qemu_domain.c 540.0 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"
29
#include "qemu_process.h"
30
#include "qemu_capabilities.h"
31
#include "qemu_hostdev.h"
32
#include "qemu_migration.h"
33
#include "qemu_migration_params.h"
34
#include "qemu_security.h"
35
#include "qemu_slirp.h"
36
#include "qemu_extdevice.h"
37
#include "qemu_blockjob.h"
38
#include "qemu_checkpoint.h"
39
#include "viralloc.h"
40
#include "virlog.h"
41
#include "virerror.h"
42
#include "cpu/cpu.h"
43
#include "viruuid.h"
E
Eric Blake 已提交
44
#include "virfile.h"
45
#include "domain_addr.h"
46
#include "domain_capabilities.h"
47
#include "domain_event.h"
48
#include "virtime.h"
49
#include "virnetdevopenvswitch.h"
50
#include "virstoragefile.h"
51
#include "virstring.h"
52
#include "virthreadjob.h"
53
#include "virprocess.h"
54
#include "vircrypto.h"
55
#include "virrandom.h"
56
#include "virsystemd.h"
57
#include "virsecret.h"
58
#include "logging/log_manager.h"
59
#include "locking/domain_lock.h"
60
#include "virdomainsnapshotobjlist.h"
61
#include "virdomaincheckpointobjlist.h"
62
#include "backup_conf.h"
J
Ján Tomko 已提交
63
#include "virutil.h"
64
#include "virdevmapper.h"
65

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

78 79
#define QEMU_QXL_VGAMEM_DEFAULT 16 * 1024

80 81
#define VIR_FROM_THIS VIR_FROM_QEMU

82 83
VIR_LOG_INIT("qemu.qemu_domain");

84 85
VIR_ENUM_IMPL(qemuDomainJob,
              QEMU_JOB_LAST,
86 87 88 89 90
              "none",
              "query",
              "destroy",
              "suspend",
              "modify",
91
              "abort",
92
              "migration operation",
93 94 95 96
              "none",   /* async job is never stored in job.active */
              "async nested",
);

97 98
VIR_ENUM_IMPL(qemuDomainAgentJob,
              QEMU_AGENT_JOB_LAST,
99 100 101 102 103
              "none",
              "query",
              "modify",
);

104 105
VIR_ENUM_IMPL(qemuDomainAsyncJob,
              QEMU_ASYNC_JOB_LAST,
106 107 108 109 110
              "none",
              "migration out",
              "migration in",
              "save",
              "dump",
111
              "snapshot",
112
              "start",
113
              "backup",
114 115
);

116 117
VIR_ENUM_IMPL(qemuDomainNamespace,
              QEMU_DOMAIN_NS_LAST,
118 119 120
              "mount",
);

121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151

/**
 * 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;
}


152
struct _qemuDomainLogContext {
153
    GObject parent;
154

155
    int writefd;
156
    int readfd; /* Only used if manager == NULL */
157
    off_t pos;
158
    ino_t inode; /* Only used if manager != NULL */
159
    char *path;
160
    virLogManagerPtr manager;
161 162
};

163
G_DEFINE_TYPE(qemuDomainLogContext, qemu_domain_log_context, G_TYPE_OBJECT);
164
static virClassPtr qemuDomainSaveCookieClass;
165

166
static void qemuDomainLogContextFinalize(GObject *obj);
167
static void qemuDomainSaveCookieDispose(void *obj);
168

169

170
static int
171
qemuDomainOnceInit(void)
172
{
173
    if (!VIR_CLASS_NEW(qemuDomainSaveCookie, virClassForObject()))
174 175
        return -1;

176 177 178
    return 0;
}

179 180 181 182 183 184 185 186 187 188 189
static void qemu_domain_log_context_init(qemuDomainLogContext *logctxt G_GNUC_UNUSED)
{
}

static void qemu_domain_log_context_class_init(qemuDomainLogContextClass *klass)
{
    GObjectClass *obj = G_OBJECT_CLASS(klass);

    obj->finalize = qemuDomainLogContextFinalize;
}

190
VIR_ONCE_GLOBAL_INIT(qemuDomain);
191 192

static void
193
qemuDomainLogContextFinalize(GObject *object)
194
{
195
    qemuDomainLogContextPtr ctxt = QEMU_DOMAIN_LOG_CONTEXT(object);
196 197 198 199 200 201
    VIR_DEBUG("ctxt=%p", ctxt);

    virLogManagerFree(ctxt->manager);
    VIR_FREE(ctxt->path);
    VIR_FORCE_CLOSE(ctxt->writefd);
    VIR_FORCE_CLOSE(ctxt->readfd);
202
    G_OBJECT_CLASS(qemu_domain_log_context_parent_class)->finalize(object);
203 204
}

J
Jiri Denemark 已提交
205
const char *
206
qemuDomainAsyncJobPhaseToString(qemuDomainAsyncJob job,
J
Ján Tomko 已提交
207
                                int phase G_GNUC_UNUSED)
J
Jiri Denemark 已提交
208 209 210 211
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
212 213
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
214 215
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
216
    case QEMU_ASYNC_JOB_SNAPSHOT:
217
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
218
    case QEMU_ASYNC_JOB_NONE:
219
    case QEMU_ASYNC_JOB_BACKUP:
220
        G_GNUC_FALLTHROUGH;
221 222
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
223 224 225 226 227 228
    }

    return "none";
}

int
229
qemuDomainAsyncJobPhaseFromString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
230 231 232 233 234 235 236 237
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
238 239
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
240 241
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
242
    case QEMU_ASYNC_JOB_SNAPSHOT:
243
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
244
    case QEMU_ASYNC_JOB_NONE:
245
    case QEMU_ASYNC_JOB_BACKUP:
246
        G_GNUC_FALLTHROUGH;
247 248
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
249 250 251 252 253 254 255 256
    }

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

257

258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
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;
}


290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
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;
        }
    }
}


306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325
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);
326
    virObjectEventStateQueue(driver->domainEventState, event);
327 328 329
}


330 331 332 333 334 335 336 337
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

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

338
    if (virCondInit(&priv->job.asyncCond) < 0) {
339
        virCondDestroy(&priv->job.cond);
340 341 342
        return -1;
    }

343 344 345 346 347 348
    return 0;
}

static void
qemuDomainObjResetJob(qemuDomainObjPrivatePtr priv)
{
349
    qemuDomainJobObjPtr job = &priv->job;
350 351

    job->active = QEMU_JOB_NONE;
352
    job->owner = 0;
353
    job->ownerAPI = NULL;
J
Jiri Denemark 已提交
354
    job->started = 0;
355 356
}

357 358 359 360 361 362 363 364 365 366 367 368 369

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

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


370 371 372
static void
qemuDomainObjResetAsyncJob(qemuDomainObjPrivatePtr priv)
{
373
    qemuDomainJobObjPtr job = &priv->job;
374 375

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
376
    job->asyncOwner = 0;
377
    job->asyncOwnerAPI = NULL;
J
Jiri Denemark 已提交
378
    job->asyncStarted = 0;
J
Jiri Denemark 已提交
379
    job->phase = 0;
380
    job->mask = QEMU_JOB_DEFAULT_MASK;
381
    job->abortJob = false;
382
    job->spiceMigration = false;
383
    job->spiceMigrated = false;
384 385
    job->dumpCompleted = false;
    VIR_FREE(job->error);
J
Jiri Denemark 已提交
386
    VIR_FREE(job->current);
387 388
    qemuMigrationParamsFree(job->migParams);
    job->migParams = NULL;
389
    job->apiFlags = 0;
390 391
}

392 393
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
394
                        qemuDomainJobObjPtr job)
395 396 397 398 399
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
400
    job->owner = priv->job.owner;
401
    job->asyncJob = priv->job.asyncJob;
402
    job->asyncOwner = priv->job.asyncOwner;
J
Jiri Denemark 已提交
403
    job->phase = priv->job.phase;
404
    job->migParams = g_steal_pointer(&priv->job.migParams);
405
    job->apiFlags = priv->job.apiFlags;
406 407 408 409 410

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

411 412 413
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
414 415
    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
J
Jiri Denemark 已提交
416
    VIR_FREE(priv->job.current);
417
    VIR_FREE(priv->job.completed);
418 419
    virCondDestroy(&priv->job.cond);
    virCondDestroy(&priv->job.asyncCond);
420 421
}

422
static bool
423
qemuDomainTrackJob(qemuDomainJob job)
424 425 426 427
{
    return (QEMU_DOMAIN_TRACK_JOBS & JOB_MASK(job)) != 0;
}

428

J
Jiri Denemark 已提交
429 430 431 432 433 434 435 436 437 438 439
int
qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->started)
        return 0;

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

440 441 442 443 444 445
    if (now < jobInfo->started) {
        VIR_WARN("Async job starts in the future");
        jobInfo->started = 0;
        return 0;
    }

J
Jiri Denemark 已提交
446 447 448 449
    jobInfo->timeElapsed = now - jobInfo->started;
    return 0;
}

450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466
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;
    }

467 468
    jobInfo->stats.mig.downtime = now - jobInfo->stopped;
    jobInfo->stats.mig.downtime_set = true;
469 470 471
    return 0;
}

472 473 474 475 476 477 478 479
static virDomainJobType
qemuDomainJobStatusToType(qemuDomainJobStatus status)
{
    switch (status) {
    case QEMU_DOMAIN_JOB_STATUS_NONE:
        break;

    case QEMU_DOMAIN_JOB_STATUS_ACTIVE:
480
    case QEMU_DOMAIN_JOB_STATUS_MIGRATING:
481
    case QEMU_DOMAIN_JOB_STATUS_QEMU_COMPLETED:
482
    case QEMU_DOMAIN_JOB_STATUS_POSTCOPY:
483
    case QEMU_DOMAIN_JOB_STATUS_PAUSED:
484 485 486 487 488 489 490 491 492 493 494 495 496 497 498
        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
499 500 501 502
int
qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                        virDomainJobInfoPtr info)
{
503
    info->type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
504 505
    info->timeElapsed = jobInfo->timeElapsed;

506 507 508 509 510 511 512 513 514 515 516 517 518
    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
519

520 521 522 523 524 525
    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;

526
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
527 528 529 530 531
        info->memTotal = jobInfo->stats.dump.total;
        info->memProcessed = jobInfo->stats.dump.completed;
        info->memRemaining = info->memTotal - info->memProcessed;
        break;

532 533 534 535 536 537
    case QEMU_DOMAIN_JOB_STATS_TYPE_BACKUP:
        info->fileTotal = jobInfo->stats.backup.total;
        info->fileProcessed = jobInfo->stats.backup.transferred;
        info->fileRemaining = info->fileTotal - info->fileProcessed;
        break;

538 539 540
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
        break;
    }
J
Jiri Denemark 已提交
541 542 543 544 545 546 547 548

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

    return 0;
}

549 550 551 552 553 554

static int
qemuDomainMigrationJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                                   int *type,
                                   virTypedParameterPtr *params,
                                   int *nparams)
J
Jiri Denemark 已提交
555
{
556
    qemuMonitorMigrationStats *stats = &jobInfo->stats.mig;
557
    qemuDomainMirrorStatsPtr mirrorStats = &jobInfo->mirrorStats;
J
Jiri Denemark 已提交
558 559 560
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;
561 562
    unsigned long long mirrorRemaining = mirrorStats->total -
                                         mirrorStats->transferred;
J
Jiri Denemark 已提交
563

564 565 566 567 568
    if (virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_OPERATION,
                             jobInfo->operation) < 0)
        goto error;

J
Jiri Denemark 已提交
569 570 571 572 573
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED,
                                jobInfo->timeElapsed) < 0)
        goto error;

574 575 576 577 578 579 580
    if (jobInfo->timeDeltaSet &&
        jobInfo->timeElapsed > jobInfo->timeDelta &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED_NET,
                                jobInfo->timeElapsed - jobInfo->timeDelta) < 0)
        goto error;

581
    if (stats->downtime_set &&
J
Jiri Denemark 已提交
582 583
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME,
584
                                stats->downtime) < 0)
J
Jiri Denemark 已提交
585 586
        goto error;

587
    if (stats->downtime_set &&
588
        jobInfo->timeDeltaSet &&
589
        stats->downtime > jobInfo->timeDelta &&
590 591
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME_NET,
592
                                stats->downtime - jobInfo->timeDelta) < 0)
593 594
        goto error;

595
    if (stats->setup_time_set &&
596 597
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_SETUP_TIME,
598
                                stats->setup_time) < 0)
599 600
        goto error;

J
Jiri Denemark 已提交
601 602
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_TOTAL,
603
                                stats->ram_total +
604 605
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
606 607
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_PROCESSED,
608
                                stats->ram_transferred +
609 610
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
611 612
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_REMAINING,
613
                                stats->ram_remaining +
614 615
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
616 617 618 619
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
620
                                stats->ram_total) < 0 ||
J
Jiri Denemark 已提交
621 622
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
623
                                stats->ram_transferred) < 0 ||
J
Jiri Denemark 已提交
624 625
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
626
                                stats->ram_remaining) < 0)
J
Jiri Denemark 已提交
627 628
        goto error;

629
    if (stats->ram_bps &&
630 631
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_BPS,
632
                                stats->ram_bps) < 0)
633 634
        goto error;

635
    if (stats->ram_duplicate_set) {
J
Jiri Denemark 已提交
636 637
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_CONSTANT,
638
                                    stats->ram_duplicate) < 0 ||
J
Jiri Denemark 已提交
639 640
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL,
641
                                    stats->ram_normal) < 0 ||
J
Jiri Denemark 已提交
642 643
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL_BYTES,
644
                                    stats->ram_normal_bytes) < 0)
J
Jiri Denemark 已提交
645 646 647
            goto error;
    }

648 649 650 651 652
    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,
653 654 655 656
                                stats->ram_iteration) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_POSTCOPY_REQS,
                                stats->ram_postcopy_reqs) < 0)
657 658
        goto error;

659 660 661 662 663 664
    if (stats->ram_page_size > 0 &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PAGE_SIZE,
                                stats->ram_page_size) < 0)
        goto error;

665 666 667 668 669
    /* 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
670 671
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_TOTAL,
672 673
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
674 675
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_PROCESSED,
676 677
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
678 679
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_REMAINING,
680 681
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
682 683
        goto error;

684
    if (stats->disk_bps &&
685 686
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_BPS,
687
                                stats->disk_bps) < 0)
688 689
        goto error;

690
    if (stats->xbzrle_set) {
J
Jiri Denemark 已提交
691 692
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE,
693
                                    stats->xbzrle_cache_size) < 0 ||
J
Jiri Denemark 已提交
694 695
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_BYTES,
696
                                    stats->xbzrle_bytes) < 0 ||
J
Jiri Denemark 已提交
697 698
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_PAGES,
699
                                    stats->xbzrle_pages) < 0 ||
J
Jiri Denemark 已提交
700 701
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE_MISSES,
702
                                    stats->xbzrle_cache_miss) < 0 ||
J
Jiri Denemark 已提交
703 704
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_OVERFLOW,
705
                                    stats->xbzrle_overflow) < 0)
J
Jiri Denemark 已提交
706 707 708
            goto error;
    }

709 710 711 712 713 714
    if (stats->cpu_throttle_percentage &&
        virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_AUTO_CONVERGE_THROTTLE,
                             stats->cpu_throttle_percentage) < 0)
        goto error;

715
 done:
716
    *type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
717 718 719 720 721 722 723 724 725 726
    *params = par;
    *nparams = npar;
    return 0;

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


727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769
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;
}


770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822
static int
qemuDomainBackupJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                                int *type,
                                virTypedParameterPtr *params,
                                int *nparams)
{
    qemuDomainBackupStats *stats = &jobInfo->stats.backup;
    g_autoptr(virTypedParamList) par = g_new0(virTypedParamList, 1);

    if (virTypedParamListAddInt(par, jobInfo->operation,
                                VIR_DOMAIN_JOB_OPERATION) < 0)
        return -1;

    if (virTypedParamListAddULLong(par, jobInfo->timeElapsed,
                                   VIR_DOMAIN_JOB_TIME_ELAPSED) < 0)
        return -1;

    if (stats->transferred > 0 || stats->total > 0) {
        if (virTypedParamListAddULLong(par, stats->total,
                                       VIR_DOMAIN_JOB_DISK_TOTAL) < 0)
            return -1;

        if (virTypedParamListAddULLong(par, stats->transferred,
                                       VIR_DOMAIN_JOB_DISK_PROCESSED) < 0)
            return -1;

        if (virTypedParamListAddULLong(par, stats->total - stats->transferred,
                                       VIR_DOMAIN_JOB_DISK_REMAINING) < 0)
            return -1;
    }

    if (stats->tmp_used > 0 || stats->tmp_total > 0) {
        if (virTypedParamListAddULLong(par, stats->tmp_used,
                                       VIR_DOMAIN_JOB_DISK_TEMP_USED) < 0)
            return -1;

        if (virTypedParamListAddULLong(par, stats->tmp_total,
                                       VIR_DOMAIN_JOB_DISK_TEMP_TOTAL) < 0)
            return -1;
    }

    if (jobInfo->status != QEMU_DOMAIN_JOB_STATUS_ACTIVE &&
        virTypedParamListAddBoolean(par,
                                    jobInfo->status == QEMU_DOMAIN_JOB_STATUS_COMPLETED,
                                    VIR_DOMAIN_JOB_SUCCESS) < 0)
        return -1;

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


823 824 825 826 827 828 829 830
int
qemuDomainJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                          int *type,
                          virTypedParameterPtr *params,
                          int *nparams)
{
    switch (jobInfo->statsType) {
    case QEMU_DOMAIN_JOB_STATS_TYPE_MIGRATION:
831
    case QEMU_DOMAIN_JOB_STATS_TYPE_SAVEDUMP:
832 833
        return qemuDomainMigrationJobInfoToParams(jobInfo, type, params, nparams);

834
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
835 836
        return qemuDomainDumpJobInfoToParams(jobInfo, type, params, nparams);

837 838 839
    case QEMU_DOMAIN_JOB_STATS_TYPE_BACKUP:
        return qemuDomainBackupJobInfoToParams(jobInfo, type, params, nparams);

840
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
841 842 843 844 845 846
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid job statistics type"));
        break;

    default:
        virReportEnumRangeError(qemuDomainJobStatsType, jobInfo->statsType);
847 848 849 850 851 852 853
        break;
    }

    return -1;
}


J
John Ferlan 已提交
854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875
/* 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:
876 877
 * @driver: qemu driver data
 * @vm: Pointer to the vm object
J
John Ferlan 已提交
878 879 880 881 882
 *
 * 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
 */
883
int
M
Martin Kletzander 已提交
884 885
qemuDomainWriteMasterKeyFile(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
J
John Ferlan 已提交
886
{
887
    g_autofree char *path = NULL;
J
John Ferlan 已提交
888 889
    int fd = -1;
    int ret = -1;
M
Martin Kletzander 已提交
890
    qemuDomainObjPrivatePtr priv = vm->privateData;
J
John Ferlan 已提交
891

892 893 894 895
    /* Only gets filled in if we have the capability */
    if (!priv->masterKey)
        return 0;

J
John Ferlan 已提交
896 897 898 899 900 901 902 903 904 905 906 907 908 909 910
    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;
    }

911
    if (qemuSecurityDomainSetPathLabel(driver, vm, path, false) < 0)
M
Martin Kletzander 已提交
912 913
        goto cleanup;

J
John Ferlan 已提交
914 915 916 917 918 919 920 921 922
    ret = 0;

 cleanup:
    VIR_FORCE_CLOSE(fd);

    return ret;
}


923 924 925 926 927 928 929 930 931
static void
qemuDomainMasterKeyFree(qemuDomainObjPrivatePtr priv)
{
    if (!priv->masterKey)
        return;

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

J
John Ferlan 已提交
932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952
/* 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)
{
953
    g_autofree char *path = NULL;
J
John Ferlan 已提交
954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021
    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);

    return 0;

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

    VIR_FORCE_CLOSE(fd);

    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)
{
1022
    g_autofree char *path = NULL;
J
John Ferlan 已提交
1023 1024 1025 1026 1027

    if (!priv->masterKey)
        return;

    /* Clear the contents */
1028
    qemuDomainMasterKeyFree(priv);
J
John Ferlan 已提交
1029 1030 1031 1032 1033 1034 1035 1036

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


/* qemuDomainMasterKeyCreate:
1037
 * @vm: Pointer to the domain object
J
John Ferlan 已提交
1038 1039 1040 1041 1042 1043 1044 1045
 *
 * 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
1046
qemuDomainMasterKeyCreate(virDomainObjPtr vm)
J
John Ferlan 已提交
1047
{
M
Martin Kletzander 已提交
1048 1049
    qemuDomainObjPrivatePtr priv = vm->privateData;

J
John Ferlan 已提交
1050 1051 1052 1053
    /* If we don't have the capability, then do nothing. */
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

1054
    if (VIR_ALLOC_N(priv->masterKey, QEMU_DOMAIN_MASTER_KEY_LEN) < 0)
1055
        return -1;
J
John Ferlan 已提交
1056 1057
    priv->masterKeyLen = QEMU_DOMAIN_MASTER_KEY_LEN;

1058
    if (virRandomBytes(priv->masterKey, priv->masterKeyLen) < 0) {
1059 1060 1061 1062
        VIR_DISPOSE_N(priv->masterKey, priv->masterKeyLen);
        return -1;
    }

J
John Ferlan 已提交
1063 1064 1065 1066
    return 0;
}


1067
static void
1068
qemuDomainSecretPlainClear(qemuDomainSecretPlainPtr secret)
1069
{
1070 1071
    VIR_FREE(secret->username);
    VIR_DISPOSE_N(secret->secret, secret->secretlen);
1072 1073 1074
}


J
John Ferlan 已提交
1075
static void
1076
qemuDomainSecretAESClear(qemuDomainSecretAESPtr secret,
1077
                         bool keepAlias)
J
John Ferlan 已提交
1078
{
1079
    if (!keepAlias)
1080
        VIR_FREE(secret->alias);
1081

1082 1083 1084
    VIR_FREE(secret->username);
    VIR_FREE(secret->iv);
    VIR_FREE(secret->ciphertext);
J
John Ferlan 已提交
1085 1086 1087
}


1088 1089 1090
static void
qemuDomainSecretInfoClear(qemuDomainSecretInfoPtr secinfo,
                          bool keepAlias)
1091
{
1092
    if (!secinfo)
1093 1094
        return;

1095
    switch ((qemuDomainSecretInfoType) secinfo->type) {
J
John Ferlan 已提交
1096
    case VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN:
1097
        qemuDomainSecretPlainClear(&secinfo->s.plain);
J
John Ferlan 已提交
1098 1099
        break;

1100
    case VIR_DOMAIN_SECRET_INFO_TYPE_AES:
1101
        qemuDomainSecretAESClear(&secinfo->s.aes, keepAlias);
J
John Ferlan 已提交
1102 1103 1104 1105 1106
        break;

    case VIR_DOMAIN_SECRET_INFO_TYPE_LAST:
        break;
    }
1107 1108 1109 1110
}


void
1111
qemuDomainSecretInfoFree(qemuDomainSecretInfoPtr secinfo)
1112
{
1113 1114
    qemuDomainSecretInfoClear(secinfo, false);
    g_free(secinfo);
1115 1116 1117
}


1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130
/**
 * qemuDomainSecretInfoDestroy:
 * @secinfo: object to destroy
 *
 * Removes any data unnecessary for further use, but keeps alias allocated.
 */
void
qemuDomainSecretInfoDestroy(qemuDomainSecretInfoPtr secinfo)
{
    qemuDomainSecretInfoClear(secinfo, true);
}


1131
static virClassPtr qemuDomainDiskPrivateClass;
1132
static void qemuDomainDiskPrivateDispose(void *obj);
1133 1134 1135 1136

static int
qemuDomainDiskPrivateOnceInit(void)
{
1137
    if (!VIR_CLASS_NEW(qemuDomainDiskPrivate, virClassForObject()))
1138
        return -1;
1139 1140

    return 0;
1141 1142
}

1143
VIR_ONCE_GLOBAL_INIT(qemuDomainDiskPrivate);
1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158

static virObjectPtr
qemuDomainDiskPrivateNew(void)
{
    qemuDomainDiskPrivatePtr priv;

    if (qemuDomainDiskPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}

1159 1160 1161 1162 1163
static void
qemuDomainDiskPrivateDispose(void *obj)
{
    qemuDomainDiskPrivatePtr priv = obj;

1164
    virObjectUnref(priv->migrSource);
1165
    VIR_FREE(priv->qomName);
1166
    VIR_FREE(priv->nodeCopyOnRead);
1167
    virObjectUnref(priv->blockjob);
1168
}
1169

1170 1171 1172 1173 1174 1175
static virClassPtr qemuDomainStorageSourcePrivateClass;
static void qemuDomainStorageSourcePrivateDispose(void *obj);

static int
qemuDomainStorageSourcePrivateOnceInit(void)
{
1176
    if (!VIR_CLASS_NEW(qemuDomainStorageSourcePrivate, virClassForObject()))
1177
        return -1;
1178 1179

    return 0;
1180 1181
}

1182
VIR_ONCE_GLOBAL_INIT(qemuDomainStorageSourcePrivate);
1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203

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;

1204 1205
    g_clear_pointer(&priv->secinfo, qemuDomainSecretInfoFree);
    g_clear_pointer(&priv->encinfo, qemuDomainSecretInfoFree);
1206 1207 1208
}


1209 1210 1211 1212 1213 1214
static virClassPtr qemuDomainVcpuPrivateClass;
static void qemuDomainVcpuPrivateDispose(void *obj);

static int
qemuDomainVcpuPrivateOnceInit(void)
{
1215
    if (!VIR_CLASS_NEW(qemuDomainVcpuPrivate, virClassForObject()))
1216
        return -1;
1217 1218

    return 0;
1219 1220
}

1221
VIR_ONCE_GLOBAL_INIT(qemuDomainVcpuPrivate);
1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238

static virObjectPtr
qemuDomainVcpuPrivateNew(void)
{
    qemuDomainVcpuPrivatePtr priv;

    if (qemuDomainVcpuPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}


static void
1239
qemuDomainVcpuPrivateDispose(void *obj)
1240
{
1241 1242 1243 1244
    qemuDomainVcpuPrivatePtr priv = obj;

    VIR_FREE(priv->type);
    VIR_FREE(priv->alias);
1245
    virJSONValueFree(priv->props);
1246 1247 1248 1249
    return;
}


1250 1251
static virClassPtr qemuDomainChrSourcePrivateClass;
static void qemuDomainChrSourcePrivateDispose(void *obj);
1252 1253

static int
1254
qemuDomainChrSourcePrivateOnceInit(void)
1255
{
1256
    if (!VIR_CLASS_NEW(qemuDomainChrSourcePrivate, virClassForObject()))
1257
        return -1;
1258 1259

    return 0;
1260 1261
}

1262
VIR_ONCE_GLOBAL_INIT(qemuDomainChrSourcePrivate);
1263 1264

static virObjectPtr
1265
qemuDomainChrSourcePrivateNew(void)
1266
{
1267
    qemuDomainChrSourcePrivatePtr priv;
1268

1269
    if (qemuDomainChrSourcePrivateInitialize() < 0)
1270 1271
        return NULL;

1272
    if (!(priv = virObjectNew(qemuDomainChrSourcePrivateClass)))
1273 1274 1275 1276 1277 1278 1279
        return NULL;

    return (virObjectPtr) priv;
}


static void
1280
qemuDomainChrSourcePrivateDispose(void *obj)
1281
{
1282
    qemuDomainChrSourcePrivatePtr priv = obj;
1283

1284
    g_clear_pointer(&priv->secinfo, qemuDomainSecretInfoFree);
1285 1286 1287
}


J
Ján Tomko 已提交
1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299
static virClassPtr qemuDomainVsockPrivateClass;
static void qemuDomainVsockPrivateDispose(void *obj);

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

    return 0;
}

1300
VIR_ONCE_GLOBAL_INIT(qemuDomainVsockPrivate);
J
Ján Tomko 已提交
1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312

static virObjectPtr
qemuDomainVsockPrivateNew(void)
{
    qemuDomainVsockPrivatePtr priv;

    if (qemuDomainVsockPrivateInitialize() < 0)
        return NULL;

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

1313 1314
    priv->vhostfd = -1;

J
Ján Tomko 已提交
1315 1316 1317 1318 1319
    return (virObjectPtr) priv;
}


static void
J
Ján Tomko 已提交
1320
qemuDomainVsockPrivateDispose(void *obj G_GNUC_UNUSED)
J
Ján Tomko 已提交
1321
{
1322 1323 1324
    qemuDomainVsockPrivatePtr priv = obj;

    VIR_FORCE_CLOSE(priv->vhostfd);
J
Ján Tomko 已提交
1325 1326 1327
}


1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339
static virClassPtr qemuDomainGraphicsPrivateClass;
static void qemuDomainGraphicsPrivateDispose(void *obj);

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

    return 0;
}

1340
VIR_ONCE_GLOBAL_INIT(qemuDomainGraphicsPrivate);
1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362

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);
1363
    g_clear_pointer(&priv->secinfo, qemuDomainSecretInfoFree);
1364 1365 1366
}


1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399
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
1400
qemuDomainNetworkPrivateDispose(void *obj G_GNUC_UNUSED)
1401
{
1402 1403 1404
    qemuDomainNetworkPrivatePtr priv = obj;

    qemuSlirpFree(priv->slirp);
1405 1406 1407
}


1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440
static virClassPtr qemuDomainFSPrivateClass;
static void qemuDomainFSPrivateDispose(void *obj);


static int
qemuDomainFSPrivateOnceInit(void)
{
    if (!VIR_CLASS_NEW(qemuDomainFSPrivate, virClassForObject()))
        return -1;

    return 0;
}


VIR_ONCE_GLOBAL_INIT(qemuDomainFSPrivate);


static virObjectPtr
qemuDomainFSPrivateNew(void)
{
    qemuDomainFSPrivatePtr priv;

    if (qemuDomainFSPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}


static void
1441
qemuDomainFSPrivateDispose(void *obj)
1442
{
1443 1444 1445
    qemuDomainFSPrivatePtr priv = obj;

    g_free(priv->vhostuser_fs_sock);
1446 1447
}

1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474
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;

1475 1476
    priv->vhost_user_fd = -1;

1477 1478 1479 1480 1481
    return (virObjectPtr) priv;
}


static void
1482
qemuDomainVideoPrivateDispose(void *obj)
1483
{
1484 1485 1486
    qemuDomainVideoPrivatePtr priv = obj;

    VIR_FORCE_CLOSE(priv->vhost_user_fd);
1487 1488 1489
}


1490 1491
/* qemuDomainSecretPlainSetup:
 * @secinfo: Pointer to secret info
J
John Ferlan 已提交
1492
 * @usageType: The virSecretUsageType
1493 1494
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1495 1496 1497 1498 1499 1500
 *
 * Taking a secinfo, fill in the plaintext information
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1501
qemuDomainSecretPlainSetup(qemuDomainSecretInfoPtr secinfo,
J
John Ferlan 已提交
1502
                           virSecretUsageType usageType,
1503 1504
                           const char *username,
                           virSecretLookupTypeDefPtr seclookupdef)
1505
{
1506
    g_autoptr(virConnect) conn = virGetConnectSecret();
1507 1508 1509 1510 1511
    int ret = -1;

    if (!conn)
        return -1;

1512
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN;
1513
    secinfo->s.plain.username = g_strdup(username);
1514

1515 1516 1517 1518 1519
    ret = virSecretGetSecretString(conn, seclookupdef, usageType,
                                   &secinfo->s.plain.secret,
                                   &secinfo->s.plain.secretlen);

    return ret;
1520 1521 1522
}


1523 1524
/* qemuDomainSecretAESSetup:
 * @priv: pointer to domain private object
1525 1526 1527 1528
 * @alias: alias of the secret
 * @username: username to use (may be NULL)
 * @secret: secret data
 * @secretlen: length of @secret
1529
 *
1530
 * Encrypts @secret for use with qemu.
1531
 *
1532
 * Returns qemuDomainSecretInfoPtr filled with the necessary information.
1533
 */
1534
static qemuDomainSecretInfoPtr
1535
qemuDomainSecretAESSetup(qemuDomainObjPrivatePtr priv,
1536
                         const char *alias,
1537
                         const char *username,
1538 1539
                         uint8_t *secret,
                         size_t secretlen)
1540
{
1541
    g_autoptr(qemuDomainSecretInfo) secinfo = NULL;
1542
    g_autofree uint8_t *raw_iv = NULL;
1543
    size_t ivlen = QEMU_DOMAIN_AES_IV_LEN;
1544
    g_autofree uint8_t *ciphertext = NULL;
1545 1546
    size_t ciphertextlen = 0;

1547 1548 1549 1550 1551 1552 1553
    if (!qemuDomainSupportsEncryptedSecret(priv)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("encrypted secrets are not supported"));
        return NULL;
    }

    secinfo = g_new0(qemuDomainSecretInfo, 1);
1554

1555
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_AES;
1556
    secinfo->s.aes.alias = g_strdup(alias);
1557
    secinfo->s.aes.username = g_strdup(username);
1558

