975

    /* if qemu doesn't support the info request, just carry on */
    if (rc == -2)
        return 0;

    if (rc < 0)
        return -1;

    for (i = 0; i < vm->def->nmems; i++) {
        virDomainMemoryDefPtr mem = vm->def->mems[i];
        qemuMonitorMemoryDeviceInfoPtr dimm;

        if (!mem->info.alias)
            continue;

        if (!(dimm = virHashLookup(meminfo, mem->info.alias)))
            continue;

        mem->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DIMM;
        mem->info.addr.dimm.slot = dimm->slot;
        mem->info.addr.dimm.base = dimm->address;
    }

    virHashFree(meminfo);
    return 0;
}


10976 10977 10978 10979
static bool
qemuDomainABIStabilityCheck(const virDomainDef *src,
                            const virDomainDef *dst)
{
10980 10981
    size_t i;

10982 10983 10984 10985 10986 10987 10988 10989 10990
    if (src->mem.source != dst->mem.source) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Target memoryBacking source '%s' doesn't "
                         "match source memoryBacking source'%s'"),
                       virDomainMemorySourceTypeToString(dst->mem.source),
                       virDomainMemorySourceTypeToString(src->mem.source));
        return false;
    }

10991 10992 10993 10994 10995 10996 10997 10998 10999 11000 11001 11002 11003
    for (i = 0; i < src->nmems; i++) {
        const char *srcAlias = src->mems[i]->info.alias;
        const char *dstAlias = dst->mems[i]->info.alias;

        if (STRNEQ_NULLABLE(srcAlias, dstAlias)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Target memory device alias '%s' doesn't "
                             "match source alias '%s'"),
                           NULLSTR(srcAlias), NULLSTR(dstAlias));
            return false;
        }
    }

11004 11005 11006 11007 11008 11009 11010 11011 11012
    return true;
}


virDomainABIStability virQEMUDriverDomainABIStability = {
    .domain = qemuDomainABIStabilityCheck,
};


11013 11014 11015 11016 11017 11018 11019 11020 11021 11022 11023 11024 11025 11026 11027 11028 11029 11030 11031 11032
static bool
qemuDomainMigratableDefCheckABIStability(virQEMUDriverPtr driver,
                                         virDomainDefPtr src,
                                         virDomainDefPtr migratableSrc,
                                         virDomainDefPtr dst,
                                         virDomainDefPtr migratableDst)
{
    if (!virDomainDefCheckABIStabilityFlags(migratableSrc,
                                            migratableDst,
                                            driver->xmlopt,
                                            VIR_DOMAIN_DEF_ABI_CHECK_SKIP_VOLATILE))
        return false;

    /* Force update any skipped values from the volatile flag */
    dst->mem.cur_balloon = src->mem.cur_balloon;

    return true;
}


11033 11034 11035
#define COPY_FLAGS (VIR_DOMAIN_XML_SECURE | \
                    VIR_DOMAIN_XML_MIGRATABLE)

11036 11037
bool
qemuDomainDefCheckABIStability(virQEMUDriverPtr driver,
11038
                               virQEMUCapsPtr qemuCaps,
11039 11040 11041 11042 11043 11044 11045
                               virDomainDefPtr src,
                               virDomainDefPtr dst)
{
    virDomainDefPtr migratableDefSrc = NULL;
    virDomainDefPtr migratableDefDst = NULL;
    bool ret = false;

11046 11047
    if (!(migratableDefSrc = qemuDomainDefCopy(driver, qemuCaps, src, COPY_FLAGS)) ||
        !(migratableDefDst = qemuDomainDefCopy(driver, qemuCaps, dst, COPY_FLAGS)))
11048 11049
        goto cleanup;

11050 11051 11052
    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   src, migratableDefSrc,
                                                   dst, migratableDefDst);
11053

11054
 cleanup:
11055 11056 11057 11058
    virDomainDefFree(migratableDefSrc);
    virDomainDefFree(migratableDefDst);
    return ret;
}
11059

11060 11061 11062 11063 11064 11065

bool
qemuDomainCheckABIStability(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
                            virDomainDefPtr dst)
{
11066
    qemuDomainObjPrivatePtr priv = vm->privateData;
11067 11068 11069 11070 11071 11072
    virDomainDefPtr migratableSrc = NULL;
    virDomainDefPtr migratableDst = NULL;
    char *xml = NULL;
    bool ret = false;

    if (!(xml = qemuDomainFormatXML(driver, vm, COPY_FLAGS)) ||
11073 11074
        !(migratableSrc = qemuDomainDefFromXML(driver, priv->qemuCaps, xml)) ||
        !(migratableDst = qemuDomainDefCopy(driver, priv->qemuCaps, dst, COPY_FLAGS)))
11075 11076 11077 11078 11079 11080 11081 11082 11083 11084 11085 11086 11087 11088 11089 11090
        goto cleanup;

    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   vm->def, migratableSrc,
                                                   dst, migratableDst);

 cleanup:
    VIR_FREE(xml);
    virDomainDefFree(migratableSrc);
    virDomainDefFree(migratableDst);
    return ret;
}

#undef COPY_FLAGS


11091
bool
11092
qemuDomainAgentAvailable(virDomainObjPtr vm,
11093 11094
                         bool reportError)
{
11095 11096
    qemuDomainObjPrivatePtr priv = vm->privateData;

11097 11098 11099 11100 11101 11102 11103
    if (virDomainObjGetState(vm, NULL) != VIR_DOMAIN_RUNNING) {
        if (reportError) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
        }
        return false;
    }
11104 11105 11106 11107 11108 11109 11110 11111 11112
    if (priv->agentError) {
        if (reportError) {
            virReportError(VIR_ERR_AGENT_UNRESPONSIVE, "%s",
                           _("QEMU guest agent is not "
                             "available due to an error"));
        }
        return false;
    }
    if (!priv->agent) {
11113 11114 11115 11116 11117 11118 11119 11120 11121 11122 11123 11124
        if (qemuFindAgentConfig(vm->def)) {
            if (reportError) {
                virReportError(VIR_ERR_AGENT_UNRESPONSIVE, "%s",
                               _("QEMU guest agent is not connected"));
            }
            return false;
        } else {
            if (reportError) {
                virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
                               _("QEMU guest agent is not configured"));
            }
            return false;
11125 11126 11127 11128
        }
    }
    return true;
}
11129

11130

11131
static unsigned long long
11132
qemuDomainGetMemorySizeAlignment(virDomainDefPtr def)
11133
{
11134 11135 11136 11137 11138
    /* PPC requires the memory sizes to be rounded to 256MiB increments, so
     * round them to the size always. */
    if (ARCH_IS_PPC64(def->os.arch))
        return 256 * 1024;

11139 11140 11141 11142 11143 11144 11145
    /* Align memory size. QEMU requires rounding to next 4KiB block.
     * We'll take the "traditional" path and round it to 1MiB*/

    return 1024;
}


11146 11147
static unsigned long long
qemuDomainGetMemoryModuleSizeAlignment(const virDomainDef *def,
J
Ján Tomko 已提交
11148
                                       const virDomainMemoryDef *mem G_GNUC_UNUSED)
11149 11150 11151 11152 11153 11154 11155 11156 11157 11158 11159 11160
{
    /* PPC requires the memory sizes to be rounded to 256MiB increments, so
     * round them to the size always. */
    if (ARCH_IS_PPC64(def->os.arch))
        return 256 * 1024;

    /* dimm memory modules require 2MiB alignment rather than the 1MiB we are
     * using elsewhere. */
    return 2048;
}


11161 11162 11163
int
qemuDomainAlignMemorySizes(virDomainDefPtr def)
{
11164 11165
    unsigned long long maxmemkb = virMemoryMaxValue(false) >> 10;
    unsigned long long maxmemcapped = virMemoryMaxValue(true) >> 10;
11166
    unsigned long long initialmem = 0;
11167
    unsigned long long hotplugmem = 0;
11168
    unsigned long long mem;
11169
    unsigned long long align = qemuDomainGetMemorySizeAlignment(def);
11170 11171 11172 11173 11174
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    /* align NUMA cell sizes if relevant */
    for (i = 0; i < ncells; i++) {
11175 11176
        mem = VIR_ROUND_UP(virDomainNumaGetNodeMemorySize(def->numa, i), align);
        initialmem += mem;
11177 11178 11179 11180 11181 11182 11183

        if (mem > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("memory size of NUMA node '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
11184
        virDomainNumaSetNodeMemorySize(def->numa, i, mem);
11185 11186
    }

11187 11188 11189 11190 11191
    /* align initial memory size, if NUMA is present calculate it as total of
     * individual aligned NUMA node sizes */
    if (initialmem == 0)
        initialmem = VIR_ROUND_UP(virDomainDefGetMemoryInitial(def), align);

11192 11193 11194 11195 11196 11197
    if (initialmem > maxmemcapped) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("initial memory size overflowed after alignment"));
        return -1;
    }

11198
    def->mem.max_memory = VIR_ROUND_UP(def->mem.max_memory, align);
11199 11200 11201 11202 11203
    if (def->mem.max_memory > maxmemkb) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("maximum memory size overflowed after alignment"));
        return -1;
    }
11204

11205
    /* Align memory module sizes */
11206 11207
    for (i = 0; i < def->nmems; i++) {
        align = qemuDomainGetMemoryModuleSizeAlignment(def, def->mems[i]);
11208
        def->mems[i]->size = VIR_ROUND_UP(def->mems[i]->size, align);
11209
        hotplugmem += def->mems[i]->size;
11210 11211 11212 11213 11214 11215 11216

        if (def->mems[i]->size > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("size of memory module '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
11217
    }
11218

11219 11220
    virDomainDefSetMemoryTotal(def, initialmem + hotplugmem);

11221 11222
    return 0;
}
11223 11224 11225 11226 11227 11228 11229 11230 11231 11232 11233


/**
 * qemuDomainMemoryDeviceAlignSize:
 * @mem: memory device definition object
 *
 * Aligns the size of the memory module as qemu enforces it. The size is updated
 * inplace. Default rounding is now to 1 MiB (qemu requires rouding to page,
 * size so this should be safe).
 */
void
11234 11235
qemuDomainMemoryDeviceAlignSize(virDomainDefPtr def,
                                virDomainMemoryDefPtr mem)
11236
{
11237
    mem->size = VIR_ROUND_UP(mem->size, qemuDomainGetMemorySizeAlignment(def));
11238
}
11239 11240 11241 11242 11243 11244 11245 11246 11247 11248 11249 11250 11251


/**
 * qemuDomainGetMonitor:
 * @vm: domain object
 *
 * Returns the monitor pointer corresponding to the domain object @vm.
 */
qemuMonitorPtr
qemuDomainGetMonitor(virDomainObjPtr vm)
{
    return ((qemuDomainObjPrivatePtr) vm->privateData)->mon;
}
11252 11253


11254 11255 11256 11257 11258 11259 11260
/**
 * qemuFindAgentConfig:
 * @def: domain definition
 *
 * Returns the pointer to the channel definition that is used to access the
 * guest agent if the agent is configured or NULL otherwise.
 */
11261
virDomainChrDefPtr
11262 11263 11264 11265 11266 11267 11268 11269 11270 11271
qemuFindAgentConfig(virDomainDefPtr def)
{
    size_t i;

    for (i = 0; i < def->nchannels; i++) {
        virDomainChrDefPtr channel = def->channels[i];

        if (channel->targetType != VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO)
            continue;

11272 11273
        if (STREQ_NULLABLE(channel->target.name, "org.qemu.guest_agent.0"))
            return channel;
11274 11275
    }

11276
    return NULL;
11277
}
11278 11279


11280
static bool
11281
qemuDomainMachineIsQ35(const char *machine,
11282
                       const virArch arch)
11283
{
11284 11285 11286
    if (!ARCH_IS_X86(arch))
        return false;

11287 11288 11289 11290 11291 11292
    if (STREQ(machine, "q35") ||
        STRPREFIX(machine, "pc-q35-")) {
        return true;
    }

    return false;
11293
}
11294 11295


11296
static bool
11297
qemuDomainMachineIsI440FX(const char *machine,
11298
                          const virArch arch)
11299
{
11300 11301 11302
    if (!ARCH_IS_X86(arch))
        return false;

11303 11304 11305 11306 11307 11308 11309 11310 11311
    if (STREQ(machine, "pc") ||
        STRPREFIX(machine, "pc-0.") ||
        STRPREFIX(machine, "pc-1.") ||
        STRPREFIX(machine, "pc-i440fx-") ||
        STRPREFIX(machine, "rhel")) {
        return true;
    }

    return false;
11312 11313 11314
}


11315
static bool
11316
qemuDomainMachineIsS390CCW(const char *machine,
11317
                           const virArch arch)
11318
{
11319 11320 11321
    if (!ARCH_IS_S390(arch))
        return false;

11322 11323 11324 11325
    if (STRPREFIX(machine, "s390-ccw"))
        return true;

    return false;
A
Andrea Bolognani 已提交
11326
}
11327

A
Andrea Bolognani 已提交
11328

11329 11330
/* You should normally avoid this function and use
 * qemuDomainIsARMVirt() instead. */
A
Andrea Bolognani 已提交
11331 11332 11333 11334 11335 11336
bool
qemuDomainMachineIsARMVirt(const char *machine,
                           const virArch arch)
{
    if (arch != VIR_ARCH_ARMV6L &&
        arch != VIR_ARCH_ARMV7L &&
11337
        arch != VIR_ARCH_AARCH64) {
11338
        return false;
11339
    }
11340

11341 11342 11343 11344
    if (STREQ(machine, "virt") ||
        STRPREFIX(machine, "virt-")) {
        return true;
    }
11345

11346
    return false;
11347 11348 11349
}


11350
static bool
A
Andrea Bolognani 已提交
11351 11352
qemuDomainMachineIsRISCVVirt(const char *machine,
                             const virArch arch)
11353
{
A
Andrea Bolognani 已提交
11354 11355
    if (!ARCH_IS_RISCV(arch))
        return false;
11356

11357 11358 11359 11360
    if (STREQ(machine, "virt") ||
        STRPREFIX(machine, "virt-")) {
        return true;
    }
11361

11362
    return false;
A
Andrea Bolognani 已提交
11363 11364 11365
}


11366 11367
/* You should normally avoid this function and use
 * qemuDomainIsPSeries() instead. */
A
Andrea Bolognani 已提交
11368 11369 11370 11371 11372 11373 11374
bool
qemuDomainMachineIsPSeries(const char *machine,
                           const virArch arch)
{
    if (!ARCH_IS_PPC64(arch))
        return false;

11375 11376 11377 11378
    if (STREQ(machine, "pseries") ||
        STRPREFIX(machine, "pseries-")) {
        return true;
    }
11379

11380
    return false;
11381 11382 11383
}


11384 11385
/* You should normally avoid this function and use
 * qemuDomainHasBuiltinIDE() instead. */
11386
bool
11387 11388
qemuDomainMachineHasBuiltinIDE(const char *machine,
                               const virArch arch)
11389
{
11390
    return qemuDomainMachineIsI440FX(machine, arch) ||
A
Andrea Bolognani 已提交
11391 11392 11393
        STREQ(machine, "malta") ||
        STREQ(machine, "sun4u") ||
        STREQ(machine, "g3beige");
11394 11395 11396
}


11397
static bool
11398
qemuDomainMachineNeedsFDC(const char *machine,
11399
                          const virArch arch)
11400 11401
{
    const char *p = STRSKIP(machine, "pc-q35-");
11402

11403 11404 11405
    if (!ARCH_IS_X86(arch))
        return false;

11406 11407 11408 11409 11410 11411 11412 11413 11414
    if (!p)
        return false;

    if (STRPREFIX(p, "1.") ||
        STREQ(p, "2.0") ||
        STREQ(p, "2.1") ||
        STREQ(p, "2.2") ||
        STREQ(p, "2.3")) {
        return false;
11415
    }
11416 11417

    return true;
11418 11419 11420
}


11421
bool
A
Andrea Bolognani 已提交
11422
qemuDomainIsQ35(const virDomainDef *def)
11423
{
11424
    return qemuDomainMachineIsQ35(def->os.machine, def->os.arch);
11425 11426
}

11427

J
Ján Tomko 已提交
11428
bool
A
Andrea Bolognani 已提交
11429
qemuDomainIsI440FX(const virDomainDef *def)
J
Ján Tomko 已提交
11430
{
11431
    return qemuDomainMachineIsI440FX(def->os.machine, def->os.arch);
11432 11433 11434 11435
}


bool
A
Andrea Bolognani 已提交
11436
qemuDomainIsS390CCW(const virDomainDef *def)
11437
{
11438
    return qemuDomainMachineIsS390CCW(def->os.machine, def->os.arch);
11439 11440 11441 11442
}


bool
A
Andrea Bolognani 已提交
11443
qemuDomainIsARMVirt(const virDomainDef *def)
11444
{
A
Andrea Bolognani 已提交
11445
    return qemuDomainMachineIsARMVirt(def->os.machine, def->os.arch);
J
Ján Tomko 已提交
11446 11447 11448
}


11449 11450 11451 11452 11453 11454 11455 11456
bool
qemuDomainIsRISCVVirt(const virDomainDef *def)
{
    return qemuDomainMachineIsRISCVVirt(def->os.machine, def->os.arch);
}


bool
A
Andrea Bolognani 已提交
11457
qemuDomainIsPSeries(const virDomainDef *def)
11458
{
A
Andrea Bolognani 已提交
11459 11460 11461 11462 11463 11464 11465 11466 11467 11468
    return qemuDomainMachineIsPSeries(def->os.machine, def->os.arch);
}


bool
qemuDomainHasPCIRoot(const virDomainDef *def)
{
    int root = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

    if (root < 0)
11469 11470
        return false;

A
Andrea Bolognani 已提交
11471
    if (def->controllers[root]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)
11472 11473 11474 11475 11476 11477
        return false;

    return true;
}


11478
bool
A
Andrea Bolognani 已提交
11479
qemuDomainHasPCIeRoot(const virDomainDef *def)
11480
{
A
Andrea Bolognani 已提交
11481 11482 11483 11484 11485 11486 11487 11488 11489
    int root = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

    if (root < 0)
        return false;

    if (def->controllers[root]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT)
        return false;

    return true;
11490 11491 11492 11493
}


bool
A
Andrea Bolognani 已提交
11494
qemuDomainHasBuiltinIDE(const virDomainDef *def)
11495
{
11496
    return qemuDomainMachineHasBuiltinIDE(def->os.machine, def->os.arch);
A
Andrea Bolognani 已提交
11497
}
11498 11499


A
Andrea Bolognani 已提交
11500 11501 11502
bool
qemuDomainNeedsFDC(const virDomainDef *def)
{
11503
    return qemuDomainMachineNeedsFDC(def->os.machine, def->os.arch);
11504 11505 11506
}


11507 11508 11509 11510
bool
qemuDomainSupportsPCI(virDomainDefPtr def,
                      virQEMUCapsPtr qemuCaps)
{
11511 11512 11513 11514
    if (def->os.arch != VIR_ARCH_ARMV6L &&
        def->os.arch != VIR_ARCH_ARMV7L &&
        def->os.arch != VIR_ARCH_AARCH64 &&
        !ARCH_IS_RISCV(def->os.arch)) {
11515
        return true;
11516
    }
11517 11518 11519 11520 11521 11522

    if (STREQ(def->os.machine, "versatilepb"))
        return true;

    if ((qemuDomainIsARMVirt(def) ||
         qemuDomainIsRISCVVirt(def)) &&
11523
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX)) {
11524
        return true;
11525
    }
11526 11527 11528 11529 11530

    return false;
}


11531 11532 11533 11534 11535 11536 11537 11538 11539 11540 11541 11542 11543 11544 11545 11546 11547 11548 11549 11550 11551 11552 11553 11554 11555 11556 11557 11558 11559 11560 11561 11562 11563 11564 11565 11566 11567 11568 11569
static bool
qemuCheckMemoryDimmConflict(const virDomainDef *def,
                            const virDomainMemoryDef *mem)
{
    size_t i;

    for (i = 0; i < def->nmems; i++) {
         virDomainMemoryDefPtr tmp = def->mems[i];

         if (tmp == mem ||
             tmp->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DIMM)
             continue;

         if (mem->info.addr.dimm.slot == tmp->info.addr.dimm.slot) {
             virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                            _("memory device slot '%u' is already being "
                              "used by another memory device"),
                            mem->info.addr.dimm.slot);
             return true;
         }

         if (mem->info.addr.dimm.base != 0 &&
             mem->info.addr.dimm.base == tmp->info.addr.dimm.base) {
             virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                            _("memory device base '0x%llx' is already being "
                              "used by another memory device"),
                            mem->info.addr.dimm.base);
             return true;
         }
    }

    return false;
}
static int
qemuDomainDefValidateMemoryHotplugDevice(const virDomainMemoryDef *mem,
                                         const virDomainDef *def)
{
    switch ((virDomainMemoryModel) mem->model) {
    case VIR_DOMAIN_MEMORY_MODEL_DIMM:
M
Michal Privoznik 已提交
11570
    case VIR_DOMAIN_MEMORY_MODEL_NVDIMM:
11571 11572 11573 11574 11575 11576 11577 11578
        if (mem->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DIMM &&
            mem->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("only 'dimm' addresses are supported for the "
                             "pc-dimm device"));
            return -1;
        }

11579 11580 11581
        if (virDomainNumaGetNodeCount(def->numa) != 0) {
            if (mem->targetNode == -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
Y
Yuri Chornoivan 已提交
11582
                               _("target NUMA node needs to be specified for "
11583 11584 11585
                                 "memory device"));
                return -1;
            }
11586 11587
        }

11588 11589 11590 11591 11592 11593 11594 11595 11596 11597 11598 11599 11600 11601 11602 11603 11604 11605 11606 11607 11608 11609 11610 11611
        if (mem->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DIMM) {
            if (mem->info.addr.dimm.slot >= def->mem.memory_slots) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("memory device slot '%u' exceeds slots "
                                 "count '%u'"),
                               mem->info.addr.dimm.slot, def->mem.memory_slots);
                return -1;
            }


            if (qemuCheckMemoryDimmConflict(def, mem))
                return -1;
        }
        break;

    case VIR_DOMAIN_MEMORY_MODEL_NONE:
    case VIR_DOMAIN_MEMORY_MODEL_LAST:
        return -1;
    }

    return 0;
}


11612 11613 11614 11615 11616 11617 11618 11619 11620 11621 11622 11623 11624 11625 11626 11627 11628 11629 11630 11631 11632
/**
 * qemuDomainDefValidateMemoryHotplug:
 * @def: domain definition
 * @qemuCaps: qemu capabilities object
 * @mem: definition of memory device that is to be added to @def with hotplug,
 *       NULL in case of regular VM startup
 *
 * Validates that the domain definition and memory modules have valid
 * configuration and are possibly able to accept @mem via hotplug if it's
 * non-NULL.
 *
 * Returns 0 on success; -1 and a libvirt error on error.
 */
int
qemuDomainDefValidateMemoryHotplug(const virDomainDef *def,
                                   virQEMUCapsPtr qemuCaps,
                                   const virDomainMemoryDef *mem)
{
    unsigned int nmems = def->nmems;
    unsigned long long hotplugSpace;
    unsigned long long hotplugMemory = 0;
M
Michal Privoznik 已提交
11633 11634
    bool needPCDimmCap = false;
    bool needNvdimmCap = false;
11635 11636 11637 11638 11639 11640 11641
    size_t i;

    hotplugSpace = def->mem.max_memory - virDomainDefGetMemoryInitial(def);

    if (mem) {
        nmems++;
        hotplugMemory = mem->size;
11642 11643 11644

        if (qemuDomainDefValidateMemoryHotplugDevice(mem, def) < 0)
            return -1;
11645 11646 11647 11648 11649 11650 11651 11652 11653 11654 11655 11656 11657
    }

    if (!virDomainDefHasMemoryHotplug(def)) {
        if (nmems) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("cannot use/hotplug a memory device when domain "
                             "'maxMemory' is not defined"));
            return -1;
        }

        return 0;
    }

11658 11659 11660 11661 11662 11663 11664 11665 11666 11667 11668
    if (!ARCH_IS_PPC64(def->os.arch)) {
        /* due to guest support, qemu would silently enable NUMA with one node
         * once the memory hotplug backend is enabled. To avoid possible
         * confusion we will enforce user originated numa configuration along
         * with memory hotplug. */
        if (virDomainNumaGetNodeCount(def->numa) == 0) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("At least one numa node has to be configured when "
                             "enabling memory hotplug"));
            return -1;
        }
11669 11670 11671 11672 11673 11674 11675 11676 11677
    }

    if (nmems > def->mem.memory_slots) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("memory device count '%u' exceeds slots count '%u'"),
                       nmems, def->mem.memory_slots);
        return -1;
    }

11678
    for (i = 0; i < def->nmems; i++) {
11679 11680
        hotplugMemory += def->mems[i]->size;

M
Michal Privoznik 已提交
11681 11682 11683 11684 11685 11686 11687 11688 11689 11690 11691 11692 11693 11694
        switch ((virDomainMemoryModel) def->mems[i]->model) {
        case VIR_DOMAIN_MEMORY_MODEL_DIMM:
            needPCDimmCap = true;
            break;

        case VIR_DOMAIN_MEMORY_MODEL_NVDIMM:
            needNvdimmCap = true;
            break;

        case VIR_DOMAIN_MEMORY_MODEL_NONE:
        case VIR_DOMAIN_MEMORY_MODEL_LAST:
            break;
        }

11695 11696 11697 11698 11699 11700
        /* already existing devices don't need to be checked on hotplug */
        if (!mem &&
            qemuDomainDefValidateMemoryHotplugDevice(def->mems[i], def) < 0)
            return -1;
    }

M
Michal Privoznik 已提交
11701 11702 11703 11704 11705 11706 11707 11708 11709 11710 11711 11712 11713 11714
    if (needPCDimmCap &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_PC_DIMM)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory hotplug isn't supported by this QEMU binary"));
        return -1;
    }

    if (needNvdimmCap &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_NVDIMM)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("nvdimm isn't supported by this QEMU binary"));
        return -1;
    }

11715 11716 11717 11718 11719 11720 11721 11722 11723 11724
    if (hotplugMemory > hotplugSpace) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory device total size exceeds hotplug space"));
        return -1;
    }

    return 0;
}


11725 11726 11727
/**
 * qemuDomainUpdateCurrentMemorySize:
 *
11728 11729
 * In case when the balloon is not present for the domain, the function
 * recalculates the maximum size to reflect possible changes.
11730
 */
11731 11732
void
qemuDomainUpdateCurrentMemorySize(virDomainObjPtr vm)
11733 11734 11735
{
    /* inactive domain doesn't need size update */
    if (!virDomainObjIsActive(vm))
11736
        return;
11737 11738 11739

    /* if no balloning is available, the current size equals to the current
     * full memory size */
11740
    if (!virDomainDefHasMemballoon(vm->def))
11741
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
11742
}
11743 11744


11745 11746 11747 11748 11749 11750 11751 11752
/**
 * ppc64VFIODeviceIsNV2Bridge:
 * @device: string with the PCI device address
 *
 * This function receives a string that represents a PCI device,
 * such as '0004:04:00.0', and tells if the device is a NVLink2
 * bridge.
 */
11753
static bool
11754 11755 11756 11757 11758 11759
ppc64VFIODeviceIsNV2Bridge(const char *device)
{
    const char *nvlink2Files[] = {"ibm,gpu", "ibm,nvlink",
                                  "ibm,nvlink-speed", "memory-region"};
    size_t i;

11760
    for (i = 0; i < G_N_ELEMENTS(nvlink2Files); i++) {
11761
        g_autofree char *file = NULL;
11762

11763 11764
        file = g_strdup_printf("/sys/bus/pci/devices/%s/of_node/%s",
                               device, nvlink2Files[i]);
11765 11766 11767 11768 11769 11770 11771 11772 11773

        if (!virFileExists(file))
            return false;
    }

    return true;
}


11774 11775 11776 11777 11778 11779 11780 11781 11782 11783 11784 11785 11786 11787 11788 11789
/**
 * getPPC64MemLockLimitBytes:
 * @def: domain definition
 *
 * A PPC64 helper that calculates the memory locking limit in order for
 * the guest to operate properly.
 */
static unsigned long long
getPPC64MemLockLimitBytes(virDomainDefPtr def)
{
    unsigned long long memKB = 0;
    unsigned long long baseLimit = 0;
    unsigned long long memory = 0;
    unsigned long long maxMemory = 0;
    unsigned long long passthroughLimit = 0;
    size_t i, nPCIHostBridges = 0;
11790
    virPCIDeviceAddressPtr pciAddr;
11791
    bool usesVFIO = false;
11792
    bool nvlink2Capable = false;
11793 11794 11795 11796 11797 11798 11799 11800 11801 11802 11803 11804 11805

    for (i = 0; i < def->ncontrollers; i++) {
        virDomainControllerDefPtr cont = def->controllers[i];

        if (!virDomainControllerIsPSeriesPHB(cont))
            continue;

        nPCIHostBridges++;
    }

    for (i = 0; i < def->nhostdevs; i++) {
        virDomainHostdevDefPtr dev = def->hostdevs[i];

11806
        if (virHostdevIsVFIODevice(dev)) {
11807
            usesVFIO = true;
11808 11809 11810

            pciAddr = &dev->source.subsys.u.pci.addr;
            if (virPCIDeviceAddressIsValid(pciAddr, false)) {
11811
                g_autofree char *pciAddrStr = NULL;
11812 11813 11814 11815 11816 11817 11818

                pciAddrStr = virPCIDeviceAddressAsString(pciAddr);
                if (ppc64VFIODeviceIsNV2Bridge(pciAddrStr)) {
                    nvlink2Capable = true;
                    break;
                }
            }
11819 11820 11821 11822 11823 11824 11825 11826 11827 11828 11829 11830 11831 11832 11833 11834 11835 11836 11837 11838 11839 11840 11841 11842 11843 11844
        }
    }

    memory = virDomainDefGetMemoryTotal(def);

    if (def->mem.max_memory)
        maxMemory = def->mem.max_memory;
    else
        maxMemory = memory;

    /* baseLimit := maxMemory / 128                                  (a)
     *              + 4 MiB * #PHBs + 8 MiB                          (b)
     *
     * (a) is the hash table
     *
     * (b) is accounting for the 32-bit DMA window - it could be either the
     * KVM accelerated TCE tables for emulated devices, or the VFIO
     * userspace view. The 4 MiB per-PHB (including the default one) covers
     * a 2GiB DMA window: default is 1GiB, but it's possible it'll be
     * increased to help performance. The 8 MiB extra should be plenty for
     * the TCE table index for any reasonable number of PHBs and several
     * spapr-vlan or spapr-vscsi devices (512kB + a tiny bit each) */
    baseLimit = maxMemory / 128 +
                4096 * nPCIHostBridges +
                8192;

11845 11846 11847 11848 11849
    /* NVLink2 support in QEMU is a special case of the passthrough
     * mechanics explained in the usesVFIO case below. The GPU RAM
     * is placed with a gap after maxMemory. The current QEMU
     * implementation puts the NVIDIA RAM above the PCI MMIO, which
     * starts at 32TiB and is the MMIO reserved for the guest main RAM.
11850
     *
11851 11852 11853
     * This window ends at 64TiB, and this is where the GPUs are being
     * placed. The next available window size is at 128TiB, and
     * 64TiB..128TiB will fit all possible NVIDIA GPUs.
11854
     *
11855 11856 11857 11858
     * The same assumption as the most common case applies here:
     * the guest will request a 64-bit DMA window, per PHB, that is
     * big enough to map all its RAM, which is now at 128TiB due
     * to the GPUs.
11859
     *
11860 11861 11862 11863 11864 11865 11866 11867 11868 11869 11870 11871 11872 11873 11874 11875 11876 11877 11878 11879 11880 11881 11882 11883 11884 11885 11886 11887 11888 11889 11890 11891 11892 11893
     * Note that the NVIDIA RAM window must be accounted for the TCE
     * table size, but *not* for the main RAM (maxMemory). This gives
     * us the following passthroughLimit for the NVLink2 case:
     *
     * passthroughLimit = maxMemory +
     *                    128TiB/512KiB * #PHBs + 8 MiB */
    if (nvlink2Capable) {
        passthroughLimit = maxMemory +
                           128 * (1ULL<<30) / 512 * nPCIHostBridges +
                           8192;
    } else if (usesVFIO) {
        /* For regular (non-NVLink2 present) VFIO passthrough, the value
         * of passthroughLimit is:
         *
         * passthroughLimit := max( 2 GiB * #PHBs,                       (c)
         *                          memory                               (d)
         *                          + memory * 1/512 * #PHBs + 8 MiB )   (e)
         *
         * (c) is the pre-DDW VFIO DMA window accounting. We're allowing 2
         * GiB rather than 1 GiB
         *
         * (d) is the with-DDW (and memory pre-registration and related
         * features) DMA window accounting - assuming that we only account
         * RAM once, even if mapped to multiple PHBs
         *
         * (e) is the with-DDW userspace view and overhead for the 64-bit
         * DMA window. This is based a bit on expected guest behaviour, but
         * there really isn't a way to completely avoid that. We assume the
         * guest requests a 64-bit DMA window (per PHB) just big enough to
         * map all its RAM. 4 kiB page size gives the 1/512; it will be
         * less with 64 kiB pages, less still if the guest is mapped with
         * hugepages (unlike the default 32-bit DMA window, DDW windows
         * can use large IOMMU pages). 8 MiB is for second and further level
         * overheads, like (b) */
11894 11895 11896
        passthroughLimit = MAX(2 * 1024 * 1024 * nPCIHostBridges,
                               memory +
                               memory / 512 * nPCIHostBridges + 8192);
11897
    }
11898 11899 11900 11901 11902 11903 11904

    memKB = baseLimit + passthroughLimit;

    return memKB << 10;
}


11905
/**
11906
 * qemuDomainGetMemLockLimitBytes:
11907 11908
 * @def: domain definition
 *
11909 11910 11911 11912 11913 11914
 * Calculate the memory locking limit that needs to be set in order for
 * the guest to operate properly. The limit depends on a number of factors,
 * including certain configuration options and less immediately apparent ones
 * such as the guest architecture or the use of certain devices.
 *
 * Returns: the memory locking limit, or 0 if setting the limit is not needed
11915 11916
 */
unsigned long long
11917
qemuDomainGetMemLockLimitBytes(virDomainDefPtr def)
11918
{
11919 11920
    unsigned long long memKB = 0;
    size_t i;
11921

11922 11923 11924 11925 11926 11927
    /* prefer the hard limit */
    if (virMemoryLimitIsSet(def->mem.hard_limit)) {
        memKB = def->mem.hard_limit;
        goto done;
    }

11928 11929 11930 11931 11932 11933 11934
    /* If the guest wants its memory to be locked, we need to raise the memory
     * locking limit so that the OS will not refuse allocation requests;
     * however, there is no reliable way for us to figure out how much memory
     * the QEMU process will allocate for its own use, so our only way out is
     * to remove the limit altogether. Use with extreme care */
    if (def->mem.locked)
        return VIR_DOMAIN_MEMORY_PARAM_UNLIMITED;
11935

11936 11937
    if (ARCH_IS_PPC64(def->os.arch) && def->virtType == VIR_DOMAIN_VIRT_KVM)
        return getPPC64MemLockLimitBytes(def);
11938

11939 11940 11941 11942 11943 11944 11945 11946 11947 11948 11949 11950 11951 11952 11953 11954 11955 11956
    /* For device passthrough using VFIO the guest memory and MMIO memory
     * regions need to be locked persistent in order to allow DMA.
     *
     * Currently the below limit is based on assumptions about the x86 platform.
     *
     * The chosen value of 1GiB below originates from x86 systems where it was
     * used as space reserved for the MMIO region for the whole system.
     *
     * On x86_64 systems the MMIO regions of the IOMMU mapped devices don't
     * count towards the locked memory limit since the memory is owned by the
     * device. Emulated devices though do count, but the regions are usually
     * small. Although it's not guaranteed that the limit will be enough for all
     * configurations it didn't pose a problem for now.
     *
     * http://www.redhat.com/archives/libvir-list/2015-November/msg00329.html
     *
     * Note that this may not be valid for all platforms.
     */
11957
    for (i = 0; i < def->nhostdevs; i++) {
11958 11959
        if (virHostdevIsVFIODevice(def->hostdevs[i]) ||
            virHostdevIsMdevDevice(def->hostdevs[i])) {
11960 11961 11962
            memKB = virDomainDefGetMemoryTotal(def) + 1024 * 1024;
            goto done;
        }
11963 11964
    }

11965 11966
 done:
    return memKB << 10;
11967
}
11968

11969

11970 11971 11972 11973 11974 11975 11976
/**
 * qemuDomainAdjustMaxMemLock:
 * @vm: domain
 *
 * Adjust the memory locking limit for the QEMU process associated to @vm, in
 * order to comply with VFIO or architecture requirements.
 *
11977 11978 11979 11980
 * The limit will not be changed unless doing so is needed; the first time
 * the limit is changed, the original (default) limit is stored in @vm and
 * that value will be restored if qemuDomainAdjustMaxMemLock() is called once
 * memory locking is no longer required.
11981 11982 11983 11984 11985 11986 11987 11988 11989
 *
 * Returns: 0 on success, <0 on failure
 */
int
qemuDomainAdjustMaxMemLock(virDomainObjPtr vm)
{
    unsigned long long bytes = 0;
    int ret = -1;

11990 11991 11992
    bytes = qemuDomainGetMemLockLimitBytes(vm->def);

    if (bytes) {
11993 11994 11995 11996 11997 11998 11999 12000 12001 12002 12003 12004 12005 12006
        /* If this is the first time adjusting the limit, save the current
         * value so that we can restore it once memory locking is no longer
         * required. Failing to obtain the current limit is not a critical
         * failure, it just means we'll be unable to lower it later */
        if (!vm->original_memlock) {
            if (virProcessGetMaxMemLock(vm->pid, &(vm->original_memlock)) < 0)
                vm->original_memlock = 0;
        }
    } else {
        /* Once memory locking is no longer required, we can restore the
         * original, usually very low, limit */
        bytes = vm->original_memlock;
        vm->original_memlock = 0;
    }
12007 12008 12009 12010 12011 12012 12013 12014 12015 12016

    /* Trying to set the memory locking limit to zero is a no-op */
    if (virProcessSetMaxMemLock(vm->pid, bytes) < 0)
        goto out;

    ret = 0;

 out:
     return ret;
}
12017

12018 12019 12020 12021 12022 12023 12024 12025 12026 12027 12028 12029 12030 12031 12032 12033 12034 12035 12036 12037 12038 12039 12040 12041 12042 12043 12044 12045 12046 12047

/**
 * qemuDomainAdjustMaxMemLockHostdev:
 * @vm: domain
 * @hostdev: device
 *
 * Temporarily add the hostdev to the domain definition. This is needed
 * because qemuDomainAdjustMaxMemLock() requires the hostdev to be already
 * part of the domain definition, but other functions like
 * qemuAssignDeviceHostdevAlias() expect it *not* to be there.
 * A better way to handle this would be nice
 *
 * Returns: 0 on success, <0 on failure
 */
int
qemuDomainAdjustMaxMemLockHostdev(virDomainObjPtr vm,
                                  virDomainHostdevDefPtr hostdev)
{
    int ret = 0;

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;
    if (qemuDomainAdjustMaxMemLock(vm) < 0)
        ret = -1;

    vm->def->hostdevs[--(vm->def->nhostdevs)] = NULL;

    return ret;
}


12048 12049 12050 12051 12052 12053 12054 12055 12056
/**
 * qemuDomainHasVcpuPids:
 * @vm: Domain object
 *
 * Returns true if we were able to successfully detect vCPU pids for the VM.
 */
bool
qemuDomainHasVcpuPids(virDomainObjPtr vm)
{
12057 12058 12059 12060 12061 12062
    size_t i;
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    virDomainVcpuDefPtr vcpu;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
12063

12064 12065 12066 12067 12068
        if (QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid > 0)
            return true;
    }

    return false;
12069
}
12070 12071 12072 12073 12074 12075 12076 12077 12078 12079 12080


/**
 * qemuDomainGetVcpuPid:
 * @vm: domain object
 * @vcpu: cpu id
 *
 * Returns the vCPU pid. If @vcpu is offline or out of range 0 is returned.
 */
pid_t
qemuDomainGetVcpuPid(virDomainObjPtr vm,
12081
                     unsigned int vcpuid)
12082
{
12083 12084
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
    return QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid;
12085
}
12086 12087