1559
    raw_iv = g_new0(uint8_t, ivlen);
1560

1561
    /* Create a random initialization vector */
1562
    if (virRandomBytes(raw_iv, ivlen) < 0)
1563
        return NULL;
1564 1565

    /* Encode the IV and save that since qemu will need it */
1566
    secinfo->s.aes.iv = g_base64_encode(raw_iv, ivlen);
1567 1568 1569 1570 1571

    if (virCryptoEncryptData(VIR_CRYPTO_CIPHER_AES256CBC,
                             priv->masterKey, QEMU_DOMAIN_MASTER_KEY_LEN,
                             raw_iv, ivlen, secret, secretlen,
                             &ciphertext, &ciphertextlen) < 0)
1572
        return NULL;
1573 1574

    /* Now encode the ciphertext and store to be passed to qemu */
1575 1576
    secinfo->s.aes.ciphertext = g_base64_encode(ciphertext,
                                                ciphertextlen);
1577

1578
    return g_steal_pointer(&secinfo);
1579 1580 1581 1582 1583 1584 1585
}


/**
 * qemuDomainSecretAESSetupFromSecret:
 * @priv: pointer to domain private object
 * @srcalias: Alias of the disk/hostdev used to generate the secret alias
1586
 * @secretuse: specific usage for the secret (may be NULL if main object is using it)
1587 1588 1589 1590 1591
 * @usageType: The virSecretUsageType
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
 *
 * Looks up a secret in the secret driver based on @usageType and @seclookupdef
1592 1593
 * and builds qemuDomainSecretInfoPtr from it. @use describes the usage of the
 * secret in case if @srcalias requires more secrets for various usage cases.
1594 1595 1596 1597
 */
static qemuDomainSecretInfoPtr
qemuDomainSecretAESSetupFromSecret(qemuDomainObjPrivatePtr priv,
                                   const char *srcalias,
1598
                                   const char *secretuse,
1599 1600
                                   virSecretUsageType usageType,
                                   const char *username,
1601
                                   virSecretLookupTypeDefPtr seclookupdef)
1602 1603 1604
{
    g_autoptr(virConnect) conn = virGetConnectSecret();
    qemuDomainSecretInfoPtr secinfo;
1605
    g_autofree char *alias = qemuAliasForSecret(srcalias, secretuse);
1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616
    uint8_t *secret = NULL;
    size_t secretlen = 0;

    if (!conn)
        return NULL;

    if (virSecretGetSecretString(conn, seclookupdef, usageType,
                                 &secret, &secretlen) < 0)
        return NULL;

    secinfo = qemuDomainSecretAESSetup(priv, alias, username, secret, secretlen);
1617 1618

    VIR_DISPOSE_N(secret, secretlen);
1619 1620

    return secinfo;
1621 1622 1623
}


1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639
/**
 * 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;
}


1640
/* qemuDomainSecretInfoNewPlain:
1641
 * @usageType: Secret usage type
1642 1643
 * @username: username
 * @lookupDef: lookup def describing secret
1644
 *
1645
 * Helper function to create a secinfo to be used for secinfo consumers. This
1646
 * sets up a 'plain' (unencrypted) secret for legacy consumers.
1647 1648 1649 1650 1651
 *
 * Returns @secinfo on success, NULL on failure. Caller is responsible
 * to eventually free @secinfo.
 */
static qemuDomainSecretInfoPtr
1652
qemuDomainSecretInfoNewPlain(virSecretUsageType usageType,
1653
                             const char *username,
1654
                             virSecretLookupTypeDefPtr lookupDef)
1655 1656 1657 1658 1659 1660
{
    qemuDomainSecretInfoPtr secinfo = NULL;

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

1661
    if (qemuDomainSecretPlainSetup(secinfo, usageType, username, lookupDef) < 0) {
1662
        g_clear_pointer(&secinfo, qemuDomainSecretInfoFree);
1663
        return NULL;
1664 1665 1666 1667 1668 1669
    }

    return secinfo;
}


1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680
/**
 * 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.
 */
1681
qemuDomainSecretInfoPtr
1682
qemuDomainSecretInfoTLSNew(qemuDomainObjPrivatePtr priv,
1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695
                           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;

1696
    return qemuDomainSecretAESSetupFromSecret(priv, srcAlias, NULL,
P
Peter Krempa 已提交
1697
                                              VIR_SECRET_USAGE_TYPE_TLS,
1698
                                              NULL, &seclookupdef);
1699 1700 1701
}


1702 1703 1704
void
qemuDomainSecretDiskDestroy(virDomainDiskDefPtr disk)
{
1705 1706
    qemuDomainStorageSourcePrivatePtr srcPriv;
    virStorageSourcePtr n;
1707

1708 1709 1710 1711 1712 1713
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if ((srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(n))) {
            qemuDomainSecretInfoDestroy(srcPriv->secinfo);
            qemuDomainSecretInfoDestroy(srcPriv->encinfo);
        }
    }
1714 1715 1716
}


1717
bool
1718
qemuDomainStorageSourceHasAuth(virStorageSourcePtr src)
1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730
{
    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;
}


1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
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;
}


1743 1744 1745 1746 1747 1748
static qemuDomainSecretInfoPtr
qemuDomainSecretStorageSourcePrepareCookies(qemuDomainObjPrivatePtr priv,
                                            virStorageSourcePtr src,
                                            const char *aliasprotocol)
{
    g_autofree char *secretalias = qemuAliasForSecret(aliasprotocol, "httpcookie");
1749
    g_autofree char *cookies = qemuBlockStorageSourceGetCookieString(src);
1750 1751 1752 1753 1754 1755

    return qemuDomainSecretAESSetup(priv, secretalias, NULL,
                                    (uint8_t *) cookies, strlen(cookies));
}


1756 1757 1758 1759 1760 1761
/**
 * 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
1762
 *
1763 1764 1765 1766
 * 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.
1767
 *
1768
 * Returns 0 on success; -1 on error while reporting an libvirt error.
1769
 */
1770
static int
1771
qemuDomainSecretStorageSourcePrepare(qemuDomainObjPrivatePtr priv,
1772
                                     virStorageSourcePtr src,
1773 1774
                                     const char *aliasprotocol,
                                     const char *aliasformat)
1775
{
1776
    qemuDomainStorageSourcePrivatePtr srcPriv;
1777
    bool iscsiHasPS = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET);
1778
    bool hasAuth = qemuDomainStorageSourceHasAuth(src);
1779 1780
    bool hasEnc = qemuDomainDiskHasEncryptionSecret(src);

1781
    if (!hasAuth && !hasEnc && src->ncookies == 0)
1782
        return 0;
1783

1784
    if (!(src->privateData = qemuDomainStorageSourcePrivateNew()))
1785 1786
        return -1;

1787
    srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
1788

1789
    if (hasAuth) {
J
John Ferlan 已提交
1790
        virSecretUsageType usageType = VIR_SECRET_USAGE_TYPE_ISCSI;
1791

1792
        if (src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD)
J
John Ferlan 已提交
1793
            usageType = VIR_SECRET_USAGE_TYPE_CEPH;
1794

1795 1796 1797 1798 1799 1800
        if (!qemuDomainSupportsEncryptedSecret(priv) ||
            (src->protocol == VIR_STORAGE_NET_PROTOCOL_ISCSI && !iscsiHasPS)) {
            srcPriv->secinfo = qemuDomainSecretInfoNewPlain(usageType,
                                                            src->auth->username,
                                                            &src->auth->seclookupdef);
        } else {
1801
            srcPriv->secinfo = qemuDomainSecretAESSetupFromSecret(priv, aliasprotocol,
1802
                                                                  "auth",
P
Peter Krempa 已提交
1803 1804
                                                                  usageType,
                                                                  src->auth->username,
1805
                                                                  &src->auth->seclookupdef);
1806 1807 1808 1809
        }

        if (!srcPriv->secinfo)
            return -1;
1810 1811
    }

1812
    if (hasEnc) {
1813
        if (!(srcPriv->encinfo = qemuDomainSecretAESSetupFromSecret(priv, aliasformat,
1814
                                                                    "encryption",
P
Peter Krempa 已提交
1815 1816
                                                                    VIR_SECRET_USAGE_TYPE_VOLUME,
                                                                    NULL,
1817
                                                                    &src->encryption->secrets[0]->seclookupdef)))
1818
              return -1;
1819 1820
    }

1821 1822 1823 1824 1825 1826 1827
    if (src->ncookies &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV) &&
        !(srcPriv->httpcookie = qemuDomainSecretStorageSourcePrepareCookies(priv,
                                                                            src,
                                                                            aliasprotocol)))
        return -1;

1828 1829 1830 1831
    return 0;
}


1832 1833 1834
void
qemuDomainSecretHostdevDestroy(virDomainHostdevDefPtr hostdev)
{
1835
    qemuDomainStorageSourcePrivatePtr srcPriv;
1836

1837 1838 1839
    if (virHostdevIsSCSIDevice(hostdev)) {
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1840

1841 1842 1843
        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI) {
            srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(iscsisrc->src);
            if (srcPriv && srcPriv->secinfo)
1844
                g_clear_pointer(&srcPriv->secinfo, qemuDomainSecretInfoFree);
1845 1846
        }
    }
1847 1848 1849 1850
}


/* qemuDomainSecretHostdevPrepare:
J
John Ferlan 已提交
1851
 * @priv: pointer to domain private object
1852 1853 1854 1855 1856 1857 1858
 * @hostdev: Pointer to a hostdev definition
 *
 * For the right host device, generate the qemuDomainSecretInfo structure.
 *
 * Returns 0 on success, -1 on failure
 */
int
1859
qemuDomainSecretHostdevPrepare(qemuDomainObjPrivatePtr priv,
1860 1861
                               virDomainHostdevDefPtr hostdev)
{
1862
    if (virHostdevIsSCSIDevice(hostdev)) {
1863 1864
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1865
        virStorageSourcePtr src = iscsisrc->src;
1866 1867

        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI &&
1868
            src->auth) {
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884
            bool iscsiHasPS = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET);
            virSecretUsageType usageType = VIR_SECRET_USAGE_TYPE_ISCSI;
            qemuDomainStorageSourcePrivatePtr srcPriv;

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

            srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);

            if (!qemuDomainSupportsEncryptedSecret(priv) || !iscsiHasPS) {
                srcPriv->secinfo = qemuDomainSecretInfoNewPlain(usageType,
                                                                src->auth->username,
                                                                &src->auth->seclookupdef);
            } else {
                srcPriv->secinfo = qemuDomainSecretAESSetupFromSecret(priv,
                                                                      hostdev->info->alias,
1885
                                                                      NULL,
1886 1887
                                                                      usageType,
                                                                      src->auth->username,
1888
                                                                      &src->auth->seclookupdef);
1889 1890 1891
            }

            if (!srcPriv->secinfo)
1892 1893 1894 1895 1896 1897 1898 1899
                return -1;
        }
    }

    return 0;
}


1900 1901 1902 1903 1904 1905 1906 1907 1908
void
qemuDomainSecretChardevDestroy(virDomainChrSourceDefPtr dev)
{
    qemuDomainChrSourcePrivatePtr chrSourcePriv =
        QEMU_DOMAIN_CHR_SOURCE_PRIVATE(dev);

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

1909
    g_clear_pointer(&chrSourcePriv->secinfo, qemuDomainSecretInfoFree);
1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924
}


/* 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
1925
qemuDomainSecretChardevPrepare(virQEMUDriverConfigPtr cfg,
1926 1927 1928 1929
                               qemuDomainObjPrivatePtr priv,
                               const char *chrAlias,
                               virDomainChrSourceDefPtr dev)
{
1930
    g_autofree char *charAlias = NULL;
1931 1932 1933 1934 1935 1936 1937 1938 1939 1940

    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)))
1941
            return -1;
1942

1943
        chrSourcePriv->secinfo =
1944
            qemuDomainSecretInfoTLSNew(priv, charAlias,
1945 1946 1947 1948
                                       cfg->chardevTLSx509secretUUID);

        if (!chrSourcePriv->secinfo)
            return -1;
1949 1950 1951 1952 1953 1954
    }

    return 0;
}


1955 1956 1957 1958 1959 1960 1961 1962 1963
static void
qemuDomainSecretGraphicsDestroy(virDomainGraphicsDefPtr graphics)
{
    qemuDomainGraphicsPrivatePtr gfxPriv = QEMU_DOMAIN_GRAPHICS_PRIVATE(graphics);

    if (!gfxPriv)
        return;

    VIR_FREE(gfxPriv->tlsAlias);
1964
    g_clear_pointer(&gfxPriv->secinfo, qemuDomainSecretInfoFree);
1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984
}


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;

1985
    gfxPriv->tlsAlias = g_strdup("vnc-tls-creds0");
1986

1987 1988 1989 1990 1991 1992 1993
    if (cfg->vncTLSx509secretUUID) {
        gfxPriv->secinfo = qemuDomainSecretInfoTLSNew(priv, gfxPriv->tlsAlias,
                                                      cfg->vncTLSx509secretUUID);
        if (!gfxPriv->secinfo)
            return -1;
    }

1994 1995 1996 1997
    return 0;
}


1998 1999 2000
/* qemuDomainSecretDestroy:
 * @vm: Domain object
 *
2001
 * Removes all unnecessary data which was needed to generate 'secret' objects.
2002 2003 2004 2005 2006 2007 2008 2009
 */
void
qemuDomainSecretDestroy(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++)
        qemuDomainSecretDiskDestroy(vm->def->disks[i]);
2010 2011 2012

    for (i = 0; i < vm->def->nhostdevs; i++)
        qemuDomainSecretHostdevDestroy(vm->def->hostdevs[i]);
2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038

    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);
2039 2040 2041

    for (i = 0; i < vm->def->ngraphics; i++)
        qemuDomainSecretGraphicsDestroy(vm->def->graphics[i]);
2042 2043 2044 2045
}


/* qemuDomainSecretPrepare:
2046
 * @driver: Pointer to driver object
2047 2048 2049
 * @vm: Domain object
 *
 * For any objects that may require an auth/secret setup, create a
2050
 * qemuDomainSecretInfo and save it in the appropriate place within
2051 2052 2053 2054 2055 2056 2057
 * 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
2058
qemuDomainSecretPrepare(virQEMUDriverPtr driver,
2059 2060
                        virDomainObjPtr vm)
{
J
John Ferlan 已提交
2061
    qemuDomainObjPrivatePtr priv = vm->privateData;
2062
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
2063 2064
    size_t i;

2065
    /* disk secrets are prepared when preparing disks */
2066

2067
    for (i = 0; i < vm->def->nhostdevs; i++) {
2068
        if (qemuDomainSecretHostdevPrepare(priv,
J
John Ferlan 已提交
2069
                                           vm->def->hostdevs[i]) < 0)
2070
            return -1;
2071 2072
    }

2073
    for (i = 0; i < vm->def->nserials; i++) {
2074
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2075 2076
                                           vm->def->serials[i]->info.alias,
                                           vm->def->serials[i]->source) < 0)
2077
            return -1;
2078 2079 2080
    }

    for (i = 0; i < vm->def->nparallels; i++) {
2081
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2082 2083
                                           vm->def->parallels[i]->info.alias,
                                           vm->def->parallels[i]->source) < 0)
2084
            return -1;
2085 2086 2087
    }

    for (i = 0; i < vm->def->nchannels; i++) {
2088
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2089 2090
                                           vm->def->channels[i]->info.alias,
                                           vm->def->channels[i]->source) < 0)
2091
            return -1;
2092 2093 2094
    }

    for (i = 0; i < vm->def->nconsoles; i++) {
2095
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2096 2097
                                           vm->def->consoles[i]->info.alias,
                                           vm->def->consoles[i]->source) < 0)
2098
            return -1;
2099 2100 2101 2102 2103
    }

    for (i = 0; i < vm->def->nsmartcards; i++)
        if (vm->def->smartcards[i]->type ==
            VIR_DOMAIN_SMARTCARD_TYPE_PASSTHROUGH &&
2104
            qemuDomainSecretChardevPrepare(cfg, priv,
2105 2106
                                           vm->def->smartcards[i]->info.alias,
                                           vm->def->smartcards[i]->data.passthru) < 0)
2107
            return -1;
2108 2109 2110

    for (i = 0; i < vm->def->nrngs; i++) {
        if (vm->def->rngs[i]->backend == VIR_DOMAIN_RNG_BACKEND_EGD &&
2111
            qemuDomainSecretChardevPrepare(cfg, priv,
2112 2113
                                           vm->def->rngs[i]->info.alias,
                                           vm->def->rngs[i]->source.chardev) < 0)
2114
            return -1;
2115 2116 2117
    }

    for (i = 0; i < vm->def->nredirdevs; i++) {
2118
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2119 2120
                                           vm->def->redirdevs[i]->info.alias,
                                           vm->def->redirdevs[i]->source) < 0)
2121
            return -1;
2122 2123
    }

2124 2125
    for (i = 0; i < vm->def->ngraphics; i++) {
        if (qemuDomainSecretGraphicsPrepare(cfg, priv, vm->def->graphics[i]) < 0)
2126
            return -1;
2127 2128
    }

2129
    return 0;
2130 2131 2132
}


2133
/* This is the old way of setting up per-domain directories */
2134
static void
2135 2136 2137 2138
qemuDomainSetPrivatePathsOld(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2139
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
2140

2141 2142
    if (!priv->libDir)
        priv->libDir = g_strdup_printf("%s/domain-%s", cfg->libDir, vm->def->name);
2143

2144 2145 2146
    if (!priv->channelTargetDir)
        priv->channelTargetDir = g_strdup_printf("%s/domain-%s",
                                                 cfg->channelTargetDir, vm->def->name);
2147 2148 2149 2150
}


int
2151 2152
qemuDomainSetPrivatePaths(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
2153
{
2154
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
2155
    qemuDomainObjPrivatePtr priv = vm->privateData;
2156
    g_autofree char *domname = virDomainDefGetShortName(vm->def);
2157

M
Martin Kletzander 已提交
2158
    if (!domname)
2159
        return -1;
M
Martin Kletzander 已提交
2160

2161 2162
    if (!priv->libDir)
        priv->libDir = g_strdup_printf("%s/domain-%s", cfg->libDir, domname);
2163

2164 2165 2166
    if (!priv->channelTargetDir)
        priv->channelTargetDir = g_strdup_printf("%s/domain-%s",
                                                 cfg->channelTargetDir, domname);
2167

2168
    return 0;
2169 2170 2171
}


2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198
int
qemuDomainObjStartWorker(virDomainObjPtr dom)
{
    qemuDomainObjPrivatePtr priv = dom->privateData;

    if (!priv->eventThread) {
        g_autofree char *threadName = g_strdup_printf("vm-%s", dom->def->name);
        if (!(priv->eventThread = virEventThreadNew(threadName)))
            return -1;
    }

    return 0;
}


void
qemuDomainObjStopWorker(virDomainObjPtr dom)
{
    qemuDomainObjPrivatePtr priv = dom->privateData;

    if (priv->eventThread) {
        g_object_unref(priv->eventThread);
        priv->eventThread = NULL;
    }
}


2199
static void *
2200
qemuDomainObjPrivateAlloc(void *opaque)
2201 2202 2203 2204 2205 2206
{
    qemuDomainObjPrivatePtr priv;

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

2207 2208 2209
    if (qemuDomainObjInitJob(priv) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to init qemu driver mutexes"));
2210
        goto error;
2211
    }
2212

2213
    if (!(priv->devs = virChrdevAlloc()))
2214 2215
        goto error;

2216 2217 2218
    if (!(priv->blockjobs = virHashCreate(5, virObjectFreeHashData)))
        goto error;

2219 2220
    /* agent commands block by default, user can choose different behavior */
    priv->agentTimeout = VIR_DOMAIN_AGENT_RESPONSE_TIMEOUT_BLOCK;
2221
    priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
2222
    priv->driver = opaque;
2223

2224
    return priv;
2225

2226
 error:
2227 2228
    VIR_FREE(priv);
    return NULL;
2229 2230
}

2231 2232 2233 2234 2235 2236 2237 2238 2239
/**
 * 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)
2240
{
2241 2242 2243
    virStringListFree(priv->qemuDevices);
    priv->qemuDevices = NULL;

2244
    virCgroupFree(&priv->cgroup);
2245 2246 2247 2248 2249

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

    VIR_FREE(priv->machineName);
2250

2251
    virObjectUnref(priv->qemuCaps);
2252
    priv->qemuCaps = NULL;
2253

2254
    VIR_FREE(priv->pidfile);
2255

2256 2257 2258
    VIR_FREE(priv->libDir);
    VIR_FREE(priv->channelTargetDir);

2259 2260
    priv->memPrealloc = false;

2261 2262 2263 2264 2265 2266 2267
    /* remove automatic pinning data */
    virBitmapFree(priv->autoNodeset);
    priv->autoNodeset = NULL;
    virBitmapFree(priv->autoCpuset);
    priv->autoCpuset = NULL;

    /* remove address data */
2268
    virDomainPCIAddressSetFree(priv->pciaddrs);
2269
    priv->pciaddrs = NULL;
J
Ján Tomko 已提交
2270
    virDomainUSBAddressSetFree(priv->usbaddrs);
2271 2272 2273 2274 2275 2276 2277 2278
    priv->usbaddrs = NULL;

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

    /* clear previously used namespaces */
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
2279

2280 2281
    priv->rememberOwner = false;

2282
    priv->reconnectBlockjobs = VIR_TRISTATE_BOOL_ABSENT;
2283
    priv->allowReboot = VIR_TRISTATE_BOOL_ABSENT;
2284 2285 2286

    virBitmapFree(priv->migrationCaps);
    priv->migrationCaps = NULL;
2287

2288
    virHashRemoveAll(priv->blockjobs);
2289 2290 2291 2292 2293

    virObjectUnref(priv->pflash0);
    priv->pflash0 = NULL;
    virObjectUnref(priv->pflash1);
    priv->pflash1 = NULL;
2294 2295 2296

    virDomainBackupDefFree(priv->backup);
    priv->backup = NULL;
2297 2298 2299

    /* reset node name allocator */
    qemuDomainStorageIdReset(priv);
2300 2301 2302 2303 2304 2305 2306 2307 2308 2309
}


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

    qemuDomainObjPrivateDataClear(priv);

2310
    virObjectUnref(priv->monConfig);
2311
    qemuDomainObjFreeJob(priv);
2312
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
2313
    VIR_FREE(priv->origname);
2314

2315
    virChrdevFree(priv->devs);
2316

2317 2318
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
2319
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
2320 2321
        qemuMonitorClose(priv->mon);
    }
D
Daniel P. Berrange 已提交
2322 2323 2324 2325
    if (priv->agent) {
        VIR_ERROR(_("Unexpected QEMU agent still active during domain deletion"));
        qemuAgentClose(priv->agent);
    }
2326
    VIR_FREE(priv->cleanupCallbacks);
2327

2328
    g_clear_pointer(&priv->migSecinfo, qemuDomainSecretInfoFree);
2329
    qemuDomainMasterKeyFree(priv);
2330

2331 2332
    virHashFree(priv->blockjobs);

2333 2334 2335 2336 2337 2338
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->eventThread) {
        VIR_ERROR(_("Unexpected event thread still active during domain deletion"));
        g_object_unref(priv->eventThread);
    }

2339 2340 2341 2342
    VIR_FREE(priv);
}


2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357
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)
2358
        (*secinfo)->s.aes.alias = g_steal_pointer(&*alias);
2359 2360 2361 2362 2363

    return 0;
}


2364 2365 2366 2367
static int
qemuStorageSourcePrivateDataParse(xmlXPathContextPtr ctxt,
                                  virStorageSourcePtr src)
{
2368
    qemuDomainStorageSourcePrivatePtr priv;
2369 2370
    g_autofree char *authalias = NULL;
    g_autofree char *encalias = NULL;
2371
    g_autofree char *httpcookiealias = NULL;
2372

2373 2374
    src->nodestorage = virXPathString("string(./nodenames/nodename[@type='storage']/@name)", ctxt);
    src->nodeformat = virXPathString("string(./nodenames/nodename[@type='format']/@name)", ctxt);
2375
    src->tlsAlias = virXPathString("string(./objects/TLSx509/@alias)", ctxt);
2376

2377 2378 2379
    if (src->sliceStorage)
        src->sliceStorage->nodename = virXPathString("string(./nodenames/nodename[@type='slice-storage']/@name)", ctxt);

2380 2381 2382
    if (src->pr)
        src->pr->mgralias = virXPathString("string(./reservations/@mgralias)", ctxt);

2383 2384
    authalias = virXPathString("string(./objects/secret[@type='auth']/@alias)", ctxt);
    encalias = virXPathString("string(./objects/secret[@type='encryption']/@alias)", ctxt);
2385
    httpcookiealias = virXPathString("string(./objects/secret[@type='httpcookie']/@alias)", ctxt);
2386

2387
    if (authalias || encalias || httpcookiealias) {
2388 2389
        if (!src->privateData &&
            !(src->privateData = qemuDomainStorageSourcePrivateNew()))
2390
            return -1;
2391 2392 2393 2394

        priv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);

        if (qemuStorageSourcePrivateDataAssignSecinfo(&priv->secinfo, &authalias) < 0)
2395
            return -1;
2396 2397

        if (qemuStorageSourcePrivateDataAssignSecinfo(&priv->encinfo, &encalias) < 0)
2398
            return -1;
2399 2400 2401

        if (qemuStorageSourcePrivateDataAssignSecinfo(&priv->httpcookie, &httpcookiealias) < 0)
            return -1;
2402 2403
    }

2404
    if (virStorageSourcePrivateDataParseRelPath(ctxt, src) < 0)
2405
        return -1;
2406

2407
    return 0;
2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422
}


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);
2423 2424 2425 2426 2427 2428 2429
}


static int
qemuStorageSourcePrivateDataFormat(virStorageSourcePtr src,
                                   virBufferPtr buf)
{
2430
    g_auto(virBuffer) tmp = VIR_BUFFER_INIT_CHILD(buf);
2431
    qemuDomainStorageSourcePrivatePtr srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
2432
    g_auto(virBuffer) nodenamesChildBuf = VIR_BUFFER_INIT_CHILD(buf);
2433

2434 2435 2436
    virBufferEscapeString(&nodenamesChildBuf, "<nodename type='storage' name='%s'/>\n", src->nodestorage);
    virBufferEscapeString(&nodenamesChildBuf, "<nodename type='format' name='%s'/>\n", src->nodeformat);

2437 2438 2439 2440
    if (src->sliceStorage)
        virBufferEscapeString(&nodenamesChildBuf, "<nodename type='slice-storage' name='%s'/>\n",
                              src->sliceStorage->nodename);

2441
    virXMLFormatElement(buf, "nodenames", NULL, &nodenamesChildBuf);
2442

2443 2444 2445
    if (src->pr)
        virBufferAsprintf(buf, "<reservations mgralias='%s'/>\n", src->pr->mgralias);

2446
    if (virStorageSourcePrivateDataFormatRelPath(src, buf) < 0)
2447
        return -1;
2448

2449 2450 2451
    if (srcPriv) {
        qemuStorageSourcePrivateDataFormatSecinfo(&tmp, srcPriv->secinfo, "auth");
        qemuStorageSourcePrivateDataFormatSecinfo(&tmp, srcPriv->encinfo, "encryption");
2452
        qemuStorageSourcePrivateDataFormatSecinfo(&tmp, srcPriv->httpcookie, "httpcookie");
2453 2454
    }

2455 2456 2457
    if (src->tlsAlias)
        virBufferAsprintf(&tmp, "<TLSx509 alias='%s'/>\n", src->tlsAlias);

2458
    virXMLFormatElement(buf, "objects", NULL, &tmp);
2459

2460
    return 0;
2461 2462 2463
}


2464 2465 2466 2467 2468 2469 2470
static int
qemuDomainDiskPrivateParse(xmlXPathContextPtr ctxt,
                           virDomainDiskDefPtr disk)
{
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);

    priv->qomName = virXPathString("string(./qom/@name)", ctxt);
2471
    priv->nodeCopyOnRead = virXPathString("string(./nodenames/nodename[@type='copyOnRead']/@name)", ctxt);
2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484

    return 0;
}


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

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

2485 2486 2487 2488 2489 2490 2491 2492 2493
    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");
    }

2494 2495 2496 2497
    return 0;
}


2498 2499
static void
qemuDomainObjPrivateXMLFormatVcpus(virBufferPtr buf,
2500
                                   virDomainDefPtr def)
2501 2502
{
    size_t i;
2503 2504 2505
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    pid_t tid;
2506 2507 2508 2509

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

2510 2511 2512 2513 2514 2515 2516 2517 2518
    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);
    }
2519 2520 2521 2522 2523 2524

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


2525
static int
2526
qemuDomainObjPrivateXMLFormatAutomaticPlacement(virBufferPtr buf,
2527
                                                qemuDomainObjPrivatePtr priv)
2528
{
2529 2530
    g_autofree char *nodeset = NULL;
    g_autofree char *cpuset = NULL;
2531

2532
    if (!priv->autoNodeset && !priv->autoCpuset)
2533 2534
        return 0;

2535 2536
    if (priv->autoNodeset &&
        !((nodeset = virBitmapFormat(priv->autoNodeset))))
2537
        return -1;
2538

2539 2540
    if (priv->autoCpuset &&
        !((cpuset = virBitmapFormat(priv->autoCpuset))))
2541
        return -1;
2542 2543 2544 2545 2546

    virBufferAddLit(buf, "<numad");
    virBufferEscapeString(buf, " nodeset='%s'", nodeset);
    virBufferEscapeString(buf, " cpuset='%s'", cpuset);
    virBufferAddLit(buf, "/>\n");
2547

2548
    return 0;
2549 2550 2551
}


2552 2553 2554 2555 2556 2557 2558
typedef struct qemuDomainPrivateBlockJobFormatData {
    virDomainXMLOptionPtr xmlopt;
    virBufferPtr buf;
} qemuDomainPrivateBlockJobFormatData;


static int
2559 2560 2561 2562 2563
qemuDomainObjPrivateXMLFormatBlockjobFormatSource(virBufferPtr buf,
                                                  const char *element,
                                                  virStorageSourcePtr src,
                                                  virDomainXMLOptionPtr xmlopt,
                                                  bool chain)
2564
{
2565
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2566
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2567 2568 2569 2570 2571 2572
    unsigned int xmlflags = VIR_DOMAIN_DEF_FORMAT_STATUS;

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

2573
    if (virDomainDiskSourceFormat(&childBuf, src, "source", 0, true, xmlflags, true, xmlopt) < 0)
2574 2575
        return -1;

2576 2577
    if (chain &&
        virDomainDiskBackingStoreFormat(&childBuf, src, xmlopt, xmlflags) < 0)
2578 2579
        return -1;

2580
    virXMLFormatElement(buf, element, &attrBuf, &childBuf);
2581 2582 2583 2584 2585

    return 0;
}


2586 2587 2588 2589
static void
qemuDomainPrivateBlockJobFormatCommit(qemuBlockJobDataPtr job,
                                      virBufferPtr buf)
{
2590 2591 2592
    g_auto(virBuffer) disabledBitmapsBuf = VIR_BUFFER_INIT_CHILD(buf);
    char **bitmaps = job->data.commit.disabledBitmapsBase;

2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603
    if (job->data.commit.base)
        virBufferAsprintf(buf, "<base node='%s'/>\n", job->data.commit.base->nodeformat);

    if (job->data.commit.top)
        virBufferAsprintf(buf, "<top node='%s'/>\n", job->data.commit.top->nodeformat);

    if (job->data.commit.topparent)
        virBufferAsprintf(buf, "<topparent node='%s'/>\n", job->data.commit.topparent->nodeformat);

    if (job->data.commit.deleteCommittedImages)
        virBufferAddLit(buf, "<deleteCommittedImages/>\n");
2604 2605 2606 2607 2608

    while (bitmaps && *bitmaps)
        virBufferEscapeString(&disabledBitmapsBuf, "<bitmap name='%s'/>\n", *(bitmaps++));

    virXMLFormatElement(buf, "disabledBaseBitmaps", NULL, &disabledBitmapsBuf);
2609 2610 2611
}


2612 2613
static int
qemuDomainObjPrivateXMLFormatBlockjobIterator(void *payload,
J
Ján Tomko 已提交
2614
                                              const void *name G_GNUC_UNUSED,
2615
                                              void *opaque)
2616
{
2617
    struct qemuDomainPrivateBlockJobFormatData *data = opaque;
2618
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2619 2620
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(data->buf);
    g_auto(virBuffer) chainsBuf = VIR_BUFFER_INIT_CHILD(&childBuf);
2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631
    qemuBlockJobDataPtr job = payload;
    const char *state = qemuBlockjobStateTypeToString(job->state);
    const char *newstate = NULL;

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

    virBufferEscapeString(&attrBuf, " name='%s'", job->name);
    virBufferEscapeString(&attrBuf, " type='%s'", qemuBlockjobTypeToString(job->type));
    virBufferEscapeString(&attrBuf, " state='%s'", state);
    virBufferEscapeString(&attrBuf, " newstate='%s'", newstate);
2632 2633
    if (job->brokentype != QEMU_BLOCKJOB_TYPE_NONE)
        virBufferEscapeString(&attrBuf, " brokentype='%s'", qemuBlockjobTypeToString(job->brokentype));
2634 2635
    if (!job->jobflagsmissing)
        virBufferAsprintf(&attrBuf, " jobflags='0x%x'", job->jobflags);
2636 2637
    virBufferEscapeString(&childBuf, "<errmsg>%s</errmsg>", job->errmsg);