12088 12089 12090 12091 12092 12093 12094 12095 12096 12097 12098 12099 12100 12101 12102 12103 12104 12105 12106 12107 12108 12109 12110 12111 12112 12113 12114 12115 12116 12117 12118 12119 12120 12121 12122 12123 12124 12125 12126 12127 12128 12129
/**
 * qemuDomainValidateVcpuInfo:
 *
 * Validates vcpu thread information. If vcpu thread IDs are reported by qemu,
 * this function validates that online vcpus have thread info present and
 * offline vcpus don't.
 *
 * Returns 0 on success -1 on error.
 */
int
qemuDomainValidateVcpuInfo(virDomainObjPtr vm)
{
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    virDomainVcpuDefPtr vcpu;
    qemuDomainVcpuPrivatePtr vcpupriv;
    size_t i;

    if (!qemuDomainHasVcpuPids(vm))
        return 0;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);

        if (vcpu->online && vcpupriv->tid == 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("qemu didn't report thread id for vcpu '%zu'"), i);
            return -1;
        }

        if (!vcpu->online && vcpupriv->tid != 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("qemu reported thread id for inactive vcpu '%zu'"),
                           i);
            return -1;
        }
    }

    return 0;
}


12130 12131 12132 12133 12134 12135 12136 12137 12138
bool
qemuDomainSupportsNewVcpuHotplug(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
}


12139
/**
12140
 * qemuDomainRefreshVcpuInfo:
12141 12142 12143
 * @driver: qemu driver data
 * @vm: domain object
 * @asyncJob: current asynchronous job type
12144
 * @state: refresh vcpu state
12145
 *
12146 12147
 * Updates vCPU information private data of @vm. Due to historical reasons this
 * function returns success even if some data were not reported by qemu.
12148
 *
12149 12150
 * If @state is true, the vcpu state is refreshed as reported by the monitor.
 *
12151
 * Returns 0 on success and -1 on fatal error.
12152 12153
 */
int
12154 12155
qemuDomainRefreshVcpuInfo(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
12156 12157
                          int asyncJob,
                          bool state)
12158
{
12159
    virDomainVcpuDefPtr vcpu;
12160 12161
    qemuDomainVcpuPrivatePtr vcpupriv;
    qemuMonitorCPUInfoPtr info = NULL;
12162
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
12163
    size_t i, j;
12164
    bool hotplug;
12165
    bool fast;
12166
    bool validTIDs = true;
12167
    int rc;
12168
    int ret = -1;
12169

12170
    hotplug = qemuDomainSupportsNewVcpuHotplug(vm);
12171 12172
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
12173

12174 12175
    VIR_DEBUG("Maxvcpus %zu hotplug %d fast query %d", maxvcpus, hotplug, fast);

12176 12177
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
12178

12179 12180
    rc = qemuMonitorGetCPUInfo(qemuDomainGetMonitor(vm), &info, maxvcpus,
                               hotplug, fast);
12181

12182
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
12183 12184
        goto cleanup;

12185
    if (rc < 0)
12186 12187
        goto cleanup;

12188 12189 12190 12191 12192 12193 12194 12195 12196 12197 12198 12199 12200 12201 12202 12203 12204 12205 12206 12207 12208 12209 12210 12211 12212 12213 12214 12215 12216 12217 12218 12219 12220
    /*
     * The query-cpus[-fast] commands return information
     * about the vCPUs, including the OS level PID that
     * is executing the vCPU.
     *
     * For KVM there is always a 1-1 mapping between
     * vCPUs and host OS PIDs.
     *
     * For TCG things are a little more complicated.
     *
     *  - In some cases the vCPUs will all have the same
     *    PID as the main emulator thread.
     *  - In some cases the first vCPU will have a distinct
     *    PID, but other vCPUs will share the emulator thread
     *
     * For MTTCG, things work the same as KVM, with each
     * vCPU getting its own PID.
     *
     * We use the Host OS PIDs for doing vCPU pinning
     * and reporting. The TCG data reporting will result
     * in bad behaviour such as pinning the wrong PID.
     * We must thus detect and discard bogus PID info
     * from TCG, while still honouring the modern MTTCG
     * impl which we can support.
     */
    for (i = 0; i < maxvcpus && validTIDs; i++) {
        if (info[i].tid == vm->pid) {
            VIR_DEBUG("vCPU[%zu] PID %llu duplicates process",
                      i, (unsigned long long)info[i].tid);
            validTIDs = false;
        }

        for (j = 0; j < i; j++) {
12221
            if (info[i].tid != 0 && info[i].tid == info[j].tid) {
12222 12223 12224 12225 12226 12227 12228 12229 12230 12231 12232 12233
                VIR_DEBUG("vCPU[%zu] PID %llu duplicates vCPU[%zu]",
                          i, (unsigned long long)info[i].tid, j);
                validTIDs = false;
            }
        }

        if (validTIDs)
            VIR_DEBUG("vCPU[%zu] PID %llu is valid",
                      i, (unsigned long long)info[i].tid);
    }

    VIR_DEBUG("Extracting vCPU information validTIDs=%d", validTIDs);
12234 12235
    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
12236
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
12237

12238
        if (validTIDs)
12239 12240 12241 12242 12243
            vcpupriv->tid = info[i].tid;

        vcpupriv->socket_id = info[i].socket_id;
        vcpupriv->core_id = info[i].core_id;
        vcpupriv->thread_id = info[i].thread_id;
12244
        vcpupriv->node_id = info[i].node_id;
12245 12246
        vcpupriv->vcpus = info[i].vcpus;
        VIR_FREE(vcpupriv->type);
12247
        vcpupriv->type = g_steal_pointer(&info[i].type);
12248
        VIR_FREE(vcpupriv->alias);
12249
        vcpupriv->alias = g_steal_pointer(&info[i].alias);
12250
        virJSONValueFree(vcpupriv->props);
12251
        vcpupriv->props = g_steal_pointer(&info[i].props);
12252
        vcpupriv->enable_id = info[i].id;
12253
        vcpupriv->qemu_id = info[i].qemu_id;
12254

12255
        if (hotplug && state) {
12256 12257 12258 12259 12260
            vcpu->online = info[i].online;
            if (info[i].hotpluggable)
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_YES;
            else
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_NO;
12261
        }
12262 12263
    }

12264
    ret = 0;
12265 12266

 cleanup:
12267
    qemuMonitorCPUInfoFree(info, maxvcpus);
12268
    return ret;
12269
}
12270

12271 12272 12273 12274 12275 12276 12277 12278 12279 12280 12281 12282 12283 12284 12285 12286 12287 12288 12289 12290 12291 12292 12293 12294 12295 12296 12297 12298 12299 12300 12301
/**
 * qemuDomainGetVcpuHalted:
 * @vm: domain object
 * @vcpu: cpu id
 *
 * Returns the vCPU halted state.
  */
bool
qemuDomainGetVcpuHalted(virDomainObjPtr vm,
                        unsigned int vcpuid)
{
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
    return QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->halted;
}

/**
 * qemuDomainRefreshVcpuHalted:
 * @driver: qemu driver data
 * @vm: domain object
 * @asyncJob: current asynchronous job type
 *
 * Updates vCPU halted state in the private data of @vm.
 *
 * Returns 0 on success and -1 on error
 */
int
qemuDomainRefreshVcpuHalted(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
                            int asyncJob)
{
    virDomainVcpuDefPtr vcpu;
12302
    qemuDomainVcpuPrivatePtr vcpupriv;
12303
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
12304
    virBitmapPtr haltedmap = NULL;
12305 12306
    size_t i;
    int ret = -1;
12307
    bool fast;
12308 12309 12310 12311 12312

    /* Not supported currently for TCG, see qemuDomainRefreshVcpuInfo */
    if (vm->def->virtType == VIR_DOMAIN_VIRT_QEMU)
        return 0;

12313
    /* The halted state is interresting only on s390(x). On other platforms
12314 12315 12316 12317 12318 12319 12320
     * the data would be stale at the time when it would be used.
     * Calling qemuMonitorGetCpuHalted() can adversely affect the running
     * VM's performance unless QEMU supports query-cpus-fast.
     */
    if (!ARCH_IS_S390(vm->def->os.arch) ||
        !virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                        QEMU_CAPS_QUERY_CPUS_FAST))
12321 12322
        return 0;

12323 12324 12325
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;

12326 12327 12328 12329
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
    haltedmap = qemuMonitorGetCpuHalted(qemuDomainGetMonitor(vm), maxvcpus,
                                        fast);
12330
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || !haltedmap)
12331 12332 12333 12334
        goto cleanup;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
12335
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
12336 12337
        vcpupriv->halted = virTristateBoolFromBool(virBitmapIsBitSet(haltedmap,
                                                                     vcpupriv->qemu_id));
12338 12339 12340 12341 12342
    }

    ret = 0;

 cleanup:
12343
    virBitmapFree(haltedmap);
12344 12345
    return ret;
}
12346 12347 12348 12349 12350 12351

bool
qemuDomainSupportsNicdev(virDomainDefPtr def,
                         virDomainNetDefPtr net)
{
    /* non-virtio ARM nics require legacy -net nic */
S
Stefan Schallenberg 已提交
12352 12353
    if (((def->os.arch == VIR_ARCH_ARMV6L) ||
        (def->os.arch == VIR_ARCH_ARMV7L) ||
12354 12355 12356 12357 12358 12359 12360 12361
        (def->os.arch == VIR_ARCH_AARCH64)) &&
        net->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_MMIO &&
        net->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)
        return false;

    return true;
}

12362 12363 12364 12365
bool
qemuDomainNetSupportsMTU(virDomainNetType type)
{
    switch (type) {
12366 12367
    case VIR_DOMAIN_NET_TYPE_NETWORK:
    case VIR_DOMAIN_NET_TYPE_BRIDGE:
12368 12369
    case VIR_DOMAIN_NET_TYPE_ETHERNET:
    case VIR_DOMAIN_NET_TYPE_VHOSTUSER:
12370 12371
        return true;
    case VIR_DOMAIN_NET_TYPE_USER:
12372 12373 12374 12375 12376 12377 12378 12379 12380 12381 12382 12383
    case VIR_DOMAIN_NET_TYPE_SERVER:
    case VIR_DOMAIN_NET_TYPE_CLIENT:
    case VIR_DOMAIN_NET_TYPE_MCAST:
    case VIR_DOMAIN_NET_TYPE_INTERNAL:
    case VIR_DOMAIN_NET_TYPE_DIRECT:
    case VIR_DOMAIN_NET_TYPE_HOSTDEV:
    case VIR_DOMAIN_NET_TYPE_UDP:
    case VIR_DOMAIN_NET_TYPE_LAST:
        break;
    }
    return false;
}
J
John Ferlan 已提交
12384

P
Peter Krempa 已提交
12385 12386 12387 12388 12389 12390 12391 12392 12393 12394 12395 12396 12397 12398 12399

virDomainDiskDefPtr
qemuDomainDiskByName(virDomainDefPtr def,
                     const char *name)
{
    virDomainDiskDefPtr ret;

    if (!(ret = virDomainDiskByName(def, name, true))) {
        virReportError(VIR_ERR_INVALID_ARG, "%s",
                       _("No device found for specified path"));
        return NULL;
    }

    return ret;
}
12400 12401 12402 12403 12404 12405 12406 12407 12408 12409 12410 12411 12412 12413 12414 12415 12416 12417 12418 12419 12420 12421 12422 12423 12424 12425 12426 12427 12428 12429 12430


/**
 * qemuDomainDefValidateDiskLunSource:
 * @src: disk source struct
 *
 * Validate whether the disk source is valid for disk device='lun'.
 *
 * Returns 0 if the configuration is valid -1 and a libvirt error if the soure
 * is invalid.
 */
int
qemuDomainDefValidateDiskLunSource(const virStorageSource *src)
{
    if (virStorageSourceGetActualType(src) == VIR_STORAGE_TYPE_NETWORK) {
        if (src->protocol != VIR_STORAGE_NET_PROTOCOL_ISCSI) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("disk device='lun' is not supported "
                             "for protocol='%s'"),
                           virStorageNetProtocolTypeToString(src->protocol));
            return -1;
        }
    } else if (!virStorageSourceIsBlockLocal(src)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("disk device='lun' is only valid for block "
                         "type disk source"));
        return -1;
    }

    return 0;
}
12431 12432 12433 12434 12435 12436


int
qemuDomainPrepareChannel(virDomainChrDefPtr channel,
                         const char *domainChannelTargetDir)
{
S
Scott Garfinkle 已提交
12437 12438 12439 12440 12441 12442
    if (channel->targetType != VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO ||
        channel->source->type != VIR_DOMAIN_CHR_TYPE_UNIX ||
        channel->source->data.nix.path)
        return 0;

    if (channel->target.name) {
12443 12444 12445
        channel->source->data.nix.path = g_strdup_printf("%s/%s",
                                                         domainChannelTargetDir,
                                                         channel->target.name);
12446 12447
    } else {
        /* Generate a unique name */
12448 12449 12450 12451 12452
        channel->source->data.nix.path = g_strdup_printf("%s/vioser-%02d-%02d-%02d.sock",
                                                         domainChannelTargetDir,
                                                         channel->info.addr.vioserial.controller,
                                                         channel->info.addr.vioserial.bus,
                                                         channel->info.addr.vioserial.port);
12453 12454 12455 12456
    }

    return 0;
}
12457 12458


12459
/* qemuDomainPrepareChardevSourceTLS:
12460 12461 12462 12463 12464 12465 12466 12467 12468 12469 12470 12471 12472 12473 12474 12475 12476
 * @source: pointer to host interface data for char devices
 * @cfg: driver configuration
 *
 * Updates host interface TLS encryption setting based on qemu.conf
 * for char devices.  This will be presented as "tls='yes|no'" in
 * live XML of a guest.
 */
void
qemuDomainPrepareChardevSourceTLS(virDomainChrSourceDefPtr source,
                                  virQEMUDriverConfigPtr cfg)
{
    if (source->type == VIR_DOMAIN_CHR_TYPE_TCP) {
        if (source->data.tcp.haveTLS == VIR_TRISTATE_BOOL_ABSENT) {
            if (cfg->chardevTLS)
                source->data.tcp.haveTLS = VIR_TRISTATE_BOOL_YES;
            else
                source->data.tcp.haveTLS = VIR_TRISTATE_BOOL_NO;
12477
            source->data.tcp.tlsFromConfig = true;
12478 12479 12480 12481 12482
        }
    }
}


12483
/* qemuDomainPrepareChardevSource:
12484
 * @def: live domain definition
12485
 * @cfg: driver configuration
12486 12487 12488 12489 12490 12491
 *
 * Iterate through all devices that use virDomainChrSourceDefPtr as host
 * interface part.
 */
void
qemuDomainPrepareChardevSource(virDomainDefPtr def,
12492
                               virQEMUDriverConfigPtr cfg)
12493 12494 12495 12496 12497 12498 12499 12500 12501 12502 12503 12504 12505 12506 12507 12508 12509 12510 12511 12512 12513 12514 12515 12516 12517 12518 12519 12520 12521
{
    size_t i;

    for (i = 0; i < def->nserials; i++)
        qemuDomainPrepareChardevSourceTLS(def->serials[i]->source, cfg);

    for (i = 0; i < def->nparallels; i++)
        qemuDomainPrepareChardevSourceTLS(def->parallels[i]->source, cfg);

    for (i = 0; i < def->nchannels; i++)
        qemuDomainPrepareChardevSourceTLS(def->channels[i]->source, cfg);

    for (i = 0; i < def->nconsoles; i++)
        qemuDomainPrepareChardevSourceTLS(def->consoles[i]->source, cfg);

    for (i = 0; i < def->nrngs; i++)
        if (def->rngs[i]->backend == VIR_DOMAIN_RNG_BACKEND_EGD)
            qemuDomainPrepareChardevSourceTLS(def->rngs[i]->source.chardev, cfg);

    for (i = 0; i < def->nsmartcards; i++)
        if (def->smartcards[i]->type == VIR_DOMAIN_SMARTCARD_TYPE_PASSTHROUGH)
            qemuDomainPrepareChardevSourceTLS(def->smartcards[i]->data.passthru,
                                              cfg);

    for (i = 0; i < def->nredirdevs; i++)
        qemuDomainPrepareChardevSourceTLS(def->redirdevs[i]->source, cfg);
}


12522 12523 12524 12525 12526 12527 12528 12529 12530 12531 12532 12533 12534 12535 12536
static int
qemuProcessPrepareStorageSourceTLSVxhs(virStorageSourcePtr src,
                                       virQEMUDriverConfigPtr cfg)
{
    /* VxHS uses only client certificates and thus has no need for
     * the server-key.pem nor a secret that could be used to decrypt
     * the it, so no need to add a secinfo for a secret UUID. */
    if (src->haveTLS == VIR_TRISTATE_BOOL_ABSENT) {
        if (cfg->vxhsTLS)
            src->haveTLS = VIR_TRISTATE_BOOL_YES;
        else
            src->haveTLS = VIR_TRISTATE_BOOL_NO;
        src->tlsFromConfig = true;
    }

12537 12538
    if (src->haveTLS == VIR_TRISTATE_BOOL_YES)
        src->tlsCertdir = g_strdup(cfg->vxhsTLSx509certdir);
12539 12540 12541 12542 12543

    return 0;
}


12544 12545 12546 12547 12548 12549 12550 12551 12552 12553 12554 12555 12556 12557 12558 12559 12560 12561 12562 12563
static int
qemuProcessPrepareStorageSourceTLSNBD(virStorageSourcePtr src,
                                      virQEMUDriverConfigPtr cfg,
                                      virQEMUCapsPtr qemuCaps)
{
    if (src->haveTLS == VIR_TRISTATE_BOOL_ABSENT) {
        if (cfg->nbdTLS)
            src->haveTLS = VIR_TRISTATE_BOOL_YES;
        else
            src->haveTLS = VIR_TRISTATE_BOOL_NO;
        src->tlsFromConfig = true;
    }

    if (src->haveTLS == VIR_TRISTATE_BOOL_YES) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_NBD_TLS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("this qemu does not support TLS transport for NBD"));
            return -1;
        }

12564
        src->tlsCertdir = g_strdup(cfg->nbdTLSx509certdir);
12565 12566 12567 12568 12569 12570
    }

    return 0;
}


12571
/* qemuProcessPrepareStorageSourceTLS:
12572
 * @source: source for a disk
12573
 * @cfg: driver configuration
12574
 * @parentAlias: alias of the parent device
12575 12576 12577 12578 12579 12580 12581
 *
 * Updates host interface TLS encryption setting based on qemu.conf
 * for disk devices.  This will be presented as "tls='yes|no'" in
 * live XML of a guest.
 *
 * Returns 0 on success, -1 on bad config/failure
 */
12582
static int
12583
qemuDomainPrepareStorageSourceTLS(virStorageSourcePtr src,
12584
                                  virQEMUDriverConfigPtr cfg,
12585 12586
                                  const char *parentAlias,
                                  virQEMUCapsPtr qemuCaps)
12587
{
12588 12589 12590 12591 12592 12593 12594 12595
    if (virStorageSourceGetActualType(src) != VIR_STORAGE_TYPE_NETWORK)
        return 0;

    switch ((virStorageNetProtocol) src->protocol) {
    case VIR_STORAGE_NET_PROTOCOL_VXHS:
        if (qemuProcessPrepareStorageSourceTLSVxhs(src, cfg) < 0)
            return -1;
        break;
12596

12597
    case VIR_STORAGE_NET_PROTOCOL_NBD:
12598 12599 12600 12601
        if (qemuProcessPrepareStorageSourceTLSNBD(src, cfg, qemuCaps) < 0)
            return -1;
        break;

12602 12603 12604 12605 12606 12607 12608 12609 12610 12611
    case VIR_STORAGE_NET_PROTOCOL_RBD:
    case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
    case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
    case VIR_STORAGE_NET_PROTOCOL_ISCSI:
    case VIR_STORAGE_NET_PROTOCOL_HTTP:
    case VIR_STORAGE_NET_PROTOCOL_HTTPS:
    case VIR_STORAGE_NET_PROTOCOL_FTP:
    case VIR_STORAGE_NET_PROTOCOL_FTPS:
    case VIR_STORAGE_NET_PROTOCOL_TFTP:
    case VIR_STORAGE_NET_PROTOCOL_SSH:
12612 12613 12614 12615 12616 12617
        if (src->haveTLS == VIR_TRISTATE_BOOL_YES) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("TLS transport is not supported for disk protocol '%s'"),
                           virStorageNetProtocolTypeToString(src->protocol));
            return -1;
        }
12618 12619 12620 12621 12622 12623
        break;

    case VIR_STORAGE_NET_PROTOCOL_NONE:
    case VIR_STORAGE_NET_PROTOCOL_LAST:
    default:
        virReportEnumRangeError(virStorageNetProtocol, src->protocol);
12624
        return -1;
12625
    }
12626