2638
    if (job->disk) {
2639 2640 2641 2642
        virBufferEscapeString(&childBuf, "<disk dst='%s'", job->disk->dst);
        if (job->mirrorChain)
            virBufferAddLit(&childBuf, " mirror='yes'");
        virBufferAddLit(&childBuf, "/>\n");
2643 2644
    } else {
        if (job->chain &&
2645 2646 2647 2648 2649
            qemuDomainObjPrivateXMLFormatBlockjobFormatSource(&chainsBuf,
                                                              "disk",
                                                              job->chain,
                                                              data->xmlopt,
                                                              true) < 0)
2650
            return -1;
2651

2652
        if (job->mirrorChain &&
2653 2654 2655 2656 2657
            qemuDomainObjPrivateXMLFormatBlockjobFormatSource(&chainsBuf,
                                                              "mirror",
                                                              job->mirrorChain,
                                                              data->xmlopt,
                                                              true) < 0)
2658 2659
            return -1;

2660
        virXMLFormatElement(&childBuf, "chains", NULL, &chainsBuf);
2661 2662
    }

2663 2664 2665 2666 2667 2668 2669 2670
    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:
2671
            qemuDomainPrivateBlockJobFormatCommit(job, &childBuf);
2672 2673
            break;

2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686
        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;

2687
        case QEMU_BLOCKJOB_TYPE_COPY:
2688 2689 2690 2691
            if (job->data.copy.shallownew)
                virBufferAddLit(&attrBuf, " shallownew='yes'");
            break;

P
Peter Krempa 已提交
2692
        case QEMU_BLOCKJOB_TYPE_BACKUP:
2693 2694 2695 2696 2697 2698 2699 2700 2701
            virBufferEscapeString(&childBuf, "<bitmap name='%s'/>\n", job->data.backup.bitmap);
            if (job->data.backup.store) {
                if (qemuDomainObjPrivateXMLFormatBlockjobFormatSource(&childBuf,
                                                                      "store",
                                                                      job->data.backup.store,
                                                                      data->xmlopt,
                                                                      false) < 0)
                    return -1;
            }
P
Peter Krempa 已提交
2702
            break;
2703 2704

        case QEMU_BLOCKJOB_TYPE_BROKEN:
2705 2706 2707 2708 2709 2710
        case QEMU_BLOCKJOB_TYPE_NONE:
        case QEMU_BLOCKJOB_TYPE_INTERNAL:
        case QEMU_BLOCKJOB_TYPE_LAST:
            break;
    }

2711 2712
    virXMLFormatElement(data->buf, "blockjob", &attrBuf, &childBuf);
    return 0;
2713 2714 2715
}


2716 2717 2718 2719
static int
qemuDomainObjPrivateXMLFormatBlockjobs(virBufferPtr buf,
                                       virDomainObjPtr vm)
{
2720
    qemuDomainObjPrivatePtr priv = vm->privateData;
2721
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2722
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2723
    bool bj = qemuDomainHasBlockjob(vm, false);
2724 2725
    struct qemuDomainPrivateBlockJobFormatData iterdata = { priv->driver->xmlopt,
                                                            &childBuf };
2726 2727 2728 2729

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

2730 2731 2732
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV) &&
        virHashForEach(priv->blockjobs,
                       qemuDomainObjPrivateXMLFormatBlockjobIterator,
2733
                       &iterdata) < 0)
2734 2735
        return -1;

2736 2737
    virXMLFormatElement(buf, "blockjobs", &attrBuf, &childBuf);
    return 0;
2738 2739 2740
}


2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760
static int
qemuDomainObjPrivateXMLFormatBackups(virBufferPtr buf,
                                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);

    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_INCREMENTAL_BACKUP))
        return 0;

    if (priv->backup &&
        virDomainBackupDefFormat(&childBuf, priv->backup, true) < 0)
        return -1;

    virXMLFormatElement(buf, "backups", &attrBuf, &childBuf);
    return 0;
}


2761
void
2762 2763 2764 2765 2766 2767 2768 2769 2770
qemuDomainObjPrivateXMLFormatAllowReboot(virBufferPtr buf,
                                         virTristateBool allowReboot)
{
    virBufferAsprintf(buf, "<allowReboot value='%s'/>\n",
                      virTristateBoolTypeToString(allowReboot));

}


2771 2772 2773 2774 2775 2776 2777 2778 2779
static void
qemuDomainObjPrivateXMLFormatPR(virBufferPtr buf,
                                qemuDomainObjPrivatePtr priv)
{
    if (priv->prDaemonRunning)
        virBufferAddLit(buf, "<prDaemon/>\n");
}


2780 2781
static int
qemuDomainObjPrivateXMLFormatNBDMigrationSource(virBufferPtr buf,
2782 2783
                                                virStorageSourcePtr src,
                                                virDomainXMLOptionPtr xmlopt)
2784
{
2785
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2786
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2787 2788 2789 2790 2791

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

2792
    if (virDomainDiskSourceFormat(&childBuf, src, "source", 0, false,
2793
                                  VIR_DOMAIN_DEF_FORMAT_STATUS, true, xmlopt) < 0)
2794
        return -1;
2795

2796
    virXMLFormatElement(buf, "migrationSource", &attrBuf, &childBuf);
2797

2798
    return 0;
2799 2800 2801 2802
}


static int
2803 2804 2805
qemuDomainObjPrivateXMLFormatNBDMigration(virBufferPtr buf,
                                          virDomainObjPtr vm)
{
2806
    qemuDomainObjPrivatePtr priv = vm->privateData;
2807 2808 2809 2810 2811
    size_t i;
    virDomainDiskDefPtr disk;
    qemuDomainDiskPrivatePtr diskPriv;

    for (i = 0; i < vm->def->ndisks; i++) {
2812 2813
        g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
        g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2814 2815
        disk = vm->def->disks[i];
        diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
2816 2817

        virBufferAsprintf(&attrBuf, " dev='%s' migrating='%s'",
2818
                          disk->dst, diskPriv->migrating ? "yes" : "no");
2819 2820 2821

        if (diskPriv->migrSource &&
            qemuDomainObjPrivateXMLFormatNBDMigrationSource(&childBuf,
2822 2823
                                                            diskPriv->migrSource,
                                                            priv->driver->xmlopt) < 0)
2824
            return -1;
2825

2826
        virXMLFormatElement(buf, "disk", &attrBuf, &childBuf);
2827
    }
2828

2829
    return 0;
2830 2831 2832
}


2833
static int
2834 2835 2836 2837
qemuDomainObjPrivateXMLFormatJob(virBufferPtr buf,
                                 virDomainObjPtr vm,
                                 qemuDomainObjPrivatePtr priv)
{
2838
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2839
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2840
    qemuDomainJob job = priv->job.active;
2841 2842

    if (!qemuDomainTrackJob(job))
2843
        job = QEMU_JOB_NONE;
2844

2845 2846
    if (job == QEMU_JOB_NONE &&
        priv->job.asyncJob == QEMU_ASYNC_JOB_NONE)
2847 2848 2849
        return 0;

    virBufferAsprintf(&attrBuf, " type='%s' async='%s'",
2850 2851
                      qemuDomainJobTypeToString(job),
                      qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
2852

2853
    if (priv->job.phase) {
2854
        virBufferAsprintf(&attrBuf, " phase='%s'",
2855 2856 2857
                          qemuDomainAsyncJobPhaseToString(priv->job.asyncJob,
                                                          priv->job.phase));
    }
2858

2859 2860 2861
    if (priv->job.asyncJob != QEMU_ASYNC_JOB_NONE)
        virBufferAsprintf(&attrBuf, " flags='0x%lx'", priv->job.apiFlags);

2862 2863
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_OUT &&
        qemuDomainObjPrivateXMLFormatNBDMigration(&childBuf, vm) < 0)
2864
        return -1;
2865

2866 2867 2868
    if (priv->job.migParams)
        qemuMigrationParamsFormat(&childBuf, priv->job.migParams);

2869
    virXMLFormatElement(buf, "job", &attrBuf, &childBuf);
2870

2871
    return 0;
2872 2873 2874
}


2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890
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;
}


2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908
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;
}


2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949
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;
}

2950
static int
2951 2952
qemuDomainObjPrivateXMLFormat(virBufferPtr buf,
                              virDomainObjPtr vm)
2953
{
2954
    qemuDomainObjPrivatePtr priv = vm->privateData;
2955 2956 2957 2958
    const char *monitorpath;

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
2959
        switch (priv->monConfig->type) {
2960
        case VIR_DOMAIN_CHR_TYPE_UNIX:
2961
            monitorpath = priv->monConfig->data.nix.path;
2962 2963 2964
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
2965
            monitorpath = priv->monConfig->data.file.path;
2966 2967 2968
            break;
        }

2969
        virBufferEscapeString(buf, "<monitor path='%s'", monitorpath);
2970
        virBufferAsprintf(buf, " type='%s'/>\n",
2971
                          virDomainChrTypeToString(priv->monConfig->type));
2972 2973
    }

2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984
    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");
    }

2985
    qemuDomainObjPrivateXMLFormatVcpus(buf, vm->def);
2986

2987
    if (priv->qemuCaps) {
2988
        size_t i;
2989 2990
        virBufferAddLit(buf, "<qemuCaps>\n");
        virBufferAdjustIndent(buf, 2);
2991
        for (i = 0; i < QEMU_CAPS_LAST; i++) {
2992
            if (virQEMUCapsGet(priv->qemuCaps, i)) {
2993
                virBufferAsprintf(buf, "<flag name='%s'/>\n",
2994
                                  virQEMUCapsTypeToString(i));
2995 2996
            }
        }
2997 2998
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</qemuCaps>\n");
2999 3000
    }

3001
    if (priv->lockState)
3002
        virBufferAsprintf(buf, "<lockstate>%s</lockstate>\n", priv->lockState);
3003

3004 3005
    if (qemuDomainObjPrivateXMLFormatJob(buf, vm, priv) < 0)
        return -1;
3006

3007
    if (priv->fakeReboot)
3008
        virBufferAddLit(buf, "<fakereboot/>\n");
3009

3010 3011
    if (priv->qemuDevices && *priv->qemuDevices) {
        char **tmp = priv->qemuDevices;
3012 3013
        virBufferAddLit(buf, "<devices>\n");
        virBufferAdjustIndent(buf, 2);
3014
        while (*tmp) {
3015
            virBufferAsprintf(buf, "<device alias='%s'/>\n", *tmp);
3016 3017
            tmp++;
        }
3018 3019
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</devices>\n");
3020 3021
    }

3022
    if (qemuDomainObjPrivateXMLFormatAutomaticPlacement(buf, priv) < 0)
3023
        return -1;
3024

3025 3026 3027 3028 3029
    /* Various per-domain paths */
    virBufferEscapeString(buf, "<libDir path='%s'/>\n", priv->libDir);
    virBufferEscapeString(buf, "<channelTargetDir path='%s'/>\n",
                          priv->channelTargetDir);

3030
    virCPUDefFormatBufFull(buf, priv->origCPU, NULL);
3031

3032
    if (priv->chardevStdioLogd)
3033
        virBufferAddLit(buf, "<chardevStdioLogd/>\n");
3034

3035 3036 3037
    if (priv->rememberOwner)
        virBufferAddLit(buf, "<rememberOwner/>\n");

3038 3039
    qemuDomainObjPrivateXMLFormatAllowReboot(buf, priv->allowReboot);

3040 3041
    qemuDomainObjPrivateXMLFormatPR(buf, priv);

3042 3043 3044
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV))
        virBufferAsprintf(buf, "<nodename index='%llu'/>\n", priv->nodenameindex);

3045 3046 3047
    if (priv->memPrealloc)
        virBufferAddLit(buf, "<memPrealloc/>\n");

3048 3049 3050
    if (qemuDomainObjPrivateXMLFormatBlockjobs(buf, vm) < 0)
        return -1;

3051 3052 3053
    if (qemuDomainObjPrivateXMLFormatSlirp(buf, vm) < 0)
        return -1;

3054 3055
    virBufferAsprintf(buf, "<agentTimeout>%i</agentTimeout>\n", priv->agentTimeout);

3056 3057 3058
    if (qemuDomainObjPrivateXMLFormatBackups(buf, vm) < 0)
        return -1;

3059 3060 3061
    return 0;
}

3062 3063 3064 3065

static int
qemuDomainObjPrivateXMLParseVcpu(xmlNodePtr node,
                                 unsigned int idx,
3066
                                 virDomainDefPtr def)
3067
{
3068
    virDomainVcpuDefPtr vcpu;
3069 3070
    g_autofree char *idstr = NULL;
    g_autofree char *pidstr = NULL;
3071
    unsigned int tmp;
3072

3073 3074
    idstr = virXMLPropString(node, "id");

3075 3076
    if (idstr &&
        (virStrToLong_uip(idstr, NULL, 10, &idx) < 0)) {
3077
        virReportError(VIR_ERR_INTERNAL_ERROR,
3078
                       _("cannot parse vcpu index '%s'"), idstr);
3079
        return -1;
3080 3081 3082 3083
    }
    if (!(vcpu = virDomainDefGetVcpu(def, idx))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("invalid vcpu index '%u'"), idx);
3084
        return -1;
3085 3086
    }

3087
    if (!(pidstr = virXMLPropString(node, "pid")))
3088
        return -1;
3089

3090
    if (virStrToLong_uip(pidstr, NULL, 10, &tmp) < 0)
3091
        return -1;
3092

3093 3094
    QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid = tmp;

3095
    return 0;
3096 3097 3098
}


3099 3100 3101 3102 3103
static int
qemuDomainObjPrivateXMLParseAutomaticPlacement(xmlXPathContextPtr ctxt,
                                               qemuDomainObjPrivatePtr priv,
                                               virQEMUDriverPtr driver)
{
3104
    g_autoptr(virCapsHostNUMA) caps = NULL;
3105 3106
    g_autofree char *nodeset = NULL;
    g_autofree char *cpuset = NULL;
3107 3108
    int nodesetSize = 0;
    size_t i;
3109 3110

    nodeset = virXPathString("string(./numad/@nodeset)", ctxt);
3111
    cpuset = virXPathString("string(./numad/@cpuset)", ctxt);
3112

3113
    if (!nodeset && !cpuset)
3114 3115
        return 0;

3116
    if (!(caps = virQEMUDriverGetHostNUMACaps(driver)))
3117
        return -1;
3118

3119 3120 3121
    /* 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 */
3122
    for (i = 0; i < caps->cells->len; i++) {
3123
        virCapsHostNUMACellPtr cell =
3124
            g_ptr_array_index(caps->cells, i);
3125 3126
        nodesetSize = MAX(nodesetSize, cell->num + 1);
    }
3127

3128
    if (nodeset &&
3129
        virBitmapParse(nodeset, &priv->autoNodeset, nodesetSize) < 0)
3130
        return -1;
3131

3132 3133
    if (cpuset) {
        if (virBitmapParse(cpuset, &priv->autoCpuset, VIR_DOMAIN_CPUMASK_LEN) < 0)
3134
            return -1;
3135 3136 3137
    } else {
        /* autoNodeset is present in this case, since otherwise we wouldn't
         * reach this code */
3138
        if (!(priv->autoCpuset = virCapabilitiesHostNUMAGetCpus(caps,
3139
                                                                priv->autoNodeset)))
3140
            return -1;
3141
    }
3142

3143
    return 0;
3144 3145 3146
}


3147 3148 3149 3150 3151 3152 3153
static virStorageSourcePtr
qemuDomainObjPrivateXMLParseBlockjobChain(xmlNodePtr node,
                                          xmlXPathContextPtr ctxt,
                                          virDomainXMLOptionPtr xmlopt)

{
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3154 3155 3156
    g_autofree char *format = NULL;
    g_autofree char *type = NULL;
    g_autofree char *index = NULL;
3157
    g_autoptr(virStorageSource) src = NULL;
3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180
    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
3181
    return g_steal_pointer(&src);
3182 3183 3184
}


3185 3186 3187 3188 3189 3190
static void
qemuDomainObjPrivateXMLParseBlockjobNodename(qemuBlockJobDataPtr job,
                                             const char *xpath,
                                             virStorageSourcePtr *src,
                                             xmlXPathContextPtr ctxt)
{
3191
    g_autofree char *nodename = NULL;
3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216

    *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;
}


3217 3218 3219 3220
static int
qemuDomainObjPrivateXMLParseBlockjobDataCommit(qemuBlockJobDataPtr job,
                                               xmlXPathContextPtr ctxt)
{
3221 3222 3223
    g_autofree xmlNodePtr *nodes = NULL;
    ssize_t nnodes;

3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249
    if (job->type == QEMU_BLOCKJOB_TYPE_COMMIT) {
        qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                     "string(./topparent/@node)",
                                                     &job->data.commit.topparent,
                                                     ctxt);

        if (!job->data.commit.topparent)
            return -1;
    }

    qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                 "string(./top/@node)",
                                                 &job->data.commit.top,
                                                 ctxt);
    qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                 "string(./base/@node)",
                                                 &job->data.commit.base,
                                                 ctxt);

    if (virXPathNode("./deleteCommittedImages", ctxt))
        job->data.commit.deleteCommittedImages = true;

    if (!job->data.commit.top ||
        !job->data.commit.base)
        return -1;

3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264
    if ((nnodes = virXPathNodeSet("./disabledBaseBitmaps/bitmap", ctxt, &nodes)) > 0) {
        size_t i;

        job->data.commit.disabledBitmapsBase = g_new0(char *, nnodes + 1);

        for (i = 0; i < nnodes; i++) {
            char *tmp;

            if (!(tmp = virXMLPropString(nodes[i], "name")))
                return -1;

            job->data.commit.disabledBitmapsBase[i] = g_steal_pointer(&tmp);
        }
    }

3265 3266 3267 3268
    return 0;
}


3269 3270
static void
qemuDomainObjPrivateXMLParseBlockjobDataSpecific(qemuBlockJobDataPtr job,
3271 3272
                                                 xmlXPathContextPtr ctxt,
                                                 virDomainXMLOptionPtr xmlopt)
3273
{
3274 3275
    g_autofree char *createmode = NULL;
    g_autofree char *shallownew = NULL;
3276 3277
    xmlNodePtr tmp;

3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288
    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:
        case QEMU_BLOCKJOB_TYPE_ACTIVE_COMMIT:
3289
            if (qemuDomainObjPrivateXMLParseBlockjobDataCommit(job, ctxt) < 0)
3290
                goto broken;
3291

3292 3293
            break;

3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306
        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;

3307
        case QEMU_BLOCKJOB_TYPE_COPY:
3308 3309 3310 3311 3312 3313 3314 3315
            if ((shallownew =  virXPathString("string(./@shallownew)", ctxt))) {
                if (STRNEQ(shallownew, "yes"))
                    goto broken;

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

P
Peter Krempa 已提交
3316
        case QEMU_BLOCKJOB_TYPE_BACKUP:
3317 3318 3319 3320 3321
            job->data.backup.bitmap =  virXPathString("string(./bitmap/@name)", ctxt);

            if (!(tmp = virXPathNode("./store", ctxt)) ||
                !(job->data.backup.store = qemuDomainObjPrivateXMLParseBlockjobChain(tmp, ctxt, xmlopt)))
                goto broken;
P
Peter Krempa 已提交
3322
            break;
3323 3324

        case QEMU_BLOCKJOB_TYPE_BROKEN:
3325 3326 3327 3328 3329 3330 3331
        case QEMU_BLOCKJOB_TYPE_NONE:
        case QEMU_BLOCKJOB_TYPE_INTERNAL:
        case QEMU_BLOCKJOB_TYPE_LAST:
            break;
    }

    return;
3332 3333 3334 3335

 broken:
    VIR_DEBUG("marking block job '%s' as invalid: malformed job data", job->name);
    job->invalidData = true;
3336 3337 3338
}


3339
static int
3340 3341
qemuDomainObjPrivateXMLParseBlockjobData(virDomainObjPtr vm,
                                         xmlNodePtr node,
3342 3343
                                         xmlXPathContextPtr ctxt,
                                         virDomainXMLOptionPtr xmlopt)
3344 3345 3346
{
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
    virDomainDiskDefPtr disk = NULL;
3347
    g_autoptr(qemuBlockJobData) job = NULL;
3348 3349
    g_autofree char *name = NULL;
    g_autofree char *typestr = NULL;
3350
    g_autofree char *brokentypestr = NULL;
3351
    int type;
3352
    g_autofree char *statestr = NULL;
3353
    int state = QEMU_BLOCKJOB_STATE_FAILED;
3354 3355 3356
    g_autofree char *diskdst = NULL;
    g_autofree char *newstatestr = NULL;
    g_autofree char *mirror = NULL;
3357 3358
    int newstate = -1;
    bool invalidData = false;
3359
    xmlNodePtr tmp;
3360
    unsigned long jobflags = 0;
3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372

    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) {
3373
        type = QEMU_BLOCKJOB_TYPE_BROKEN;
3374 3375 3376 3377 3378 3379
        invalidData = true;
    }

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

3380 3381 3382 3383
    if ((brokentypestr = virXPathString("string(./@brokentype)", ctxt)) &&
        (job->brokentype = qemuBlockjobTypeFromString(brokentypestr)) < 0)
        job->brokentype = QEMU_BLOCKJOB_TYPE_NONE;

3384 3385 3386 3387 3388 3389 3390 3391 3392
    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)) &&
3393
        !(disk = virDomainDiskByTarget(vm->def, diskdst)))
3394 3395
        invalidData = true;

3396 3397 3398 3399
    if ((mirror = virXPathString("string(./disk/@mirror)", ctxt)) &&
        STRNEQ(mirror, "yes"))
        invalidData = true;

3400 3401 3402
    if (virXPathULongHex("string(./@jobflags)", ctxt, &jobflags) != 0)
        job->jobflagsmissing = true;

3403 3404 3405 3406 3407 3408 3409 3410 3411 3412
    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;
    }

3413 3414 3415 3416 3417 3418 3419
    if (mirror) {
        if (disk)
            job->mirrorChain = virObjectRef(disk->mirror);
        else
            invalidData = true;
    }

3420 3421
    job->state = state;
    job->newstate = newstate;
3422
    job->jobflags = jobflags;
3423 3424
    job->errmsg = virXPathString("string(./errmsg)", ctxt);
    job->invalidData = invalidData;
3425 3426
    job->disk = disk;

3427
    qemuDomainObjPrivateXMLParseBlockjobDataSpecific(job, ctxt, xmlopt);
3428

3429
    if (qemuBlockJobRegister(job, vm, disk, false) < 0)
3430 3431 3432 3433 3434 3435 3436 3437 3438
        return -1;

    return 0;
}


static int
qemuDomainObjPrivateXMLParseBlockjobs(virDomainObjPtr vm,
                                      qemuDomainObjPrivatePtr priv,
3439 3440
                                      xmlXPathContextPtr ctxt)
{
3441
    g_autofree xmlNodePtr *nodes = NULL;
3442
    ssize_t nnodes = 0;
3443
    g_autofree char *active = NULL;
3444
    int tmp;
3445
    size_t i;
3446 3447 3448 3449 3450

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

3451 3452 3453 3454 3455
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if ((nnodes = virXPathNodeSet("./blockjobs/blockjob", ctxt, &nodes)) < 0)
            return -1;

        for (i = 0; i < nnodes; i++) {
3456 3457
            if (qemuDomainObjPrivateXMLParseBlockjobData(vm, nodes[i], ctxt,
                                                         priv->driver->xmlopt) < 0)
3458 3459 3460 3461
                return -1;
        }
    }

3462 3463 3464 3465
    return 0;
}


3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493
static int
qemuDomainObjPrivateXMLParseBackups(qemuDomainObjPrivatePtr priv,
                                    xmlXPathContextPtr ctxt)
{
    g_autofree xmlNodePtr *nodes = NULL;
    ssize_t nnodes = 0;

    if ((nnodes = virXPathNodeSet("./backups/domainbackup", ctxt, &nodes)) < 0)
        return -1;

    if (nnodes > 1) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("only one backup job is supported"));
        return -1;
    }

    if (nnodes == 0)
        return 0;

    if (!(priv->backup = virDomainBackupDefParseNode(ctxt->doc, nodes[0],
                                                     priv->driver->xmlopt,
                                                     VIR_DOMAIN_BACKUP_PARSE_INTERNAL)))
        return -1;

    return 0;
}


3494
int
3495 3496 3497 3498
qemuDomainObjPrivateXMLParseAllowReboot(xmlXPathContextPtr ctxt,
                                        virTristateBool *allowReboot)
{
    int val;
3499
    g_autofree char *valStr = NULL;
3500 3501 3502 3503 3504

    if ((valStr = virXPathString("string(./allowReboot/@value)", ctxt))) {
        if ((val = virTristateBoolTypeFromString(valStr)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("invalid allowReboot value '%s'"), valStr);
3505
            return -1;
3506 3507 3508 3509
        }
        *allowReboot = val;
    }

3510
    return 0;
3511 3512 3513
}


3514 3515 3516 3517 3518 3519 3520 3521
static void
qemuDomainObjPrivateXMLParsePR(xmlXPathContextPtr ctxt,
                               bool *prDaemonRunning)
{
    *prDaemonRunning = virXPathBoolean("boolean(./prDaemon)", ctxt) > 0;
}


3522 3523 3524
static int
qemuDomainObjPrivateXMLParseJobNBDSource(xmlNodePtr node,
                                         xmlXPathContextPtr ctxt,
3525 3526
                                         virDomainDiskDefPtr disk,
                                         virDomainXMLOptionPtr xmlopt)
3527
{
3528
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3529
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
3530 3531
    g_autofree char *format = NULL;
    g_autofree char *type = NULL;
3532
    g_autoptr(virStorageSource) migrSource = NULL;
3533
    xmlNodePtr sourceNode;
3534 3535 3536

    ctxt->node = node;

3537 3538
    if (!(ctxt->node = virXPathNode("./migrationSource", ctxt)))
        return 0;
3539 3540 3541 3542

    if (!(type = virXMLPropString(ctxt->node, "type"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing storage source type"));
3543
        return -1;
3544 3545 3546 3547
    }

    if (!(format = virXMLPropString(ctxt->node, "format"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
3548
                       _("missing storage source format"));
3549
        return -1;
3550 3551
    }

3552
    if (!(migrSource = virDomainStorageSourceParseBase(type, format, NULL)))
3553
        return -1;
3554

3555 3556 3557 3558 3559
    /* newer libvirt uses the <source> subelement instead of formatting the
     * source directly into <migrationSource> */
    if ((sourceNode = virXPathNode("./source", ctxt)))
        ctxt->node = sourceNode;

3560
    if (virDomainStorageSourceParse(ctxt->node, ctxt, migrSource,
3561
                                    VIR_DOMAIN_DEF_PARSE_STATUS, xmlopt) < 0)
3562
        return -1;
3563

3564
    diskPriv->migrSource = g_steal_pointer(&migrSource);
3565
    return 0;
3566 3567 3568
}


3569 3570 3571 3572 3573
static int
qemuDomainObjPrivateXMLParseJobNBD(virDomainObjPtr vm,
                                   qemuDomainObjPrivatePtr priv,
                                   xmlXPathContextPtr ctxt)
{
3574
    g_autofree xmlNodePtr *nodes = NULL;
3575 3576 3577 3578
    size_t i;
    int n;

    if ((n = virXPathNodeSet("./disk[@migrating='yes']", ctxt, &nodes)) < 0)
3579
        return -1;
3580 3581 3582 3583 3584 3585 3586 3587 3588

    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;
3589
            g_autofree char *dst = NULL;
3590

3591
            if ((dst = virXMLPropString(nodes[i], "dev")) &&
3592
                (disk = virDomainDiskByTarget(vm->def, dst))) {
3593
                QEMU_DOMAIN_DISK_PRIVATE(disk)->migrating = true;
3594 3595

                if (qemuDomainObjPrivateXMLParseJobNBDSource(nodes[i], ctxt,
3596 3597
                                                             disk,
                                                             priv->driver->xmlopt) < 0)
3598
                    return -1;
3599
            }
3600 3601 3602
        }
    }

3603
    return 0;
3604 3605 3606
}


3607 3608 3609 3610 3611
static int
qemuDomainObjPrivateXMLParseJob(virDomainObjPtr vm,
                                qemuDomainObjPrivatePtr priv,
                                xmlXPathContextPtr ctxt)
{
3612
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3613
    g_autofree char *tmp = NULL;
3614

3615 3616
    if (!(ctxt->node = virXPathNode("./job[1]", ctxt)))
        return 0;
3617 3618

    if ((tmp = virXPathString("string(@type)", ctxt))) {
3619 3620 3621 3622 3623
        int type;

        if ((type = qemuDomainJobTypeFromString(tmp)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown job type %s"), tmp);
3624
            return -1;
3625 3626 3627 3628 3629
        }
        VIR_FREE(tmp);
        priv->job.active = type;
    }

3630
    if ((tmp = virXPathString("string(@async)", ctxt))) {
3631 3632 3633 3634 3635
        int async;

        if ((async = qemuDomainAsyncJobTypeFromString(tmp)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown async job type %s"), tmp);
3636
            return -1;
3637 3638 3639 3640
        }
        VIR_FREE(tmp);
        priv->job.asyncJob = async;

3641
        if ((tmp = virXPathString("string(@phase)", ctxt))) {
3642 3643 3644 3645
            priv->job.phase = qemuDomainAsyncJobPhaseFromString(async, tmp);
            if (priv->job.phase < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Unknown job phase %s"), tmp);
3646
                return -1;
3647 3648 3649 3650 3651
            }
            VIR_FREE(tmp);
        }
    }

3652 3653
    if (virXPathULongHex("string(@flags)", ctxt, &priv->job.apiFlags) == -2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Invalid job flags"));
3654
        return -1;
3655 3656
    }

3657
    if (qemuDomainObjPrivateXMLParseJobNBD(vm, priv, ctxt) < 0)
3658
        return -1;
3659

3660
    if (qemuMigrationParamsParse(ctxt, &priv->job.migParams) < 0)
3661
        return -1;
3662

3663
    return 0;
3664 3665 3666
}


3667 3668 3669 3670 3671 3672
static int
qemuDomainObjPrivateXMLParseSlirpFeatures(xmlNodePtr featuresNode,
                                          xmlXPathContextPtr ctxt,
                                          qemuSlirpPtr slirp)
{
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3673
    g_autofree xmlNodePtr *nodes = NULL;
3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685
    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++) {
3686
        g_autofree char *str = virXMLPropString(nodes[i], "name");
3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705
        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;
}


3706
static int
3707
qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt,
3708
                             virDomainObjPtr vm,
3709
                             virDomainDefParserConfigPtr config)
3710
{
3711
    qemuDomainObjPrivatePtr priv = vm->privateData;
3712
    virQEMUDriverPtr driver = config->priv;
3713
    char *monitorpath;
3714
    g_autofree char *tmp = NULL;
3715 3716
    int n;
    size_t i;
3717
    g_autofree xmlNodePtr *nodes = NULL;
3718
    xmlNodePtr node = NULL;
3719
    g_autoptr(virQEMUCaps) qemuCaps = NULL;
3720

3721
    if (!(priv->monConfig = virDomainChrSourceDefNew(NULL)))
3722 3723 3724 3725
        goto error;

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
3726 3727
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("no monitor path"));
3728 3729 3730 3731 3732
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
3733
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
3734
    else
3735
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
3736 3737
    VIR_FREE(tmp);

3738
    switch (priv->monConfig->type) {
3739
    case VIR_DOMAIN_CHR_TYPE_PTY:
3740
        priv->monConfig->data.file.path = monitorpath;
3741 3742
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
3743
        priv->monConfig->data.nix.path = monitorpath;
3744 3745 3746
        break;
    default:
        VIR_FREE(monitorpath);
3747 3748 3749
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported monitor type '%s'"),
                       virDomainChrTypeToString(priv->monConfig->type));
3750 3751 3752
        goto error;
    }

3753 3754 3755 3756 3757 3758
    if (virXPathInt("string(./agentTimeout)", ctxt, &priv->agentTimeout) == -2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse agent timeout"));
        goto error;
    }

3759 3760 3761 3762
    if ((node = virXPathNode("./namespaces", ctxt))) {
        xmlNodePtr next;

        for (next = node->children; next; next = next->next) {
3763
            int ns = qemuDomainNamespaceTypeFromString((const char *)next->name);
3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782

            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;
    }

3783 3784
    priv->rememberOwner = virXPathBoolean("count(./rememberOwner) > 0", ctxt);

3785
    if ((n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes)) < 0)
3786 3787
        goto error;

3788
    for (i = 0; i < n; i++) {
3789
        if (qemuDomainObjPrivateXMLParseVcpu(nodes[i], i, vm->def) < 0)
3790
            goto error;
3791
    }
3792
    VIR_FREE(nodes);
3793