12627 12628 12629 12630
    if (src->haveTLS == VIR_TRISTATE_BOOL_YES &&
        !(src->tlsAlias = qemuAliasTLSObjFromSrcAlias(parentAlias)))
        return -1;

12631 12632 12633 12634
    return 0;
}


12635
void
12636 12637 12638 12639
qemuDomainPrepareShmemChardev(virDomainShmemDefPtr shmem)
{
    if (!shmem->server.enabled ||
        shmem->server.chr.data.nix.path)
12640
        return;
12641

12642 12643
    shmem->server.chr.data.nix.path = g_strdup_printf("/var/lib/libvirt/shmem-%s-sock",
                                                      shmem->name);
12644 12645 12646
}


12647 12648 12649 12650 12651 12652 12653 12654 12655 12656 12657 12658 12659 12660 12661 12662 12663 12664 12665 12666 12667 12668 12669 12670 12671 12672 12673 12674 12675 12676 12677 12678 12679
/**
 * qemuDomainVcpuHotplugIsInOrder:
 * @def: domain definition
 *
 * Returns true if online vcpus were added in order (clustered behind vcpu0
 * with increasing order).
 */
bool
qemuDomainVcpuHotplugIsInOrder(virDomainDefPtr def)
{
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    unsigned int prevorder = 0;
    size_t seenonlinevcpus = 0;
    size_t i;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(def, i);

        if (!vcpu->online)
            break;

        if (vcpu->order < prevorder)
            break;

        if (vcpu->order > prevorder)
            prevorder = vcpu->order;

        seenonlinevcpus++;
    }

    return seenonlinevcpus == virDomainDefGetVcpus(def);
}
12680 12681 12682 12683 12684 12685 12686 12687 12688 12689 12690 12691 12692 12693 12694 12695 12696 12697 12698 12699 12700 12701 12702 12703 12704 12705 12706 12707 12708 12709 12710 12711 12712 12713


/**
 * qemuDomainVcpuPersistOrder:
 * @def: domain definition
 *
 * Saves the order of vcpus detected from qemu to the domain definition.
 * The private data note the order only for the entry describing the
 * hotpluggable entity. This function copies the order into the definition part
 * of all sub entities.
 */
void
qemuDomainVcpuPersistOrder(virDomainDefPtr def)
{
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    qemuDomainVcpuPrivatePtr vcpupriv;
    unsigned int prevorder = 0;
    size_t i;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(def, i);
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);

        if (!vcpu->online) {
            vcpu->order = 0;
        } else {
            if (vcpupriv->enable_id != 0)
                prevorder = vcpupriv->enable_id;

            vcpu->order = prevorder;
        }
    }
}
12714 12715 12716 12717 12718 12719 12720 12721 12722 12723 12724 12725 12726 12727 12728 12729 12730 12731 12732 12733


int
qemuDomainCheckMonitor(virQEMUDriverPtr driver,
                       virDomainObjPtr vm,
                       qemuDomainAsyncJob asyncJob)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;

    ret = qemuMonitorCheck(priv->mon);

    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;

    return ret;
}
12734 12735 12736 12737 12738 12739


bool
qemuDomainSupportsVideoVga(virDomainVideoDefPtr video,
                           virQEMUCapsPtr qemuCaps)
{
12740 12741 12742 12743 12744 12745 12746 12747
    if (video->type == VIR_DOMAIN_VIDEO_TYPE_VIRTIO) {
        if (video->backend == VIR_DOMAIN_VIDEO_BACKEND_TYPE_VHOSTUSER) {
            if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VHOST_USER_VGA))
                return false;
        } else if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_VGA)) {
            return false;
        }
    }
12748 12749 12750

    return true;
}
12751 12752


12753 12754 12755 12756 12757 12758 12759 12760
bool
qemuDomainNeedsVFIO(const virDomainDef *def)
{
    return virDomainDefHasVFIOHostdev(def) ||
        virDomainDefHasMdevHostdev(def);
}


12761 12762
/**
 * qemuDomainGetHostdevPath:
12763
 * @def: domain definition
12764
 * @dev: host device definition
12765
 * @teardown: true if device will be removed
12766
 * @npaths: number of items in @path and @perms arrays
12767 12768 12769
 * @path: resulting path to @dev
 * @perms: Optional pointer to VIR_CGROUP_DEVICE_* perms
 *
12770 12771 12772 12773 12774 12775
 * For given device @dev fetch its host path and store it at
 * @path. If a device requires other paths to be present/allowed
 * they are stored in the @path array after the actual path.
 * Optionally, caller can get @perms on the path (e.g. rw/ro).
 *
 * The caller is responsible for freeing the memory.
12776 12777 12778 12779
 *
 * Returns 0 on success, -1 otherwise.
 */
int
12780 12781 12782
qemuDomainGetHostdevPath(virDomainDefPtr def,
                         virDomainHostdevDefPtr dev,
                         bool teardown,
12783 12784 12785
                         size_t *npaths,
                         char ***path,
                         int **perms)
12786 12787 12788 12789 12790 12791
{
    int ret = -1;
    virDomainHostdevSubsysUSBPtr usbsrc = &dev->source.subsys.u.usb;
    virDomainHostdevSubsysPCIPtr pcisrc = &dev->source.subsys.u.pci;
    virDomainHostdevSubsysSCSIPtr scsisrc = &dev->source.subsys.u.scsi;
    virDomainHostdevSubsysSCSIVHostPtr hostsrc = &dev->source.subsys.u.scsi_host;
12792
    virDomainHostdevSubsysMediatedDevPtr mdevsrc = &dev->source.subsys.u.mdev;
12793 12794 12795 12796 12797
    virPCIDevicePtr pci = NULL;
    virUSBDevicePtr usb = NULL;
    virSCSIDevicePtr scsi = NULL;
    virSCSIVHostDevicePtr host = NULL;
    char *tmpPath = NULL;
12798 12799 12800
    bool includeVFIO = false;
    char **tmpPaths = NULL;
    int *tmpPerms = NULL;
12801
    size_t tmpNpaths = 0;
12802
    int perm = 0;
12803

12804
    *npaths = 0;
12805 12806 12807

    switch ((virDomainHostdevMode) dev->mode) {
    case VIR_DOMAIN_HOSTDEV_MODE_SUBSYS:
12808
        switch ((virDomainHostdevSubsysType)dev->source.subsys.type) {
12809 12810 12811 12812 12813 12814 12815 12816 12817 12818 12819
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
            if (pcisrc->backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO) {
                pci = virPCIDeviceNew(pcisrc->addr.domain,
                                      pcisrc->addr.bus,
                                      pcisrc->addr.slot,
                                      pcisrc->addr.function);
                if (!pci)
                    goto cleanup;

                if (!(tmpPath = virPCIDeviceGetIOMMUGroupDev(pci)))
                    goto cleanup;
12820 12821

                perm = VIR_CGROUP_DEVICE_RW;
12822
                if (teardown) {
12823
                    if (!virDomainDefHasVFIOHostdev(def))
12824 12825 12826 12827
                        includeVFIO = true;
                } else {
                    includeVFIO = true;
                }
12828 12829 12830 12831 12832 12833 12834 12835 12836 12837 12838 12839
            }
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
            if (dev->missing)
                break;
            usb = virUSBDeviceNew(usbsrc->bus,
                                  usbsrc->device,
                                  NULL);
            if (!usb)
                goto cleanup;

12840 12841
            if (VIR_STRDUP(tmpPath, virUSBDeviceGetPath(usb)) < 0)
                goto cleanup;
12842
            perm = VIR_CGROUP_DEVICE_RW;
12843 12844 12845 12846 12847
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI:
            if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI) {
                virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
12848
                VIR_DEBUG("Not updating /dev for hostdev iSCSI path '%s'", iscsisrc->src->path);
12849 12850 12851 12852 12853 12854 12855 12856 12857 12858 12859 12860 12861
            } else {
                virDomainHostdevSubsysSCSIHostPtr scsihostsrc = &scsisrc->u.host;
                scsi = virSCSIDeviceNew(NULL,
                                        scsihostsrc->adapter,
                                        scsihostsrc->bus,
                                        scsihostsrc->target,
                                        scsihostsrc->unit,
                                        dev->readonly,
                                        dev->shareable);

                if (!scsi)
                    goto cleanup;

12862 12863
                if (VIR_STRDUP(tmpPath, virSCSIDeviceGetPath(scsi)) < 0)
                    goto cleanup;
12864 12865
                perm = virSCSIDeviceGetReadonly(scsi) ?
                    VIR_CGROUP_DEVICE_READ : VIR_CGROUP_DEVICE_RW;
12866 12867 12868 12869 12870 12871 12872 12873 12874
            }
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI_HOST: {
            if (hostsrc->protocol ==
                VIR_DOMAIN_HOSTDEV_SUBSYS_SCSI_HOST_PROTOCOL_TYPE_VHOST) {
                if (!(host = virSCSIVHostDeviceNew(hostsrc->wwpn)))
                    goto cleanup;

12875 12876
                if (VIR_STRDUP(tmpPath, virSCSIVHostDeviceGetPath(host)) < 0)
                    goto cleanup;
12877
                perm = VIR_CGROUP_DEVICE_RW;
12878 12879 12880 12881
            }
            break;
        }

12882
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
12883
            if (!(tmpPath = virMediatedDeviceGetIOMMUGroupDev(mdevsrc->uuidstr)))
12884 12885 12886 12887 12888
                goto cleanup;

            includeVFIO = true;
            perm = VIR_CGROUP_DEVICE_RW;
            break;
12889 12890 12891 12892 12893 12894 12895 12896 12897 12898 12899
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
            break;
        }
        break;

    case VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES:
    case VIR_DOMAIN_HOSTDEV_MODE_LAST:
        /* nada */
        break;
    }

12900 12901 12902 12903 12904
    if (tmpPath) {
        size_t toAlloc = 1;

        if (includeVFIO)
            toAlloc = 2;
12905

12906
        if (VIR_ALLOC_N(tmpPaths, toAlloc) < 0 ||
12907
            VIR_ALLOC_N(tmpPerms, toAlloc) < 0)
12908
            goto cleanup;
12909
        tmpPaths[0] = g_strdup(tmpPath);
12910 12911 12912 12913
        tmpNpaths = toAlloc;
        tmpPerms[0] = perm;

        if (includeVFIO) {
12914
            tmpPaths[1] = g_strdup(QEMU_DEV_VFIO);
12915 12916 12917 12918 12919 12920 12921 12922 12923 12924 12925 12926
            tmpPerms[1] = VIR_CGROUP_DEVICE_RW;
        }
    }

    *npaths = tmpNpaths;
    tmpNpaths = 0;
    *path = tmpPaths;
    tmpPaths = NULL;
    if (perms) {
        *perms = tmpPerms;
        tmpPerms = NULL;
    }
12927 12928
    ret = 0;
 cleanup:
12929
    virStringListFreeCount(tmpPaths, tmpNpaths);
12930
    VIR_FREE(tmpPerms);
12931 12932 12933 12934
    virPCIDeviceFree(pci);
    virUSBDeviceFree(usb);
    virSCSIDeviceFree(scsi);
    virSCSIVHostDeviceFree(host);
12935
    VIR_FREE(tmpPath);
12936 12937 12938 12939
    return ret;
}


12940 12941 12942 12943
/**
 * qemuDomainGetPreservedMountPath:
 * @cfg: driver configuration data
 * @vm: domain object
12944
 * @mountpoint: mount point path to convert
12945
 *
12946
 * For given @mountpoint return new path where the mount point
12947 12948 12949 12950 12951 12952 12953 12954
 * should be moved temporarily whilst building the namespace.
 *
 * Returns: allocated string on success which the caller must free,
 *          NULL on failure.
 */
static char *
qemuDomainGetPreservedMountPath(virQEMUDriverConfigPtr cfg,
                                virDomainObjPtr vm,
12955
                                const char *mountpoint)
12956 12957 12958
{
    char *path = NULL;
    char *tmp;
12959
    const char *suffix = mountpoint + strlen(QEMU_DEVPREFIX);
12960
    char *domname = virDomainDefGetShortName(vm->def);
12961 12962
    size_t off;

12963 12964 12965
    if (!domname)
        return NULL;

12966
    if (STREQ(mountpoint, "/dev"))
12967 12968
        suffix = "dev";

12969
    path = g_strdup_printf("%s/%s.%s", cfg->stateDir, domname, suffix);
12970

12971
    /* Now consider that @mountpoint is "/dev/blah/blah2".
12972 12973 12974 12975 12976 12977 12978 12979 12980 12981 12982 12983 12984
     * @suffix then points to "blah/blah2". However, caller
     * expects all the @paths to be the same depth. The
     * caller doesn't always do `mkdir -p` but sometimes bare
     * `touch`. Therefore fix all the suffixes. */
    off = strlen(path) - strlen(suffix);

    tmp = path + off;
    while (*tmp) {
        if (*tmp == '/')
            *tmp = '.';
        tmp++;
    }

12985
    VIR_FREE(domname);
12986 12987 12988 12989
    return path;
}


12990 12991 12992 12993 12994 12995 12996 12997 12998 12999 13000 13001
/**
 * qemuDomainGetPreservedMounts:
 *
 * Process list of mounted filesystems and:
 * a) save all FSs mounted under /dev to @devPath
 * b) generate backup path for all the entries in a)
 *
 * Any of the return pointers can be NULL.
 *
 * Returns 0 on success, -1 otherwise (with error reported)
 */
static int
13002
qemuDomainGetPreservedMounts(virQEMUDriverConfigPtr cfg,
13003 13004 13005 13006 13007 13008
                             virDomainObjPtr vm,
                             char ***devPath,
                             char ***devSavePath,
                             size_t *ndevPath)
{
    char **paths = NULL, **mounts = NULL;
13009
    size_t i, j, nmounts;
13010

13011
    if (virFileGetMountSubtree(QEMU_PROC_MOUNTS, "/dev",
13012 13013 13014 13015 13016 13017 13018 13019 13020
                               &mounts, &nmounts) < 0)
        goto error;

    if (!nmounts) {
        if (ndevPath)
            *ndevPath = 0;
        return 0;
    }

13021 13022 13023 13024 13025 13026 13027 13028 13029 13030 13031 13032
    /* There can be nested mount points. For instance
     * /dev/shm/blah can be a mount point and /dev/shm too. It
     * doesn't make much sense to return the former path because
     * caller preserves the latter (and with that the former
     * too). Therefore prune nested mount points.
     * NB mounts[0] is "/dev". Should we start the outer loop
     * from the beginning of the array all we'd be left with is
     * just the first element. Think about it.
     */
    for (i = 1; i < nmounts; i++) {
        j = i + 1;
        while (j < nmounts) {
13033 13034
            char *c = STRSKIP(mounts[j], mounts[i]);

13035
            if (c && (*c == '/' || *c == '\0')) {
13036 13037 13038 13039 13040 13041 13042 13043
                VIR_DEBUG("Dropping path %s because of %s", mounts[j], mounts[i]);
                VIR_DELETE_ELEMENT(mounts, j, nmounts);
            } else {
                j++;
            }
        }
    }

13044 13045 13046 13047
    if (VIR_ALLOC_N(paths, nmounts) < 0)
        goto error;

    for (i = 0; i < nmounts; i++) {
13048
        if (!(paths[i] = qemuDomainGetPreservedMountPath(cfg, vm, mounts[i])))
13049 13050 13051 13052 13053 13054 13055 13056 13057 13058 13059 13060 13061 13062 13063 13064 13065 13066 13067 13068 13069 13070 13071 13072 13073
            goto error;
    }

    if (devPath)
        *devPath = mounts;
    else
        virStringListFreeCount(mounts, nmounts);

    if (devSavePath)
        *devSavePath = paths;
    else
        virStringListFreeCount(paths, nmounts);

    if (ndevPath)
        *ndevPath = nmounts;

    return 0;

 error:
    virStringListFreeCount(mounts, nmounts);
    virStringListFreeCount(paths, nmounts);
    return -1;
}


13074 13075
struct qemuDomainCreateDeviceData {
    const char *path;     /* Path to temp new /dev location */
13076 13077
    char * const *devMountsPath;
    size_t ndevMountsPath;
13078 13079 13080
};


13081
static int
13082
qemuDomainCreateDeviceRecursive(const char *device,
13083
                                const struct qemuDomainCreateDeviceData *data,
13084 13085
                                bool allow_noent,
                                unsigned int ttl)
13086 13087
{
    char *devicePath = NULL;
13088
    char *target = NULL;
13089 13090
    struct stat sb;
    int ret = -1;
13091
    bool isLink = false;
13092
    bool isDev = false;
13093
    bool isReg = false;
13094
    bool isDir = false;
13095
    bool create = false;
13096 13097 13098
#ifdef WITH_SELINUX
    char *tcon = NULL;
#endif
13099

13100 13101 13102 13103 13104 13105 13106
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             device);
        return ret;
    }

13107
    if (lstat(device, &sb) < 0) {
13108 13109
        if (errno == ENOENT && allow_noent) {
            /* Ignore non-existent device. */
13110
            return 0;
13111
        }
13112 13113
        virReportSystemError(errno, _("Unable to stat %s"), device);
        return ret;
13114 13115
    }

13116
    isLink = S_ISLNK(sb.st_mode);
13117
    isDev = S_ISCHR(sb.st_mode) || S_ISBLK(sb.st_mode);
13118
    isReg = S_ISREG(sb.st_mode) || S_ISFIFO(sb.st_mode) || S_ISSOCK(sb.st_mode);
13119
    isDir = S_ISDIR(sb.st_mode);
13120 13121 13122 13123 13124 13125 13126 13127 13128 13129 13130 13131 13132 13133 13134 13135

    /* Here, @device might be whatever path in the system. We
     * should create the path in the namespace iff it's "/dev"
     * prefixed. However, if it is a symlink, we need to traverse
     * it too (it might point to something in "/dev"). Just
     * consider:
     *
     *   /var/sym1 -> /var/sym2 -> /dev/sda  (because users can)
     *
     * This means, "/var/sym1" is not created (it's shared with
     * the parent namespace), nor "/var/sym2", but "/dev/sda".
     *
     * TODO Remove all `.' and `..' from the @device path.
     * Otherwise we might get fooled with `/dev/../var/my_image'.
     * For now, lets hope callers play nice.
     */
13136
    if (STRPREFIX(device, QEMU_DEVPREFIX)) {
13137
        size_t i;
13138

13139 13140 13141 13142 13143 13144 13145 13146 13147 13148
        for (i = 0; i < data->ndevMountsPath; i++) {
            if (STREQ(data->devMountsPath[i], "/dev"))
                continue;
            if (STRPREFIX(device, data->devMountsPath[i]))
                break;
        }

        if (i == data->ndevMountsPath) {
            /* Okay, @device is in /dev but not in any mount point under /dev.
             * Create it. */
13149 13150
            devicePath = g_strdup_printf("%s/%s", data->path,
                                         device + strlen(QEMU_DEVPREFIX));
13151 13152 13153 13154 13155 13156 13157 13158 13159 13160 13161 13162

            if (virFileMakeParentPath(devicePath) < 0) {
                virReportSystemError(errno,
                                     _("Unable to create %s"),
                                     devicePath);
                goto cleanup;
            }
            VIR_DEBUG("Creating dev %s", device);
            create = true;
        } else {
            VIR_DEBUG("Skipping dev %s because of %s mount point",
                      device, data->devMountsPath[i]);
13163
        }
13164 13165
    }

13166 13167 13168 13169 13170 13171 13172 13173 13174
    if (isLink) {
        /* We are dealing with a symlink. Create a dangling symlink and descend
         * down one level which hopefully creates the symlink's target. */
        if (virFileReadLink(device, &target) < 0) {
            virReportSystemError(errno,
                                 _("unable to resolve symlink %s"),
                                 device);
            goto cleanup;
        }
13175

13176 13177 13178 13179 13180 13181 13182 13183 13184
        if (create &&
            symlink(target, devicePath) < 0) {
            if (errno == EEXIST) {
                ret = 0;
            } else {
                virReportSystemError(errno,
                                     _("unable to create symlink %s"),
                                     devicePath);
            }
13185 13186 13187
            goto cleanup;
        }

13188 13189 13190 13191 13192 13193 13194 13195 13196 13197
        /* Tricky part. If the target starts with a slash then we need to take
         * it as it is. Otherwise we need to replace the last component in the
         * original path with the link target:
         * /dev/rtc -> rtc0 (want /dev/rtc0)
         * /dev/disk/by-id/ata-SanDisk_SDSSDXPS480G_161101402485 -> ../../sda
         *   (want /dev/disk/by-id/../../sda)
         * /dev/stdout -> /proc/self/fd/1 (no change needed)
         */
        if (IS_RELATIVE_FILE_NAME(target)) {
            char *c = NULL, *tmp = NULL, *devTmp = NULL;
13198