3794
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
3795 3796
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse qemu capabilities flags"));
3797 3798 3799
        goto error;
    }
    if (n > 0) {
3800
        if (!(qemuCaps = virQEMUCapsNew()))
3801 3802
            goto error;

3803
        for (i = 0; i < n; i++) {
3804
            g_autofree char *str = virXMLPropString(nodes[i], "name");
3805
            if (str) {
3806
                int flag = virQEMUCapsTypeFromString(str);
3807
                if (flag < 0) {
3808 3809
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Unknown qemu capabilities flag %s"), str);
3810 3811
                    goto error;
                }
3812
                virQEMUCapsSet(qemuCaps, flag);
3813 3814 3815
            }
        }

3816
        priv->qemuCaps = g_steal_pointer(&qemuCaps);
3817 3818 3819
    }
    VIR_FREE(nodes);

3820
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
3821

3822
    if (qemuDomainObjPrivateXMLParseJob(vm, priv, ctxt) < 0)
3823 3824
        goto error;

3825 3826
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847
    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);

3848 3849 3850 3851 3852 3853
    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++) {
3854 3855
        g_autofree char *alias = virXMLPropString(nodes[i], "alias");
        g_autofree char *pid = virXMLPropString(nodes[i], "pid");
J
Ján Tomko 已提交
3856
        g_autoptr(qemuSlirp) slirp = qemuSlirpNew();
3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872
        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;

3873
        QEMU_DOMAIN_NETWORK_PRIVATE(dev.data.net)->slirp = g_steal_pointer(&slirp);
3874 3875 3876
    }
    VIR_FREE(nodes);

3877
    if (qemuDomainObjPrivateXMLParseAutomaticPlacement(ctxt, priv, driver) < 0)
3878 3879
        goto error;

3880 3881 3882 3883 3884 3885
    if ((tmp = virXPathString("string(./libDir/@path)", ctxt)))
        priv->libDir = tmp;
    if ((tmp = virXPathString("string(./channelTargetDir/@path)", ctxt)))
        priv->channelTargetDir = tmp;
    tmp = NULL;

3886
    qemuDomainSetPrivatePathsOld(driver, vm);
3887

3888 3889 3890
    if (virCPUDefParseXML(ctxt, "./cpu", VIR_CPU_TYPE_GUEST, &priv->origCPU) < 0)
        goto error;

3891 3892 3893
    priv->chardevStdioLogd = virXPathBoolean("boolean(./chardevStdioLogd)",
                                             ctxt) == 1;

3894 3895
    qemuDomainObjPrivateXMLParseAllowReboot(ctxt, &priv->allowReboot);

3896 3897
    qemuDomainObjPrivateXMLParsePR(ctxt, &priv->prDaemonRunning);

3898
    if (qemuDomainObjPrivateXMLParseBlockjobs(vm, priv, ctxt) < 0)
3899 3900
        goto error;

3901 3902 3903
    if (qemuDomainObjPrivateXMLParseBackups(priv, ctxt) < 0)
        goto error;

3904 3905 3906 3907 3908 3909 3910 3911
    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;
    }

3912 3913
    priv->memPrealloc = virXPathBoolean("boolean(./memPrealloc)", ctxt) == 1;

3914 3915
    return 0;

3916
 error:
3917 3918
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
3919
    virObjectUnref(priv->monConfig);
3920
    priv->monConfig = NULL;
3921
    virStringListFree(priv->qemuDevices);
3922
    priv->qemuDevices = NULL;
3923 3924 3925 3926
    return -1;
}


3927 3928 3929 3930 3931 3932 3933 3934 3935
static void *
qemuDomainObjPrivateXMLGetParseOpaque(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return priv->qemuCaps;
}


3936 3937 3938
virDomainXMLPrivateDataCallbacks virQEMUDriverPrivateDataCallbacks = {
    .alloc = qemuDomainObjPrivateAlloc,
    .free = qemuDomainObjPrivateFree,
3939
    .diskNew = qemuDomainDiskPrivateNew,
3940 3941
    .diskParse = qemuDomainDiskPrivateParse,
    .diskFormat = qemuDomainDiskPrivateFormat,
3942
    .vcpuNew = qemuDomainVcpuPrivateNew,
3943
    .chrSourceNew = qemuDomainChrSourcePrivateNew,
J
Ján Tomko 已提交
3944
    .vsockNew = qemuDomainVsockPrivateNew,
3945
    .graphicsNew = qemuDomainGraphicsPrivateNew,
3946
    .networkNew = qemuDomainNetworkPrivateNew,
3947
    .videoNew = qemuDomainVideoPrivateNew,
3948
    .fsNew = qemuDomainFSPrivateNew,
3949 3950
    .parse = qemuDomainObjPrivateXMLParse,
    .format = qemuDomainObjPrivateXMLFormat,
3951
    .getParseOpaque = qemuDomainObjPrivateXMLGetParseOpaque,
3952 3953
    .storageParse = qemuStorageSourcePrivateDataParse,
    .storageFormat = qemuStorageSourcePrivateDataFormat,
3954 3955 3956
};


3957 3958 3959 3960 3961 3962
static void
qemuDomainXmlNsDefFree(qemuDomainXmlNsDefPtr def)
{
    if (!def)
        return;

3963 3964 3965
    virStringListFreeCount(def->args, def->num_args);
    virStringListFreeCount(def->env_name, def->num_env);
    virStringListFreeCount(def->env_value, def->num_env);
3966 3967
    virStringListFreeCount(def->capsadd, def->ncapsadd);
    virStringListFreeCount(def->capsdel, def->ncapsdel);
3968

3969 3970 3971 3972
    VIR_FREE(def);
}


3973 3974 3975
static void
qemuDomainDefNamespaceFree(void *nsdata)
{
3976
    qemuDomainXmlNsDefPtr cmd = nsdata;
3977

3978
    qemuDomainXmlNsDefFree(cmd);
3979 3980
}

3981 3982 3983 3984 3985

static int
qemuDomainDefNamespaceParseCommandlineArgs(qemuDomainXmlNsDefPtr nsdef,
                                           xmlXPathContextPtr ctxt)
{
3986
    g_autofree xmlNodePtr *nodes = NULL;
3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010
    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;
}


4011 4012 4013
static int
qemuDomainDefNamespaceParseCommandlineEnvNameValidate(const char *envname)
{
4014
    if (!g_ascii_isalpha(envname[0]) && envname[0] != '_') {
4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033
        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)
{
4034
    g_autofree xmlNodePtr *nodes = NULL;
4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066
    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;
}


4067 4068 4069 4070
static int
qemuDomainDefNamespaceParseCaps(qemuDomainXmlNsDefPtr nsdef,
                                xmlXPathContextPtr ctxt)
{
4071
    g_autofree xmlNodePtr *nodesadd = NULL;
4072
    ssize_t nnodesadd;
4073
    g_autofree xmlNodePtr *nodesdel = NULL;
4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110
    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;
}


4111
static int
4112
qemuDomainDefNamespaceParse(xmlXPathContextPtr ctxt,
4113 4114
                            void **data)
{
4115 4116
    qemuDomainXmlNsDefPtr nsdata = NULL;
    int ret = -1;
4117

4118
    if (VIR_ALLOC(nsdata) < 0)
4119 4120
        return -1;

4121
    if (qemuDomainDefNamespaceParseCommandlineArgs(nsdata, ctxt) < 0 ||
4122 4123
        qemuDomainDefNamespaceParseCommandlineEnv(nsdata, ctxt) < 0 ||
        qemuDomainDefNamespaceParseCaps(nsdata, ctxt) < 0)
4124
        goto cleanup;
4125

4126 4127
    if (nsdata->num_args > 0 || nsdata->num_env > 0 ||
        nsdata->ncapsadd > 0 || nsdata->ncapsdel > 0)
4128
        *data = g_steal_pointer(&nsdata);
4129

4130
    ret = 0;
4131

4132 4133 4134
 cleanup:
    qemuDomainDefNamespaceFree(nsdata);
    return ret;
4135 4136
}

4137 4138 4139 4140

static void
qemuDomainDefNamespaceFormatXMLCommandline(virBufferPtr buf,
                                           qemuDomainXmlNsDefPtr cmd)
4141
{
4142
    size_t i;
4143 4144

    if (!cmd->num_args && !cmd->num_env)
4145
        return;
4146

4147 4148 4149
    virBufferAddLit(buf, "<qemu:commandline>\n");
    virBufferAdjustIndent(buf, 2);

4150
    for (i = 0; i < cmd->num_args; i++)
4151
        virBufferEscapeString(buf, "<qemu:arg value='%s'/>\n",
4152 4153
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
4154
        virBufferAsprintf(buf, "<qemu:env name='%s'", cmd->env_name[i]);
4155 4156 4157 4158 4159
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }

4160 4161
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:commandline>\n");
4162 4163 4164
}


4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187
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");
}


4188 4189 4190 4191 4192 4193 4194
static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainXmlNsDefPtr cmd = nsdata;

    qemuDomainDefNamespaceFormatXMLCommandline(buf, cmd);
4195
    qemuDomainDefNamespaceFormatXMLCaps(buf, cmd);
4196

4197 4198 4199 4200
    return 0;
}


4201
virXMLNamespace virQEMUDriverDomainXMLNamespace = {
4202 4203 4204
    .parse = qemuDomainDefNamespaceParse,
    .free = qemuDomainDefNamespaceFree,
    .format = qemuDomainDefNamespaceFormatXML,
4205
    .prefix = "qemu",
4206
    .uri = "http://libvirt.org/schemas/domain/qemu/1.0",
4207
};
4208

4209

P
Pavel Hrdina 已提交
4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228
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;
}


4229
static int
4230 4231
qemuDomainDefAddDefaultDevices(virDomainDefPtr def,
                               virQEMUCapsPtr qemuCaps)
4232
{
4233
    bool addDefaultUSB = true;
4234
    int usbModel = -1; /* "default for machinetype" */
4235
    int pciRoot;       /* index within def->controllers */
4236
    bool addImplicitSATA = false;
4237
    bool addPCIRoot = false;
L
Laine Stump 已提交
4238
    bool addPCIeRoot = false;
4239
    bool addDefaultMemballoon = true;
4240 4241
    bool addDefaultUSBKBD = false;
    bool addDefaultUSBMouse = false;
4242
    bool addPanicDevice = false;
4243

P
Pavel Hrdina 已提交
4244 4245
    /* add implicit input devices */
    if (qemuDomainDefAddImplicitInputDevice(def) < 0)
4246
        return -1;
P
Pavel Hrdina 已提交
4247

4248 4249 4250 4251
    /* 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
4252
        if (STREQ(def->os.machine, "isapc")) {
4253
            addDefaultUSB = false;
4254
            break;
4255
        }
4256
        if (qemuDomainIsQ35(def)) {
4257 4258
            addPCIeRoot = true;
            addImplicitSATA = true;
4259

4260 4261 4262
            /* 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.
4263
             */
4264 4265 4266
            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))
4267 4268
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
            else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_USB_EHCI1))
4269 4270 4271
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_ICH9_EHCI1;
            else
                addDefaultUSB = false;
4272
            break;
L
Laine Stump 已提交
4273
        }
4274
        if (qemuDomainIsI440FX(def))
4275
            addPCIRoot = true;
4276 4277
        break;

4278 4279 4280 4281 4282 4283 4284
    case VIR_ARCH_ARMV6L:
        addDefaultUSB = false;
        addDefaultMemballoon = false;
        if (STREQ(def->os.machine, "versatilepb"))
            addPCIRoot = true;
        break;

4285
    case VIR_ARCH_ARMV7L:
4286
    case VIR_ARCH_AARCH64:
4287 4288
        addDefaultUSB = false;
        addDefaultMemballoon = false;
4289
        if (qemuDomainIsARMVirt(def))
4290 4291
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
        break;
4292

4293
    case VIR_ARCH_PPC64:
4294
    case VIR_ARCH_PPC64LE:
4295 4296 4297
        addPCIRoot = true;
        addDefaultUSBKBD = true;
        addDefaultUSBMouse = true;
4298 4299 4300
        /* For pSeries guests, the firmware provides the same
         * functionality as the pvpanic device, so automatically
         * add the definition if not already present */
4301
        if (qemuDomainIsPSeries(def))
4302
            addPanicDevice = true;
4303 4304
        break;

4305 4306 4307 4308 4309 4310 4311
    case VIR_ARCH_ALPHA:
    case VIR_ARCH_PPC:
    case VIR_ARCH_PPCEMB:
    case VIR_ARCH_SH4:
    case VIR_ARCH_SH4EB:
        addPCIRoot = true;
        break;
4312

4313 4314 4315
    case VIR_ARCH_RISCV32:
    case VIR_ARCH_RISCV64:
        addDefaultUSB = false;
4316 4317
        if (qemuDomainIsRISCVVirt(def))
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
4318 4319
        break;

4320 4321 4322
    case VIR_ARCH_S390:
    case VIR_ARCH_S390X:
        addDefaultUSB = false;
4323
        addPanicDevice = true;
4324
        addPCIRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_ZPCI);
4325
        break;
4326 4327 4328 4329 4330 4331

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

4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351
    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:
4352 4353 4354 4355
    default:
        break;
    }

4356
    if (addDefaultUSB &&
4357 4358
        virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_USB, 0) < 0 &&
        virDomainDefAddUSBController(def, 0, usbModel) < 0)
4359
        return -1;
4360

4361 4362 4363
    if (addImplicitSATA &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_SATA, 0, -1) < 0)
4364
        return -1;
4365

4366 4367
    pciRoot = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

4368 4369 4370
    /* NB: any machine that sets addPCIRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
     */
4371 4372 4373 4374 4375 4376 4377 4378
    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));
4379
                return -1;
4380 4381 4382
            }
        } else if (!virDomainDefAddController(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
                                              VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)) {
4383
            return -1;
4384 4385
        }
    }
4386

4387 4388 4389
    /* 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
4390 4391 4392
     *
     * NB: any machine that sets addPCIeRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
4393 4394
     */
    if (addPCIeRoot) {
4395 4396 4397 4398 4399 4400 4401
        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));
4402
                return -1;
4403 4404 4405
            }
        } else if (!virDomainDefAddController(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
                                             VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT)) {
4406
            return -1;
4407
        }
4408
    }
4409

4410
    if (addDefaultMemballoon && !def->memballoon) {
4411 4412
        virDomainMemballoonDefPtr memballoon;
        if (VIR_ALLOC(memballoon) < 0)
4413
            return -1;
4414 4415 4416 4417 4418

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

4419 4420 4421 4422
    if (STRPREFIX(def->os.machine, "s390-virtio") &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_S390) && def->memballoon)
        def->memballoon->model = VIR_DOMAIN_MEMBALLOON_MODEL_NONE;

4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442
    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;
    }

4443 4444 4445 4446 4447
    if (addDefaultUSBKBD &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_KBD,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
4448
        return -1;
4449 4450 4451 4452 4453 4454

    if (addDefaultUSBMouse &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
4455
        return -1;
4456

D
Dmitry Andreev 已提交
4457 4458 4459 4460
    if (addPanicDevice) {
        size_t j;
        for (j = 0; j < def->npanics; j++) {
            if (def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT ||
4461 4462 4463 4464
                (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
4465 4466
                break;
        }
4467

D
Dmitry Andreev 已提交
4468 4469 4470 4471 4472 4473
        if (j == def->npanics) {
            virDomainPanicDefPtr panic;
            if (VIR_ALLOC(panic) < 0 ||
                VIR_APPEND_ELEMENT_COPY(def->panics,
                                        def->npanics, panic) < 0) {
                VIR_FREE(panic);
4474
                return -1;
D
Dmitry Andreev 已提交
4475 4476
            }
        }
4477 4478
    }

4479
    return 0;
4480 4481 4482
}


A
Andrea Bolognani 已提交
4483 4484 4485
/**
 * qemuDomainDefEnableDefaultFeatures:
 * @def: domain definition
4486
 * @qemuCaps: QEMU capabilities
A
Andrea Bolognani 已提交
4487 4488 4489 4490 4491
 *
 * Make sure that features that should be enabled by default are actually
 * enabled and configure default values related to those features.
 */
static void
4492 4493
qemuDomainDefEnableDefaultFeatures(virDomainDefPtr def,
                                   virQEMUCapsPtr qemuCaps)
A
Andrea Bolognani 已提交
4494
{
4495
    /* The virt machine type always uses GIC: if the relevant information
4496 4497
     * was not included in the domain XML, we need to choose a suitable
     * GIC version ourselves */
4498
    if ((def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ABSENT &&
4499
         qemuDomainIsARMVirt(def)) ||
4500 4501 4502
        (def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ON &&
         def->gic_version == VIR_GIC_VERSION_NONE)) {
        virGICVersion version;
4503 4504 4505 4506 4507

        VIR_DEBUG("Looking for usable GIC version in domain capabilities");
        for (version = VIR_GIC_VERSION_LAST - 1;
             version > VIR_GIC_VERSION_NONE;
             version--) {
4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522

            /* 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;
            }

4523 4524 4525 4526 4527 4528 4529 4530
            if (virQEMUCapsSupportsGICVersion(qemuCaps,
                                              def->virtType,
                                              version)) {
                VIR_DEBUG("Using GIC version %s",
                          virGICVersionTypeToString(version));
                def->gic_version = version;
                break;
            }
4531 4532
        }

4533 4534 4535 4536 4537 4538 4539
        /* 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;
        }

4540 4541 4542
        /* 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;
4543
    }
A
Andrea Bolognani 已提交
4544 4545 4546
}


4547 4548 4549 4550 4551
static int
qemuCanonicalizeMachine(virDomainDefPtr def, virQEMUCapsPtr qemuCaps)
{
    const char *canon;

4552 4553
    if (!(canon = virQEMUCapsGetCanonicalMachine(qemuCaps, def->virtType,
                                                 def->os.machine)))
4554 4555 4556 4557
        return 0;

    if (STRNEQ(canon, def->os.machine)) {
        char *tmp;
4558
        tmp = g_strdup(canon);
4559 4560 4561 4562 4563 4564 4565 4566
        VIR_FREE(def->os.machine);
        def->os.machine = tmp;
    }

    return 0;
}


4567
static int
4568 4569 4570 4571 4572
qemuDomainRecheckInternalPaths(virDomainDefPtr def,
                               virQEMUDriverConfigPtr cfg,
                               unsigned int flags)
{
    size_t i = 0;
4573
    size_t j = 0;
4574 4575 4576 4577

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

4578 4579 4580 4581 4582 4583
        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
4584
             * the auto-generated socket based on config option from qemu.conf
4585 4586 4587 4588
             * 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 &&
4589
                !glisten->autoGenerated &&
4590 4591 4592 4593 4594 4595 4596
                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;
                }
4597
            }
4598 4599
        }
    }
4600 4601

    return 0;
4602 4603 4604
}


4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674
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;
}


4675 4676
static int
qemuDomainDefSetDefaultCPU(virDomainDefPtr def,
4677
                           virArch hostarch,
4678 4679 4680 4681 4682 4683 4684 4685 4686
                           virQEMUCapsPtr qemuCaps)
{
    const char *model;

    if (def->cpu &&
        (def->cpu->mode != VIR_CPU_MODE_CUSTOM ||
         def->cpu->model))
        return 0;

4687 4688 4689
    if (!virCPUArchIsSupported(def->os.arch))
        return 0;

4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711
    /* Default CPU model info from QEMU is usable for TCG only except for
     * x86, s390, and ppc64. */
    if (!ARCH_IS_X86(def->os.arch) &&
        !ARCH_IS_S390(def->os.arch) &&
        !ARCH_IS_PPC64(def->os.arch) &&
        def->virtType != VIR_DOMAIN_VIRT_QEMU)
        return 0;

    model = virQEMUCapsGetMachineDefaultCPU(qemuCaps, def->os.machine, def->virtType);
    if (!model) {
        VIR_DEBUG("Unknown default CPU model for domain '%s'", def->name);
        return 0;
    }

    if (STREQ(model, "host") && def->virtType != VIR_DOMAIN_VIRT_KVM) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("QEMU reports invalid default CPU model \"host\" "
                         "for non-kvm domain virt type"));
        return -1;
    }

    if (!def->cpu)
4712
        def->cpu = virCPUDefNew();
4713 4714 4715 4716

    def->cpu->type = VIR_CPU_TYPE_GUEST;

    if (STREQ(model, "host")) {
4717
        if (ARCH_IS_S390(def->os.arch) &&
4718
            virQEMUCapsIsCPUModeSupported(qemuCaps, hostarch, def->virtType,
4719 4720
                                          VIR_CPU_MODE_HOST_MODEL,
                                          def->os.machine)) {
4721 4722 4723 4724 4725 4726 4727
            def->cpu->mode = VIR_CPU_MODE_HOST_MODEL;
        } else {
            def->cpu->mode = VIR_CPU_MODE_HOST_PASSTHROUGH;
        }

        VIR_DEBUG("Setting default CPU mode for domain '%s' to %s",
                  def->name, virCPUModeTypeToString(def->cpu->mode));
4728
    } else {
4729 4730 4731 4732 4733 4734
        /* We need to turn off all CPU checks when the domain is started
         * because the default CPU (e.g., qemu64) may not be runnable on any
         * host. QEMU will just disable the unavailable features and we will
         * update the CPU definition accordingly and set check to FULL when
         * starting the domain. */
        def->cpu->check = VIR_CPU_CHECK_NONE;
4735 4736 4737 4738
        def->cpu->mode = VIR_CPU_MODE_CUSTOM;
        def->cpu->match = VIR_CPU_MATCH_EXACT;
        def->cpu->fallback = VIR_CPU_FALLBACK_FORBID;
        def->cpu->model = g_strdup(model);
4739 4740 4741

        VIR_DEBUG("Setting default CPU model for domain '%s' to %s",
                  def->name, model);
4742 4743 4744 4745 4746 4747
    }

    return 0;
}


4748 4749 4750
static int
qemuDomainDefCPUPostParse(virDomainDefPtr def)
{
A
Andrea Bolognani 已提交
4751 4752 4753 4754
    virCPUFeatureDefPtr sveFeature = NULL;
    bool sveVectorLengthsProvided = false;
    size_t i;

4755 4756 4757
    if (!def->cpu)
        return 0;

4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811
    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
4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844
    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++;
        }
    }

4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877
    /* 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;
}


4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909
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;
}


4910 4911
static int
qemuDomainDefPostParseBasic(virDomainDefPtr def,
J
Ján Tomko 已提交
4912
                            void *opaque G_GNUC_UNUSED)
4913
{
4914 4915
    virQEMUDriverPtr driver = opaque;

4916
    /* check for emulator and create a default one if needed */
4917 4918
    if (!def->emulator) {
        if (!(def->emulator = virQEMUCapsGetDefaultEmulator(
4919 4920 4921 4922
                  driver->hostarch, def->os.arch))) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("No emulator found for arch '%s'"),
                           virArchToString(def->os.arch));
4923
            return 1;
4924
        }
4925
    }
4926 4927 4928 4929 4930

    return 0;
}


4931 4932
static int
qemuDomainDefPostParse(virDomainDefPtr def,
4933
                       unsigned int parseFlags,
4934
                       void *opaque,
4935
                       void *parseOpaque)
4936 4937
{
    virQEMUDriverPtr driver = opaque;
4938
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
4939
    virQEMUCapsPtr qemuCaps = parseOpaque;
4940

4941 4942 4943 4944 4945
    /* 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.
     */
    if (!qemuCaps)
4946 4947
        return 1;

4948 4949 4950
    if (def->os.bootloader || def->os.bootloaderArgs) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bootloader is not supported by QEMU"));
4951
        return -1;
4952 4953
    }

4954
    if (!def->os.machine) {
4955 4956 4957
        const char *machine = virQEMUCapsGetPreferredMachine(qemuCaps,
                                                             def->virtType);
        def->os.machine = g_strdup(machine);
4958 4959
    }

4960
    qemuDomainNVRAMPathGenerate(cfg, def);
4961

4962
    if (qemuDomainDefAddDefaultDevices(def, qemuCaps) < 0)
4963
        return -1;
4964

4965
    if (qemuCanonicalizeMachine(def, qemuCaps) < 0)
4966
        return -1;
4967

4968
    if (qemuDomainDefSetDefaultCPU(def, driver->hostarch, qemuCaps) < 0)
4969 4970
        return -1;

4971
    qemuDomainDefEnableDefaultFeatures(def, qemuCaps);
A
Andrea Bolognani 已提交
4972

4973
    if (qemuDomainRecheckInternalPaths(def, cfg, parseFlags) < 0)
4974
        return -1;
4975

4976
    if (qemuSecurityVerify(driver->securityManager, def) < 0)
4977
        return -1;
4978

4979
    if (qemuDomainDefVcpusPostParse(def) < 0)
4980
        return -1;
4981

4982
    if (qemuDomainDefCPUPostParse(def) < 0)
4983
        return -1;
4984

4985
    if (qemuDomainDefTsegPostParse(def, qemuCaps) < 0)
4986
        return -1;
4987

4988
    return 0;
4989 4990
}

4991

4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021
/**
 * 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;
}


5022 5023 5024
#define QEMU_MAX_VCPUS_WITHOUT_EIM 255


5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 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 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 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132
static int
qemuDomainDefValidatePSeriesFeature(const virDomainDef *def,
                                    virQEMUCapsPtr qemuCaps,
                                    int feature)
{
    const char *str;

    if (def->features[feature] != VIR_TRISTATE_SWITCH_ABSENT &&
        !qemuDomainIsPSeries(def)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("The '%s' feature is not supported for "
                         "architecture '%s' or machine type '%s'"),
                       virDomainFeatureTypeToString(feature),
                       virArchToString(def->os.arch),
                       def->os.machine);
        return -1;
    }

    if (def->features[feature] == VIR_TRISTATE_SWITCH_ABSENT)
        return 0;

    switch (feature) {
    case VIR_DOMAIN_FEATURE_HPT:
        if (def->features[feature] != VIR_TRISTATE_SWITCH_ON)
            break;

        if (def->hpt_resizing != VIR_DOMAIN_HPT_RESIZING_NONE) {
            if (!virQEMUCapsGet(qemuCaps,
                                QEMU_CAPS_MACHINE_PSERIES_RESIZE_HPT)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("HTP resizing is not supported by this "
                                "QEMU binary"));
                return -1;
            }

            str = virDomainHPTResizingTypeToString(def->hpt_resizing);
            if (!str) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("Invalid setting for HPT resizing"));
                return -1;
            }
        }

        if (def->hpt_maxpagesize > 0 &&
            !virQEMUCapsGet(qemuCaps,
                            QEMU_CAPS_MACHINE_PSERIES_CAP_HPT_MAX_PAGE_SIZE)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Configuring the page size for HPT guests "
                             "is not supported by this QEMU binary"));
            return -1;
        }
        break;

    case VIR_DOMAIN_FEATURE_HTM:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_PSERIES_CAP_HTM)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("HTM configuration is not supported by this "
                             "QEMU binary"));
            return -1;
        }

        str = virTristateSwitchTypeToString(def->features[feature]);
        if (!str) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Invalid setting for HTM state"));
            return -1;
        }

        break;

    case VIR_DOMAIN_FEATURE_NESTED_HV:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_PSERIES_CAP_NESTED_HV)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Nested HV configuration is not supported by "
                             "this QEMU binary"));
            return -1;
        }

        str = virTristateSwitchTypeToString(def->features[feature]);
        if (!str) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Invalid setting for nested HV state"));
            return -1;
        }

        break;

    case VIR_DOMAIN_FEATURE_CCF_ASSIST:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_PSERIES_CAP_CCF_ASSIST)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("ccf-assist configuration is not supported by "
                           "this QEMU binary"));
            return -1;
        }

        str = virTristateSwitchTypeToString(def->features[feature]);
        if (!str) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Invalid setting for ccf-assist state"));
            return -1;
        }

        break;
    }

    return 0;
}

5133
static int
5134 5135
qemuDomainDefValidateFeatures(const virDomainDef *def,
                              virQEMUCapsPtr qemuCaps)
5136
{
5137
    size_t i;
5138

5139 5140 5141 5142 5143
    for (i = 0; i < VIR_DOMAIN_FEATURE_LAST; i++) {
        const char *featureName = virDomainFeatureTypeToString(i);

        switch ((virDomainFeature) i) {
        case VIR_DOMAIN_FEATURE_IOAPIC:
5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176
            if (def->features[i] != VIR_DOMAIN_IOAPIC_NONE) {
                if (!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;
                }

                if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_KERNEL_IRQCHIP)) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                                   _("I/O APIC tuning is not supported by "
                                     "this QEMU binary"));
                    return -1;
                }

                switch ((virDomainIOAPIC) def->features[i]) {
                case VIR_DOMAIN_IOAPIC_QEMU:
                    if (!virQEMUCapsGet(qemuCaps,
                                        QEMU_CAPS_MACHINE_KERNEL_IRQCHIP_SPLIT)) {
                        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                                       _("split I/O APIC is not supported by this "
                                         "QEMU binary"));
                        return -1;
                    }
                    break;
                case VIR_DOMAIN_IOAPIC_KVM:
                case VIR_DOMAIN_IOAPIC_NONE:
                case VIR_DOMAIN_IOAPIC_LAST:
                    break;
                }
5177 5178 5179 5180
            }
            break;

        case VIR_DOMAIN_FEATURE_HPT:
5181
        case VIR_DOMAIN_FEATURE_HTM:
5182
        case VIR_DOMAIN_FEATURE_NESTED_HV:
5183
        case VIR_DOMAIN_FEATURE_CCF_ASSIST:
5184
            if (qemuDomainDefValidatePSeriesFeature(def, qemuCaps, i) < 0)
5185 5186 5187
                return -1;
            break;

5188 5189
        case VIR_DOMAIN_FEATURE_GIC:
            if (def->features[i] == VIR_TRISTATE_SWITCH_ON &&
5190
                !qemuDomainIsARMVirt(def)) {
5191 5192 5193 5194 5195 5196 5197 5198 5199 5200
                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;

5201 5202
        case VIR_DOMAIN_FEATURE_SMM:
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
5203
                !virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT)) {
5204 5205 5206 5207 5208 5209
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("smm is not available with this QEMU binary"));
                return -1;
            }
            break;

5210 5211 5212 5213 5214 5215 5216 5217 5218 5219
        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;

5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230
        case VIR_DOMAIN_FEATURE_VMPORT:
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
                !virQEMUCapsSupportsVmport(qemuCaps, def)) {

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("vmport is not available "
                                 "with this QEMU binary"));
                return -1;
            }
            break;

5231 5232 5233 5234 5235 5236 5237 5238 5239 5240
        case VIR_DOMAIN_FEATURE_VMCOREINFO:
            if (def->features[i] == VIR_TRISTATE_SWITCH_ON &&
                !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMCOREINFO)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                              _("vmcoreinfo is not available "
                                "with this QEMU binary"));
                return -1;
            }
            break;

5241 5242 5243 5244 5245 5246 5247 5248 5249 5250
        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:
5251
        case VIR_DOMAIN_FEATURE_MSRS:
5252 5253 5254
        case VIR_DOMAIN_FEATURE_LAST:
            break;
        }
5255 5256 5257 5258 5259 5260
    }

    return 0;
}


5261
static int
M
Marc-André Lureau 已提交
5262 5263
qemuDomainDefValidateMemory(const virDomainDef *def,
                            virQEMUCapsPtr qemuCaps)
5264 5265
{
    const long system_page_size = virGetSystemPageSizeKB();
5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283
    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;
    }
5284

M
Marc-André Lureau 已提交
5285 5286 5287
    if (mem->source == VIR_DOMAIN_MEMORY_SOURCE_MEMFD &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
5288
                       _("hugepages is not supported with memfd memory source"));
M
Marc-André Lureau 已提交
5289 5290 5291
        return -1;
    }

5292 5293
    /* We can't guarantee any other mem.access
     * if no guest NUMA nodes are defined. */
5294
    if (mem->hugepages[0].size != system_page_size &&
5295
        virDomainNumaGetNodeCount(def->numa) == 0 &&
5296 5297
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_DEFAULT &&
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_PRIVATE) {
5298 5299 5300
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("memory access mode '%s' not supported "
                         "without guest numa node"),
5301
                       virDomainMemoryAccessTypeToString(mem->access));
5302 5303 5304
        return -1;
    }

5305 5306 5307 5308 5309 5310 5311
    if (mem->nosharepages && !virQEMUCapsGet(qemuCaps, QEMU_CAPS_MEM_MERGE)) {
         virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                        _("disable shared memory is not available "
                          "with this QEMU binary"));
        return -1;
    }

5312 5313 5314 5315
    return 0;
}


5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378
static int
qemuDomainDefValidateNuma(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
{
    const long system_page_size = virGetSystemPageSizeKB();
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;
    bool hasMemoryCap = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_MEMORY_RAM) ||
                        virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_MEMORY_FILE) ||
                        virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_MEMORY_MEMFD);

    if (virDomainNumatuneHasPerNodeBinding(def->numa) && !hasMemoryCap) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Per-node memory binding is not supported "
                         "with this QEMU"));
        return -1;
    }

    if (def->mem.nhugepages &&
        def->mem.hugepages[0].size != system_page_size &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_MEMORY_FILE)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("huge pages per NUMA node are not "
                         "supported with this QEMU"));
        return -1;
    }

    for (i = 0; i < ncells; i++) {
        g_autofree char * cpumask = NULL;

        if (!hasMemoryCap &&
            virDomainNumaGetNodeMemoryAccessMode(def->numa, i)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Shared memory mapping is not supported "
                             "with this QEMU"));
            return -1;
        }

        if (!(cpumask = virBitmapFormat(virDomainNumaGetNodeCpumask(def->numa, i))))
            return -1;

        if (strchr(cpumask, ',') &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_NUMA)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("disjoint NUMA cpu ranges are not supported "
                             "with this QEMU"));
            return -1;
        }