13199
            devTmp = g_strdup(device);
13200

13201 13202 13203
            if ((c = strrchr(devTmp, '/')))
                *(c + 1) = '\0';

13204
            tmp = g_strdup_printf("%s%s", devTmp, target);
13205 13206
            VIR_FREE(devTmp);
            VIR_FREE(target);
13207
            target = g_steal_pointer(&tmp);
13208 13209
        }

13210
        if (qemuDomainCreateDeviceRecursive(target, data,
13211
                                            allow_noent, ttl - 1) < 0)
13212
            goto cleanup;
13213
    } else if (isDev) {
13214 13215 13216 13217 13218 13219 13220 13221 13222 13223
        if (create &&
            mknod(devicePath, sb.st_mode, sb.st_rdev) < 0) {
            if (errno == EEXIST) {
                ret = 0;
            } else {
                virReportSystemError(errno,
                                     _("Failed to make device %s"),
                                     devicePath);
            }
            goto cleanup;
13224
        }
13225
    } else if (isReg) {
13226
        if (create &&
13227
            virFileTouch(devicePath, sb.st_mode) < 0)
13228
            goto cleanup;
13229 13230
        /* Just create the file here so that code below sets
         * proper owner and mode. Bind mount only after that. */
13231 13232 13233 13234
    } else if (isDir) {
        if (create &&
            virFileMakePathWithMode(devicePath, sb.st_mode) < 0)
            goto cleanup;
13235 13236 13237 13238 13239
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       device, sb.st_mode);
        goto cleanup;
13240 13241 13242 13243
    }

    if (!create) {
        ret = 0;
13244 13245 13246
        goto cleanup;
    }

13247
    if (lchown(devicePath, sb.st_uid, sb.st_gid) < 0) {
13248 13249 13250 13251 13252 13253
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             devicePath);
        goto cleanup;
    }

13254 13255 13256 13257 13258 13259 13260 13261 13262
    /* Symlinks don't have mode */
    if (!isLink &&
        chmod(devicePath, sb.st_mode) < 0) {
        virReportSystemError(errno,
                             _("Failed to set permissions for device %s"),
                             devicePath);
        goto cleanup;
    }

13263 13264 13265
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileCopyACLs(device, devicePath) < 0 &&
13266 13267 13268 13269 13270 13271 13272
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Failed to copy ACLs on device %s"),
                             devicePath);
        goto cleanup;
    }

13273
#ifdef WITH_SELINUX
13274
    if (lgetfilecon_raw(device, &tcon) < 0 &&
13275 13276 13277
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"),
13278
                             device);
13279 13280 13281 13282
        goto cleanup;
    }

    if (tcon &&
13283
        lsetfilecon_raw(devicePath, (const char *)tcon) < 0) {
13284 13285 13286 13287 13288 13289 13290 13291 13292 13293 13294
        VIR_WARNINGS_NO_WLOGICALOP_EQUAL_EXPR
        if (errno != EOPNOTSUPP && errno != ENOTSUP) {
        VIR_WARNINGS_RESET
            virReportSystemError(errno,
                                 _("Unable to set SELinux label on %s"),
                                 devicePath);
            goto cleanup;
        }
    }
#endif

13295
    /* Finish mount process started earlier. */
13296
    if ((isReg || isDir) &&
13297 13298 13299
        virFileBindMountDevice(device, devicePath) < 0)
        goto cleanup;

13300 13301
    ret = 0;
 cleanup:
13302
    VIR_FREE(target);
13303
    VIR_FREE(devicePath);
13304 13305 13306
#ifdef WITH_SELINUX
    freecon(tcon);
#endif
13307 13308 13309 13310
    return ret;
}


13311 13312
static int
qemuDomainCreateDevice(const char *device,
13313
                       const struct qemuDomainCreateDeviceData *data,
13314 13315 13316 13317
                       bool allow_noent)
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

13318
    return qemuDomainCreateDeviceRecursive(device, data,
13319 13320 13321
                                           allow_noent, symloop_max);
}

13322 13323

static int
13324
qemuDomainPopulateDevices(virQEMUDriverConfigPtr cfg,
J
Ján Tomko 已提交
13325
                          virDomainObjPtr vm G_GNUC_UNUSED,
13326
                          const struct qemuDomainCreateDeviceData *data)
13327 13328 13329 13330 13331 13332 13333 13334 13335
{
    const char *const *devices = (const char *const *) cfg->cgroupDeviceACL;
    size_t i;
    int ret = -1;

    if (!devices)
        devices = defaultDeviceACL;

    for (i = 0; devices[i]; i++) {
13336
        if (qemuDomainCreateDevice(devices[i], data, true) < 0)
13337 13338 13339 13340 13341 13342 13343 13344 13345 13346
            goto cleanup;
    }

    ret = 0;
 cleanup:
    return ret;
}


static int
13347 13348
qemuDomainSetupDev(virQEMUDriverConfigPtr cfg,
                   virSecurityManagerPtr mgr,
13349
                   virDomainObjPtr vm,
13350
                   const struct qemuDomainCreateDeviceData *data)
13351 13352 13353 13354 13355 13356 13357
{
    char *mount_options = NULL;
    char *opts = NULL;
    int ret = -1;

    VIR_DEBUG("Setting up /dev/ for domain %s", vm->def->name);

13358
    mount_options = qemuSecurityGetMountOptions(mgr, vm->def);
13359

13360 13361
    if (!mount_options)
        mount_options = g_strdup("");
13362 13363 13364 13365 13366

    /*
     * tmpfs is limited to 64kb, since we only have device nodes in there
     * and don't want to DOS the entire OS RAM usage
     */
13367
    opts = g_strdup_printf("mode=755,size=65536%s", mount_options);
13368

13369
    if (virFileSetupDev(data->path, opts) < 0)
13370 13371
        goto cleanup;

13372
    if (qemuDomainPopulateDevices(cfg, vm, data) < 0)
13373 13374 13375 13376 13377 13378 13379 13380 13381 13382
        goto cleanup;

    ret = 0;
 cleanup:
    VIR_FREE(opts);
    VIR_FREE(mount_options);
    return ret;
}


13383
static int
J
Ján Tomko 已提交
13384
qemuDomainSetupDisk(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
13385
                    virDomainDiskDefPtr disk,
13386
                    const struct qemuDomainCreateDeviceData *data)
13387 13388 13389 13390 13391
{
    virStorageSourcePtr next;
    char *dst = NULL;
    int ret = -1;

13392
    for (next = disk->src; virStorageSourceIsBacking(next); next = next->backingStore) {
13393
        if (!next->path || !virStorageSourceIsLocalStorage(next)) {
13394 13395 13396 13397
            /* Not creating device. Just continue. */
            continue;
        }

13398
        if (qemuDomainCreateDevice(next->path, data, false) < 0)
13399 13400 13401
            goto cleanup;
    }

13402
    /* qemu-pr-helper might require access to /dev/mapper/control. */
13403
    if (disk->src->pr &&
13404
        qemuDomainCreateDevice(QEMU_DEVICE_MAPPER_CONTROL_PATH, data, true) < 0)
13405 13406
        goto cleanup;

13407 13408 13409 13410 13411 13412 13413 13414
    ret = 0;
 cleanup:
    VIR_FREE(dst);
    return ret;
}


static int
13415
qemuDomainSetupAllDisks(virQEMUDriverConfigPtr cfg,
13416
                        virDomainObjPtr vm,
13417
                        const struct qemuDomainCreateDeviceData *data)
13418 13419 13420 13421 13422
{
    size_t i;
    VIR_DEBUG("Setting up disks");

    for (i = 0; i < vm->def->ndisks; i++) {
13423
        if (qemuDomainSetupDisk(cfg,
13424
                                vm->def->disks[i],
13425
                                data) < 0)
13426 13427 13428 13429 13430 13431 13432 13433
            return -1;
    }

    VIR_DEBUG("Setup all disks");
    return 0;
}


13434
static int
J
Ján Tomko 已提交
13435
qemuDomainSetupHostdev(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
13436
                       virDomainHostdevDefPtr dev,
13437
                       const struct qemuDomainCreateDeviceData *data)
13438 13439
{
    int ret = -1;
13440 13441
    char **path = NULL;
    size_t i, npaths = 0;
13442

13443
    if (qemuDomainGetHostdevPath(NULL, dev, false, &npaths, &path, NULL) < 0)
13444 13445
        goto cleanup;

13446
    for (i = 0; i < npaths; i++) {
13447
        if (qemuDomainCreateDevice(path[i], data, false) < 0)
13448
            goto cleanup;
13449 13450 13451 13452
    }

    ret = 0;
 cleanup:
13453 13454
    for (i = 0; i < npaths; i++)
        VIR_FREE(path[i]);
13455 13456 13457 13458 13459 13460
    VIR_FREE(path);
    return ret;
}


static int
13461
qemuDomainSetupAllHostdevs(virQEMUDriverConfigPtr cfg,
13462
                           virDomainObjPtr vm,
13463
                           const struct qemuDomainCreateDeviceData *data)
13464 13465 13466 13467 13468
{
    size_t i;

    VIR_DEBUG("Setting up hostdevs");
    for (i = 0; i < vm->def->nhostdevs; i++) {
13469
        if (qemuDomainSetupHostdev(cfg,
13470
                                   vm->def->hostdevs[i],
13471
                                   data) < 0)
13472 13473 13474 13475 13476 13477 13478
            return -1;
    }
    VIR_DEBUG("Setup all hostdevs");
    return 0;
}


M
Michal Privoznik 已提交
13479
static int
J
Ján Tomko 已提交
13480
qemuDomainSetupMemory(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
M
Michal Privoznik 已提交
13481
                      virDomainMemoryDefPtr mem,
13482
                      const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
13483 13484 13485 13486
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

13487
    return qemuDomainCreateDevice(mem->nvdimmPath, data, false);
M
Michal Privoznik 已提交
13488 13489 13490 13491 13492 13493
}


static int
qemuDomainSetupAllMemories(virQEMUDriverConfigPtr cfg,
                           virDomainObjPtr vm,
13494
                           const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
13495 13496 13497 13498 13499 13500 13501
{
    size_t i;

    VIR_DEBUG("Setting up memories");
    for (i = 0; i < vm->def->nmems; i++) {
        if (qemuDomainSetupMemory(cfg,
                                  vm->def->mems[i],
13502
                                  data) < 0)
M
Michal Privoznik 已提交
13503 13504 13505 13506 13507 13508 13509
            return -1;
    }
    VIR_DEBUG("Setup all memories");
    return 0;
}


13510
static int
J
Ján Tomko 已提交
13511
qemuDomainSetupChardev(virDomainDefPtr def G_GNUC_UNUSED,
13512 13513 13514
                       virDomainChrDefPtr dev,
                       void *opaque)
{
13515
    const struct qemuDomainCreateDeviceData *data = opaque;
13516
    const char *path = NULL;
13517

13518
    if (!(path = virDomainChrSourceDefGetPath(dev->source)))
13519 13520
        return 0;

13521 13522 13523 13524 13525 13526
    /* Socket created by qemu. It doesn't exist upfront. */
    if (dev->source->type == VIR_DOMAIN_CHR_TYPE_UNIX &&
        dev->source->data.nix.listen)
        return 0;

    return qemuDomainCreateDevice(path, data, true);
13527 13528 13529 13530
}


static int
J
Ján Tomko 已提交
13531
qemuDomainSetupAllChardevs(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
13532
                           virDomainObjPtr vm,
13533
                           const struct qemuDomainCreateDeviceData *data)
13534 13535 13536 13537 13538 13539
{
    VIR_DEBUG("Setting up chardevs");

    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuDomainSetupChardev,
13540
                               (void *)data) < 0)
13541 13542 13543 13544 13545 13546 13547
        return -1;

    VIR_DEBUG("Setup all chardevs");
    return 0;
}


13548
static int
J
Ján Tomko 已提交
13549
qemuDomainSetupTPM(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
13550
                   virDomainObjPtr vm,
13551
                   const struct qemuDomainCreateDeviceData *data)
13552 13553 13554 13555 13556 13557 13558 13559 13560 13561 13562
{
    virDomainTPMDefPtr dev = vm->def->tpm;

    if (!dev)
        return 0;

    VIR_DEBUG("Setting up TPM");

    switch (dev->type) {
    case VIR_DOMAIN_TPM_TYPE_PASSTHROUGH:
        if (qemuDomainCreateDevice(dev->data.passthrough.source.data.file.path,
13563
                                   data, false) < 0)
13564 13565 13566
            return -1;
        break;

13567
    case VIR_DOMAIN_TPM_TYPE_EMULATOR:
13568 13569 13570 13571 13572 13573 13574 13575 13576 13577
    case VIR_DOMAIN_TPM_TYPE_LAST:
        /* nada */
        break;
    }

    VIR_DEBUG("Setup TPM");
    return 0;
}


13578
static int
J
Ján Tomko 已提交
13579
qemuDomainSetupGraphics(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
13580
                        virDomainGraphicsDefPtr gfx,
13581
                        const struct qemuDomainCreateDeviceData *data)
13582
{
13583
    const char *rendernode = virDomainGraphicsGetRenderNode(gfx);
13584

13585
    if (!rendernode)
13586 13587
        return 0;

13588
    return qemuDomainCreateDevice(rendernode, data, false);
13589 13590 13591 13592
}


static int
13593
qemuDomainSetupAllGraphics(virQEMUDriverConfigPtr cfg,
13594
                           virDomainObjPtr vm,
13595
                           const struct qemuDomainCreateDeviceData *data)
13596 13597 13598 13599 13600
{
    size_t i;

    VIR_DEBUG("Setting up graphics");
    for (i = 0; i < vm->def->ngraphics; i++) {
13601
        if (qemuDomainSetupGraphics(cfg,
13602
                                    vm->def->graphics[i],
13603
                                    data) < 0)
13604 13605 13606 13607 13608 13609 13610 13611
            return -1;
    }

    VIR_DEBUG("Setup all graphics");
    return 0;
}


13612
static int
J
Ján Tomko 已提交
13613
qemuDomainSetupInput(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
13614
                     virDomainInputDefPtr input,
13615
                     const struct qemuDomainCreateDeviceData *data)
13616
{
J
Ján Tomko 已提交
13617
    const char *path = virDomainInputDefGetPath(input);
13618

J
Ján Tomko 已提交
13619 13620
    if (path && qemuDomainCreateDevice(path, data, false) < 0)
        return -1;
13621

J
Ján Tomko 已提交
13622
    return 0;
13623 13624 13625 13626
}


static int
13627
qemuDomainSetupAllInputs(virQEMUDriverConfigPtr cfg,
13628
                         virDomainObjPtr vm,
13629
                         const struct qemuDomainCreateDeviceData *data)
13630 13631 13632
{
    size_t i;

13633
    VIR_DEBUG("Setting up inputs");
13634
    for (i = 0; i < vm->def->ninputs; i++) {
13635
        if (qemuDomainSetupInput(cfg,
13636
                                 vm->def->inputs[i],
13637
                                 data) < 0)
13638 13639
            return -1;
    }
13640
    VIR_DEBUG("Setup all inputs");
13641 13642 13643 13644
    return 0;
}


13645
static int
J
Ján Tomko 已提交
13646
qemuDomainSetupRNG(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
13647
                   virDomainRNGDefPtr rng,
13648
                   const struct qemuDomainCreateDeviceData *data)
13649 13650 13651
{
    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
13652
        if (qemuDomainCreateDevice(rng->source.file, data, false) < 0)
13653 13654 13655 13656 13657 13658 13659 13660 13661 13662 13663 13664 13665
            return -1;

    case VIR_DOMAIN_RNG_BACKEND_EGD:
    case VIR_DOMAIN_RNG_BACKEND_LAST:
        /* nada */
        break;
    }

    return 0;
}


static int
13666
qemuDomainSetupAllRNGs(virQEMUDriverConfigPtr cfg,
13667
                       virDomainObjPtr vm,
13668
                       const struct qemuDomainCreateDeviceData *data)
13669 13670 13671 13672 13673
{
    size_t i;

    VIR_DEBUG("Setting up RNGs");
    for (i = 0; i < vm->def->nrngs; i++) {
13674
        if (qemuDomainSetupRNG(cfg,
13675
                               vm->def->rngs[i],
13676
                               data) < 0)
13677 13678 13679 13680 13681 13682 13683 13684
            return -1;
    }

    VIR_DEBUG("Setup all RNGs");
    return 0;
}


13685
static int
J
Ján Tomko 已提交
13686
qemuDomainSetupLoader(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
13687 13688 13689 13690 13691 13692 13693 13694 13695 13696 13697 13698 13699 13700 13701 13702 13703 13704 13705 13706 13707 13708 13709 13710
                      virDomainObjPtr vm,
                      const struct qemuDomainCreateDeviceData *data)
{
    virDomainLoaderDefPtr loader = vm->def->os.loader;
    int ret = -1;

    VIR_DEBUG("Setting up loader");

    if (loader) {
        switch ((virDomainLoader) loader->type) {
        case VIR_DOMAIN_LOADER_TYPE_ROM:
            if (qemuDomainCreateDevice(loader->path, data, false) < 0)
                goto cleanup;
            break;

        case VIR_DOMAIN_LOADER_TYPE_PFLASH:
            if (qemuDomainCreateDevice(loader->path, data, false) < 0)
                goto cleanup;

            if (loader->nvram &&
                qemuDomainCreateDevice(loader->nvram, data, false) < 0)
                goto cleanup;
            break;

13711
        case VIR_DOMAIN_LOADER_TYPE_NONE:
13712 13713 13714 13715 13716 13717 13718 13719 13720 13721 13722 13723
        case VIR_DOMAIN_LOADER_TYPE_LAST:
            break;
        }
    }

    VIR_DEBUG("Setup loader");
    ret = 0;
 cleanup:
    return ret;
}


13724
static int
J
Ján Tomko 已提交
13725
qemuDomainSetupLaunchSecurity(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
13726 13727 13728 13729 13730 13731 13732 13733 13734 13735
                              virDomainObjPtr vm,
                              const struct qemuDomainCreateDeviceData *data)
{
    virDomainSEVDefPtr sev = vm->def->sev;

    if (!sev || sev->sectype != VIR_DOMAIN_LAUNCH_SECURITY_SEV)
        return 0;

    VIR_DEBUG("Setting up launch security");

13736
    if (qemuDomainCreateDevice(QEMU_DEV_SEV, data, false) < 0)
13737 13738 13739 13740 13741 13742 13743
        return -1;

    VIR_DEBUG("Set up launch security");
    return 0;
}


13744
int
13745 13746
qemuDomainBuildNamespace(virQEMUDriverConfigPtr cfg,
                         virSecurityManagerPtr mgr,
13747 13748
                         virDomainObjPtr vm)
{
13749
    struct qemuDomainCreateDeviceData data;
13750
    char *devPath = NULL;
13751
    char **devMountsPath = NULL, **devMountsSavePath = NULL;
13752 13753 13754 13755 13756 13757 13758 13759
    size_t ndevMountsPath = 0, i;
    int ret = -1;

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT)) {
        ret = 0;
        goto cleanup;
    }

13760
    if (qemuDomainGetPreservedMounts(cfg, vm,
13761 13762
                                     &devMountsPath, &devMountsSavePath,
                                     &ndevMountsPath) < 0)
13763 13764
        goto cleanup;

13765 13766 13767 13768 13769 13770 13771 13772 13773 13774 13775 13776 13777
    for (i = 0; i < ndevMountsPath; i++) {
        if (STREQ(devMountsPath[i], "/dev")) {
            devPath = devMountsSavePath[i];
            break;
        }
    }

    if (!devPath) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Unable to find any /dev mount"));
        goto cleanup;
    }

13778
    data.path = devPath;
13779 13780
    data.devMountsPath = devMountsPath;
    data.ndevMountsPath = ndevMountsPath;
13781

13782 13783 13784
    if (virProcessSetupPrivateMountNS() < 0)
        goto cleanup;

13785
    if (qemuDomainSetupDev(cfg, mgr, vm, &data) < 0)
13786 13787
        goto cleanup;

13788
    if (qemuDomainSetupAllDisks(cfg, vm, &data) < 0)
13789 13790
        goto cleanup;

13791
    if (qemuDomainSetupAllHostdevs(cfg, vm, &data) < 0)
13792 13793
        goto cleanup;

13794
    if (qemuDomainSetupAllMemories(cfg, vm, &data) < 0)
13795 13796
        goto cleanup;

13797
    if (qemuDomainSetupAllChardevs(cfg, vm, &data) < 0)
13798 13799
        goto cleanup;

13800
    if (qemuDomainSetupTPM(cfg, vm, &data) < 0)
13801 13802
        goto cleanup;

13803
    if (qemuDomainSetupAllGraphics(cfg, vm, &data) < 0)
13804 13805
        goto cleanup;

13806
    if (qemuDomainSetupAllInputs(cfg, vm, &data) < 0)
13807 13808
        goto cleanup;

13809
    if (qemuDomainSetupAllRNGs(cfg, vm, &data) < 0)
13810 13811
        goto cleanup;

13812 13813 13814
    if (qemuDomainSetupLoader(cfg, vm, &data) < 0)
        goto cleanup;

13815 13816 13817
    if (qemuDomainSetupLaunchSecurity(cfg, vm, &data) < 0)
        goto cleanup;

13818 13819
    /* Save some mount points because we want to share them with the host */
    for (i = 0; i < ndevMountsPath; i++) {
13820 13821
        struct stat sb;

13822 13823 13824
        if (devMountsSavePath[i] == devPath)
            continue;

13825 13826 13827 13828 13829 13830 13831
        if (stat(devMountsPath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsPath[i]);
            goto cleanup;
        }

13832 13833 13834
        /* At this point, devMountsPath is either:
         * a file (regular or special), or
         * a directory. */
13835
        if ((S_ISDIR(sb.st_mode) && virFileMakePath(devMountsSavePath[i]) < 0) ||
13836
            (!S_ISDIR(sb.st_mode) && virFileTouch(devMountsSavePath[i], sb.st_mode) < 0)) {
13837 13838 13839 13840 13841 13842
            virReportSystemError(errno,
                                 _("Failed to create %s"),
                                 devMountsSavePath[i]);
            goto cleanup;
        }

13843
        if (virFileMoveMount(devMountsPath[i], devMountsSavePath[i]) < 0)
13844 13845 13846
            goto cleanup;
    }

13847
    if (virFileMoveMount(devPath, "/dev") < 0)
13848 13849 13850
        goto cleanup;

    for (i = 0; i < ndevMountsPath; i++) {
13851 13852
        struct stat sb;

13853 13854 13855
        if (devMountsSavePath[i] == devPath)
            continue;

13856 13857 13858 13859
        if (stat(devMountsSavePath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsSavePath[i]);
13860 13861 13862
            goto cleanup;
        }

13863 13864 13865 13866 13867 13868 13869 13870 13871 13872 13873 13874 13875 13876 13877
        if (S_ISDIR(sb.st_mode)) {
            if (virFileMakePath(devMountsPath[i]) < 0) {
                virReportSystemError(errno, _("Cannot create %s"),
                                     devMountsPath[i]);
                goto cleanup;
            }
        } else {
            if (virFileMakeParentPath(devMountsPath[i]) < 0 ||
                virFileTouch(devMountsPath[i], sb.st_mode) < 0) {
                virReportSystemError(errno, _("Cannot create %s"),
                                     devMountsPath[i]);
                goto cleanup;
            }
        }

13878
        if (virFileMoveMount(devMountsSavePath[i], devMountsPath[i]) < 0)
13879 13880 13881 13882 13883
            goto cleanup;
    }

    ret = 0;
 cleanup:
13884 13885 13886 13887 13888 13889 13890
    for (i = 0; i < ndevMountsPath; i++) {
        /* The path can be either a regular file or a dir. */
        if (virFileIsDir(devMountsSavePath[i]))
            rmdir(devMountsSavePath[i]);
        else
            unlink(devMountsSavePath[i]);
    }
13891
    virStringListFreeCount(devMountsPath, ndevMountsPath);
13892
    virStringListFreeCount(devMountsSavePath, ndevMountsPath);
13893 13894 13895 13896 13897 13898 13899 13900 13901 13902 13903
    return ret;
}


int
qemuDomainCreateNamespace(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

13904 13905
    if (virBitmapIsBitSet(cfg->namespaces, QEMU_DOMAIN_NS_MOUNT) &&
        qemuDomainEnableNamespace(vm, QEMU_DOMAIN_NS_MOUNT) < 0)
M
Michal Privoznik 已提交
13906
        goto cleanup;
13907 13908 13909 13910 13911 13912 13913 13914

    ret = 0;
 cleanup:
    virObjectUnref(cfg);
    return ret;
}


13915
void
J
Ján Tomko 已提交
13916
qemuDomainDestroyNamespace(virQEMUDriverPtr driver G_GNUC_UNUSED,
13917 13918 13919 13920 13921 13922 13923
                           virDomainObjPtr vm)
{
    if (qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        qemuDomainDisableNamespace(vm, QEMU_DOMAIN_NS_MOUNT);
}


13924
bool
J
Ján Tomko 已提交
13925
qemuDomainNamespaceAvailable(qemuDomainNamespace ns G_GNUC_UNUSED)
13926 13927 13928 13929 13930 13931 13932 13933 13934 13935 13936 13937 13938 13939 13940 13941 13942 13943 13944 13945 13946 13947 13948 13949 13950 13951 13952
{
#if !defined(__linux__)
    /* Namespaces are Linux specific. */
    return false;

#else /* defined(__linux__) */

    switch (ns) {
    case QEMU_DOMAIN_NS_MOUNT:
# if !defined(HAVE_SYS_ACL_H) || !defined(WITH_SELINUX)
        /* We can't create the exact copy of paths if either of
         * these is not available. */
        return false;
# else
        if (virProcessNamespaceAvailable(VIR_PROCESS_NAMESPACE_MNT) < 0)
            return false;
# endif
        break;
    case QEMU_DOMAIN_NS_LAST:
        break;
    }

    return true;
#endif /* defined(__linux__) */
}


13953 13954 13955 13956
struct qemuDomainAttachDeviceMknodData {
    virQEMUDriverPtr driver;
    virDomainObjPtr vm;
    const char *file;
13957
    const char *target;
13958 13959
    struct stat sb;
    void *acl;
13960 13961 13962
#ifdef WITH_SELINUX
    char *tcon;
#endif
13963 13964 13965
};


13966 13967
/* Our way of creating devices is highly linux specific */
#if defined(__linux__)
13968
static int
J
Ján Tomko 已提交
13969
qemuDomainAttachDeviceMknodHelper(pid_t pid G_GNUC_UNUSED,
13970 13971 13972 13973
                                  void *opaque)
{
    struct qemuDomainAttachDeviceMknodData *data = opaque;
    int ret = -1;
13974
    bool delDevice = false;
13975
    bool isLink = S_ISLNK(data->sb.st_mode);
13976
    bool isDev = S_ISCHR(data->sb.st_mode) || S_ISBLK(data->sb.st_mode);
13977
    bool isReg = S_ISREG(data->sb.st_mode) || S_ISFIFO(data->sb.st_mode) || S_ISSOCK(data->sb.st_mode);
13978
    bool isDir = S_ISDIR(data->sb.st_mode);
13979

13980
    qemuSecurityPostFork(data->driver->securityManager);
13981 13982 13983 13984 13985 13986 13987

    if (virFileMakeParentPath(data->file) < 0) {
        virReportSystemError(errno,
                             _("Unable to create %s"), data->file);
        goto cleanup;
    }

13988 13989
    if (isLink) {
        VIR_DEBUG("Creating symlink %s -> %s", data->file, data->target);
13990 13991 13992 13993 13994 13995 13996 13997 13998 13999 14000 14001

        /* First, unlink the symlink target. Symlinks change and
         * therefore we have no guarantees that pre-existing
         * symlink is still valid. */
        if (unlink(data->file) < 0 &&
            errno != ENOENT) {
            virReportSystemError(errno,
                                 _("Unable to remove symlink %s"),
                                 data->file);
            goto cleanup;
        }

14002
        if (symlink(data->target, data->file) < 0) {
14003 14004 14005 14006
            virReportSystemError(errno,
                                 _("Unable to create symlink %s (pointing to %s)"),
                                 data->file, data->target);
            goto cleanup;
14007 14008
        } else {
            delDevice = true;
14009
        }
14010
    } else if (isDev) {
14011 14012 14013 14014 14015 14016 14017 14018 14019 14020 14021 14022 14023 14024 14025 14026
        VIR_DEBUG("Creating dev %s (%d,%d)",
                  data->file, major(data->sb.st_rdev), minor(data->sb.st_rdev));
        if (mknod(data->file, data->sb.st_mode, data->sb.st_rdev) < 0) {
            /* Because we are not removing devices on hotunplug, or
             * we might be creating part of backing chain that
             * already exist due to a different disk plugged to
             * domain, accept EEXIST. */
            if (errno != EEXIST) {
                virReportSystemError(errno,
                                     _("Unable to create device %s"),
                                     data->file);
                goto cleanup;
            }
        } else {
            delDevice = true;
        }
14027
    } else if (isReg || isDir) {
14028 14029 14030 14031 14032
        /* We are not cleaning up disks on virDomainDetachDevice
         * because disk might be still in use by different disk
         * as its backing chain. This might however clash here.
         * Therefore do the cleanup here. */
        if (umount(data->file) < 0 &&
14033
            errno != ENOENT && errno != EINVAL) {
14034 14035 14036 14037 14038
            virReportSystemError(errno,
                                 _("Unable to umount %s"),
                                 data->file);
            goto cleanup;
        }
14039 14040
        if ((isReg && virFileTouch(data->file, data->sb.st_mode) < 0) ||
            (isDir && virFileMakePathWithMode(data->file, data->sb.st_mode) < 0))
14041 14042 14043 14044
            goto cleanup;
        delDevice = true;
        /* Just create the file here so that code below sets
         * proper owner and mode. Move the mount only after that. */
14045 14046 14047 14048 14049
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       data->file, data->sb.st_mode);
        goto cleanup;
14050 14051
    }

14052 14053 14054 14055 14056 14057 14058
    if (lchown(data->file, data->sb.st_uid, data->sb.st_gid) < 0) {
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             data->file);
        goto cleanup;
    }

14059 14060 14061 14062 14063 14064 14065 14066 14067
    /* Symlinks don't have mode */
    if (!isLink &&
        chmod(data->file, data->sb.st_mode) < 0) {
        virReportSystemError(errno,
                             _("Failed to set permissions for device %s"),
                             data->file);
        goto cleanup;
    }

14068 14069 14070
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileSetACLs(data->file, data->acl) < 0 &&
14071 14072 14073 14074 14075 14076
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to set ACLs on %s"), data->file);
        goto cleanup;
    }

14077
# ifdef WITH_SELINUX
14078
    if (data->tcon &&
14079
        lsetfilecon_raw(data->file, (const char *)data->tcon) < 0) {
14080 14081 14082 14083 14084 14085 14086 14087 14088
        VIR_WARNINGS_NO_WLOGICALOP_EQUAL_EXPR
        if (errno != EOPNOTSUPP && errno != ENOTSUP) {
        VIR_WARNINGS_RESET
            virReportSystemError(errno,
                                 _("Unable to set SELinux label on %s"),
                                 data->file);
            goto cleanup;
        }
    }
14089
# endif
14090

14091
    /* Finish mount process started earlier. */
14092
    if ((isReg || isDir) &&
14093 14094 14095
        virFileMoveMount(data->target, data->file) < 0)
        goto cleanup;

14096 14097
    ret = 0;
 cleanup:
14098 14099 14100 14101 14102 14103
    if (ret < 0 && delDevice) {
        if (isDir)
            virFileDeleteTree(data->file);
        else
            unlink(data->file);
    }
14104
# ifdef WITH_SELINUX
14105
    freecon(data->tcon);
14106
# endif
14107 14108 14109 14110 14111 14112
    virFileFreeACLs(&data->acl);
    return ret;
}


static int
14113 14114 14115
qemuDomainAttachDeviceMknodRecursive(virQEMUDriverPtr driver,
                                     virDomainObjPtr vm,
                                     const char *file,
14116 14117
                                     char * const *devMountsPath,
                                     size_t ndevMountsPath,
14118
                                     unsigned int ttl)
14119
{
14120
    virQEMUDriverConfigPtr cfg = NULL;
14121 14122
    struct qemuDomainAttachDeviceMknodData data;
    int ret = -1;
14123 14124
    char *target = NULL;
    bool isLink;
14125
    bool isReg;
14126
    bool isDir;
14127

14128 14129 14130 14131 14132 14133 14134
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             file);
        return ret;
    }

14135 14136 14137 14138 14139 14140
    memset(&data, 0, sizeof(data));

    data.driver = driver;
    data.vm = vm;
    data.file = file;

14141
    if (lstat(file, &data.sb) < 0) {
14142 14143 14144 14145 14146
        virReportSystemError(errno,
                             _("Unable to access %s"), file);
        return ret;
    }

14147
    isLink = S_ISLNK(data.sb.st_mode);
14148
    isReg = S_ISREG(data.sb.st_mode) || S_ISFIFO(data.sb.st_mode) || S_ISSOCK(data.sb.st_mode);
14149
    isDir = S_ISDIR(data.sb.st_mode);
14150

14151
    if ((isReg || isDir) && STRPREFIX(file, QEMU_DEVPREFIX)) {
14152 14153 14154 14155 14156 14157 14158 14159 14160
        cfg = virQEMUDriverGetConfig(driver);
        if (!(target = qemuDomainGetPreservedMountPath(cfg, vm, file)))
            goto cleanup;

        if (virFileBindMountDevice(file, target) < 0)
            goto cleanup;

        data.target = target;
    } else if (isLink) {
14161 14162 14163 14164 14165 14166 14167 14168 14169 14170
        if (virFileReadLink(file, &target) < 0) {
            virReportSystemError(errno,
                                 _("unable to resolve symlink %s"),
                                 file);
            return ret;
        }

        if (IS_RELATIVE_FILE_NAME(target)) {
            char *c = NULL, *tmp = NULL, *fileTmp = NULL;

14171
            fileTmp = g_strdup(file);
14172 14173 14174 14175

            if ((c = strrchr(fileTmp, '/')))
                *(c + 1) = '\0';

14176
            tmp = g_strdup_printf("%s%s", fileTmp, target);
14177 14178
            VIR_FREE(fileTmp);
            VIR_FREE(target);
14179
            target = g_steal_pointer(&tmp);
14180 14181 14182 14183 14184 14185 14186 14187
        }

        data.target = target;
    }

    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileGetACLs(file, &data.acl) < 0 &&
14188 14189 14190
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to get ACLs on %s"), file);
14191
        goto cleanup;
14192 14193
    }

14194
# ifdef WITH_SELINUX
14195
    if (lgetfilecon_raw(file, &data.tcon) < 0 &&
14196 14197 14198 14199 14200
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"), file);
        goto cleanup;
    }
14201
# endif
14202

14203
    if (STRPREFIX(file, QEMU_DEVPREFIX)) {
14204 14205 14206 14207 14208 14209 14210 14211 14212 14213 14214 14215
        size_t i;

        for (i = 0; i < ndevMountsPath; i++) {
            if (STREQ(devMountsPath[i], "/dev"))
                continue;
            if (STRPREFIX(file, devMountsPath[i]))
                break;
        }

        if (i == ndevMountsPath) {
            if (qemuSecurityPreFork(driver->securityManager) < 0)
                goto cleanup;
14216

14217 14218 14219 14220 14221 14222
            if (virProcessRunInMountNamespace(vm->pid,
                                              qemuDomainAttachDeviceMknodHelper,
                                              &data) < 0) {
                qemuSecurityPostFork(driver->securityManager);
                goto cleanup;
            }
14223
            qemuSecurityPostFork(driver->securityManager);
14224 14225 14226
        } else {
            VIR_DEBUG("Skipping dev %s because of %s mount point",
                      file, devMountsPath[i]);
14227
        }
14228 14229
    }

14230
    if (isLink &&
14231 14232 14233
        qemuDomainAttachDeviceMknodRecursive(driver, vm, target,
                                             devMountsPath, ndevMountsPath,
                                             ttl -1) < 0)
14234
        goto cleanup;
14235 14236 14237

    ret = 0;
 cleanup:
14238
# ifdef WITH_SELINUX
14239
    freecon(data.tcon);
14240
# endif
14241
    virFileFreeACLs(&data.acl);
14242 14243
    if (isReg && target)
        umount(target);
14244
    VIR_FREE(target);
14245
    virObjectUnref(cfg);
14246
    return ret;
14247 14248 14249
}


14250 14251 14252 14253
#else /* !defined(__linux__) */


static int
J
Ján Tomko 已提交
14254 14255 14256 14257 14258 14259
qemuDomainAttachDeviceMknodRecursive(virQEMUDriverPtr driver G_GNUC_UNUSED,
                                     virDomainObjPtr vm G_GNUC_UNUSED,
                                     const char *file G_GNUC_UNUSED,
                                     char * const *devMountsPath G_GNUC_UNUSED,
                                     size_t ndevMountsPath G_GNUC_UNUSED,
                                     unsigned int ttl G_GNUC_UNUSED)
14260 14261 14262 14263 14264 14265 14266 14267 14268 14269
{
    virReportSystemError(ENOSYS, "%s",
                         _("Namespaces are not supported on this platform."));
    return -1;
}


#endif /* !defined(__linux__) */


14270 14271 14272
static int
qemuDomainAttachDeviceMknod(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
14273 14274 14275
                            const char *file,
                            char * const *devMountsPath,
                            size_t ndevMountsPath)
14276 14277 14278
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

14279 14280 14281
    return qemuDomainAttachDeviceMknodRecursive(driver, vm, file,
                                                devMountsPath, ndevMountsPath,
                                                symloop_max);
14282 14283 14284
}


14285
static int
J
Ján Tomko 已提交
14286
qemuDomainDetachDeviceUnlinkHelper(pid_t pid G_GNUC_UNUSED,
14287 14288 14289 14290 14291 14292 14293 14294 14295 14296 14297 14298 14299 14300 14301 14302
                                   void *opaque)
{
    const char *path = opaque;

    VIR_DEBUG("Unlinking %s", path);
    if (unlink(path) < 0 && errno != ENOENT) {
        virReportSystemError(errno,
                             _("Unable to remove device %s"), path);
        return -1;
    }

    return 0;
}


static int
J
Ján Tomko 已提交
14303
qemuDomainDetachDeviceUnlink(virQEMUDriverPtr driver G_GNUC_UNUSED,
14304
                             virDomainObjPtr vm,
14305 14306 14307
                             const char *file,
                             char * const *devMountsPath,
                             size_t ndevMountsPath)
14308
{
14309 14310
    int ret = -1;
    size_t i;
14311

14312
    if (STRPREFIX(file, QEMU_DEVPREFIX)) {
14313 14314 14315 14316 14317 14318 14319 14320 14321 14322 14323 14324 14325 14326 14327 14328 14329 14330
        for (i = 0; i < ndevMountsPath; i++) {
            if (STREQ(devMountsPath[i], "/dev"))
                continue;
            if (STRPREFIX(file, devMountsPath[i]))
                break;
        }

        if (i == ndevMountsPath) {
            if (virProcessRunInMountNamespace(vm->pid,
                                              qemuDomainDetachDeviceUnlinkHelper,
                                              (void *)file) < 0)
                goto cleanup;
        }
    }

    ret = 0;
 cleanup:
    return ret;
14331 14332 14333
}


14334 14335 14336 14337
static int
qemuDomainNamespaceMknodPaths(virDomainObjPtr vm,
                              const char **paths,
                              size_t npaths)
14338
{
14339 14340 14341
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    virQEMUDriverConfigPtr cfg;
14342 14343
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0;
14344
    int ret = -1;
14345
    size_t i;
14346

14347 14348
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
14349 14350
        return 0;

14351 14352 14353 14354 14355 14356
    cfg = virQEMUDriverGetConfig(driver);
    if (qemuDomainGetPreservedMounts(cfg, vm,
                                     &devMountsPath, NULL,
                                     &ndevMountsPath) < 0)
        goto cleanup;

14357 14358 14359 14360 14361 14362 14363 14364 14365 14366 14367 14368 14369 14370 14371 14372
    for (i = 0; i < npaths; i++) {
        if (qemuDomainAttachDeviceMknod(driver,
                                        vm,
                                        paths[i],
                                        devMountsPath, ndevMountsPath) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    virObjectUnref(cfg);
    return ret;
}


14373 14374 14375 14376 14377 14378 14379 14380 14381 14382
static int
qemuDomainNamespaceMknodPath(virDomainObjPtr vm,
                             const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceMknodPaths(vm, paths, 1);
}


14383 14384 14385 14386 14387 14388 14389 14390 14391 14392 14393 14394 14395
static int
qemuDomainNamespaceUnlinkPaths(virDomainObjPtr vm,
                               const char **paths,
                               size_t npaths)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    virQEMUDriverConfigPtr cfg;
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0;
    size_t i;
    int ret = -1;