    }

    if (virDomainNumaNodesDistancesAreBeingSet(def->numa) &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_NUMA_DIST)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("setting NUMA distances is not "
                         "supported with this qemu"));
        return -1;
    }

    return 0;
}


5379 5380 5381 5382
static int
qemuDomainValidateCpuCount(const virDomainDef *def,
                            virQEMUCapsPtr qemuCaps)
{
5383 5384
    unsigned int maxCpus = virQEMUCapsGetMachineMaxCpus(qemuCaps, def->virtType,
                                                        def->os.machine);
5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402

    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;
}


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
static int
qemuDomainDeviceDefValidateNVRAM(virDomainNVRAMDefPtr nvram,
                                 const virDomainDef *def,
                                 virQEMUCapsPtr qemuCaps)
{
    if (!nvram)
        return 0;

    if (qemuDomainIsPSeries(def)) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_NVRAM)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("nvram device is not supported by "
                             "this QEMU binary"));
            return -1;
        }
    } else {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("nvram device is only supported for PPC64"));
        return -1;
    }

    if (!(nvram->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO &&
          nvram->info.addr.spaprvio.has_reg)) {

        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("nvram address type must be spaprvio"));
        return -1;
    }

    return 0;
}


5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456
static int
qemuDomainDeviceDefValidateHub(virDomainHubDefPtr hub,
                               virQEMUCapsPtr qemuCaps)
{
    if (hub->type != VIR_DOMAIN_HUB_TYPE_USB) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("hub type %s not supported"),
                       virDomainHubTypeToString(hub->type));
        return -1;
    }

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_USB_HUB)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("usb-hub not supported by QEMU binary"));
        return -1;
    }

    return 0;
}


5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546
static int
qemuDomainDefValidateClockTimers(const virDomainDef *def,
                                 virQEMUCapsPtr qemuCaps)
{
    size_t i;

    for (i = 0; i < def->clock.ntimers; i++) {
        virDomainTimerDefPtr timer = def->clock.timers[i];

        switch ((virDomainTimerNameType)timer->name) {
        case VIR_DOMAIN_TIMER_NAME_PLATFORM:
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("unsupported timer type (name) '%s'"),
                           virDomainTimerNameTypeToString(timer->name));
            return -1;

        case VIR_DOMAIN_TIMER_NAME_TSC:
        case VIR_DOMAIN_TIMER_NAME_KVMCLOCK:
        case VIR_DOMAIN_TIMER_NAME_HYPERVCLOCK:
        case VIR_DOMAIN_TIMER_NAME_LAST:
            break;

        case VIR_DOMAIN_TIMER_NAME_RTC:
            switch (timer->track) {
            case -1: /* unspecified - use hypervisor default */
            case VIR_DOMAIN_TIMER_TRACK_GUEST:
            case VIR_DOMAIN_TIMER_TRACK_WALL:
                break;
            case VIR_DOMAIN_TIMER_TRACK_BOOT:
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("unsupported rtc timer track '%s'"),
                               virDomainTimerTrackTypeToString(timer->track));
                return -1;
            }

            switch (timer->tickpolicy) {
            case -1:
            case VIR_DOMAIN_TIMER_TICKPOLICY_DELAY:
                /* This is the default - missed ticks delivered when
                   next scheduled, at normal rate */
                break;
            case VIR_DOMAIN_TIMER_TICKPOLICY_CATCHUP:
                /* deliver ticks at a faster rate until caught up */
                break;
            case VIR_DOMAIN_TIMER_TICKPOLICY_MERGE:
            case VIR_DOMAIN_TIMER_TICKPOLICY_DISCARD:
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("unsupported rtc timer tickpolicy '%s'"),
                               virDomainTimerTickpolicyTypeToString(
                                   timer->tickpolicy));
                return -1;
            }
            break;

        case VIR_DOMAIN_TIMER_NAME_PIT:
            switch (timer->tickpolicy) {
            case -1:
            case VIR_DOMAIN_TIMER_TICKPOLICY_DELAY:
            case VIR_DOMAIN_TIMER_TICKPOLICY_DISCARD:
                break;
            case VIR_DOMAIN_TIMER_TICKPOLICY_CATCHUP:
                if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM_PIT_TICK_POLICY)) {
                    /* can't catchup if we don't have kvm-pit */
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                                   _("unsupported pit tickpolicy '%s'"),
                                   virDomainTimerTickpolicyTypeToString(
                                       timer->tickpolicy));
                    return -1;
                }
                break;
            case VIR_DOMAIN_TIMER_TICKPOLICY_MERGE:
                /* no way to support this mode for pit in qemu */
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("unsupported pit tickpolicy '%s'"),
                               virDomainTimerTickpolicyTypeToString(
                                   timer->tickpolicy));
                return -1;
            }
            break;

        case VIR_DOMAIN_TIMER_NAME_HPET:
            /* no hpet timer available. The only possible action
              is to raise an error if present="yes" */
            if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_NO_HPET) &&
                timer->present == 1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               "%s", _("hpet timer is not supported"));
                return -1;
            }
            break;
5547 5548

        case VIR_DOMAIN_TIMER_NAME_ARMVTIMER:
5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586
            if (def->virtType != VIR_DOMAIN_VIRT_KVM ||
                !qemuDomainIsARMVirt(def)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Configuring the '%s' timer is not supported "
                                 "for virtType=%s arch=%s machine=%s guests"),
                               virDomainTimerNameTypeToString(timer->name),
                               virDomainVirtTypeToString(def->virtType),
                               virArchToString(def->os.arch),
                               def->os.machine);
                return -1;
            }
            if (timer->present == 0) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("The '%s' timer can't be disabled"),
                               virDomainTimerNameTypeToString(timer->name));
                return -1;
            }
            if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CPU_KVM_NO_ADJVTIME)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Configuring the '%s' timer is not supported "
                                 "with this QEMU binary"),
                               virDomainTimerNameTypeToString(timer->name));
                return -1;
            }

            switch (timer->tickpolicy) {
            case -1:
            case VIR_DOMAIN_TIMER_TICKPOLICY_DELAY:
            case VIR_DOMAIN_TIMER_TICKPOLICY_DISCARD:
                break;
            case VIR_DOMAIN_TIMER_TICKPOLICY_CATCHUP:
            case VIR_DOMAIN_TIMER_TICKPOLICY_MERGE:
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("The '%s' timer does not support tickpolicy '%s'"),
                               virDomainTimerNameTypeToString(timer->name),
                               virDomainTimerTickpolicyTypeToString(timer->tickpolicy));
                return -1;
            }
5587
            break;
5588 5589 5590 5591 5592 5593 5594
        }
    }

    return 0;
}


5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626
static int
qemuDomainDefValidatePM(const virDomainDef *def,
                        virQEMUCapsPtr qemuCaps)
{
    bool q35Dom = qemuDomainIsQ35(def);

    if (def->pm.s3) {
        bool q35ICH9_S3 = q35Dom &&
                          virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_DISABLE_S3);

        if (!q35ICH9_S3 && !virQEMUCapsGet(qemuCaps, QEMU_CAPS_PIIX_DISABLE_S3)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           "%s", _("setting ACPI S3 not supported"));
            return -1;
        }
    }

    if (def->pm.s4) {
        bool q35ICH9_S4 = q35Dom &&
                          virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_DISABLE_S4);

        if (!q35ICH9_S4 && !virQEMUCapsGet(qemuCaps, QEMU_CAPS_PIIX_DISABLE_S4)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           "%s", _("setting ACPI S4 not supported"));
            return -1;
        }
    }

    return 0;
}


5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651
static int
qemuDomainDefValidateBoot(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
{
    if (def->os.bios.rt_set) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_REBOOT_TIMEOUT)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("reboot timeout is not supported "
                             "by this QEMU binary"));
            return -1;
        }
    }

    if (def->os.bm_timeout_set) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SPLASH_TIMEOUT)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("splash timeout is not supported "
                             "by this QEMU binary"));
            return -1;
        }
    }

    return 0;
}

5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693
static int
qemuDomainDefValidateConsole(const virDomainDef *def,
                             virQEMUCapsPtr qemuCaps)
{
    size_t i;

    /* Explicit console devices */
    for (i = 0; i < def->nconsoles; i++) {
        virDomainChrDefPtr console = def->consoles[i];

        switch (console->targetType) {
        case VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_SCLP:
            if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_SCLPCONSOLE)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("sclpconsole is not supported in this QEMU binary"));
                return -1;
            }
            break;

        case VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_SCLPLM:
            if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_SCLPLMCONSOLE)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("sclplmconsole is not supported in this QEMU binary"));
                return -1;
            }
            break;

        case VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_VIRTIO:
        case VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_SERIAL:
            break;

        default:
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("unsupported console target type %s"),
                           NULLSTR(virDomainChrConsoleTargetTypeToString(console->targetType)));
            return -1;
        }
    }

    return 0;
}

5694

5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710
static int
qemuSoundCodecTypeToCaps(int type)
{
    switch (type) {
    case VIR_DOMAIN_SOUND_CODEC_TYPE_DUPLEX:
        return QEMU_CAPS_HDA_DUPLEX;
    case VIR_DOMAIN_SOUND_CODEC_TYPE_MICRO:
        return QEMU_CAPS_HDA_MICRO;
    case VIR_DOMAIN_SOUND_CODEC_TYPE_OUTPUT:
        return QEMU_CAPS_HDA_OUTPUT;
    default:
        return -1;
    }
}


5711 5712 5713 5714
static int
qemuDomainDeviceDefValidateSound(virDomainSoundDefPtr sound,
                                 virQEMUCapsPtr qemuCaps)
{
5715 5716
    size_t i;

5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747
    switch ((virDomainSoundModel) sound->model) {
    case VIR_DOMAIN_SOUND_MODEL_USB:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_USB_AUDIO)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("usb-audio controller is not supported "
                             "by this QEMU binary"));
            return -1;
        }
        break;
    case VIR_DOMAIN_SOUND_MODEL_ICH9:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_ICH9_INTEL_HDA)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("The ich9-intel-hda audio controller "
                             "is not supported in this QEMU binary"));
            return -1;
        }
        break;

    case VIR_DOMAIN_SOUND_MODEL_ES1370:
    case VIR_DOMAIN_SOUND_MODEL_AC97:
    case VIR_DOMAIN_SOUND_MODEL_ICH6:
    case VIR_DOMAIN_SOUND_MODEL_SB16:
    case VIR_DOMAIN_SOUND_MODEL_PCSPK:
        break;
    case VIR_DOMAIN_SOUND_MODEL_LAST:
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("sound card model '%s' is not supported by qemu"),
                       virDomainSoundModelTypeToString(sound->model));
        return -1;
    }

5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765
    if (sound->model == VIR_DOMAIN_SOUND_MODEL_ICH6 ||
        sound->model == VIR_DOMAIN_SOUND_MODEL_ICH9) {
        for (i = 0; i < sound->ncodecs; i++) {
            const char *stype;
            int type, flags;

            type = sound->codecs[i]->type;
            stype = qemuSoundCodecTypeToString(type);
            flags = qemuSoundCodecTypeToCaps(type);

            if (flags == -1 || !virQEMUCapsGet(qemuCaps, flags)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("%s not supported in this QEMU binary"), stype);
                return -1;
            }
        }
    }

5766 5767 5768
    return 0;
}

5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781
static int
qemuDomainDeviceDefValidateMemory(virDomainMemoryDefPtr mem,
                                  virQEMUCapsPtr qemuCaps)
{
    if (mem->model == VIR_DOMAIN_MEMORY_MODEL_NVDIMM &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_NVDIMM)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("nvdimm isn't supported by this QEMU binary"));
        return -1;
    }

    return 0;
}
5782

5783 5784
static int
qemuDomainDefValidate(const virDomainDef *def,
5785
                      void *opaque)
5786
{
5787
    virQEMUDriverPtr driver = opaque;
5788 5789
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
    g_autoptr(virQEMUCaps) qemuCaps = NULL;
5790
    size_t i;
5791

5792
    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
5793
                                            def->emulator)))
5794
        return -1;
5795

5796 5797 5798 5799
    if (def->os.type != VIR_DOMAIN_OSTYPE_HVM) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Emulator '%s' does not support os type '%s'"),
                       def->emulator, virDomainOSTypeToString(def->os.type));
5800
        return -1;
5801 5802 5803 5804 5805 5806
    }

    if (!virQEMUCapsIsArchSupported(qemuCaps, def->os.arch)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Emulator '%s' does not support arch '%s'"),
                       def->emulator, virArchToString(def->os.arch));
5807
        return -1;
5808 5809 5810 5811 5812 5813
    }

    if (!virQEMUCapsIsVirtTypeSupported(qemuCaps, def->virtType)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Emulator '%s' does not support virt type '%s'"),
                       def->emulator, virDomainVirtTypeToString(def->virtType));
5814
        return -1;
5815 5816
    }

5817 5818 5819 5820 5821
    if (qemuCaps &&
        !virQEMUCapsIsMachineSupported(qemuCaps, def->virtType, def->os.machine)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Emulator '%s' does not support machine type '%s'"),
                       def->emulator, def->os.machine);
5822
        return -1;
5823 5824
    }

5825 5826 5827
    if (def->mem.min_guarantee) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Parameter 'min_guarantee' not supported by QEMU."));
5828
        return -1;
5829 5830 5831
    }

    /* On x86, UEFI requires ACPI */
5832
    if ((def->os.firmware == VIR_DOMAIN_OS_DEF_FIRMWARE_EFI ||
5833
         virDomainDefHasOldStyleUEFI(def)) &&
5834 5835 5836 5837
        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"));
5838
        return -1;
5839 5840 5841 5842 5843
    }

    /* On aarch64, ACPI requires UEFI */
    if (def->features[VIR_DOMAIN_FEATURE_ACPI] == VIR_TRISTATE_SWITCH_ON &&
        def->os.arch == VIR_ARCH_AARCH64 &&
5844
        (def->os.firmware != VIR_DOMAIN_OS_DEF_FIRMWARE_EFI &&
5845
         !virDomainDefHasOldStyleUEFI(def))) {
5846 5847
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("ACPI requires UEFI on this architecture"));
5848
        return -1;
5849 5850
    }

M
Michal Privoznik 已提交
5851 5852 5853 5854 5855
    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. */

5856
        if (!qemuDomainIsQ35(def)) {
M
Michal Privoznik 已提交
5857 5858
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot is supported with q35 machine types only"));
5859
            return -1;
M
Michal Privoznik 已提交
5860 5861 5862 5863 5864 5865 5866 5867
        }

        /* 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"));
5868
            return -1;
M
Michal Privoznik 已提交
5869 5870
        }

5871 5872 5873
        /* 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
5874 5875
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot requires SMM feature enabled"));
5876
            return -1;
M
Michal Privoznik 已提交
5877 5878 5879
        }
    }

5880 5881 5882 5883
    if (def->genidRequested &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMGENID)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("this QEMU does not support the 'genid' capability"));
5884
        return -1;
5885 5886
    }

5887 5888 5889 5890 5891
    /* Serial graphics adapter */
    if (def->os.bios.useserial == VIR_TRISTATE_BOOL_YES) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SGA)) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("qemu does not support SGA"));
5892
            return -1;
5893 5894 5895 5896
        }
        if (!def->nserials) {
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("need at least one serial port to use SGA"));
5897
            return -1;
5898 5899 5900
        }
    }

5901
    if (qemuDomainDefValidateClockTimers(def, qemuCaps) < 0)
5902
        return -1;
5903

5904
    if (qemuDomainDefValidatePM(def, qemuCaps) < 0)
5905
        return -1;
5906

5907
    if (qemuDomainDefValidateBoot(def, qemuCaps) < 0)
5908
        return -1;
5909

5910 5911 5912 5913 5914
    /* 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;
5915
        unsigned int granularity;
5916
        unsigned int numacpus;
5917 5918 5919 5920

        /* 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 */
5921 5922 5923 5924
        if (virDomainDefGetVcpusTopology(def, &topologycpus) == 0) {
            if (topologycpus != virDomainDefGetVcpusMax(def)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("CPU topology doesn't match maximum vcpu count"));
5925
                return -1;
5926 5927 5928 5929 5930 5931 5932 5933
            }

            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");
            }
5934
        }
5935 5936 5937 5938 5939 5940 5941 5942

        /* 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);
5943
            return -1;
5944
        }
5945 5946
    }

5947
    if (qemuDomainValidateCpuCount(def, qemuCaps) < 0)
5948
        return -1;
5949

5950 5951 5952 5953 5954 5955 5956
    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);
5957
            return -1;
5958 5959 5960 5961 5962 5963
        }
        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);
5964
            return -1;
5965 5966 5967
        }
    }

B
Bing Niu 已提交
5968
    if (def->nresctrls &&
5969 5970 5971
        def->virtType != VIR_DOMAIN_VIRT_KVM) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("cachetune is only supported for KVM domains"));
5972
        return -1;
5973 5974
    }

5975
    if (qemuDomainDefValidateFeatures(def, qemuCaps) < 0)
5976
        return -1;
5977

M
Marc-André Lureau 已提交
5978
    if (qemuDomainDefValidateMemory(def, qemuCaps) < 0)
5979
        return -1;
5980

5981
    if (qemuDomainDefValidateNuma(def, qemuCaps) < 0)
5982
        return -1;
5983 5984

    if (qemuDomainDefValidateConsole(def, qemuCaps) < 0)
5985
        return -1;
5986

5987 5988 5989 5990 5991 5992 5993
    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"));
5994
                return -1;
5995 5996 5997 5998
            }
        }
    }

5999
    return 0;
6000 6001 6002
}


6003 6004 6005 6006 6007 6008 6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026
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;
}


6027 6028 6029 6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040 6041
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
6042 6043
qemuDomainChrSourceDefValidate(const virDomainChrSourceDef *def,
                               virQEMUCapsPtr qemuCaps)
6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055
{
    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;

6056 6057 6058 6059 6060 6061 6062 6063 6064
    case VIR_DOMAIN_CHR_TYPE_FILE:
        if (def->data.file.append != VIR_TRISTATE_SWITCH_ABSENT &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_CHARDEV_FILE_APPEND)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("append not supported in this QEMU binary"));
            return -1;
        }
        break;

6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078
    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_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;
    }

6079 6080 6081 6082 6083 6084 6085 6086
    if (def->logfile) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CHARDEV_LOGFILE)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("logfile not supported in this QEMU binary"));
            return -1;
        }
    }

6087 6088 6089 6090
    return 0;
}


6091 6092 6093
static int
qemuDomainChrSerialTargetTypeToAddressType(int targetType)
{
6094
    switch ((virDomainChrSerialTargetType)targetType) {
6095 6096 6097 6098 6099 6100
    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;
6101 6102
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
        return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO;
6103
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
6104
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
6105 6106 6107 6108 6109 6110 6111 6112 6113
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        break;
    }

    return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE;
}


6114 6115 6116 6117 6118 6119 6120 6121 6122 6123
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;
6124 6125
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
6126
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
6127
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
6128
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
6129 6130 6131
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
6132 6133 6134 6135 6136 6137 6138 6139 6140
    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;
}


6141
static int
6142
qemuDomainChrTargetDefValidate(const virDomainChrDef *chr)
6143
{
6144 6145
    int expected;

6146
    switch ((virDomainChrDeviceType)chr->deviceType) {
6147 6148 6149
    case VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL:

        /* Validate target type */
6150
        switch ((virDomainChrSerialTargetType)chr->targetType) {
6151
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA:
6152 6153
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI:
6154
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
6155

6156
            expected = qemuDomainChrSerialTargetTypeToAddressType(chr->targetType);
6157 6158

            if (chr->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
6159 6160 6161 6162 6163
                chr->info.type != expected) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Target type '%s' requires address type '%s'"),
                               virDomainChrSerialTargetTypeToString(chr->targetType),
                               virDomainDeviceAddressTypeToString(expected));
6164 6165 6166 6167
                return -1;
            }
            break;

6168
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
6169
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
6170 6171 6172 6173 6174 6175 6176 6177 6178
            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;

6179 6180 6181 6182
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            break;
        }
6183 6184 6185 6186 6187 6188

        /* 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:
6189
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
6190
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
6191 6192
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
6193
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209

            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;
        }
6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221 6222 6223
        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;
}


6224
static int
6225
qemuDomainChrDefValidate(const virDomainChrDef *dev,
6226 6227
                         const virDomainDef *def,
                         virQEMUCapsPtr qemuCaps)
6228
{
6229
    if (qemuDomainChrSourceDefValidate(dev->source, qemuCaps) < 0)
6230 6231
        return -1;

6232
    if (qemuDomainChrTargetDefValidate(dev) < 0)
6233 6234
        return -1;

6235
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL &&
6236
        (ARCH_IS_S390(def->os.arch) || qemuDomainIsPSeries(def))) {
6237 6238 6239 6240 6241
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("parallel ports are not supported"));
            return -1;
    }

6242 6243 6244
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL) {
        bool isCompatible = true;

6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255
        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;
            }
        }

6256 6257 6258 6259 6260 6261
        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;
        }

6262 6263 6264 6265 6266 6267 6268
        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;
        }

6269 6270 6271 6272 6273 6274 6275 6276 6277 6278 6279
        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;
        }
    }

6280 6281 6282 6283 6284
    return 0;
}


static int
6285 6286
qemuDomainSmartcardDefValidate(const virDomainSmartcardDef *def,
                               virQEMUCapsPtr qemuCaps)
6287
{
6288 6289 6290 6291 6292 6293 6294 6295 6296 6297 6298 6299 6300 6301 6302 6303 6304 6305 6306 6307 6308 6309 6310 6311 6312 6313 6314 6315 6316 6317 6318 6319 6320
    switch (def->type) {
    case VIR_DOMAIN_SMARTCARD_TYPE_HOST:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCID_EMULATED)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("this QEMU binary lacks smartcard host "
                             "mode support"));
            return -1;
        }
        break;

    case VIR_DOMAIN_SMARTCARD_TYPE_HOST_CERTIFICATES:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCID_EMULATED)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("this QEMU binary lacks smartcard host "
                             "mode support"));
            return -1;
        }
        break;

    case VIR_DOMAIN_SMARTCARD_TYPE_PASSTHROUGH:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCID_PASSTHRU)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("this QEMU binary lacks smartcard "
                             "passthrough mode support"));
            return -1;
        }
        break;

    default:
        virReportEnumRangeError(virDomainSmartcardType, def->type);
        return -1;
    }

6321
    if (def->type == VIR_DOMAIN_SMARTCARD_TYPE_PASSTHROUGH &&
6322
        qemuDomainChrSourceDefValidate(def->data.passthru, qemuCaps) < 0)
6323 6324 6325 6326 6327 6328 6329
        return -1;

    return 0;
}


static int
6330
qemuDomainRNGDefValidate(const virDomainRNGDef *def,
J
Ján Tomko 已提交
6331
                         virQEMUCapsPtr qemuCaps G_GNUC_UNUSED)
6332 6333
{
    if (def->backend == VIR_DOMAIN_RNG_BACKEND_EGD &&
6334
        qemuDomainChrSourceDefValidate(def->source.chardev, qemuCaps) < 0)
6335 6336 6337 6338 6339 6340 6341
        return -1;

    return 0;
}


static int
6342 6343
qemuDomainRedirdevDefValidate(const virDomainRedirdevDef *def,
                              virQEMUCapsPtr qemuCaps)
6344
{
6345
    if (qemuDomainChrSourceDefValidate(def->source, qemuCaps) < 0)
6346 6347 6348 6349 6350 6351
        return -1;

    return 0;
}


6352 6353 6354 6355 6356 6357 6358 6359 6360
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
6361
                           _("%s model of watchdog can go only on PCI bus"),
6362 6363 6364 6365 6366 6367 6368 6369 6370
                           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
6371
                           _("%s model of watchdog can go only on ISA bus"),
6372 6373 6374 6375 6376 6377 6378 6379
                           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
6380
                           _("%s model of watchdog is virtual and cannot go on any bus."),
6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397 6398 6399
                           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;
}


6400 6401 6402 6403 6404 6405 6406 6407 6408 6409 6410 6411 6412 6413
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'.)
     */
6414
    char macstr[VIR_MAC_STRING_BUFLEN];
6415 6416
    virDomainNetType actualType = virDomainNetGetActualType(net);

6417 6418
    virMacAddrFormat(&net->mac, macstr);

6419 6420 6421 6422 6423 6424
    /* hypervisor-agnostic validation */
    if (virDomainActualNetDefValidate(net) < 0)
        return -1;

    /* QEMU-specific validation */

6425 6426 6427 6428 6429 6430 6431 6432 6433
    /* 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,
6434 6435
                           _("interface %s - multiqueue is not supported for network interfaces of type %s"),
                           macstr, virDomainNetTypeToString(actualType));
6436 6437 6438 6439 6440 6441
            return -1;
        }

        if (net->driver.virtio.queues > 1 &&
            actualType == VIR_DOMAIN_NET_TYPE_VHOSTUSER &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_VHOSTUSER_MULTIQUEUE)) {
6442 6443 6444
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("interface %s - multiqueue is not supported for network interfaces of type vhost-user with this QEMU binary"),
                           macstr);
6445 6446 6447 6448 6449 6450 6451 6452 6453 6454
            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) {
6455
        const virNetDevVPortProfile *vport = virDomainNetGetActualVirtPortProfile(net);
6456

6457 6458 6459 6460
        if (!(actualType == VIR_DOMAIN_NET_TYPE_NETWORK ||
              actualType == VIR_DOMAIN_NET_TYPE_BRIDGE ||
              actualType == VIR_DOMAIN_NET_TYPE_ETHERNET)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
6461 6462
                           _("interface %s - filterref is not supported for network interfaces of type %s"),
                           macstr, virDomainNetTypeToString(actualType));
6463 6464 6465 6466 6467
            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,
6468 6469
                           _("interface %s - filterref is not supported for network interfaces with virtualport type %s"),
                           macstr, virNetDevVPortTypeToString(vport->virtPortType));
6470 6471 6472 6473 6474 6475 6476 6477 6478
            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,
6479 6480
                       _("interface %s - custom tap device path is not supported for network interfaces of type %s"),
                       macstr, virDomainNetTypeToString(actualType));
6481 6482 6483
        return -1;
    }

6484 6485 6486 6487 6488 6489 6490
    if (net->teaming.type == VIR_DOMAIN_NET_TEAMING_TYPE_TRANSIENT &&
        actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("interface %s - teaming transient device must be type='hostdev', not '%s'"),
                       macstr, virDomainNetTypeToString(actualType));
        return -1;
    }
6491 6492 6493 6494
    return 0;
}


6495
static int
6496 6497
qemuDomainDeviceDefValidateNetwork(const virDomainNetDef *net,
                                   virQEMUCapsPtr qemuCaps)
6498
{
6499 6500
    bool hasIPv4 = false;
    bool hasIPv6 = false;
6501
    size_t i;
6502

6503 6504 6505 6506 6507 6508 6509 6510 6511 6512
    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];
6513

6514
            if (VIR_SOCKET_ADDR_VALID(&net->guestIP.ips[i]->peer)) {
6515
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
6516 6517
                               _("Invalid attempt to set peer IP for guest"));
                return -1;
6518
            }
6519

6520 6521
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ip->address, AF_INET)) {
                if (hasIPv4) {
6522
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
6523 6524 6525
                                   _("Only one IPv4 address per "
                                     "interface is allowed"));
                    return -1;
6526
                }
6527
                hasIPv4 = true;
6528

6529 6530
                if (ip->prefix > 0 &&
                    (ip->prefix < 4 || ip->prefix > 27)) {
6531
                    virReportError(VIR_ERR_XML_ERROR, "%s",
6532
                                   _("invalid prefix, must be in range of 4-27"));
6533
                    return -1;
6534
                }
6535
            }
6536

6537 6538 6539 6540 6541 6542
            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;
6543
                }
6544
                hasIPv6 = true;
6545

6546 6547 6548 6549 6550
                if (ip->prefix > 120) {
                    virReportError(VIR_ERR_XML_ERROR, "%s",
                                   _("prefix too long"));
                    return -1;
                }
6551
            }
6552
        }
6553 6554 6555 6556 6557 6558 6559
    } 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;
    }
6560

6561
    if (virDomainNetIsVirtioModel(net)) {
6562 6563 6564 6565 6566 6567 6568 6569 6570
        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;
6571
        }
6572
    }
6573

6574 6575 6576 6577 6578 6579 6580 6581
    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;
    }

6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604
    if (net->teaming.type != VIR_DOMAIN_NET_TEAMING_TYPE_NONE &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_NET_FAILOVER)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("virtio-net failover (teaming) is not supported with this QEMU binary"));
        return -1;
    }
    if (net->teaming.type == VIR_DOMAIN_NET_TEAMING_TYPE_PERSISTENT
        && !virDomainNetIsVirtioModel(net)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("virtio-net teaming persistent interface must be <model type='virtio'/>, not '%s'"),
                       virDomainNetGetModelString(net));
        return -1;
    }
    if (net->teaming.type == VIR_DOMAIN_NET_TEAMING_TYPE_TRANSIENT &&
        net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
        net->type != VIR_DOMAIN_NET_TYPE_NETWORK) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("virtio-net teaming transient interface must be type='hostdev', not '%s'"),
                       virDomainNetTypeToString(net->type));
        return -1;
    }

   if (net->coalesce && !qemuDomainNetSupportsCoalesce(net->type)) {
6605 6606 6607 6608 6609 6610 6611 6612 6613 6614
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("coalesce settings on interface type %s are not supported"),
                       virDomainNetTypeToString(net->type));
        return -1;
    }

    return 0;
}


6615
static int
6616
qemuDomainMdevDefVFIOPCIValidate(const virDomainHostdevDef *hostdev,
6617 6618
                                 const virDomainDef *def,
                                 virQEMUCapsPtr qemuCaps)
6619
{
6620 6621
    const virDomainHostdevSubsysMediatedDev *dev;

6622 6623 6624 6625 6626 6627 6628
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VFIO_PCI)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("VFIO PCI device assignment is not "
                         "supported by this version of QEMU"));
        return -1;
    }

6629 6630 6631 6632 6633 6634 6635 6636 6637
    /* 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;
    }

6638
    dev = &hostdev->source.subsys.u.mdev;
6639
    if (dev->display == VIR_TRISTATE_SWITCH_ABSENT)
6640 6641 6642 6643 6644 6645 6646 6647 6648
        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;
    }

6649
    if (dev->model != VIR_MDEV_MODEL_TYPE_VFIO_PCI) {
6650 6651 6652 6653 6654 6655 6656
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("<hostdev> attribute 'display' is only supported"
                         " with model='vfio-pci'"));

        return -1;
    }

6657
    if (dev->display == VIR_TRISTATE_SWITCH_ON) {
6658 6659 6660 6661 6662 6663 6664 6665 6666 6667 6668 6669
        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;
}


6670
static int
6671
qemuDomainMdevDefVFIOAPValidate(const virDomainHostdevDef *hostdev,
6672 6673
                                const virDomainDef *def,
                                virQEMUCapsPtr qemuCaps)
6674 6675 6676 6677
{
    size_t i;
    bool vfioap_found = false;

6678 6679 6680 6681 6682 6683 6684
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VFIO_AP)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("VFIO AP device assignment is not "
                         "supported by this version of QEMU"));
        return -1;
    }

6685 6686 6687 6688 6689 6690 6691 6692 6693
    /* 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;
    }

6694 6695
    /* VFIO-AP is restricted to a single mediated device only */
    for (i = 0; i < def->nhostdevs; i++) {
6696
        virDomainHostdevDefPtr hdev = def->hostdevs[i];
6697

6698 6699
        if (virHostdevIsMdevDevice(hdev) &&
            hdev->source.subsys.u.mdev.model == VIR_MDEV_MODEL_TYPE_VFIO_AP) {
6700 6701 6702 6703 6704 6705 6706 6707 6708 6709 6710 6711 6712 6713
            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;
}


6714
static int
6715
qemuDomainMdevDefValidate(const virDomainHostdevDef *hostdev,
6716 6717 6718
                          const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
{
6719 6720 6721
    const virDomainHostdevSubsysMediatedDev *mdevsrc;

    mdevsrc = &hostdev->source.subsys.u.mdev;
6722 6723
    switch ((virMediatedDeviceModelType) mdevsrc->model) {
    case VIR_MDEV_MODEL_TYPE_VFIO_PCI:
6724
        return qemuDomainMdevDefVFIOPCIValidate(hostdev, def, qemuCaps);
6725
    case VIR_MDEV_MODEL_TYPE_VFIO_AP:
6726
        return qemuDomainMdevDefVFIOAPValidate(hostdev, def, qemuCaps);
6727
    case VIR_MDEV_MODEL_TYPE_VFIO_CCW:
6728 6729 6730 6731 6732 6733
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VFIO_CCW)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("VFIO CCW device assignment is not "
                             "supported by this version of QEMU"));
            return -1;
        }
6734 6735 6736 6737 6738 6739 6740 6741 6742 6743 6744 6745
        break;
    case VIR_MDEV_MODEL_TYPE_LAST:
    default:
        virReportEnumRangeError(virMediatedDeviceModelType,
                                mdevsrc->model);
        return -1;
    }

    return 0;
}


6746 6747
static int
qemuDomainDeviceDefValidateHostdev(const virDomainHostdevDef *hostdev,
6748 6749
                                   const virDomainDef *def,
                                   virQEMUCapsPtr qemuCaps)
6750
{
6751 6752
    int backend;

6753 6754 6755 6756 6757 6758 6759 6760 6761
    /* 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;
    }

6762 6763 6764 6765
    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_SCSI:
6766
            break;
6767 6768 6769 6770 6771 6772 6773 6774 6775 6776 6777 6778 6779 6780

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
            backend = hostdev->source.subsys.u.pci.backend;

            if (backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO) {
                if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VFIO_PCI)) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                                   _("VFIO PCI device assignment is not "
                                     "supported by this version of qemu"));
                    return -1;
                }
            }
            break;

6781
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI_HOST:
6782 6783
            if (hostdev->info->bootIndex) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
6784 6785
                               _("booting from assigned devices is not "
                                 "supported by vhost SCSI devices"));
6786 6787
                return -1;
            }
6788 6789
            break;
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
6790
            return qemuDomainMdevDefValidate(hostdev, def, qemuCaps);
6791 6792 6793 6794 6795 6796 6797 6798
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
        default:
            virReportEnumRangeError(virDomainHostdevSubsysType,
                                    hostdev->source.subsys.type);
            return -1;
        }
    }

6799 6800 6801 6802
    return 0;
}


6803
static int
6804 6805
qemuDomainDeviceDefValidateVideo(const virDomainVideoDef *video,
                                 virQEMUCapsPtr qemuCaps)
6806
{
6807 6808
    /* there's no properties to validate for NONE video devices */
    if (video->type == VIR_DOMAIN_VIDEO_TYPE_NONE)
6809
        return 0;
6810 6811 6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836 6837

    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);
6838 6839
            return -1;
        }
6840 6841 6842 6843 6844 6845 6846
        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;
            }
6847

6848 6849 6850 6851 6852 6853 6854 6855
            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;
            }
        }
    }

6856 6857 6858 6859 6860 6861 6862 6863 6864 6865 6866
    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;
        }
    }

6867 6868 6869 6870 6871 6872
    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)"));
6873 6874 6875 6876
            return -1;
        }
    }

6877 6878 6879 6880 6881 6882 6883
    if (video->backend == VIR_DOMAIN_VIDEO_BACKEND_TYPE_VHOSTUSER) {
        if (video->type == VIR_DOMAIN_VIDEO_TYPE_VIRTIO &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VHOST_USER_GPU)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("this QEMU does not support 'vhost-user' video device"));
            return -1;
        }
6884 6885 6886 6887 6888 6889 6890 6891 6892
    } else if (video->accel) {
        if (video->accel->accel3d == VIR_TRISTATE_SWITCH_ON &&
            (video->type != VIR_DOMAIN_VIDEO_TYPE_VIRTIO ||
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_GPU_VIRGL))) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("%s 3d acceleration is not supported"),
                           virDomainVideoTypeToString(video->type));
            return -1;
        }
6893 6894
    }

6895 6896 6897 6898
    return 0;
}


6899
int
6900 6901
qemuDomainValidateStorageSource(virStorageSourcePtr src,
                                virQEMUCapsPtr qemuCaps)
6902
{
6903 6904
    int actualType = virStorageSourceGetActualType(src);

6905 6906 6907 6908 6909 6910
    if (src->format == VIR_STORAGE_FILE_COW) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                      _("'cow' storage format is not supported"));
        return -1;
    }

6911 6912 6913 6914 6915 6916 6917
    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;
    }

6918 6919 6920 6921 6922 6923 6924
    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;
    }

6925 6926 6927 6928 6929 6930 6931 6932 6933 6934
    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;
    }

6935 6936 6937 6938 6939
    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",
6940
                       _("LUKS encrypted QCOW2 images are not supported by this QEMU"));
6941 6942 6943
        return -1;
    }

6944
    if (src->format == VIR_STORAGE_FILE_FAT &&
6945
        actualType != VIR_STORAGE_TYPE_VOLUME &&
6946 6947 6948 6949 6950 6951 6952 6953 6954 6955 6956 6957 6958 6959 6960 6961 6962 6963 6964 6965 6966 6967
        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;
        }
    }

6968 6969 6970 6971 6972
    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;
6973 6974
    }

6975 6976 6977 6978 6979 6980 6981 6982 6983
    /* 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;
    }

6984 6985 6986 6987 6988 6989 6990 6991 6992 6993 6994 6995
    if (src->sliceStorage) {
        /* In pre-blockdev era we can't configure the slice so we can allow them
         * only for detected backing store entries as they are populated
         * from a place that qemu would be able to read */
        if (!src->detected &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCKDEV)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("storage slice is not supported by this QEMU binary"));
            return -1;
        }
    }

6996 6997 6998 6999 7000 7001 7002 7003 7004 7005 7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020 7021 7022 7023 7024 7025 7026 7027 7028 7029 7030 7031 7032 7033 7034 7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052 7053 7054 7055 7056 7057 7058 7059 7060 7061 7062 7063 7064 7065 7066 7067 7068 7069 7070
    if (src->sslverify != VIR_TRISTATE_BOOL_ABSENT) {
        if (actualType != VIR_STORAGE_TYPE_NETWORK ||
            (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
             src->protocol != VIR_STORAGE_NET_PROTOCOL_FTPS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("ssl verification is supported only with HTTPS/FTPS protocol"));
            return -1;
        }

        if (!src->detected &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCKDEV)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("ssl verification setting is not supported by this QEMU binary"));
            return -1;
        }
    }

    if (src->ncookies > 0) {
        if (actualType != VIR_STORAGE_TYPE_NETWORK ||
            (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
             src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("http cookies are supported only with HTTP(S) protocol"));
            return -1;
        }

        if (!src->detected &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCKDEV)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("http cookies are not supported by this QEMU binary"));
            return -1;
        }

        if (virStorageSourceNetCookiesValidate(src) < 0)
            return -1;
    }

    if (src->readahead > 0) {
        if (actualType != VIR_STORAGE_TYPE_NETWORK ||
            (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
             src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP &&
             src->protocol != VIR_STORAGE_NET_PROTOCOL_FTP &&
             src->protocol != VIR_STORAGE_NET_PROTOCOL_FTPS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("readaehad is supported only with HTTP(S)/FTP(s) protocols"));
            return -1;
        }

        if (!src->detected &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCKDEV)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("readahead setting is not supported with this QEMU binary"));
            return -1;
        }
    }

    if (src->timeout > 0) {
        if (actualType != VIR_STORAGE_TYPE_NETWORK ||
            (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
             src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP &&
             src->protocol != VIR_STORAGE_NET_PROTOCOL_FTP &&
             src->protocol != VIR_STORAGE_NET_PROTOCOL_FTPS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("timeout is supported only with HTTP(S)/FTP(s) protocols"));
            return -1;
        }

        if (!src->detected &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCKDEV)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("timeout setting is not supported with this QEMU binary"));
            return -1;
        }
    }

7071 7072 7073 7074
    return 0;
}


7075
int
7076 7077
qemuDomainDeviceDefValidateDisk(const virDomainDiskDef *disk,
                                virQEMUCapsPtr qemuCaps)
7078
{
7079
    const char *driverName = virDomainDiskGetDriver(disk);
7080
    virStorageSourcePtr n;
7081 7082
    int idx;
    int partition;
7083

7084 7085 7086 7087 7088 7089
    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;
7090 7091
    }

7092 7093 7094 7095 7096 7097 7098 7099 7100 7101 7102 7103 7104 7105 7106
    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;
        }
7107 7108
    }

7109 7110 7111 7112 7113 7114 7115 7116 7117 7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131 7132 7133 7134 7135
    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;
    }

7136 7137 7138 7139 7140 7141 7142
    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
7143 7144 7145 7146 7147 7148 7149 7150
    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;
    }

7151 7152 7153 7154 7155 7156 7157 7158 7159 7160 7161 7162 7163
    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;
    }

7164 7165 7166 7167 7168
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if (qemuDomainValidateStorageSource(n, qemuCaps) < 0)
            return -1;
    }

7169 7170 7171 7172
    return 0;
}


7173 7174 7175 7176
static int
qemuDomainDeviceDefValidateControllerAttributes(const virDomainControllerDef *controller)
{
    if (!(controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
7177 7178 7179
          (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))) {
7180 7181 7182 7183 7184 7185 7186 7187 7188 7189 7190 7191 7192 7193 7194 7195 7196 7197 7198 7199
        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;
        }
7200 7201 7202 7203 7204
        if (controller->iothread) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'iothread' is only supported for virtio-scsi controller"));
            return -1;
        }
7205 7206 7207 7208 7209 7210
    }

    return 0;
}


7211 7212 7213 7214 7215 7216
/**
 * @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
7217
 * Returns true if acceptable, false otherwise with error message set.
7218 7219 7220 7221 7222 7223 7224 7225 7226 7227 7228 7229 7230 7231 7232
 */
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:
7233 7234
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL:
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL:
7235 7236 7237 7238 7239 7240 7241 7242 7243 7244 7245 7246 7247 7248 7249 7250 7251 7252 7253 7254 7255 7256 7257 7258 7259 7260 7261 7262 7263 7264 7265 7266 7267
        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;
7268 7269 7270 7271 7272 7273
    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;
7274 7275 7276 7277 7278 7279
    }

    return true;
}


7280 7281 7282 7283 7284 7285 7286 7287 7288 7289 7290 7291 7292 7293 7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305
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;
}


7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320 7321 7322
/* 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;

7323 7324
    if (controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI &&
7325 7326
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
       virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
7327 7328
                       _("virtio-scsi IOThreads only available for virtio "
                         "pci and virtio ccw controllers"));
7329 7330 7331 7332 7333 7334 7335 7336 7337 7338 7339 7340 7341 7342 7343 7344 7345 7346 7347 7348 7349
       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:
7350 7351
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL:
7352 7353 7354 7355 7356 7357 7358 7359 7360 7361 7362
            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:
7363
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT:
7364 7365 7366 7367 7368 7369 7370 7371
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LAST:
            break;
    }

    return 0;
}


7372 7373 7374 7375 7376 7377 7378 7379 7380 7381 7382 7383 7384 7385 7386 7387 7388 7389 7390 7391 7392 7393 7394 7395 7396 7397 7398 7399 7400 7401 7402 7403
/**
 * 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;
7404 7405
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCIE_PCI_BRIDGE:
        return QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE;
7406 7407 7408 7409 7410
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE:
        return 0;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_LAST:
    default:
        return -1;
7411 7412
    }

7413
    return -1;
7414 7415 7416
}


7417 7418 7419 7420 7421 7422 7423 7424 7425 7426 7427 7428 7429 7430 7431 7432 7433
#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));


7434 7435 7436 7437 7438 7439 7440 7441 7442
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);
7443
    int cap = virDomainControllerPCIModelNameToQEMUCaps(pciopts->modelName);
7444 7445 7446 7447 7448 7449 7450 7451 7452 7453

    if (!model) {
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }
    if (!modelName) {
        virReportEnumRangeError(virDomainControllerPCIModelName, pciopts->modelName);
        return -1;
    }

7454 7455 7456 7457 7458 7459 7460 7461 7462
    /* 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:
7463
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
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 7494 7495 7496 7497 7498 7499 7500 7501 7502 7503 7504 7505 7506 7507 7508 7509 7510 7511 7512 7513 7514 7515 7516 7517 7518 7519 7520 7521 7522 7523 7524 7525 7526 7527 7528 7529 7530 7531 7532 7533 7534 7535 7536 7537 7538 7539 7540 7541 7542 7543 7544 7545 7546 7547 7548 7549 7550 7551 7552 7553 7554 7555 7556 7557 7558 7559 7560 7561 7562 7563 7564 7565 7566 7567
        /* 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;

7568 7569 7570 7571 7572 7573 7574
    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;

7575 7576 7577 7578 7579 7580 7581
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

7582 7583 7584 7585 7586 7587 7588 7589 7590
    /* 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:
7591
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7592 7593 7594 7595 7596 7597 7598 7599 7600 7601 7602 7603 7604 7605 7606 7607 7608 7609
        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,
J
Ján Tomko 已提交
7610
         * the index must be zero */
7611 7612 7613 7614 7615 7616 7617 7618 7619 7620 7621 7622 7623 7624 7625
        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;
    }

7626 7627 7628 7629 7630 7631 7632 7633 7634 7635 7636 7637 7638 7639 7640 7641 7642 7643 7644 7645 7646 7647 7648 7649 7650 7651 7652
    /* 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:
7653
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7654 7655 7656 7657 7658 7659 7660 7661 7662 7663 7664 7665 7666
        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;
    }

7667 7668 7669
    /* pcihole64 */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
7670 7671 7672 7673 7674 7675 7676 7677 7678 7679 7680 7681 7682 7683 7684 7685 7686
        if (pciopts->pcihole64 ||  pciopts->pcihole64size != 0) {
            if (!qemuDomainIsI440FX(def)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Setting the 64-bit PCI hole size is not "
                                 "supported for machine '%s'"), def->os.machine);
                return -1;
            }

            if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_I440FX_PCI_HOLE64_SIZE)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("64-bit PCI hole size setting is not supported "
                                 "with this QEMU binary"));
                return -1;
            }
        }
        break;

7687
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
7688 7689 7690 7691 7692 7693 7694 7695 7696 7697 7698 7699 7700 7701
        if (pciopts->pcihole64 || pciopts->pcihole64size != 0) {
            if (!qemuDomainIsQ35(def)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Setting the 64-bit PCI hole size is not "
                                 "supported for machine '%s'"), def->os.machine);
                return -1;
            }

            if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_Q35_PCI_HOLE64_SIZE)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("64-bit PCI hole size setting is not supported "
                                 "with this QEMU binary"));
                return -1;
            }
7702 7703 7704 7705 7706 7707 7708 7709 7710 7711
        }
        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:
7712
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7713 7714 7715 7716 7717 7718 7719 7720 7721 7722 7723 7724 7725 7726
        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;
    }

7727 7728 7729 7730 7731 7732 7733 7734 7735 7736 7737 7738 7739 7740 7741 7742 7743
    /* 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:
7744
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7745 7746 7747 7748 7749 7750 7751 7752 7753 7754 7755 7756 7757
        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;
    }

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
    /* 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:
7789
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7790 7791 7792 7793 7794 7795 7796 7797 7798 7799 7800 7801 7802
        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;
    }

7803 7804 7805 7806 7807 7808 7809 7810 7811 7812 7813 7814 7815 7816 7817 7818 7819
    /* 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:
7820
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7821 7822 7823 7824 7825 7826 7827 7828 7829 7830 7831 7832 7833
        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;
    }

7834 7835 7836 7837 7838 7839 7840 7841 7842 7843 7844 7845 7846 7847 7848 7849 7850 7851 7852 7853 7854
    /* 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:
7855
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7856 7857 7858 7859 7860 7861 7862 7863 7864 7865 7866 7867 7868 7869 7870 7871
        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);
    }

7872 7873 7874 7875 7876 7877 7878 7879 7880 7881 7882 7883 7884 7885 7886 7887 7888 7889 7890 7891 7892 7893 7894 7895 7896 7897
    /* 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;
    }

7898
    return 0;
7899 7900 7901
}


7902 7903 7904 7905 7906
#undef virReportControllerInvalidValue
#undef virReportControllerInvalidOption
#undef virReportControllerMissingOption


7907 7908 7909 7910 7911 7912 7913 7914 7915 7916 7917 7918 7919 7920 7921 7922 7923 7924
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;
}


7925
static int
7926 7927 7928
qemuDomainDeviceDefValidateController(const virDomainControllerDef *controller,
                                      const virDomainDef *def,
                                      virQEMUCapsPtr qemuCaps)
7929
{
7930 7931
    int ret = 0;

7932
    if (!qemuDomainCheckCCWS390AddressSupport(def, &controller->info, qemuCaps,
7933 7934 7935
                                              "controller"))
        return -1;

7936 7937 7938 7939
    if (controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
        !qemuDomainCheckSCSIControllerModel(qemuCaps, controller->model))
        return -1;

7940 7941 7942
    if (qemuDomainDeviceDefValidateControllerAttributes(controller) < 0)
        return -1;

7943
    switch ((virDomainControllerType)controller->type) {
7944
    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
7945 7946 7947
        ret = qemuDomainDeviceDefValidateControllerIDE(controller, def);
        break;

7948
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
7949 7950 7951
        ret = qemuDomainDeviceDefValidateControllerSCSI(controller, def);
        break;

7952
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
7953 7954
        ret = qemuDomainDeviceDefValidateControllerPCI(controller, def,
                                                       qemuCaps);
7955 7956
        break;

7957
    case VIR_DOMAIN_CONTROLLER_TYPE_SATA:
7958 7959 7960 7961 7962
        ret = qemuDomainDeviceDefValidateControllerSATA(controller, def,
                                                        qemuCaps);
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
7963 7964 7965
    case VIR_DOMAIN_CONTROLLER_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_CONTROLLER_TYPE_CCID:
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
7966
    case VIR_DOMAIN_CONTROLLER_TYPE_XENBUS:
7967 7968 7969 7970
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
    }

7971
    return ret;
7972 7973 7974
}


7975
static int
7976 7977
qemuDomainDeviceDefValidateVsock(const virDomainVsockDef *vsock,
                                 const virDomainDef *def,
7978 7979 7980 7981 7982 7983 7984 7985
                                 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;
    }
7986

7987
    if (!qemuDomainCheckCCWS390AddressSupport(def, &vsock->info, qemuCaps,
7988 7989 7990
                                              "vsock"))
        return -1;

7991 7992
    return 0;
}
7993

7994 7995 7996

static int
qemuDomainDeviceDefValidateTPM(virDomainTPMDef *tpm,
7997 7998
                               const virDomainDef *def,
                               virQEMUCapsPtr qemuCaps)
7999
{
8000 8001
    virQEMUCapsFlags flag;

8002 8003 8004 8005 8006 8007
    /* 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:
8008
        /* TPM 1.2 + CRB do not work */
8009
        if (tpm->type == VIR_DOMAIN_TPM_TYPE_EMULATOR &&
8010
            tpm->model == VIR_DOMAIN_TPM_MODEL_CRB) {
8011 8012 8013 8014 8015 8016 8017 8018 8019 8020 8021
            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;
    }
8022 8023 8024 8025 8026 8027 8028 8029 8030 8031 8032 8033 8034 8035 8036 8037

    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;
    }

8038 8039 8040 8041 8042 8043 8044
    switch (tpm->model) {
    case VIR_DOMAIN_TPM_MODEL_TIS:
        flag = QEMU_CAPS_DEVICE_TPM_TIS;
        break;
    case VIR_DOMAIN_TPM_MODEL_CRB:
        flag = QEMU_CAPS_DEVICE_TPM_CRB;
        break;
8045
    case VIR_DOMAIN_TPM_MODEL_SPAPR:
8046 8047
        flag = QEMU_CAPS_DEVICE_TPM_SPAPR;
        break;
8048 8049 8050 8051 8052 8053 8054 8055 8056 8057 8058 8059 8060 8061 8062
    case VIR_DOMAIN_TPM_MODEL_LAST:
    default:
        virReportEnumRangeError(virDomainTPMModel, tpm->model);
        return -1;
    }

    if (!virQEMUCapsGet(qemuCaps, flag)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("The QEMU executable %s does not support TPM "
                         "model %s"),
                       def->emulator,
                       virDomainTPMModelTypeToString(tpm->model));
        return -1;
    }

8063
    return 0;
8064 8065 8066 8067 8068 8069 8070 8071

 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;
8072 8073 8074
}


8075 8076 8077 8078 8079 8080 8081 8082 8083 8084 8085 8086 8087 8088 8089 8090 8091 8092 8093 8094 8095 8096 8097 8098 8099 8100 8101 8102 8103 8104 8105 8106 8107 8108 8109 8110 8111 8112 8113 8114 8115 8116 8117 8118 8119 8120 8121 8122 8123 8124 8125 8126 8127 8128 8129 8130 8131 8132 8133 8134 8135 8136 8137 8138 8139 8140 8141 8142 8143 8144 8145 8146 8147 8148
static int
qemuDomainDeviceDefValidateSPICEGraphics(const virDomainGraphicsDef *graphics,
                                         virQEMUDriverPtr driver,
                                         virQEMUCapsPtr qemuCaps)
{
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
    virDomainGraphicsListenDefPtr glisten = NULL;
    int tlsPort = graphics->data.spice.tlsPort;

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SPICE)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("spice graphics are not supported with this QEMU"));
        return -1;
    }

    glisten = virDomainGraphicsGetListen((virDomainGraphicsDefPtr)graphics, 0);
    if (!glisten) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing listen element"));
        return -1;
    }

    switch (glisten->type) {
    case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_SOCKET:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SPICE_UNIX)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("unix socket for spice graphics are not supported "
                             "with this QEMU"));
            return -1;
        }
        break;

    case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_ADDRESS:
    case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_NETWORK:
        if (tlsPort > 0 && !cfg->spiceTLS) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("spice TLS port set in XML configuration, "
                             "but TLS is disabled in qemu.conf"));
            return -1;
        }
        break;

    case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_NONE:
        break;
    case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_LAST:
        break;
    }

    if (graphics->data.spice.filetransfer == VIR_TRISTATE_BOOL_NO &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_SPICE_FILE_XFER_DISABLE)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU can't disable file transfers through spice"));
            return -1;
    }

    if (graphics->data.spice.gl == VIR_TRISTATE_BOOL_YES) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SPICE_GL)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support spice OpenGL"));
            return -1;
        }

        if (graphics->data.spice.rendernode &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_SPICE_RENDERNODE)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support spice OpenGL rendernode"));
            return -1;
        }
    }

    return 0;
}


8149 8150 8151
static int
qemuDomainDeviceDefValidateGraphics(const virDomainGraphicsDef *graphics,
                                    const virDomainDef *def,
8152
                                    virQEMUDriverPtr driver,
8153 8154 8155 8156 8157 8158
                                    virQEMUCapsPtr qemuCaps)
{
    bool have_egl_headless = false;
    size_t i;

    for (i = 0; i < def->ngraphics; i++) {
8159
        if (def->graphics[i]->type == VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS) {
8160 8161 8162 8163 8164 8165 8166 8167 8168 8169 8170 8171 8172 8173 8174 8175 8176 8177 8178 8179 8180 8181 8182 8183 8184 8185 8186 8187 8188 8189 8190 8191 8192 8193 8194 8195 8196 8197
            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;
        }
    }

8198 8199 8200 8201 8202 8203 8204 8205 8206 8207 8208 8209 8210
    switch (graphics->type) {
    case VIR_DOMAIN_GRAPHICS_TYPE_SDL:
        if (graphics->data.sdl.gl != VIR_TRISTATE_BOOL_ABSENT) {
            if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SDL_GL)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("OpenGL for SDL is not supported with this QEMU "
                                 "binary"));
                return -1;
            }
        }
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_VNC:
8211 8212 8213 8214 8215
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VNC)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("vnc graphics are not supported with this QEMU"));
            return -1;
        }
8216
        break;
8217

8218
    case VIR_DOMAIN_GRAPHICS_TYPE_SPICE:
8219 8220 8221 8222 8223 8224
        if (qemuDomainDeviceDefValidateSPICEGraphics(graphics, driver,
                                                     qemuCaps) < 0)
            return -1;

        break;

8225
    case VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS:
8226 8227 8228 8229 8230 8231 8232 8233
        if (graphics->data.egl_headless.rendernode &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_EGL_HEADLESS_RENDERNODE)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support OpenGL rendernode "
                             "with egl-headless graphics type"));
            return -1;
        }

8234 8235 8236 8237 8238 8239 8240 8241 8242 8243 8244 8245
        break;
    case VIR_DOMAIN_GRAPHICS_TYPE_RDP:
    case VIR_DOMAIN_GRAPHICS_TYPE_DESKTOP:
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("unsupported graphics type '%s'"),
                       virDomainGraphicsTypeToString(graphics->type));
        return -1;
    case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
    default:
        return -1;
    }

8246 8247 8248 8249
    return 0;
}


8250 8251
static int
qemuDomainDeviceDefValidateInput(const virDomainInputDef *input,
8252
                                 const virDomainDef *def,
8253 8254
                                 virQEMUCapsPtr qemuCaps)
{
8255 8256 8257 8258
    const char *baseName;
    int cap;
    int ccwCap;

8259 8260 8261 8262 8263 8264 8265 8266
    if (input->bus == VIR_DOMAIN_INPUT_BUS_PS2 && !ARCH_IS_X86(def->os.arch) &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_I8042)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("%s is not supported by this QEMU binary"),
                       virDomainInputBusTypeToString(input->bus));
        return -1;
    }

8267 8268 8269
    if (input->bus != VIR_DOMAIN_INPUT_BUS_VIRTIO)
        return 0;

8270 8271 8272 8273 8274 8275 8276 8277 8278 8279 8280 8281 8282 8283 8284 8285 8286 8287 8288 8289 8290 8291 8292 8293 8294 8295 8296 8297 8298 8299 8300 8301
    /* 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;
    }

8302 8303
    switch ((virDomainInputType)input->type) {
    case VIR_DOMAIN_INPUT_TYPE_MOUSE:
8304 8305 8306
        baseName = "virtio-mouse";
        cap = QEMU_CAPS_VIRTIO_MOUSE;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_MOUSE_CCW;
8307 8308
        break;
    case VIR_DOMAIN_INPUT_TYPE_TABLET:
8309 8310 8311
        baseName = "virtio-tablet";
        cap = QEMU_CAPS_VIRTIO_TABLET;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_TABLET_CCW;
8312 8313
        break;
    case VIR_DOMAIN_INPUT_TYPE_KBD:
8314 8315 8316
        baseName = "virtio-keyboard";
        cap = QEMU_CAPS_VIRTIO_KEYBOARD;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_KEYBOARD_CCW;
8317 8318
        break;
    case VIR_DOMAIN_INPUT_TYPE_PASSTHROUGH:
8319 8320 8321
        baseName = "virtio-input-host";
        cap = QEMU_CAPS_VIRTIO_INPUT_HOST;
        ccwCap = QEMU_CAPS_LAST;
8322 8323 8324 8325 8326 8327 8328 8329
        break;
    case VIR_DOMAIN_INPUT_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainInputType,
                                input->type);
        return -1;
    }

8330 8331 8332 8333 8334 8335 8336 8337 8338
    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;
    }

8339 8340 8341 8342
    return 0;
}


8343 8344 8345 8346 8347 8348 8349 8350 8351
static int
qemuDomainDeviceDefValidateMemballoon(const virDomainMemballoonDef *memballoon,
                                      virQEMUCapsPtr qemuCaps)
{
    if (!memballoon ||
        memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_NONE) {
        return 0;
    }

8352 8353 8354
    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) {
8355 8356 8357 8358 8359 8360 8361 8362 8363 8364 8365 8366 8367 8368 8369 8370 8371
        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;
}


8372 8373 8374 8375 8376 8377 8378 8379 8380 8381 8382 8383 8384 8385 8386 8387 8388 8389 8390 8391 8392 8393 8394 8395
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;

8396
    case VIR_DOMAIN_IOMMU_MODEL_SMMUV3:
8397 8398 8399 8400 8401 8402 8403 8404 8405 8406 8407 8408 8409 8410
        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;
        }
8411 8412
        break;

8413 8414 8415
    case VIR_DOMAIN_IOMMU_MODEL_LAST:
    default:
        virReportEnumRangeError(virDomainIOMMUModel, iommu->model);
8416
        return -1;
8417 8418
    }

8419 8420 8421 8422 8423 8424 8425 8426 8427 8428 8429 8430 8431 8432 8433 8434 8435 8436 8437 8438 8439 8440 8441 8442 8443 8444 8445 8446 8447 8448 8449 8450 8451 8452
    /* 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;
    }

8453 8454 8455
    return 0;
}

J
Ján Tomko 已提交
8456 8457 8458 8459 8460 8461 8462 8463 8464 8465 8466 8467 8468 8469 8470 8471 8472 8473 8474 8475 8476 8477 8478 8479 8480 8481 8482 8483 8484 8485 8486 8487 8488 8489 8490 8491 8492 8493 8494 8495
static int
qemuDomainDefValidateVirtioFSSharedMemory(const virDomainDef *def)
{
    size_t numa_nodes = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    if (numa_nodes == 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("virtiofs requires one or more NUMA nodes"));
        return -1;
    }

    for (i = 0; i < numa_nodes; i++) {
        virDomainMemoryAccess node_access =
            virDomainNumaGetNodeMemoryAccessMode(def->numa, i);

        switch (node_access) {
        case VIR_DOMAIN_MEMORY_ACCESS_DEFAULT:
            if (def->mem.access != VIR_DOMAIN_MEMORY_ACCESS_SHARED) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("virtiofs requires shared memory"));
                return -1;
            }
            break;
        case VIR_DOMAIN_MEMORY_ACCESS_SHARED:
            break;
        case VIR_DOMAIN_MEMORY_ACCESS_PRIVATE:
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtiofs requires shared memory"));
            return -1;

        case VIR_DOMAIN_MEMORY_ACCESS_LAST:
        default:
            virReportEnumRangeError(virDomainMemoryAccess, node_access);
            return -1;

        }
    }
    return 0;
}
8496

8497 8498
static int
qemuDomainDeviceDefValidateFS(virDomainFSDefPtr fs,
J
Ján Tomko 已提交
8499 8500 8501
                              const virDomainDef *def,
                              virQEMUDriverPtr driver,
                              virQEMUCapsPtr qemuCaps)
8502 8503 8504 8505 8506 8507 8508 8509 8510 8511 8512 8513 8514 8515 8516 8517 8518 8519 8520 8521 8522 8523 8524 8525 8526 8527 8528
{
    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;

8529
    case VIR_DOMAIN_FS_DRIVER_TYPE_VIRTIOFS:
J
Ján Tomko 已提交
8530 8531 8532 8533 8534 8535 8536 8537 8538 8539 8540 8541 8542 8543 8544 8545 8546 8547 8548 8549 8550 8551 8552 8553 8554 8555 8556 8557 8558 8559 8560 8561 8562
        if (!virQEMUDriverIsPrivileged(driver)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtiofs is not yet supported in session mode"));
            return -1;
        }
        if (fs->accessmode != VIR_DOMAIN_FS_ACCESSMODE_PASSTHROUGH) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtiofs only supports passthrough accessmode"));
            return -1;
        }
        if (fs->wrpolicy != VIR_DOMAIN_FS_WRPOLICY_DEFAULT) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtiofs does not support wrpolicy"));
            return -1;
        }
        if (fs->model != VIR_DOMAIN_FS_MODEL_DEFAULT) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtiofs does not support model"));
            return -1;
        }
        if (fs->format != VIR_STORAGE_FILE_NONE) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtiofs does not support format"));
            return -1;
        }
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VHOST_USER_FS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtiofs is not supported with this QEMU binary"));
            return -1;
        }
        if (qemuDomainDefValidateVirtioFSSharedMemory(def) < 0)
            return -1;
        break;
8563

8564 8565 8566 8567 8568 8569 8570 8571 8572 8573
    case VIR_DOMAIN_FS_DRIVER_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainFSDriverType, fs->fsdriver);
        return -1;
    }

    return 0;
}


8574 8575 8576 8577 8578 8579 8580 8581 8582 8583 8584 8585 8586 8587 8588 8589 8590 8591 8592 8593 8594 8595 8596 8597 8598
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;

8599 8600
    switch ((virDomainDeviceAddressType) info->type) {
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI:
8601 8602
        return qemuDomainDeviceDefValidateZPCIAddress(info, qemuCaps);

8603 8604 8605 8606 8607 8608
    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;

8609 8610 8611 8612 8613 8614 8615 8616 8617 8618 8619 8620 8621 8622
    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;
        }

8623 8624 8625 8626 8627 8628 8629 8630 8631
    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:
8632
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_UNASSIGNED:
8633 8634 8635 8636 8637 8638 8639 8640 8641
        /* No validation for these address types yet */
        break;

    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainDeviceAddressType, info->type);
        return -1;
    }

8642 8643 8644 8645
    return 0;
}


8646 8647 8648
static int
qemuDomainDeviceDefValidate(const virDomainDeviceDef *dev,
                            const virDomainDef *def,
8649
                            void *opaque)
8650
{
8651
    int ret = 0;
8652
    virQEMUDriverPtr driver = opaque;
8653 8654
    g_autoptr(virQEMUCaps) qemuCaps = NULL;
    g_autoptr(virDomainCaps) domCaps = NULL;
8655 8656 8657 8658

    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
                                            def->emulator)))
        return -1;
8659