14396 14397
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
14398 14399 14400 14401 14402 14403 14404 14405 14406 14407 14408 14409 14410 14411 14412
        return 0;

    cfg = virQEMUDriverGetConfig(driver);

    if (qemuDomainGetPreservedMounts(cfg, vm,
                                     &devMountsPath, NULL,
                                     &ndevMountsPath) < 0)
        goto cleanup;

    for (i = 0; i < npaths; i++) {
        if (qemuDomainDetachDeviceUnlink(driver, vm, paths[i],
                                         devMountsPath, ndevMountsPath) < 0)
            goto cleanup;
    }

14413
    ret = 0;
14414 14415 14416 14417 14418 14419 14420
 cleanup:
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    virObjectUnref(cfg);
    return ret;
}


14421 14422 14423 14424 14425 14426 14427 14428 14429 14430
static int
qemuDomainNamespaceUnlinkPath(virDomainObjPtr vm,
                              const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceUnlinkPaths(vm, paths, 1);
}


14431
int
14432
qemuDomainNamespaceSetupDisk(virDomainObjPtr vm,
14433 14434 14435
                             virStorageSourcePtr src)
{
    virStorageSourcePtr next;
14436
    const char **paths = NULL;
14437
    size_t npaths = 0;
14438
    char *dmPath = NULL;
14439 14440
    int ret = -1;

14441
    for (next = src; virStorageSourceIsBacking(next); next = next->backingStore) {
14442 14443
        if (virStorageSourceIsEmpty(next) ||
            !virStorageSourceIsLocalStorage(next)) {
14444 14445 14446 14447
            /* Not creating device. Just continue. */
            continue;
        }

14448
        if (VIR_APPEND_ELEMENT_COPY(paths, npaths, next->path) < 0)
14449 14450 14451
            goto cleanup;
    }

14452
    /* qemu-pr-helper might require access to /dev/mapper/control. */
14453 14454 14455 14456 14457
    if (src->pr) {
        dmPath = g_strdup(QEMU_DEVICE_MAPPER_CONTROL_PATH);
        if (VIR_APPEND_ELEMENT_COPY(paths, npaths, dmPath) < 0)
            goto cleanup;
    }
14458

14459
    if (qemuDomainNamespaceMknodPaths(vm, paths, npaths) < 0)
14460
        goto cleanup;
14461

14462 14463
    ret = 0;
 cleanup:
14464
    VIR_FREE(dmPath);
14465
    VIR_FREE(paths);
14466 14467 14468 14469 14470
    return ret;
}


int
J
Ján Tomko 已提交
14471 14472
qemuDomainNamespaceTeardownDisk(virDomainObjPtr vm G_GNUC_UNUSED,
                                virStorageSourcePtr src G_GNUC_UNUSED)
14473 14474 14475 14476 14477 14478 14479 14480 14481
{
    /* While in hotplug case we create the whole backing chain,
     * here we must limit ourselves. The disk we want to remove
     * might be a part of backing chain of another disk.
     * If you are reading these lines and have some spare time
     * you can come up with and algorithm that checks for that.
     * I don't, therefore: */
    return 0;
}
14482 14483 14484


int
14485
qemuDomainNamespaceSetupHostdev(virDomainObjPtr vm,
14486 14487 14488
                                virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
14489
    char **paths = NULL;
14490
    size_t i, npaths = 0;
14491

14492
    if (qemuDomainGetHostdevPath(NULL, hostdev, false, &npaths, &paths, NULL) < 0)
14493 14494
        goto cleanup;

14495
    if (qemuDomainNamespaceMknodPaths(vm, (const char **)paths, npaths) < 0)
14496 14497
        goto cleanup;

14498 14499
    ret = 0;
 cleanup:
14500
    for (i = 0; i < npaths; i++)
14501 14502
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
14503 14504 14505 14506 14507
    return ret;
}


int
14508
qemuDomainNamespaceTeardownHostdev(virDomainObjPtr vm,
14509 14510 14511
                                   virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
14512
    char **paths = NULL;
14513
    size_t i, npaths = 0;
14514

14515
    if (qemuDomainGetHostdevPath(vm->def, hostdev, true,
14516
                                 &npaths, &paths, NULL) < 0)
14517 14518
        goto cleanup;

14519
    if (qemuDomainNamespaceUnlinkPaths(vm, (const char **)paths, npaths) < 0)
14520 14521
        goto cleanup;

14522 14523
    ret = 0;
 cleanup:
14524
    for (i = 0; i < npaths; i++)
14525 14526
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
14527 14528
    return ret;
}
14529 14530


M
Michal Privoznik 已提交
14531
int
14532
qemuDomainNamespaceSetupMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
14533 14534 14535 14536 14537
                               virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

14538 14539
    if (qemuDomainNamespaceMknodPath(vm, mem->nvdimmPath) < 0)
        return -1;
14540

14541
    return 0;
M
Michal Privoznik 已提交
14542 14543 14544 14545
}


int
14546
qemuDomainNamespaceTeardownMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
14547 14548 14549 14550 14551
                                  virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

14552 14553
    if (qemuDomainNamespaceUnlinkPath(vm, mem->nvdimmPath) < 0)
        return -1;
14554

14555
    return 0;
M
Michal Privoznik 已提交
14556 14557 14558
}


14559
int
14560
qemuDomainNamespaceSetupChardev(virDomainObjPtr vm,
14561 14562 14563 14564
                                virDomainChrDefPtr chr)
{
    const char *path;

14565
    if (!(path = virDomainChrSourceDefGetPath(chr->source)))
14566 14567
        return 0;

14568 14569 14570 14571
    /* Socket created by qemu. It doesn't exist upfront. */
    if (chr->source->type == VIR_DOMAIN_CHR_TYPE_UNIX &&
        chr->source->data.nix.listen)
        return 0;
14572

14573 14574
    if (qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
14575

14576
    return 0;
14577 14578 14579 14580
}


int
14581
qemuDomainNamespaceTeardownChardev(virDomainObjPtr vm,
14582 14583 14584 14585 14586 14587 14588 14589 14590
                                   virDomainChrDefPtr chr)
{
    const char *path = NULL;

    if (chr->source->type != VIR_DOMAIN_CHR_TYPE_DEV)
        return 0;

    path = chr->source->data.file.path;

14591 14592
    if (qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
14593

14594
    return 0;
14595
}
14596 14597 14598


int
14599
qemuDomainNamespaceSetupRNG(virDomainObjPtr vm,
14600 14601 14602 14603 14604 14605 14606 14607 14608 14609 14610
                            virDomainRNGDefPtr rng)
{
    const char *path = NULL;

    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
        path = rng->source.file;
        break;

    case VIR_DOMAIN_RNG_BACKEND_EGD:
    case VIR_DOMAIN_RNG_BACKEND_LAST:
14611
        break;
14612 14613
    }

14614 14615
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
14616

14617
    return 0;
14618 14619 14620 14621
}


int
14622
qemuDomainNamespaceTeardownRNG(virDomainObjPtr vm,
14623 14624 14625 14626 14627 14628 14629 14630 14631 14632 14633
                               virDomainRNGDefPtr rng)
{
    const char *path = NULL;

    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
        path = rng->source.file;
        break;

    case VIR_DOMAIN_RNG_BACKEND_EGD:
    case VIR_DOMAIN_RNG_BACKEND_LAST:
14634
        break;
14635 14636
    }

14637 14638
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
14639

14640
    return 0;
14641 14642 14643 14644 14645 14646 14647 14648 14649 14650 14651 14652
}


int
qemuDomainNamespaceSetupInput(virDomainObjPtr vm,
                              virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

14653 14654 14655
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
    return 0;
14656 14657 14658 14659 14660 14661 14662 14663 14664 14665 14666 14667
}


int
qemuDomainNamespaceTeardownInput(virDomainObjPtr vm,
                                 virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

14668 14669
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
14670

14671
    return 0;
14672
}
14673 14674 14675 14676 14677 14678 14679 14680 14681 14682 14683 14684 14685 14686 14687 14688 14689 14690 14691 14692 14693 14694 14695 14696 14697 14698 14699 14700 14701 14702 14703 14704 14705 14706 14707 14708 14709 14710 14711 14712 14713 14714 14715


/**
 * qemuDomainDiskLookupByNodename:
 * @def: domain definition to look for the disk
 * @nodename: block backend node name to find
 * @src: filled with the specific backing store element if provided
 * @idx: index of @src in the backing chain, if provided
 *
 * Looks up the disk in the domain via @nodename and returns its definition.
 * Optionally fills @src and @idx if provided with the specific backing chain
 * element which corresponds to the node name.
 */
virDomainDiskDefPtr
qemuDomainDiskLookupByNodename(virDomainDefPtr def,
                               const char *nodename,
                               virStorageSourcePtr *src,
                               unsigned int *idx)
{
    size_t i;
    unsigned int srcindex;
    virStorageSourcePtr tmp = NULL;

    if (!idx)
        idx = &srcindex;

    if (src)
        *src = NULL;

    *idx = 0;

    for (i = 0; i < def->ndisks; i++) {
        if ((tmp = virStorageSourceFindByNodeName(def->disks[i]->src,
                                                  nodename, idx))) {
            if (src)
                *src = tmp;

            return def->disks[i];
        }
    }

    return NULL;
}
14716 14717 14718 14719 14720 14721 14722 14723 14724 14725


/**
 * qemuDomainDiskBackingStoreGetName:
 *
 * Creates a name using the indexed syntax (vda[1])for the given backing store
 * entry for a disk.
 */
char *
qemuDomainDiskBackingStoreGetName(virDomainDiskDefPtr disk,
J
Ján Tomko 已提交
14726
                                  virStorageSourcePtr src G_GNUC_UNUSED,
14727 14728 14729 14730 14731
                                  unsigned int idx)
{
    char *ret = NULL;

    if (idx)
14732
        ret = g_strdup_printf("%s[%d]", disk->dst, idx);
14733
    else
14734
        ret = g_strdup(disk->dst);
14735 14736 14737

    return ret;
}
14738 14739 14740 14741 14742 14743 14744 14745 14746 14747 14748 14749 14750 14751 14752 14753 14754 14755 14756 14757 14758 14759 14760 14761 14762 14763 14764


virStorageSourcePtr
qemuDomainGetStorageSourceByDevstr(const char *devstr,
                                   virDomainDefPtr def)
{
    virDomainDiskDefPtr disk = NULL;
    virStorageSourcePtr src = NULL;
    char *target = NULL;
    unsigned int idx;
    size_t i;

    if (virStorageFileParseBackingStoreStr(devstr, &target, &idx) < 0) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("failed to parse block device '%s'"), devstr);
        return NULL;
    }

    for (i = 0; i < def->ndisks; i++) {
        if (STREQ(target, def->disks[i]->dst)) {
            disk = def->disks[i];
            break;
        }
    }

    if (!disk) {
        virReportError(VIR_ERR_INVALID_ARG,
Y
Yuri Chornoivan 已提交
14765
                       _("failed to find disk '%s'"), target);
14766 14767 14768
        goto cleanup;
    }

14769 14770 14771 14772
    if (idx == 0)
        src = disk->src;
    else
        src = virStorageFileChainLookup(disk->src, NULL, NULL, idx, NULL);
14773 14774 14775 14776 14777

 cleanup:
    VIR_FREE(target);
    return src;
}
14778 14779 14780 14781 14782 14783 14784 14785


static void
qemuDomainSaveCookieDispose(void *obj)
{
    qemuDomainSaveCookiePtr cookie = obj;

    VIR_DEBUG("cookie=%p", cookie);
14786 14787

    virCPUDefFree(cookie->cpu);
14788 14789 14790 14791
}


qemuDomainSaveCookiePtr
14792
qemuDomainSaveCookieNew(virDomainObjPtr vm)
14793
{
14794
    qemuDomainObjPrivatePtr priv = vm->privateData;
14795 14796 14797 14798 14799 14800 14801 14802
    qemuDomainSaveCookiePtr cookie = NULL;

    if (qemuDomainInitialize() < 0)
        goto error;

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
        goto error;

14803 14804 14805
    if (priv->origCPU && !(cookie->cpu = virCPUDefCopy(vm->def->cpu)))
        goto error;

14806 14807 14808 14809
    cookie->slirpHelper = qemuDomainGetSlirpHelperOk(vm);

    VIR_DEBUG("Save cookie %p, cpu=%p, slirpHelper=%d",
              cookie, cookie->cpu, cookie->slirpHelper);
14810 14811 14812 14813 14814 14815 14816 14817 14818 14819

    return cookie;

 error:
    virObjectUnref(cookie);
    return NULL;
}


static int
J
Ján Tomko 已提交
14820
qemuDomainSaveCookieParse(xmlXPathContextPtr ctxt G_GNUC_UNUSED,
14821 14822 14823 14824 14825 14826 14827 14828 14829 14830
                          virObjectPtr *obj)
{
    qemuDomainSaveCookiePtr cookie = NULL;

    if (qemuDomainInitialize() < 0)
        goto error;

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
        goto error;

14831 14832 14833 14834
    if (virCPUDefParseXML(ctxt, "./cpu[1]", VIR_CPU_TYPE_GUEST,
                          &cookie->cpu) < 0)
        goto error;

14835 14836
    cookie->slirpHelper = virXPathBoolean("boolean(./slirpHelper)", ctxt) > 0;

14837 14838 14839 14840 14841 14842 14843 14844 14845 14846
    *obj = (virObjectPtr) cookie;
    return 0;

 error:
    virObjectUnref(cookie);
    return -1;
}


static int
14847 14848
qemuDomainSaveCookieFormat(virBufferPtr buf,
                           virObjectPtr obj)
14849
{
14850 14851 14852
    qemuDomainSaveCookiePtr cookie = (qemuDomainSaveCookiePtr) obj;

    if (cookie->cpu &&
14853
        virCPUDefFormatBufFull(buf, cookie->cpu, NULL) < 0)
14854 14855
        return -1;

14856 14857 14858
    if (cookie->slirpHelper)
        virBufferAddLit(buf, "<slirpHelper/>\n");

14859 14860 14861 14862 14863 14864 14865 14866
    return 0;
}


virSaveCookieCallbacks virQEMUDriverDomainSaveCookie = {
    .parse = qemuDomainSaveCookieParse,
    .format = qemuDomainSaveCookieFormat,
};
14867 14868 14869 14870 14871 14872 14873 14874 14875 14876 14877 14878 14879 14880 14881 14882 14883 14884 14885 14886 14887 14888 14889 14890 14891 14892 14893 14894 14895 14896 14897 14898 14899 14900 14901 14902 14903 14904 14905 14906


/**
 * qemuDomainUpdateCPU:
 * @vm: domain which is being started
 * @cpu: CPU updated when the domain was running previously (before migration,
 *       snapshot, or save)
 * @origCPU: where to store the original CPU from vm->def in case @cpu was
 *           used instead
 *
 * Replace the CPU definition with the updated one when QEMU is new enough to
 * allow us to check extra features it is about to enable or disable when
 * starting a domain. The original CPU is stored in @origCPU.
 *
 * Returns 0 on success, -1 on error.
 */
int
qemuDomainUpdateCPU(virDomainObjPtr vm,
                    virCPUDefPtr cpu,
                    virCPUDefPtr *origCPU)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    *origCPU = NULL;

    if (!cpu || !vm->def->cpu ||
        !virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION) ||
        virCPUDefIsEqual(vm->def->cpu, cpu, false))
        return 0;

    if (!(cpu = virCPUDefCopy(cpu)))
        return -1;

    VIR_DEBUG("Replacing CPU def with the updated one");

    *origCPU = vm->def->cpu;
    vm->def->cpu = cpu;

    return 0;
}
14907

14908 14909 14910 14911 14912 14913 14914 14915 14916 14917 14918 14919 14920 14921 14922 14923 14924 14925 14926 14927 14928 14929 14930 14931 14932 14933 14934 14935

/**
 * qemuDomainFixupCPUS:
 * @vm: domain object
 * @origCPU: original CPU used when the domain was started
 *
 * Libvirt older than 3.9.0 could have messed up the expansion of host-model
 * CPU when reconnecting to a running domain by adding features QEMU does not
 * support (such as cmt). This API fixes both the actual CPU provided by QEMU
 * (stored in the domain object) and the @origCPU used when starting the
 * domain.
 *
 * This is safe even if the original CPU definition used mode='custom' (rather
 * than host-model) since we know QEMU was able to start the domain and thus
 * the CPU definitions do not contain any features unknown to QEMU.
 *
 * This function can only be used on an active domain or when restoring a
 * domain which was running.
 *
 * Returns 0 on success, -1 on error.
 */
int
qemuDomainFixupCPUs(virDomainObjPtr vm,
                    virCPUDefPtr *origCPU)
{
    virCPUDefPtr fixedCPU = NULL;
    virCPUDefPtr fixedOrig = NULL;
    virArch arch = vm->def->os.arch;
14936
    int ret = -1;
14937 14938 14939 14940 14941 14942 14943 14944 14945 14946 14947 14948 14949 14950 14951 14952 14953 14954 14955 14956 14957 14958 14959 14960 14961 14962 14963 14964 14965 14966

    if (!ARCH_IS_X86(arch))
        return 0;

    if (!vm->def->cpu ||
        vm->def->cpu->mode != VIR_CPU_MODE_CUSTOM ||
        !vm->def->cpu->model)
        return 0;

    /* Missing origCPU means QEMU created exactly the same virtual CPU which
     * we asked for or libvirt was too old to mess up the translation from
     * host-model.
     */
    if (!*origCPU)
        return 0;

    if (virCPUDefFindFeature(vm->def->cpu, "cmt") &&
        (!(fixedCPU = virCPUDefCopyWithoutModel(vm->def->cpu)) ||
         virCPUDefCopyModelFilter(fixedCPU, vm->def->cpu, false,
                                  virQEMUCapsCPUFilterFeatures, &arch) < 0))
        goto cleanup;

    if (virCPUDefFindFeature(*origCPU, "cmt") &&
        (!(fixedOrig = virCPUDefCopyWithoutModel(*origCPU)) ||
         virCPUDefCopyModelFilter(fixedOrig, *origCPU, false,
                                  virQEMUCapsCPUFilterFeatures, &arch) < 0))
        goto cleanup;

    if (fixedCPU) {
        virCPUDefFree(vm->def->cpu);
14967
        vm->def->cpu = g_steal_pointer(&fixedCPU);
14968 14969 14970 14971
    }

    if (fixedOrig) {
        virCPUDefFree(*origCPU);
14972
        *origCPU = g_steal_pointer(&fixedOrig);
14973 14974 14975 14976 14977 14978 14979 14980 14981 14982 14983
    }

    ret = 0;

 cleanup:
    virCPUDefFree(fixedCPU);
    virCPUDefFree(fixedOrig);
    return ret;
}


14984 14985 14986 14987 14988 14989 14990 14991 14992 14993 14994 14995 14996 14997 14998 14999 15000 15001 15002 15003 15004 15005 15006 15007 15008
/**
 * qemuDomainUpdateQEMUCaps:
 * @vm: domain object
 * @qemuCapsCache: cache of QEMU capabilities
 *
 * This function updates the used QEMU capabilities of @vm by querying
 * the QEMU capabilities cache.
 *
 * Returns 0 on success, -1 on error.
 */
int
qemuDomainUpdateQEMUCaps(virDomainObjPtr vm,
                         virFileCachePtr qemuCapsCache)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    virObjectUnref(priv->qemuCaps);
    if (!(priv->qemuCaps = virQEMUCapsCacheLookupCopy(qemuCapsCache,
                                                      vm->def->emulator,
                                                      vm->def->os.machine)))
        return -1;
    return 0;
}


15009 15010 15011 15012 15013 15014 15015
char *
qemuDomainGetMachineName(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    char *ret = NULL;

15016
    if (vm->pid > 0) {
15017 15018 15019 15020 15021 15022 15023 15024 15025 15026 15027
        ret = virSystemdGetMachineNameByPID(vm->pid);
        if (!ret)
            virResetLastError();
    }

    if (!ret)
        ret = virDomainGenerateMachineName("qemu", vm->def->id, vm->def->name,
                                           virQEMUDriverIsPrivileged(driver));

    return ret;
}
15028 15029 15030 15031 15032 15033 15034 15035 15036 15037 15038 15039 15040