8660 8661 8662 8663
    if (!(domCaps = virQEMUDriverGetDomainCapabilities(driver, qemuCaps,
                                                       def->os.machine,
                                                       def->os.arch,
                                                       def->virtType)))
8664
        return -1;
8665

8666
    if ((ret = qemuDomainDeviceDefValidateAddress(dev, qemuCaps)) < 0)
8667
        return ret;
8668

8669
    if ((ret = virDomainCapsDeviceDefValidate(domCaps, dev, def)) < 0)
8670
        return ret;
8671

8672
    switch ((virDomainDeviceType)dev->type) {
8673
    case VIR_DOMAIN_DEVICE_NET:
8674
        ret = qemuDomainDeviceDefValidateNetwork(dev->data.net, qemuCaps);
8675 8676 8677
        break;

    case VIR_DOMAIN_DEVICE_CHR:
8678
        ret = qemuDomainChrDefValidate(dev->data.chr, def, qemuCaps);
8679 8680 8681
        break;

    case VIR_DOMAIN_DEVICE_SMARTCARD:
8682
        ret = qemuDomainSmartcardDefValidate(dev->data.smartcard, qemuCaps);
8683 8684 8685
        break;

    case VIR_DOMAIN_DEVICE_RNG:
8686
        ret = qemuDomainRNGDefValidate(dev->data.rng, qemuCaps);
8687 8688 8689
        break;

    case VIR_DOMAIN_DEVICE_REDIRDEV:
8690
        ret = qemuDomainRedirdevDefValidate(dev->data.redirdev, qemuCaps);
8691 8692 8693 8694 8695 8696 8697
        break;

    case VIR_DOMAIN_DEVICE_WATCHDOG:
        ret = qemuDomainWatchdogDefValidate(dev->data.watchdog, def);
        break;

    case VIR_DOMAIN_DEVICE_HOSTDEV:
8698 8699
        ret = qemuDomainDeviceDefValidateHostdev(dev->data.hostdev, def,
                                                 qemuCaps);
8700 8701 8702
        break;

    case VIR_DOMAIN_DEVICE_VIDEO:
8703
        ret = qemuDomainDeviceDefValidateVideo(dev->data.video, qemuCaps);
8704 8705 8706
        break;

    case VIR_DOMAIN_DEVICE_DISK:
8707
        ret = qemuDomainDeviceDefValidateDisk(dev->data.disk, qemuCaps);
8708 8709
        break;

8710
    case VIR_DOMAIN_DEVICE_CONTROLLER:
8711 8712
        ret = qemuDomainDeviceDefValidateController(dev->data.controller, def,
                                                    qemuCaps);
8713 8714
        break;

8715
    case VIR_DOMAIN_DEVICE_VSOCK:
8716
        ret = qemuDomainDeviceDefValidateVsock(dev->data.vsock, def, qemuCaps);
8717 8718
        break;

8719
    case VIR_DOMAIN_DEVICE_TPM:
8720
        ret = qemuDomainDeviceDefValidateTPM(dev->data.tpm, def, qemuCaps);
8721 8722
        break;

8723 8724
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainDeviceDefValidateGraphics(dev->data.graphics, def,
8725
                                                  driver, qemuCaps);
8726 8727
        break;

8728 8729 8730 8731
    case VIR_DOMAIN_DEVICE_INPUT:
        ret = qemuDomainDeviceDefValidateInput(dev->data.input, def, qemuCaps);
        break;

8732 8733 8734 8735
    case VIR_DOMAIN_DEVICE_MEMBALLOON:
        ret = qemuDomainDeviceDefValidateMemballoon(dev->data.memballoon, qemuCaps);
        break;

8736 8737 8738 8739
    case VIR_DOMAIN_DEVICE_IOMMU:
        ret = qemuDomainDeviceDefValidateIOMMU(dev->data.iommu, def, qemuCaps);
        break;

8740
    case VIR_DOMAIN_DEVICE_FS:
J
Ján Tomko 已提交
8741
        ret = qemuDomainDeviceDefValidateFS(dev->data.fs, def, driver, qemuCaps);
8742 8743
        break;

8744 8745 8746 8747
    case VIR_DOMAIN_DEVICE_NVRAM:
        ret = qemuDomainDeviceDefValidateNVRAM(dev->data.nvram, def, qemuCaps);
        break;

8748 8749 8750 8751
    case VIR_DOMAIN_DEVICE_HUB:
        ret = qemuDomainDeviceDefValidateHub(dev->data.hub, qemuCaps);
        break;

8752
    case VIR_DOMAIN_DEVICE_SOUND:
8753 8754 8755
        ret = qemuDomainDeviceDefValidateSound(dev->data.sound, qemuCaps);
        break;

8756 8757 8758 8759
    case VIR_DOMAIN_DEVICE_MEMORY:
        ret = qemuDomainDeviceDefValidateMemory(dev->data.memory, qemuCaps);
        break;

8760
    case VIR_DOMAIN_DEVICE_LEASE:
8761 8762 8763 8764 8765
    case VIR_DOMAIN_DEVICE_SHMEM:
    case VIR_DOMAIN_DEVICE_PANIC:
    case VIR_DOMAIN_DEVICE_NONE:
    case VIR_DOMAIN_DEVICE_LAST:
        break;
8766 8767
    }

8768 8769 8770 8771
    return ret;
}


8772 8773 8774 8775 8776 8777 8778 8779 8780
/**
 * 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.
 */
8781
static int
8782 8783
qemuDomainDefaultNetModel(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
8784
{
8785
    if (ARCH_IS_S390(def->os.arch))
8786
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
8787

S
Stefan Schallenberg 已提交
8788 8789
    if (def->os.arch == VIR_ARCH_ARMV6L ||
        def->os.arch == VIR_ARCH_ARMV7L ||
8790
        def->os.arch == VIR_ARCH_AARCH64) {
8791
        if (STREQ(def->os.machine, "versatilepb"))
8792
            return VIR_DOMAIN_NET_MODEL_SMC91C111;
8793

8794
        if (qemuDomainIsARMVirt(def))
8795
            return VIR_DOMAIN_NET_MODEL_VIRTIO;
8796

8797 8798
        /* Incomplete. vexpress (and a few others) use this, but not all
         * arm boards */
8799
        return VIR_DOMAIN_NET_MODEL_LAN9118;
8800 8801
    }

8802 8803
    /* virtio is a sensible default for RISC-V virt guests */
    if (qemuDomainIsRISCVVirt(def))
8804
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
8805

8806 8807 8808
    /* In all other cases the model depends on the capabilities. If they were
     * not provided don't report any default. */
    if (!qemuCaps)
8809
        return VIR_DOMAIN_NET_MODEL_UNKNOWN;
8810

8811 8812 8813 8814
    /* 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))
8815
        return VIR_DOMAIN_NET_MODEL_RTL8139;
8816
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_E1000))
8817
        return VIR_DOMAIN_NET_MODEL_E1000;
8818
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_NET))
8819
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
8820 8821 8822

    /* We've had no luck detecting support for any network device,
     * but we have to return something: might as well be rtl8139 */
8823
    return VIR_DOMAIN_NET_MODEL_RTL8139;
8824
}
8825

8826 8827

/*
8828 8829 8830 8831 8832 8833 8834 8835 8836 8837 8838 8839 8840 8841 8842 8843
 * 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.
8844
 */
8845
static void
8846 8847 8848
qemuDomainChrDefDropDefaultPath(virDomainChrDefPtr chr,
                                virQEMUDriverPtr driver)
{
8849
    g_autoptr(virQEMUDriverConfig) cfg = NULL;
8850
    virBuffer buf = VIR_BUFFER_INITIALIZER;
8851
    g_autofree char *regexp = NULL;
8852

8853 8854 8855 8856
    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) {
8857
        return;
8858 8859
    }

8860 8861 8862 8863 8864 8865 8866 8867 8868 8869
    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);
8870 8871 8872
}


8873 8874 8875 8876 8877 8878 8879 8880 8881 8882 8883 8884 8885 8886 8887 8888 8889 8890 8891 8892 8893 8894 8895 8896 8897 8898 8899 8900 8901 8902 8903 8904 8905 8906 8907 8908 8909 8910 8911 8912 8913 8914 8915 8916 8917 8918 8919 8920 8921 8922
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;
}


8923
#define QEMU_USB_XHCI_MAXPORTS 15
8924 8925


8926 8927 8928
static int
qemuDomainControllerDefPostParse(virDomainControllerDefPtr cont,
                                 const virDomainDef *def,
8929 8930
                                 virQEMUCapsPtr qemuCaps,
                                 unsigned int parseFlags)
8931
{
8932 8933
    switch ((virDomainControllerType)cont->type) {
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
8934 8935 8936
        /* Set the default SCSI controller model if not already set */
        if (qemuDomainSetSCSIControllerModel(def, cont, qemuCaps) < 0)
            return -1;
8937
        break;
8938

8939
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
8940
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT && qemuCaps) {
8941
            /* Pick a suitable default model for the USB controller if none
8942 8943
             * has been selected by the user and we have the qemuCaps for
             * figuring out which contollers are supported.
8944 8945 8946 8947 8948 8949 8950
             *
             * 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() */
8951 8952 8953 8954 8955 8956 8957 8958 8959 8960 8961

            /* 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;
                }
8962
            } else if (ARCH_IS_PPC64(def->os.arch)) {
8963 8964
                /* To not break migration we need to set default USB controller
                 * for ppc64 to pci-ohci if we cannot change ABI of the VM.
8965 8966
                 * The nec-usb-xhci or qemu-xhci controller is used as default
                 * only for newly defined domains or devices. */
8967
                if ((parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE) &&
8968 8969 8970
                    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) &&
8971 8972 8973
                    virQEMUCapsGet(qemuCaps, QEMU_CAPS_NEC_USB_XHCI)) {
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
                } else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_OHCI)) {
8974
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_PCI_OHCI;
8975 8976 8977
                } else {
                    /* Explicitly fallback to legacy USB controller for PPC64. */
                    cont->model = -1;
8978
                }
8979
            } else if (def->os.arch == VIR_ARCH_AARCH64) {
8980 8981 8982
                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))
8983
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
8984 8985 8986 8987 8988 8989 8990 8991 8992 8993 8994
            }
        }
        /* 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;
        }
8995 8996 8997
        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) {
8998
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8999 9000 9001
                           _("'%s' controller only supports up to '%u' ports"),
                           virDomainControllerModelUSBTypeToString(cont->model),
                           QEMU_USB_XHCI_MAXPORTS);
9002 9003 9004
            return -1;
        }
        break;
9005

9006
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
9007 9008 9009 9010 9011 9012 9013

        /* 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) {
9014 9015 9016 9017 9018 9019
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("pci-root and pcie-root controllers "
                             "should have index 0"));
            return -1;
        }

9020
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS &&
9021
            !qemuDomainIsI440FX(def)) {
9022 9023 9024 9025 9026 9027
            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 &&
9028
            !qemuDomainIsQ35(def)) {
9029 9030 9031 9032 9033 9034
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("pcie-expander-bus controllers are only supported "
                             "on q35-based machinetypes"));
            return -1;
        }

9035 9036 9037
        /* 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
9038 9039
         * from 0 .. size-1.
         */
9040 9041
        if (cont->opts.pciopts.numaNode >= 0 &&
            cont->opts.pciopts.numaNode >=
9042
            (int)virDomainNumaGetNodeCount(def->numa)) {
9043 9044 9045 9046 9047 9048 9049 9050 9051 9052
            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;
        }
9053 9054 9055 9056 9057 9058 9059
        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:
9060
    case VIR_DOMAIN_CONTROLLER_TYPE_XENBUS:
9061 9062
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
9063 9064 9065 9066 9067
    }

    return 0;
}

9068 9069 9070 9071 9072 9073
static int
qemuDomainChrDefPostParse(virDomainChrDefPtr chr,
                          const virDomainDef *def,
                          virQEMUDriverPtr driver,
                          unsigned int parseFlags)
{
9074 9075 9076 9077 9078 9079 9080 9081 9082 9083 9084 9085 9086 9087
    /* 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)) {
9088
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
9089
        } else if (qemuDomainIsARMVirt(def) || qemuDomainIsRISCVVirt(def)) {
9090
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
9091 9092
        } else if (ARCH_IS_S390(def->os.arch)) {
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
9093 9094 9095
        }
    }

9096 9097 9098
    /* Set the default target model */
    if (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL &&
        chr->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_NONE) {
9099
        switch ((virDomainChrSerialTargetType)chr->targetType) {
9100 9101 9102 9103 9104 9105 9106 9107 9108
        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;
9109 9110 9111
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY;
            break;
9112
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
9113 9114 9115 9116 9117
            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;
            }
9118
            break;
9119 9120 9121
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE;
            break;
9122 9123 9124 9125 9126 9127 9128
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            /* Nothing to do */
            break;
        }
    }

9129 9130
    /* clear auto generated unix socket path for inactive definitions */
    if (parseFlags & VIR_DOMAIN_DEF_PARSE_INACTIVE) {
9131
        qemuDomainChrDefDropDefaultPath(chr, driver);
9132 9133 9134 9135 9136 9137 9138 9139 9140 9141 9142 9143

        /* 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;
}
9144

9145

9146 9147 9148 9149 9150 9151 9152 9153 9154 9155 9156 9157 9158 9159 9160 9161
/**
 * 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;
9162 9163
    g_autofree char *authalias = NULL;
    g_autofree char *encalias = NULL;
9164 9165 9166 9167 9168 9169 9170 9171 9172 9173 9174 9175 9176 9177 9178 9179 9180 9181 9182 9183 9184 9185 9186 9187 9188 9189 9190 9191 9192 9193 9194 9195 9196 9197 9198 9199 9200 9201

    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) {
9202
        authalias = g_strdup_printf("%s-secret0", disk->info.alias);
9203 9204

        if (qemuStorageSourcePrivateDataAssignSecinfo(&priv->secinfo, &authalias) < 0)
9205
            return -1;
9206 9207 9208
    }

    if (restoreEncSecret) {
9209
        encalias = g_strdup_printf("%s-luks-secret0", disk->info.alias);
9210 9211

        if (qemuStorageSourcePrivateDataAssignSecinfo(&priv->encinfo, &encalias) < 0)
9212
            return -1;
9213 9214
    }

9215
    return 0;
9216 9217 9218
}


9219 9220
static int
qemuDomainDeviceDiskDefPostParse(virDomainDiskDefPtr disk,
9221
                                 virQEMUCapsPtr qemuCaps,
9222
                                 unsigned int parseFlags)
9223 9224
{
    /* set default disk types and drivers */
9225 9226 9227
    if (!virDomainDiskGetDriver(disk) &&
        virDomainDiskSetDriver(disk, "qemu") < 0)
        return -1;
9228

9229 9230
    /* default disk format for drives */
    if (virDomainDiskGetFormat(disk) == VIR_STORAGE_FILE_NONE &&
9231
        virDomainDiskGetType(disk) != VIR_STORAGE_TYPE_VOLUME)
9232
        virDomainDiskSetFormat(disk, VIR_STORAGE_FILE_RAW);
9233

9234 9235 9236 9237
    /* default disk format for mirrored drive */
    if (disk->mirror &&
        disk->mirror->format == VIR_STORAGE_FILE_NONE)
        disk->mirror->format = VIR_STORAGE_FILE_RAW;
9238

9239 9240 9241 9242
    if (qemuDomainDeviceDiskDefPostParseRestoreSecAlias(disk, qemuCaps,
                                                        parseFlags) < 0)
        return -1;

9243 9244 9245 9246 9247 9248 9249
    /* 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;

9250 9251 9252 9253
    return 0;
}


9254 9255 9256 9257 9258 9259
static int
qemuDomainDeviceNetDefPostParse(virDomainNetDefPtr net,
                                const virDomainDef *def,
                                virQEMUCapsPtr qemuCaps)
{
    if (net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
9260 9261
        !virDomainNetGetModelString(net))
        net->model = qemuDomainDefaultNetModel(def, qemuCaps);
9262 9263 9264 9265 9266

    return 0;
}


9267
static int
9268 9269
qemuDomainDefaultVideoDevice(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
9270
{
9271
    if (ARCH_IS_PPC64(def->os.arch))
9272
        return VIR_DOMAIN_VIDEO_TYPE_VGA;
9273 9274 9275
    if (qemuDomainIsARMVirt(def) ||
        qemuDomainIsRISCVVirt(def) ||
        ARCH_IS_S390(def->os.arch)) {
9276
        return VIR_DOMAIN_VIDEO_TYPE_VIRTIO;
9277
    }
9278 9279 9280 9281 9282
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_CIRRUS_VGA))
        return VIR_DOMAIN_VIDEO_TYPE_CIRRUS;
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VGA))
        return VIR_DOMAIN_VIDEO_TYPE_VGA;
    return VIR_DOMAIN_VIDEO_TYPE_DEFAULT;
9283 9284 9285
}


9286 9287
static int
qemuDomainDeviceVideoDefPostParse(virDomainVideoDefPtr video,
9288 9289
                                  const virDomainDef *def,
                                  virQEMUCapsPtr qemuCaps)
9290
{
9291
    if (video->type == VIR_DOMAIN_VIDEO_TYPE_DEFAULT)
9292
        video->type = qemuDomainDefaultVideoDevice(def, qemuCaps);
9293 9294 9295 9296 9297 9298 9299 9300 9301 9302

    if (video->type == VIR_DOMAIN_VIDEO_TYPE_QXL &&
        !video->vgamem) {
        video->vgamem = QEMU_QXL_VGAMEM_DEFAULT;
    }

    return 0;
}


9303 9304 9305 9306 9307 9308 9309 9310 9311 9312 9313 9314 9315 9316 9317 9318 9319
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
9320 9321 9322 9323 9324 9325 9326 9327 9328 9329
static int
qemuDomainVsockDefPostParse(virDomainVsockDefPtr vsock)
{
    if (vsock->model == VIR_DOMAIN_VSOCK_MODEL_DEFAULT)
        vsock->model = VIR_DOMAIN_VSOCK_MODEL_VIRTIO;

    return 0;
}


9330 9331 9332 9333 9334 9335 9336
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) &&
9337
        mdevsrc->model == VIR_MDEV_MODEL_TYPE_VFIO_PCI &&
9338 9339 9340 9341 9342 9343 9344 9345 9346 9347 9348 9349 9350 9351 9352 9353 9354 9355 9356 9357 9358 9359
        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;
}


9360
static int
9361 9362
qemuDomainTPMDefPostParse(virDomainTPMDefPtr tpm,
                          virArch arch)
9363
{
9364 9365 9366 9367 9368 9369 9370
    if (tpm->model == VIR_DOMAIN_TPM_MODEL_DEFAULT) {
        if (ARCH_IS_PPC64(arch))
            tpm->model = VIR_DOMAIN_TPM_MODEL_SPAPR;
        else
            tpm->model = VIR_DOMAIN_TPM_MODEL_TIS;
    }

9371 9372 9373 9374
    return 0;
}


9375 9376
static int
qemuDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
9377
                             const virDomainDef *def,
9378
                             unsigned int parseFlags,
9379
                             void *opaque,
9380
                             void *parseOpaque)
9381
{
9382
    virQEMUDriverPtr driver = opaque;
9383 9384 9385
    /* 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. */
9386
    virQEMUCapsPtr qemuCaps = parseOpaque;
9387
    int ret = -1;
9388

9389 9390 9391 9392
    switch ((virDomainDeviceType) dev->type) {
    case VIR_DOMAIN_DEVICE_NET:
        ret = qemuDomainDeviceNetDefPostParse(dev->data.net, def, qemuCaps);
        break;
9393

9394
    case VIR_DOMAIN_DEVICE_DISK:
9395
        ret = qemuDomainDeviceDiskDefPostParse(dev->data.disk, qemuCaps,
9396
                                               parseFlags);
9397
        break;
9398

9399
    case VIR_DOMAIN_DEVICE_VIDEO:
9400
        ret = qemuDomainDeviceVideoDefPostParse(dev->data.video, def, qemuCaps);
9401
        break;
9402

9403 9404 9405
    case VIR_DOMAIN_DEVICE_PANIC:
        ret = qemuDomainDevicePanicDefPostParse(dev->data.panic, def);
        break;
9406

9407 9408 9409 9410
    case VIR_DOMAIN_DEVICE_CONTROLLER:
        ret = qemuDomainControllerDefPostParse(dev->data.controller, def,
                                               qemuCaps, parseFlags);
        break;
9411

9412 9413 9414
    case VIR_DOMAIN_DEVICE_SHMEM:
        ret = qemuDomainShmemDefPostParse(dev->data.shmem);
        break;
9415

9416 9417 9418
    case VIR_DOMAIN_DEVICE_CHR:
        ret = qemuDomainChrDefPostParse(dev->data.chr, def, driver, parseFlags);
        break;
9419

J
Ján Tomko 已提交
9420 9421 9422 9423
    case VIR_DOMAIN_DEVICE_VSOCK:
        ret = qemuDomainVsockDefPostParse(dev->data.vsock);
        break;

9424 9425 9426 9427
    case VIR_DOMAIN_DEVICE_HOSTDEV:
        ret = qemuDomainHostdevDefPostParse(dev->data.hostdev, qemuCaps);
        break;

9428
    case VIR_DOMAIN_DEVICE_TPM:
9429
        ret = qemuDomainTPMDefPostParse(dev->data.tpm, def->os.arch);
9430 9431
        break;

9432 9433 9434 9435 9436 9437 9438 9439 9440 9441 9442 9443 9444 9445 9446 9447 9448 9449 9450 9451 9452 9453 9454 9455 9456 9457 9458
    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_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;
    }
9459 9460

    return ret;
9461 9462 9463
}


9464 9465
static int
qemuDomainDefAssignAddresses(virDomainDef *def,
J
Ján Tomko 已提交
9466
                             unsigned int parseFlags G_GNUC_UNUSED,
9467
                             void *opaque,
9468
                             void *parseOpaque)
9469 9470
{
    virQEMUDriverPtr driver = opaque;
9471 9472 9473
    /* 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. */
9474
    virQEMUCapsPtr qemuCaps = parseOpaque;
9475
    bool newDomain = parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE;
9476

9477 9478 9479 9480 9481 9482 9483
    /* 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;

9484 9485 9486 9487 9488 9489
    return qemuDomainAssignAddresses(def, qemuCaps, driver, NULL, newDomain);
}


static int
qemuDomainPostParseDataAlloc(const virDomainDef *def,
J
Ján Tomko 已提交
9490
                             unsigned int parseFlags G_GNUC_UNUSED,
9491 9492 9493 9494 9495 9496
                             void *opaque,
                             void **parseOpaque)
{
    virQEMUDriverPtr driver = opaque;

    if (!(*parseOpaque = virQEMUCapsCacheLookup(driver->qemuCapsCache,
9497
                                                def->emulator)))
9498
        return 1;
9499

9500 9501 9502 9503 9504 9505 9506 9507
    return 0;
}


static void
qemuDomainPostParseDataFree(void *parseOpaque)
{
    virQEMUCapsPtr qemuCaps = parseOpaque;
9508 9509 9510 9511 9512

    virObjectUnref(qemuCaps);
}


9513
virDomainDefParserConfig virQEMUDriverDomainDefParserConfig = {
9514
    .domainPostParseBasicCallback = qemuDomainDefPostParseBasic,
9515 9516
    .domainPostParseDataAlloc = qemuDomainPostParseDataAlloc,
    .domainPostParseDataFree = qemuDomainPostParseDataFree,
9517
    .devicesPostParseCallback = qemuDomainDeviceDefPostParse,
9518
    .domainPostParseCallback = qemuDomainDefPostParse,
9519
    .assignAddressesCallback = qemuDomainDefAssignAddresses,
9520
    .domainValidateCallback = qemuDomainDefValidate,
9521 9522
    .deviceValidateCallback = qemuDomainDeviceDefValidate,

9523
    .features = VIR_DOMAIN_DEF_FEATURE_MEMORY_HOTPLUG |
9524
                VIR_DOMAIN_DEF_FEATURE_OFFLINE_VCPUPIN |
9525
                VIR_DOMAIN_DEF_FEATURE_INDIVIDUAL_VCPUS |
9526
                VIR_DOMAIN_DEF_FEATURE_USER_ALIAS |
9527 9528
                VIR_DOMAIN_DEF_FEATURE_FW_AUTOSELECT |
                VIR_DOMAIN_DEF_FEATURE_NET_MODEL_STRING,
9529 9530 9531
};


9532
static void
9533 9534
qemuDomainObjSaveStatus(virQEMUDriverPtr driver,
                        virDomainObjPtr obj)
9535
{
9536
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
9537 9538

    if (virDomainObjIsActive(obj)) {
9539
        if (virDomainObjSave(obj, driver->xmlopt, cfg->stateDir) < 0)
9540
            VIR_WARN("Failed to save status on vm %s", obj->def->name);
9541
    }
9542 9543
}

9544 9545 9546 9547 9548 9549 9550 9551

void
qemuDomainSaveStatus(virDomainObjPtr obj)
{
    qemuDomainObjSaveStatus(QEMU_DOMAIN_PRIVATE(obj)->driver, obj);
}


9552 9553 9554 9555
void
qemuDomainSaveConfig(virDomainObjPtr obj)
{
    virQEMUDriverPtr driver = QEMU_DOMAIN_PRIVATE(obj)->driver;
9556
    g_autoptr(virQEMUDriverConfig) cfg = NULL;
9557 9558 9559 9560 9561 9562 9563 9564 9565 9566 9567 9568
    virDomainDefPtr def = NULL;

    if (virDomainObjIsActive(obj))
        def = obj->newDef;
    else
        def = obj->def;

    if (!def)
        return;

    cfg = virQEMUDriverGetConfig(driver);

9569
    if (virDomainDefSave(def, driver->xmlopt, cfg->configDir) < 0)
9570 9571 9572 9573
        VIR_WARN("Failed to save config of vm %s", obj->def->name);
}


J
Jiri Denemark 已提交
9574
void
9575
qemuDomainObjSetJobPhase(virQEMUDriverPtr driver,
J
Jiri Denemark 已提交
9576 9577 9578 9579
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
9580
    unsigned long long me = virThreadSelfID();
J
Jiri Denemark 已提交
9581 9582 9583 9584

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

9585 9586 9587 9588 9589
    VIR_DEBUG("Setting '%s' phase to '%s'",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob, phase));

    if (priv->job.asyncOwner && me != priv->job.asyncOwner) {
9590
        VIR_WARN("'%s' async job is owned by thread %llu",
9591 9592 9593 9594
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }

J
Jiri Denemark 已提交
9595
    priv->job.phase = phase;
9596
    priv->job.asyncOwner = me;
9597
    qemuDomainObjSaveStatus(driver, obj);
J
Jiri Denemark 已提交
9598 9599
}

9600
void
9601 9602
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
9603 9604 9605
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

9606 9607 9608 9609 9610 9611 9612
    if (!priv->job.asyncJob)
        return;

    priv->job.mask = allowedJobs | JOB_MASK(QEMU_JOB_DESTROY);
}

void
9613
qemuDomainObjDiscardAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
9614 9615 9616 9617 9618 9619
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
9620
    qemuDomainObjSaveStatus(driver, obj);
9621 9622
}

9623 9624 9625 9626 9627 9628 9629 9630 9631
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()) {
9632
        VIR_WARN("'%s' async job is owned by thread %llu",
9633 9634 9635 9636 9637 9638
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }
    priv->job.asyncOwner = 0;
}

9639
static bool
9640
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
9641
{
9642 9643 9644
    return !priv->job.asyncJob ||
           job == QEMU_JOB_NONE ||
           (priv->job.mask & JOB_MASK(job)) != 0;
9645 9646
}

9647
bool
9648
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
9649 9650 9651 9652
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

9653 9654 9655 9656 9657 9658 9659 9660 9661 9662 9663
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));
}

9664 9665 9666
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

9667 9668 9669 9670 9671 9672
/**
 * qemuDomainObjBeginJobInternal:
 * @driver: qemu driver
 * @obj: domain object
 * @job: qemuDomainJob to start
 * @asyncJob: qemuDomainAsyncJob to start
9673
 * @nowait: don't wait trying to acquire @job
9674 9675 9676 9677
 *
 * 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
9678 9679 9680 9681 9682
 * 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.
9683 9684 9685 9686 9687
 *
 * Returns: 0 on success,
 *         -2 if unable to start job because of timeout or
 *            maxQueuedJobs limit,
 *         -1 otherwise.
9688
 */
9689
static int ATTRIBUTE_NONNULL(1)
9690
qemuDomainObjBeginJobInternal(virQEMUDriverPtr driver,
9691
                              virDomainObjPtr obj,
9692
                              qemuDomainJob job,
9693
                              qemuDomainAgentJob agentJob,
9694 9695
                              qemuDomainAsyncJob asyncJob,
                              bool nowait)
9696 9697
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
9698
    unsigned long long now;
9699
    unsigned long long then;
9700
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
J
Jiri Denemark 已提交
9701
    bool async = job == QEMU_JOB_ASYNC;
9702
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
9703
    const char *blocker = NULL;
9704
    const char *agentBlocker = NULL;
J
Jiri Denemark 已提交
9705
    int ret = -1;
J
Jiri Denemark 已提交
9706
    unsigned long long duration = 0;
9707
    unsigned long long agentDuration = 0;
J
Jiri Denemark 已提交
9708
    unsigned long long asyncDuration = 0;
9709

9710 9711 9712 9713 9714 9715
    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
9716
              qemuDomainJobTypeToString(priv->job.active),
9717
              qemuDomainAgentJobTypeToString(priv->job.agentActive),
J
Jiri Denemark 已提交
9718
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
9719

9720
    if (virTimeMillisNow(&now) < 0)
9721
        return -1;
9722

9723
    priv->jobs_queued++;
J
Jiri Denemark 已提交
9724
    then = now + QEMU_JOB_WAIT_TIME;
9725

9726
 retry:
9727 9728
    if ((!async && job != QEMU_JOB_DESTROY) &&
        cfg->maxQueuedJobs &&
9729
        priv->jobs_queued > cfg->maxQueuedJobs) {
9730 9731 9732
        goto error;
    }

9733
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
9734 9735 9736
        if (nowait)
            goto cleanup;

9737
        VIR_DEBUG("Waiting for async job (vm=%p name=%s)", obj, obj->def->name);
9738
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->parent.lock, then) < 0)
9739 9740 9741
            goto error;
    }

9742
    while (!qemuDomainObjCanSetJob(priv, job, agentJob)) {
9743 9744 9745
        if (nowait)
            goto cleanup;

9746
        VIR_DEBUG("Waiting for job (vm=%p name=%s)", obj, obj->def->name);
9747
        if (virCondWaitUntil(&priv->job.cond, &obj->parent.lock, then) < 0)
9748
            goto error;
9749
    }
9750 9751 9752

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
9753
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
9754 9755
        goto retry;

J
Jiri Denemark 已提交
9756 9757
    ignore_value(virTimeMillisNow(&now));

9758 9759 9760 9761 9762 9763 9764 9765 9766 9767 9768 9769 9770 9771 9772 9773 9774 9775 9776 9777 9778 9779 9780 9781 9782 9783 9784 9785 9786 9787 9788 9789 9790 9791 9792 9793 9794 9795 9796 9797
    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;
9798
    }
9799

9800
    if (qemuDomainTrackJob(job))
9801
        qemuDomainObjSaveStatus(driver, obj);
9802

9803
    return 0;
9804

9805
 error:
J
Jiri Denemark 已提交
9806 9807 9808
    ignore_value(virTimeMillisNow(&now));
    if (priv->job.active && priv->job.started)
        duration = now - priv->job.started;
9809 9810
    if (priv->job.agentActive && priv->job.agentStarted)
        agentDuration = now - priv->job.agentStarted;
J
Jiri Denemark 已提交
9811 9812 9813
    if (priv->job.asyncJob && priv->job.asyncStarted)
        asyncDuration = now - priv->job.asyncStarted;

9814 9815 9816 9817
    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)",
9818
             qemuDomainJobTypeToString(job),
9819
             qemuDomainAgentJobTypeToString(agentJob),
9820 9821 9822
             qemuDomainAsyncJobTypeToString(asyncJob),
             obj->def->name,
             qemuDomainJobTypeToString(priv->job.active),
9823
             qemuDomainAgentJobTypeToString(priv->job.agentActive),
9824
             qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
9825
             priv->job.owner, NULLSTR(priv->job.ownerAPI),
9826
             priv->job.agentOwner, NULLSTR(priv->job.agentOwnerAPI),
J
Jiri Denemark 已提交
9827
             priv->job.asyncOwner, NULLSTR(priv->job.asyncOwnerAPI),
9828
             priv->job.apiFlags,
9829
             duration / 1000, agentDuration / 1000, asyncDuration / 1000);
9830