/* Check whether the device address is using either 'ccw' or default s390
 * address format and whether that's "legal" for the current qemu and/or
 * guest os.machine type. This is the corollary to the code which doesn't
 * find the address type set using an emulator that supports either 'ccw'
 * or s390 and sets the address type based on the capabilities.
 *
 * If the address is using 'ccw' or s390 and it's not supported, generate
 * an error and return false; otherwise, return true.
 */
bool
qemuDomainCheckCCWS390AddressSupport(const virDomainDef *def,
15041
                                     const virDomainDeviceInfo *info,
15042 15043 15044
                                     virQEMUCapsPtr qemuCaps,
                                     const char *devicename)
{
15045
    if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
15046 15047 15048 15049 15050 15051
        if (!qemuDomainIsS390CCW(def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("cannot use CCW address type for device "
                             "'%s' using machine type '%s'"),
                       devicename, def->os.machine);
            return false;
15052
        } else if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW)) {
15053 15054 15055 15056 15057
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CCW address type is not supported by "
                             "this QEMU"));
            return false;
        }
15058
    } else if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390) {
15059 15060 15061 15062 15063 15064 15065 15066 15067
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_S390)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtio S390 address type is not supported by "
                             "this QEMU"));
            return false;
        }
    }
    return true;
}
15068 15069


15070
/**
15071
 * qemuDomainPrepareDiskSourceData:
15072 15073 15074 15075
 *
 * @disk: Disk config object
 * @src: source to start from
 *
15076
 * Prepares various aspects of a storage source belonging to a disk backing
15077 15078
 * chain based on the disk configuration. This function should be also called
 * for detected backing chain members.
15079
 */
15080
void
15081
qemuDomainPrepareDiskSourceData(virDomainDiskDefPtr disk,
15082
                                virStorageSourcePtr src)
15083
{
15084 15085 15086
    if (!disk)
        return;

15087
    /* transfer properties valid only for the top level image */
15088 15089
    if (src == disk->src)
        src->detect_zeroes = disk->detect_zeroes;
15090

15091 15092 15093 15094
    /* transfer properties valid for the full chain */
    src->iomode = disk->iomode;
    src->cachemode = disk->cachemode;
    src->discard = disk->discard;
15095

15096 15097
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY)
        src->floppyimg = true;
15098 15099 15100
}


15101 15102 15103 15104 15105 15106 15107 15108 15109
static void
qemuDomainPrepareDiskCachemode(virDomainDiskDefPtr disk)
{
    if (disk->cachemode == VIR_DOMAIN_DISK_CACHE_DEFAULT &&
        disk->src->shared && !disk->src->readonly)
        disk->cachemode = VIR_DOMAIN_DISK_CACHE_DISABLE;
}


15110 15111
static int
qemuDomainPrepareStorageSourcePR(virStorageSourcePtr src,
15112 15113
                                 qemuDomainObjPrivatePtr priv,
                                 const char *parentalias)
15114 15115 15116 15117 15118
{
    if (!src->pr)
        return 0;

    if (virStoragePRDefIsManaged(src->pr)) {
15119
        VIR_FREE(src->pr->path);
15120 15121
        if (!(src->pr->path = qemuDomainGetManagedPRSocketPath(priv)))
            return -1;
15122
        src->pr->mgralias = g_strdup(qemuDomainGetManagedPRAlias());
15123 15124 15125
    } else {
        if (!(src->pr->mgralias = qemuDomainGetUnmanagedPRAlias(parentalias)))
            return -1;
15126 15127 15128 15129 15130 15131
    }

    return 0;
}


15132 15133 15134 15135 15136 15137 15138 15139 15140 15141 15142 15143
/**
 * qemuDomainPrepareDiskSourceLegacy:
 * @disk: disk to prepare
 * @priv: VM private data
 * @cfg: qemu driver config
 *
 * Prepare any disk source relevant data for use with the -drive command line.
 */
static int
qemuDomainPrepareDiskSourceLegacy(virDomainDiskDefPtr disk,
                                  qemuDomainObjPrivatePtr priv,
                                  virQEMUDriverConfigPtr cfg)
15144
{
15145 15146 15147
    if (qemuDomainValidateStorageSource(disk->src, priv->qemuCaps) < 0)
        return -1;

15148
    qemuDomainPrepareStorageSourceConfig(disk->src, cfg, priv->qemuCaps);
15149
    qemuDomainPrepareDiskSourceData(disk, disk->src);
15150

15151 15152 15153
    if (qemuDomainSecretStorageSourcePrepare(priv, disk->src,
                                             disk->info.alias,
                                             disk->info.alias) < 0)
15154 15155
        return -1;

15156
    if (qemuDomainPrepareStorageSourcePR(disk->src, priv, disk->info.alias) < 0)
15157
        return -1;
15158

15159 15160
    if (qemuDomainPrepareStorageSourceTLS(disk->src, cfg, disk->info.alias,
                                          priv->qemuCaps) < 0)
15161 15162
        return -1;

15163 15164 15165 15166
    return 0;
}


15167
int
15168 15169 15170 15171 15172 15173 15174
qemuDomainPrepareStorageSourceBlockdev(virDomainDiskDefPtr disk,
                                       virStorageSourcePtr src,
                                       qemuDomainObjPrivatePtr priv,
                                       virQEMUDriverConfigPtr cfg)
{
    src->id = qemuDomainStorageIdNew(priv);

15175 15176
    src->nodestorage = g_strdup_printf("libvirt-%u-storage", src->id);
    src->nodeformat = g_strdup_printf("libvirt-%u-format", src->id);
15177 15178 15179 15180

    if (qemuDomainValidateStorageSource(src, priv->qemuCaps) < 0)
        return -1;

15181
    qemuDomainPrepareStorageSourceConfig(src, cfg, priv->qemuCaps);
15182
    qemuDomainPrepareDiskSourceData(disk, src);
15183 15184 15185 15186 15187 15188

    if (qemuDomainSecretStorageSourcePrepare(priv, src,
                                             src->nodestorage,
                                             src->nodeformat) < 0)
        return -1;

15189
    if (qemuDomainPrepareStorageSourcePR(src, priv, src->nodestorage) < 0)
15190 15191
        return -1;

15192
    if (qemuDomainPrepareStorageSourceTLS(src, cfg, src->nodestorage,
15193 15194 15195 15196 15197 15198 15199 15200 15201 15202 15203 15204 15205 15206 15207 15208
                                          priv->qemuCaps) < 0)
        return -1;

    return 0;
}


static int
qemuDomainPrepareDiskSourceBlockdev(virDomainDiskDefPtr disk,
                                    qemuDomainObjPrivatePtr priv,
                                    virQEMUDriverConfigPtr cfg)
{
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
    virStorageSourcePtr n;

    if (disk->copy_on_read == VIR_TRISTATE_SWITCH_ON &&
15209 15210
        !diskPriv->nodeCopyOnRead)
        diskPriv->nodeCopyOnRead = g_strdup_printf("libvirt-CoR-%s", disk->dst);
15211 15212 15213 15214 15215 15216 15217 15218 15219 15220

    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if (qemuDomainPrepareStorageSourceBlockdev(disk, n, priv, cfg) < 0)
            return -1;
    }

    return 0;
}


15221 15222 15223 15224 15225 15226 15227
int
qemuDomainPrepareDiskSource(virDomainDiskDefPtr disk,
                            qemuDomainObjPrivatePtr priv,
                            virQEMUDriverConfigPtr cfg)
{
    qemuDomainPrepareDiskCachemode(disk);

15228
    /* set default format for storage pool based disks */
15229
    if (disk->src->type == VIR_STORAGE_TYPE_VOLUME &&
15230 15231 15232 15233 15234 15235 15236 15237
        disk->src->format <= VIR_STORAGE_FILE_NONE) {
        int actualType = virStorageSourceGetActualType(disk->src);

        if (actualType == VIR_STORAGE_TYPE_DIR)
            disk->src->format = VIR_STORAGE_FILE_FAT;
        else
            disk->src->format = VIR_STORAGE_FILE_RAW;
    }
15238

15239 15240 15241 15242 15243 15244 15245
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if (qemuDomainPrepareDiskSourceBlockdev(disk, priv, cfg) < 0)
            return -1;
    } else {
        if (qemuDomainPrepareDiskSourceLegacy(disk, priv, cfg) < 0)
            return -1;
    }
15246

15247 15248
    return 0;
}
15249 15250


15251 15252 15253 15254 15255 15256 15257 15258 15259 15260 15261 15262 15263 15264 15265 15266 15267 15268 15269 15270 15271 15272 15273 15274 15275 15276 15277 15278 15279 15280 15281 15282 15283 15284 15285 15286 15287 15288 15289 15290 15291 15292 15293 15294 15295 15296 15297 15298 15299 15300 15301 15302 15303 15304 15305 15306 15307 15308 15309 15310 15311 15312 15313 15314 15315 15316 15317 15318 15319 15320 15321 15322 15323 15324 15325
/**
 * qemuDomainDiskCachemodeFlags:
 *
 * Converts disk cachemode to the cache mode options for qemu. Returns -1 for
 * invalid @cachemode values and fills the flags and returns 0 on success.
 * Flags may be NULL.
 */
int
qemuDomainDiskCachemodeFlags(int cachemode,
                             bool *writeback,
                             bool *direct,
                             bool *noflush)
{
    bool dummy;

    if (!writeback)
        writeback = &dummy;

    if (!direct)
        direct = &dummy;

    if (!noflush)
        noflush = &dummy;

    /* Mapping of cache modes to the attributes according to qemu-options.hx
     *              │ cache.writeback   cache.direct   cache.no-flush
     * ─────────────┼─────────────────────────────────────────────────
     * writeback    │ true              false          false
     * none         │ true              true           false
     * writethrough │ false             false          false
     * directsync   │ false             true           false
     * unsafe       │ true              false          true
     */
    switch ((virDomainDiskCache) cachemode) {
    case VIR_DOMAIN_DISK_CACHE_DISABLE: /* 'none' */
        *writeback = true;
        *direct = true;
        *noflush = false;
        break;

    case VIR_DOMAIN_DISK_CACHE_WRITETHRU:
        *writeback = false;
        *direct = false;
        *noflush = false;
        break;

    case VIR_DOMAIN_DISK_CACHE_WRITEBACK:
        *writeback = true;
        *direct = false;
        *noflush = false;
        break;

    case VIR_DOMAIN_DISK_CACHE_DIRECTSYNC:
        *writeback = false;
        *direct = true;
        *noflush = false;
        break;

    case VIR_DOMAIN_DISK_CACHE_UNSAFE:
        *writeback = true;
        *direct = false;
        *noflush = true;
        break;

    case VIR_DOMAIN_DISK_CACHE_DEFAULT:
    case VIR_DOMAIN_DISK_CACHE_LAST:
    default:
        virReportEnumRangeError(virDomainDiskCache, cachemode);
        return -1;
    }

    return 0;
}


15326 15327 15328 15329 15330 15331 15332 15333 15334 15335
void
qemuProcessEventFree(struct qemuProcessEvent *event)
{
    if (!event)
        return;

    switch (event->eventType) {
    case QEMU_PROCESS_EVENT_GUESTPANIC:
        qemuMonitorEventPanicInfoFree(event->data);
        break;
15336 15337 15338
    case QEMU_PROCESS_EVENT_RDMA_GID_STATUS_CHANGED:
        qemuMonitorEventRdmaGidStatusFree(event->data);
        break;
15339 15340 15341 15342 15343 15344 15345 15346
    case QEMU_PROCESS_EVENT_WATCHDOG:
    case QEMU_PROCESS_EVENT_DEVICE_DELETED:
    case QEMU_PROCESS_EVENT_NIC_RX_FILTER_CHANGED:
    case QEMU_PROCESS_EVENT_SERIAL_CHANGED:
    case QEMU_PROCESS_EVENT_BLOCK_JOB:
    case QEMU_PROCESS_EVENT_MONITOR_EOF:
        VIR_FREE(event->data);
        break;
15347 15348 15349
    case QEMU_PROCESS_EVENT_JOB_STATUS_CHANGE:
        virObjectUnref(event->data);
        break;
15350
    case QEMU_PROCESS_EVENT_PR_DISCONNECT:
15351 15352 15353 15354 15355
    case QEMU_PROCESS_EVENT_LAST:
        break;
    }
    VIR_FREE(event);
}
15356 15357 15358


char *
15359
qemuDomainGetManagedPRSocketPath(qemuDomainObjPrivatePtr priv)
15360 15361 15362
{
    char *ret = NULL;

15363
    ret = g_strdup_printf("%s/%s.sock", priv->libDir, qemuDomainGetManagedPRAlias());
15364 15365 15366

    return ret;
}
15367 15368 15369 15370 15371 15372 15373 15374 15375 15376 15377 15378 15379 15380 15381 15382 15383 15384 15385 15386 15387 15388 15389 15390 15391 15392 15393


/**
 * qemuDomainStorageIdNew:
 * @priv: qemu VM private data object.
 *
 * Generate a new unique id for a storage object. Useful for node name generation.
 */
unsigned int
qemuDomainStorageIdNew(qemuDomainObjPrivatePtr priv)
{
    return ++priv->nodenameindex;
}


/**
 * qemuDomainStorageIdReset:
 * @priv: qemu VM private data object.
 *
 * Resets the data for the node name generator. The node names need to be unique
 * for a single instance, so can be reset on VM shutdown.
 */
void
qemuDomainStorageIdReset(qemuDomainObjPrivatePtr priv)
{
    priv->nodenameindex = 0;
}
15394 15395 15396 15397 15398 15399 15400 15401 15402 15403 15404 15405 15406 15407 15408 15409 15410 15411 15412 15413 15414 15415 15416 15417 15418 15419 15420 15421 15422


virDomainEventResumedDetailType
qemuDomainRunningReasonToResumeEvent(virDomainRunningReason reason)
{
    switch (reason) {
    case VIR_DOMAIN_RUNNING_RESTORED:
    case VIR_DOMAIN_RUNNING_FROM_SNAPSHOT:
        return VIR_DOMAIN_EVENT_RESUMED_FROM_SNAPSHOT;

    case VIR_DOMAIN_RUNNING_MIGRATED:
    case VIR_DOMAIN_RUNNING_MIGRATION_CANCELED:
        return VIR_DOMAIN_EVENT_RESUMED_MIGRATED;

    case VIR_DOMAIN_RUNNING_POSTCOPY:
        return VIR_DOMAIN_EVENT_RESUMED_POSTCOPY;

    case VIR_DOMAIN_RUNNING_UNKNOWN:
    case VIR_DOMAIN_RUNNING_SAVE_CANCELED:
    case VIR_DOMAIN_RUNNING_BOOTED:
    case VIR_DOMAIN_RUNNING_UNPAUSED:
    case VIR_DOMAIN_RUNNING_WAKEUP:
    case VIR_DOMAIN_RUNNING_CRASHED:
    case VIR_DOMAIN_RUNNING_LAST:
        break;
    }

    return VIR_DOMAIN_EVENT_RESUMED_UNPAUSED;
}
J
John Ferlan 已提交
15423 15424 15425 15426 15427


/* qemuDomainIsUsingNoShutdown:
 * @priv: Domain private data
 *
J
Ján Tomko 已提交
15428 15429
 * We can receive an event when QEMU stops. If we use no-shutdown, then
 * we can watch for this event and do a soft/warm reboot.
J
John Ferlan 已提交
15430 15431 15432 15433 15434 15435 15436
 *
 * Returns: @true when -no-shutdown either should be or was added to the
 * command line.
 */
bool
qemuDomainIsUsingNoShutdown(qemuDomainObjPrivatePtr priv)
{
J
Ján Tomko 已提交
15437
    return priv->allowReboot == VIR_TRISTATE_BOOL_YES;
J
John Ferlan 已提交
15438
}
15439 15440 15441 15442 15443 15444 15445 15446 15447


bool
qemuDomainDiskIsMissingLocalOptional(virDomainDiskDefPtr disk)
{
    return disk->startupPolicy == VIR_DOMAIN_STARTUP_POLICY_OPTIONAL &&
           virStorageSourceIsLocalStorage(disk->src) && disk->src->path &&
           !virFileExists(disk->src->path);
}
15448 15449


15450
void
15451 15452 15453 15454
qemuDomainNVRAMPathFormat(virQEMUDriverConfigPtr cfg,
                            virDomainDefPtr def,
                            char **path)
{
15455
    *path = g_strdup_printf("%s/%s_VARS.fd", cfg->nvramDir, def->name);
15456 15457 15458
}


15459
void
15460 15461 15462 15463 15464
qemuDomainNVRAMPathGenerate(virQEMUDriverConfigPtr cfg,
                            virDomainDefPtr def)
{
    if (def->os.loader &&
        def->os.loader->type == VIR_DOMAIN_LOADER_TYPE_PFLASH &&
15465
        def->os.loader->readonly == VIR_TRISTATE_BOOL_YES &&
15466 15467
        !def->os.loader->nvram)
        qemuDomainNVRAMPathFormat(cfg, def, &def->os.loader->nvram);
15468 15469

}
15470 15471 15472 15473 15474 15475 15476 15477 15478 15479 15480 15481 15482 15483 15484 15485 15486 15487 15488 15489 15490 15491 15492 15493 15494 15495 15496 15497 15498 15499 15500 15501 15502 15503


virDomainEventSuspendedDetailType
qemuDomainPausedReasonToSuspendedEvent(virDomainPausedReason reason)
{
    switch (reason) {
    case VIR_DOMAIN_PAUSED_MIGRATION:
        return VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED;

    case VIR_DOMAIN_PAUSED_FROM_SNAPSHOT:
        return VIR_DOMAIN_EVENT_SUSPENDED_FROM_SNAPSHOT;

    case VIR_DOMAIN_PAUSED_POSTCOPY_FAILED:
        return VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY_FAILED;

    case VIR_DOMAIN_PAUSED_POSTCOPY:
        return VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY;

    case VIR_DOMAIN_PAUSED_UNKNOWN:
    case VIR_DOMAIN_PAUSED_USER:
    case VIR_DOMAIN_PAUSED_SAVE:
    case VIR_DOMAIN_PAUSED_DUMP:
    case VIR_DOMAIN_PAUSED_IOERROR:
    case VIR_DOMAIN_PAUSED_WATCHDOG:
    case VIR_DOMAIN_PAUSED_SHUTTING_DOWN:
    case VIR_DOMAIN_PAUSED_SNAPSHOT:
    case VIR_DOMAIN_PAUSED_CRASHED:
    case VIR_DOMAIN_PAUSED_STARTING_UP:
    case VIR_DOMAIN_PAUSED_LAST:
        break;
    }

    return VIR_DOMAIN_EVENT_SUSPENDED_PAUSED;
}
15504 15505 15506 15507


static int
qemuDomainDefHasManagedPRBlockjobIterator(void *payload,
J
Ján Tomko 已提交
15508
                                          const void *name G_GNUC_UNUSED,
15509 15510 15511 15512 15513 15514 15515 15516 15517 15518 15519 15520 15521 15522 15523 15524 15525 15526 15527 15528 15529 15530 15531 15532 15533 15534 15535 15536 15537 15538 15539 15540 15541 15542 15543 15544
                                          void *opaque)
{
    qemuBlockJobDataPtr job = payload;
    bool *hasPR = opaque;

    if (job->disk)
        return 0;

    if ((job->chain && virStorageSourceChainHasManagedPR(job->chain)) ||
        (job->mirrorChain && virStorageSourceChainHasManagedPR(job->mirrorChain)))
        *hasPR = true;

    return 0;
}


/**
 * qemuDomainDefHasManagedPR:
 * @vm: domain object
 *
 * @vm must be an active VM. Returns true if @vm has any storage source with
 * managed persistent reservations.
 */
bool
qemuDomainDefHasManagedPR(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    bool jobPR = false;

    if (virDomainDefHasManagedPR(vm->def))
        return true;

    virHashForEach(priv->blockjobs, qemuDomainDefHasManagedPRBlockjobIterator, &jobPR);

    return jobPR;
}
15545 15546 15547 15548 15549 15550 15551 15552 15553 15554 15555 15556 15557


/**
 * qemuDomainSupportsCheckpointsBlockjobs:
 * @vm: domain object
 *
 * Checks whether a block job is supported in possible combination with
 * checkpoints (qcow2 bitmaps). Returns -1 if unsupported and reports an error
 * 0 in case everything is supported.
 */
int
qemuDomainSupportsCheckpointsBlockjobs(virDomainObjPtr vm)
{
15558 15559 15560 15561
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_INCREMENTAL_BACKUP) &&
        virDomainListCheckpoints(vm->checkpoints, NULL, NULL, NULL, 0) > 0) {
15562 15563 15564 15565 15566 15567 15568
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s",
                       _("cannot perform block operations while checkpoint exists"));
        return -1;
    }

    return 0;
}