9831 9832 9833 9834 9835 9836 9837 9838 9839
    if (job) {
        if (nested || qemuDomainNestedJobAllowed(priv, job))
            blocker = priv->job.ownerAPI;
        else
            blocker = priv->job.asyncOwnerAPI;
    }

    if (agentJob)
        agentBlocker = priv->job.agentOwnerAPI;
9840

9841
    if (errno == ETIMEDOUT) {
9842 9843 9844 9845 9846 9847
        if (blocker && agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by monitor=%s agent=%s)"),
                           blocker, agentBlocker);
        } else if (blocker) {
9848
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
9849 9850
                           _("cannot acquire state change "
                             "lock (held by monitor=%s)"),
9851
                           blocker);
9852 9853 9854 9855 9856
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by agent=%s)"),
                           agentBlocker);
9857 9858 9859 9860
        } else {
            virReportError(VIR_ERR_OPERATION_TIMEOUT, "%s",
                           _("cannot acquire state change lock"));
        }
9861 9862 9863
        ret = -2;
    } else if (cfg->maxQueuedJobs &&
               priv->jobs_queued > cfg->maxQueuedJobs) {
9864 9865 9866 9867 9868 9869 9870
        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) {
9871
            virReportError(VIR_ERR_OPERATION_FAILED,
9872 9873
                           _("cannot acquire state change "
                             "lock (held by monitor=%s) "
9874 9875
                             "due to max_queued limit"),
                           blocker);
9876 9877 9878 9879 9880 9881
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_FAILED,
                           _("cannot acquire state change "
                             "lock (held by agent=%s) "
                             "due to max_queued limit"),
                           agentBlocker);
9882 9883 9884 9885 9886
        } else {
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("cannot acquire state change lock "
                             "due to max_queued limit"));
        }
9887 9888
        ret = -2;
    } else {
9889
        virReportSystemError(errno, "%s", _("cannot acquire job mutex"));
9890
    }
J
Jiri Denemark 已提交
9891 9892

 cleanup:
9893
    priv->jobs_queued--;
9894
    return ret;
9895 9896 9897
}

/*
9898
 * obj must be locked before calling
9899 9900 9901 9902
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
9903
 * Successful calls must be followed by EndJob eventually
9904
 */
9905
int qemuDomainObjBeginJob(virQEMUDriverPtr driver,
9906
                          virDomainObjPtr obj,
9907
                          qemuDomainJob job)
9908
{
9909
    if (qemuDomainObjBeginJobInternal(driver, obj, job,
9910
                                      QEMU_AGENT_JOB_NONE,
9911
                                      QEMU_ASYNC_JOB_NONE, false) < 0)
9912 9913 9914
        return -1;
    else
        return 0;
9915 9916
}

9917 9918 9919 9920 9921 9922 9923 9924 9925 9926 9927 9928 9929 9930 9931 9932 9933
/**
 * 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);
}

9934
int qemuDomainObjBeginAsyncJob(virQEMUDriverPtr driver,
9935
                               virDomainObjPtr obj,
9936
                               qemuDomainAsyncJob asyncJob,
9937 9938
                               virDomainJobOperation operation,
                               unsigned long apiFlags)
9939
{
9940 9941
    qemuDomainObjPrivatePtr priv;

9942
    if (qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_ASYNC,
9943
                                      QEMU_AGENT_JOB_NONE,
9944
                                      asyncJob, false) < 0)
9945
        return -1;
9946 9947 9948

    priv = obj->privateData;
    priv->job.current->operation = operation;
9949
    priv->job.apiFlags = apiFlags;
9950
    return 0;
9951 9952
}

9953
int
9954 9955
qemuDomainObjBeginNestedJob(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
9956
                            qemuDomainAsyncJob asyncJob)
9957 9958 9959 9960 9961
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (asyncJob != priv->job.asyncJob) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
9962 9963
                       _("unexpected async job %d type expected %d"),
                       asyncJob, priv->job.asyncJob);
9964 9965 9966 9967
        return -1;
    }

    if (priv->job.asyncOwner != virThreadSelfID()) {
9968
        VIR_WARN("This thread doesn't seem to be the async job owner: %llu",
9969 9970 9971 9972 9973
                 priv->job.asyncOwner);
    }

    return qemuDomainObjBeginJobInternal(driver, obj,
                                         QEMU_JOB_ASYNC_NESTED,
9974
                                         QEMU_AGENT_JOB_NONE,
9975 9976
                                         QEMU_ASYNC_JOB_NONE,
                                         false);
9977 9978
}

9979 9980 9981 9982 9983 9984 9985 9986 9987 9988 9989 9990 9991 9992 9993 9994 9995 9996 9997
/**
 * 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,
9998
                                         QEMU_AGENT_JOB_NONE,
9999 10000
                                         QEMU_ASYNC_JOB_NONE, true);
}
10001

10002
/*
10003
 * obj must be locked and have a reference before calling
10004 10005 10006 10007
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 */
10008 10009
void
qemuDomainObjEndJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
10010 10011
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
10012
    qemuDomainJob job = priv->job.active;
10013

10014 10015
    priv->jobs_queued--;

10016
    VIR_DEBUG("Stopping job: %s (async=%s vm=%p name=%s)",
10017
              qemuDomainJobTypeToString(job),
10018 10019
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
10020

10021
    qemuDomainObjResetJob(priv);
10022
    if (qemuDomainTrackJob(job))
10023
        qemuDomainObjSaveStatus(driver, obj);
10024 10025 10026 10027 10028 10029 10030 10031 10032 10033 10034 10035 10036 10037 10038 10039 10040 10041 10042 10043 10044 10045 10046 10047
    /* 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);
}

10048
void
10049
qemuDomainObjEndAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
10050 10051
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
10052

10053 10054
    priv->jobs_queued--;

10055 10056 10057
    VIR_DEBUG("Stopping async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
10058

10059
    qemuDomainObjResetAsyncJob(priv);
10060
    qemuDomainObjSaveStatus(driver, obj);
10061 10062 10063
    virCondBroadcast(&priv->job.asyncCond);
}

10064 10065 10066 10067 10068
void
qemuDomainObjAbortAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

10069 10070 10071
    VIR_DEBUG("Requesting abort of async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
10072

10073 10074
    priv->job.abortJob = true;
    virDomainObjBroadcast(obj);
10075 10076
}

10077 10078 10079 10080
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU monitor API call
10081 10082 10083
 * Must have already called qemuDomainObjBeginJob() and checked
 * that the VM is still active; may not be used for nested async
 * jobs.
10084 10085 10086
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
10087
static int
10088
qemuDomainObjEnterMonitorInternal(virQEMUDriverPtr driver,
10089
                                  virDomainObjPtr obj,
10090
                                  qemuDomainAsyncJob asyncJob)
10091 10092 10093
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

10094
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
10095 10096 10097
        int ret;
        if ((ret = qemuDomainObjBeginNestedJob(driver, obj, asyncJob)) < 0)
            return ret;
10098
        if (!virDomainObjIsActive(obj)) {
10099 10100
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
10101
            qemuDomainObjEndJob(driver, obj);
10102 10103
            return -1;
        }
10104 10105 10106
    } 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");
10107 10108 10109 10110 10111
    } 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);
10112 10113
    }

10114 10115
    VIR_DEBUG("Entering monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
10116
    virObjectLock(priv->mon);
10117
    virObjectRef(priv->mon);
10118
    ignore_value(virTimeMillisNow(&priv->monStart));
10119
    virObjectUnlock(obj);
10120 10121

    return 0;
10122 10123
}

10124
static void ATTRIBUTE_NONNULL(1)
10125
qemuDomainObjExitMonitorInternal(virQEMUDriverPtr driver,
10126
                                 virDomainObjPtr obj)
10127 10128
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
10129
    bool hasRefs;
10130

10131
    hasRefs = virObjectUnref(priv->mon);
10132

10133
    if (hasRefs)
10134
        virObjectUnlock(priv->mon);
10135

10136
    virObjectLock(obj);
10137 10138
    VIR_DEBUG("Exited monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
10139

10140
    priv->monStart = 0;
10141
    if (!hasRefs)
10142
        priv->mon = NULL;
10143

J
Jiri Denemark 已提交
10144 10145
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjEndJob(driver, obj);
10146 10147
}

10148
void qemuDomainObjEnterMonitor(virQEMUDriverPtr driver,
10149
                               virDomainObjPtr obj)
10150
{
10151
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, obj,
10152
                                                   QEMU_ASYNC_JOB_NONE));
10153 10154
}

10155
/* obj must NOT be locked before calling
10156 10157
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
10158 10159 10160 10161 10162 10163
 *
 * 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.
10164
 */
10165 10166
int qemuDomainObjExitMonitor(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
10167
{
10168
    qemuDomainObjExitMonitorInternal(driver, obj);
10169
    if (!virDomainObjIsActive(obj)) {
10170
        if (virGetLastErrorCode() == VIR_ERR_OK)
10171 10172
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
10173 10174 10175
        return -1;
    }
    return 0;
10176
}
10177 10178

/*
10179
 * obj must be locked before calling
10180 10181
 *
 * To be called immediately before any QEMU monitor API call.
10182
 * Must have already either called qemuDomainObjBeginJob()
10183 10184 10185 10186 10187
 * 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
10188 10189 10190
 * 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).
10191 10192
 */
int
10193
qemuDomainObjEnterMonitorAsync(virQEMUDriverPtr driver,
10194
                               virDomainObjPtr obj,
10195
                               qemuDomainAsyncJob asyncJob)
10196
{
10197
    return qemuDomainObjEnterMonitorInternal(driver, obj, asyncJob);
10198 10199
}

D
Daniel P. Berrange 已提交
10200

10201 10202 10203 10204
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU agent API call.
10205 10206
 * Must have already called qemuDomainObjBeginAgentJob() and
 * checked that the VM is still active.
10207 10208 10209
 *
 * To be followed with qemuDomainObjExitAgent() once complete
 */
10210
qemuAgentPtr
10211
qemuDomainObjEnterAgent(virDomainObjPtr obj)
D
Daniel P. Berrange 已提交
10212 10213
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
10214
    qemuAgentPtr agent = priv->agent;
D
Daniel P. Berrange 已提交
10215

10216 10217
    VIR_DEBUG("Entering agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
10218 10219 10220

    virObjectLock(agent);
    virObjectRef(agent);
10221
    virObjectUnlock(obj);
10222 10223

    return agent;
D
Daniel P. Berrange 已提交
10224 10225
}

10226 10227 10228 10229 10230 10231

/* obj must NOT be locked before calling
 *
 * Should be paired with an earlier qemuDomainObjEnterAgent() call
 */
void
10232
qemuDomainObjExitAgent(virDomainObjPtr obj, qemuAgentPtr agent)
D
Daniel P. Berrange 已提交
10233
{
10234 10235
    virObjectUnlock(agent);
    virObjectUnref(agent);
10236
    virObjectLock(obj);
D
Daniel P. Berrange 已提交
10237

10238 10239
    VIR_DEBUG("Exited agent (agent=%p vm=%p name=%s)",
              agent, obj, obj->def->name);
D
Daniel P. Berrange 已提交
10240 10241
}

10242
void qemuDomainObjEnterRemote(virDomainObjPtr obj)
10243
{
10244 10245
    VIR_DEBUG("Entering remote (vm=%p name=%s)",
              obj, obj->def->name);
10246
    virObjectUnlock(obj);
10247 10248
}

10249 10250 10251 10252

int
qemuDomainObjExitRemote(virDomainObjPtr obj,
                        bool checkActive)
10253
{
10254
    virObjectLock(obj);
10255 10256
    VIR_DEBUG("Exited remote (vm=%p name=%s)",
              obj, obj->def->name);
10257 10258 10259 10260 10261 10262 10263 10264 10265

    if (checkActive && !virDomainObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("domain '%s' is not running"),
                       obj->def->name);
        return -1;
    }

    return 0;
10266
}
10267 10268


10269 10270
static virDomainDefPtr
qemuDomainDefFromXML(virQEMUDriverPtr driver,
10271
                     virQEMUCapsPtr qemuCaps,
10272 10273 10274 10275
                     const char *xml)
{
    virDomainDefPtr def;

10276
    def = virDomainDefParseString(xml, driver->xmlopt, qemuCaps,
10277 10278 10279 10280 10281 10282 10283
                                  VIR_DOMAIN_DEF_PARSE_INACTIVE |
                                  VIR_DOMAIN_DEF_PARSE_SKIP_VALIDATE);

    return def;
}


10284 10285
virDomainDefPtr
qemuDomainDefCopy(virQEMUDriverPtr driver,
10286
                  virQEMUCapsPtr qemuCaps,
10287 10288 10289 10290
                  virDomainDefPtr src,
                  unsigned int flags)
{
    virDomainDefPtr ret = NULL;
10291
    g_autofree char *xml = NULL;
10292

10293
    if (!(xml = qemuDomainDefFormatXML(driver, qemuCaps, src, flags)))
10294
        return NULL;
10295

10296
    ret = qemuDomainDefFromXML(driver, qemuCaps, xml);
10297 10298 10299 10300

    return ret;
}

10301

10302 10303 10304 10305 10306 10307 10308 10309 10310 10311 10312 10313 10314 10315 10316 10317 10318 10319 10320 10321
int
qemuDomainMakeCPUMigratable(virCPUDefPtr cpu)
{
    if (cpu->mode == VIR_CPU_MODE_CUSTOM &&
        STREQ_NULLABLE(cpu->model, "Icelake-Server")) {
        /* Originally Icelake-Server CPU model contained pconfig CPU feature.
         * It was never actually enabled and thus it was removed. To enable
         * migration to QEMU 3.1.0 (with both new and old libvirt), we
         * explicitly disable pconfig in migration XML (otherwise old libvirt
         * would think it was implicitly enabled on the source). New libvirt
         * will drop it from the XML before starting the domain on new QEMU.
         */
        if (virCPUDefUpdateFeature(cpu, "pconfig", VIR_CPU_FEATURE_DISABLE) < 0)
            return -1;
    }

    return 0;
}


10322 10323
static int
qemuDomainDefFormatBufInternal(virQEMUDriverPtr driver,
10324
                               virQEMUCapsPtr qemuCaps,
10325 10326 10327 10328
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags,
                               virBuffer *buf)
10329
{
10330
    int ret = -1;
10331
    virDomainDefPtr copy = NULL;
10332

10333 10334
    virCheckFlags(VIR_DOMAIN_XML_COMMON_FLAGS | VIR_DOMAIN_XML_UPDATE_CPU, -1);

10335 10336 10337
    if (!(flags & (VIR_DOMAIN_XML_UPDATE_CPU | VIR_DOMAIN_XML_MIGRATABLE)))
        goto format;

10338
    if (!(copy = virDomainDefCopy(def, driver->xmlopt, qemuCaps,
10339 10340 10341 10342 10343
                                  flags & VIR_DOMAIN_XML_MIGRATABLE)))
        goto cleanup;

    def = copy;

10344
    /* Update guest CPU requirements according to host CPU */
10345
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) &&
10346
        def->cpu &&
10347 10348
        (def->cpu->mode != VIR_CPU_MODE_CUSTOM ||
         def->cpu->model)) {
10349
        g_autoptr(virQEMUCaps) qCaps = NULL;
10350 10351 10352 10353 10354

        if (qemuCaps) {
            qCaps = virObjectRef(qemuCaps);
        } else {
            if (!(qCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
10355
                                                     def->virtType,
10356 10357 10358 10359
                                                     def->emulator,
                                                     def->os.machine)))
                goto cleanup;
        }
10360 10361

        if (virCPUUpdate(def->os.arch, def->cpu,
10362
                         virQEMUCapsGetHostModel(qCaps, def->virtType,
10363
                                                 VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE)) < 0)
10364 10365 10366
            goto cleanup;
    }

10367
    if ((flags & VIR_DOMAIN_XML_MIGRATABLE)) {
10368
        size_t i;
10369
        int toremove = 0;
10370
        virDomainControllerDefPtr usb = NULL, pci = NULL;
10371 10372 10373 10374 10375 10376 10377 10378 10379 10380 10381 10382 10383 10384 10385

        /* 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];
            }
        }
10386 10387 10388 10389 10390 10391 10392 10393 10394

        /* 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'/>.
10395
         */
10396
        if (qemuDomainIsI440FX(def) &&
10397
            usb && usb->idx == 0 &&
10398
            (usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT ||
10399 10400
             usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_PIIX3_UHCI) &&
            !virDomainDeviceAliasIsUserAlias(usb->info.alias)) {
10401 10402
            VIR_DEBUG("Removing default USB controller from domain '%s'"
                      " for migration compatibility", def->name);
10403
            toremove++;
10404 10405 10406 10407 10408 10409 10410 10411 10412 10413 10414 10415 10416 10417 10418 10419 10420
        } 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 &&
10421 10422 10423
            pci->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT &&
            !virDomainDeviceAliasIsUserAlias(pci->info.alias) &&
            !pci->opts.pciopts.pcihole64) {
L
Laine Stump 已提交
10424
            VIR_DEBUG("Removing default pci-root from domain '%s'"
10425
                      " for migration compatibility", def->name);
10426
            toremove++;
10427 10428 10429 10430
        } else {
            pci = NULL;
        }

10431
        if (toremove) {
10432 10433 10434
            virDomainControllerDefPtr *controllers = def->controllers;
            int ncontrollers = def->ncontrollers;

10435
            if (VIR_ALLOC_N(def->controllers, ncontrollers - toremove) < 0) {
10436
                def->controllers = controllers;
10437 10438 10439 10440 10441
                goto cleanup;
            }

            def->ncontrollers = 0;
            for (i = 0; i < ncontrollers; i++) {
10442
                if (controllers[i] != usb && controllers[i] != pci)
10443 10444
                    def->controllers[def->ncontrollers++] = controllers[i];
            }
10445 10446 10447 10448

            VIR_FREE(controllers);
            virDomainControllerDefFree(pci);
            virDomainControllerDefFree(usb);
10449
        }
10450

10451 10452 10453 10454 10455 10456 10457 10458 10459
        /* 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;
            }
        }

10460 10461
        for (i = 0; i < def->nchannels; i++)
            qemuDomainChrDefDropDefaultPath(def->channels[i], driver);
10462

10463 10464 10465 10466 10467 10468 10469 10470 10471
        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) {
10472
                switch ((virDomainChrSerialTargetType)serial->targetType) {
10473
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
10474
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
10475 10476 10477 10478 10479 10480
                    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:
10481
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
10482 10483 10484 10485 10486 10487 10488 10489
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
                    /* Nothing to do */
                    break;
                }
            }
        }

10490 10491 10492 10493 10494 10495 10496 10497 10498
        /* 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;
        }
10499

10500
        if (def->cpu && qemuDomainMakeCPUMigratable(def->cpu) < 0)
10501
            goto cleanup;
10502 10503
    }

10504
 format:
10505
    ret = virDomainDefFormatInternal(def, driver->xmlopt, buf,
10506
                                     virDomainDefFormatConvertXMLFlags(flags));
10507

10508
 cleanup:
10509
    virDomainDefFree(copy);
10510 10511
    return ret;
}
10512

10513 10514 10515

int
qemuDomainDefFormatBuf(virQEMUDriverPtr driver,
10516
                       virQEMUCapsPtr qemuCaps,
10517 10518 10519 10520
                       virDomainDefPtr def,
                       unsigned int flags,
                       virBufferPtr buf)
{
10521
    return qemuDomainDefFormatBufInternal(driver, qemuCaps, def, NULL, flags, buf);
10522 10523 10524 10525 10526
}


static char *
qemuDomainDefFormatXMLInternal(virQEMUDriverPtr driver,
10527
                               virQEMUCapsPtr qemuCaps,
10528 10529 10530
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags)
10531 10532 10533
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

10534
    if (qemuDomainDefFormatBufInternal(driver, qemuCaps, def, origCPU, flags, &buf) < 0)
10535 10536 10537 10538 10539
        return NULL;

    return virBufferContentAndReset(&buf);
}

10540 10541 10542

char *
qemuDomainDefFormatXML(virQEMUDriverPtr driver,
10543
                       virQEMUCapsPtr qemuCaps,
10544 10545 10546
                       virDomainDefPtr def,
                       unsigned int flags)
{
10547
    return qemuDomainDefFormatXMLInternal(driver, qemuCaps, def, NULL, flags);
10548 10549 10550
}


10551
char *qemuDomainFormatXML(virQEMUDriverPtr driver,
10552
                          virDomainObjPtr vm,
10553
                          unsigned int flags)
10554 10555
{
    virDomainDefPtr def;
10556 10557
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCPUDefPtr origCPU = NULL;
10558

10559
    if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef) {
10560
        def = vm->newDef;
10561
    } else {
10562
        def = vm->def;
10563
        origCPU = priv->origCPU;
10564
    }
10565

10566
    return qemuDomainDefFormatXMLInternal(driver, priv->qemuCaps, def, origCPU, flags);
10567 10568
}

10569
char *
10570
qemuDomainDefFormatLive(virQEMUDriverPtr driver,
10571
                        virQEMUCapsPtr qemuCaps,
10572
                        virDomainDefPtr def,
10573
                        virCPUDefPtr origCPU,
10574 10575
                        bool inactive,
                        bool compatible)
10576 10577 10578 10579 10580
{
    unsigned int flags = QEMU_DOMAIN_FORMAT_LIVE_FLAGS;

    if (inactive)
        flags |= VIR_DOMAIN_XML_INACTIVE;
10581 10582
    if (compatible)
        flags |= VIR_DOMAIN_XML_MIGRATABLE;
10583

10584
    return qemuDomainDefFormatXMLInternal(driver, qemuCaps, def, origCPU, flags);
10585 10586
}

10587

10588
void qemuDomainObjTaint(virQEMUDriverPtr driver,
10589
                        virDomainObjPtr obj,
10590
                        virDomainTaintFlags taint,
10591
                        qemuDomainLogContextPtr logCtxt)
10592
{
10593
    virErrorPtr orig_err = NULL;
10594
    g_autofree char *timestamp = NULL;
10595
    char uuidstr[VIR_UUID_STRING_BUFLEN];
10596
    int rc;
10597

10598 10599 10600 10601 10602 10603 10604 10605 10606 10607 10608 10609 10610 10611
    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 */
10612
    virErrorPreserveLast(&orig_err);
10613

10614 10615 10616
    if (!(timestamp = virTimeStringNow()))
        goto cleanup;

10617 10618 10619 10620 10621 10622 10623 10624 10625 10626 10627 10628
    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));
10629 10630
    }

10631
    if (rc < 0)
10632 10633 10634
        virResetLastError();

 cleanup:
10635
    virErrorRestore(&orig_err);
10636 10637 10638
}


10639
void qemuDomainObjCheckTaint(virQEMUDriverPtr driver,
10640
                             virDomainObjPtr obj,
10641
                             qemuDomainLogContextPtr logCtxt)
10642
{
10643
    size_t i;
10644
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
10645
    qemuDomainObjPrivatePtr priv = obj->privateData;
10646
    bool custom_hypervisor_feat = false;
10647

10648
    if (virQEMUDriverIsPrivileged(driver) &&
10649
        (cfg->user == 0 ||
10650
         cfg->group == 0))
10651
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
10652

10653
    if (priv->hookRun)
10654
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOOK, logCtxt);
10655

10656
    if (obj->def->namespaceData) {
10657 10658
        qemuDomainXmlNsDefPtr qemuxmlns = obj->def->namespaceData;
        if (qemuxmlns->num_args || qemuxmlns->num_env)
10659
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logCtxt);
10660 10661 10662 10663
        if (qemuxmlns->ncapsadd > 0 || qemuxmlns->ncapsdel > 0)
            custom_hypervisor_feat = true;
    }

10664 10665
    if (custom_hypervisor_feat ||
        (cfg->capabilityfilters && *cfg->capabilityfilters)) {
10666 10667
        qemuDomainObjTaint(driver, obj,
                           VIR_DOMAIN_TAINT_CUSTOM_HYPERVISOR_FEATURE, logCtxt);
10668 10669
    }

10670
    if (obj->def->cpu && obj->def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
10671
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOST_CPU, logCtxt);
10672

10673
    for (i = 0; i < obj->def->ndisks; i++)
10674
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logCtxt);
10675

10676 10677
    for (i = 0; i < obj->def->nhostdevs; i++)
        qemuDomainObjCheckHostdevTaint(driver, obj, obj->def->hostdevs[i],
10678
                                       logCtxt);
10679

10680
    for (i = 0; i < obj->def->nnets; i++)
10681
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logCtxt);
10682

10683
    if (obj->def->os.dtb)
10684
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_DTB, logCtxt);
10685 10686 10687
}


10688
void qemuDomainObjCheckDiskTaint(virQEMUDriverPtr driver,
10689
                                 virDomainObjPtr obj,
10690
                                 virDomainDiskDefPtr disk,
10691
                                 qemuDomainLogContextPtr logCtxt)
10692
{
10693 10694
    if (disk->rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
10695
                           logCtxt);
10696

10697 10698
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
        virStorageSourceGetActualType(disk->src) == VIR_STORAGE_TYPE_BLOCK &&
10699
        disk->src->path && virFileIsCDROM(disk->src->path) == 1)
10700
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CDROM_PASSTHROUGH,
10701
                           logCtxt);
10702 10703 10704
}


10705 10706 10707
void qemuDomainObjCheckHostdevTaint(virQEMUDriverPtr driver,
                                    virDomainObjPtr obj,
                                    virDomainHostdevDefPtr hostdev,
10708
                                    qemuDomainLogContextPtr logCtxt)
10709
{
10710
    if (!virHostdevIsSCSIDevice(hostdev))
10711
        return;
10712

10713 10714
    if (hostdev->source.subsys.u.scsi.rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
10715 10716 10717
}


10718
void qemuDomainObjCheckNetTaint(virQEMUDriverPtr driver,
10719
                                virDomainObjPtr obj,
10720
                                virDomainNetDefPtr net,
10721
                                qemuDomainLogContextPtr logCtxt)
10722
{
10723 10724 10725 10726 10727 10728
    /* 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)
10729
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logCtxt);
10730
}
10731 10732


10733 10734 10735 10736
qemuDomainLogContextPtr qemuDomainLogContextNew(virQEMUDriverPtr driver,
                                                virDomainObjPtr vm,
                                                qemuDomainLogContextMode mode)
{
10737
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
10738
    qemuDomainLogContextPtr ctxt = QEMU_DOMAIN_LOG_CONTEXT(g_object_new(QEMU_TYPE_DOMAIN_LOG_CONTEXT, NULL));
10739

10740
    VIR_DEBUG("Context new %p stdioLogD=%d", ctxt, cfg->stdioLogD);
10741 10742 10743
    ctxt->writefd = -1;
    ctxt->readfd = -1;

10744
    ctxt->path = g_strdup_printf("%s/%s.log", cfg->logDir, vm->def->name);
10745

10746 10747 10748 10749
    if (cfg->stdioLogD) {
        ctxt->manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver));
        if (!ctxt->manager)
            goto error;
10750

10751 10752 10753 10754
        ctxt->writefd = virLogManagerDomainOpenLogFile(ctxt->manager,
                                                       "qemu",
                                                       vm->def->uuid,
                                                       vm->def->name,
10755
                                                       ctxt->path,
10756 10757 10758 10759 10760 10761
                                                       0,
                                                       &ctxt->inode,
                                                       &ctxt->pos);
        if (ctxt->writefd < 0)
            goto error;
    } else {
10762
        if ((ctxt->writefd = open(ctxt->path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
10763
            virReportSystemError(errno, _("failed to create logfile %s"),
10764
                                 ctxt->path);
10765 10766
            goto error;
        }
10767
        if (virSetCloseExec(ctxt->writefd) < 0) {
10768
            virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
10769
                                 ctxt->path);
10770 10771 10772
            goto error;
        }

10773 10774 10775 10776 10777 10778 10779
        /* 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"),
10780
                                 ctxt->path);
10781 10782 10783 10784
            goto error;
        }

        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START) {
10785
            if ((ctxt->readfd = open(ctxt->path, O_RDONLY, S_IRUSR | S_IWUSR)) < 0) {
10786
                virReportSystemError(errno, _("failed to open logfile %s"),
10787
                                     ctxt->path);
10788 10789 10790 10791
                goto error;
            }
            if (virSetCloseExec(ctxt->readfd) < 0) {
                virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
10792
                                     ctxt->path);
10793 10794 10795 10796
                goto error;
            }
        }

10797 10798
        if ((ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END)) < 0) {
            virReportSystemError(errno, _("failed to seek in log file %s"),
10799
                                 ctxt->path);
10800 10801 10802 10803 10804 10805 10806
            goto error;
        }
    }

    return ctxt;

 error:
10807
    g_clear_object(&ctxt);
10808
    return NULL;
10809 10810 10811 10812 10813 10814 10815
}


int qemuDomainLogContextWrite(qemuDomainLogContextPtr ctxt,
                              const char *fmt, ...)
{
    va_list argptr;
10816
    g_autofree char *message = NULL;
10817 10818 10819 10820
    int ret = -1;

    va_start(argptr, fmt);

10821
    message = g_strdup_vprintf(fmt, argptr);
10822 10823
    if (!ctxt->manager &&
        lseek(ctxt->writefd, 0, SEEK_END) < 0) {
10824
        virReportSystemError(errno, "%s",
10825
                             _("Unable to seek to end of domain logfile"));
10826 10827 10828 10829 10830 10831 10832 10833 10834 10835 10836 10837 10838 10839 10840 10841 10842 10843 10844
        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);
    return ret;
}


ssize_t qemuDomainLogContextRead(qemuDomainLogContextPtr ctxt,
                                 char **msg)
{
10845 10846 10847 10848
    VIR_DEBUG("Context read %p manager=%p inode=%llu pos=%llu",
              ctxt, ctxt->manager,
              (unsigned long long)ctxt->inode,
              (unsigned long long)ctxt->pos);
10849
    char *buf;
10850 10851 10852
    size_t buflen;
    if (ctxt->manager) {
        buf = virLogManagerDomainReadLogFile(ctxt->manager,
10853
                                             ctxt->path,
10854 10855 10856 10857 10858 10859 10860 10861 10862
                                             ctxt->inode,
                                             ctxt->pos,
                                             1024 * 128,
                                             0);
        if (!buf)
            return -1;
        buflen = strlen(buf);
    } else {
        ssize_t got;
10863

10864
        buflen = 1024 * 128;
10865

10866 10867
        /* Best effort jump to start of messages */
        ignore_value(lseek(ctxt->readfd, ctxt->pos, SEEK_SET));
10868

10869 10870
        if (VIR_ALLOC_N(buf, buflen) < 0)
            return -1;
10871

10872 10873 10874 10875 10876 10877 10878 10879 10880 10881 10882 10883 10884
        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;
    }
10885 10886 10887

    *msg = buf;

10888
    return buflen;
10889 10890 10891
}


10892 10893 10894 10895 10896 10897 10898 10899 10900 10901 10902
/**
 * 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,
                           ...)
{
10903
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
10904 10905
    virLogManagerPtr manager = NULL;
    va_list ap;
10906
    g_autofree char *path = NULL;
10907
    int writefd = -1;
10908
    g_autofree char *message = NULL;
10909 10910 10911 10912
    int ret = -1;

    va_start(ap, fmt);

10913
    message = g_strdup_vprintf(fmt, ap);
10914 10915 10916 10917

    VIR_DEBUG("Append log message (vm='%s' message='%s) stdioLogD=%d",
              vm->def->name, message, cfg->stdioLogD);

10918
    path = g_strdup_printf("%s/%s.log", cfg->logDir, vm->def->name);
10919 10920 10921 10922 10923 10924 10925 10926 10927 10928 10929 10930 10931 10932 10933 10934 10935 10936 10937 10938 10939 10940 10941 10942 10943 10944 10945 10946 10947 10948

    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_FORCE_CLOSE(writefd);
    virLogManagerFree(manager);

    return ret;
}


10949 10950 10951 10952 10953 10954 10955 10956
int qemuDomainLogContextGetWriteFD(qemuDomainLogContextPtr ctxt)
{
    return ctxt->writefd;
}


void qemuDomainLogContextMarkPosition(qemuDomainLogContextPtr ctxt)
{
10957 10958
    if (ctxt->manager)
        virLogManagerDomainGetLogFilePosition(ctxt->manager,
10959
                                              ctxt->path,
10960 10961 10962 10963 10964
                                              0,
                                              &ctxt->inode,
                                              &ctxt->pos);
    else
        ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END);
10965 10966 10967
}


10968 10969 10970 10971 10972 10973
virLogManagerPtr qemuDomainLogContextGetManager(qemuDomainLogContextPtr ctxt)
{
    return ctxt->manager;
}


10974 10975
/* Locate an appropriate 'qemu-img' binary.  */
const char *
10976
qemuFindQemuImgBinary(virQEMUDriverPtr driver)
10977
{
10978 10979
    if (!driver->qemuImgBinary)
        virReportError(VIR_ERR_INTERNAL_ERROR,
10980
                       "%s", _("unable to find qemu-img"));
10981 10982 10983 10984 10985 10986

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
10987
                                virDomainMomentObjPtr snapshot,
10988
                                virDomainXMLOptionPtr xmlopt,
10989
                                const char *snapshotDir)
10990
{
10991 10992 10993
    g_autofree char *newxml = NULL;
    g_autofree char *snapDir = NULL;
    g_autofree char *snapFile = NULL;
10994
    char uuidstr[VIR_UUID_STRING_BUFLEN];
10995 10996
    unsigned int flags = VIR_DOMAIN_SNAPSHOT_FORMAT_SECURE |
        VIR_DOMAIN_SNAPSHOT_FORMAT_INTERNAL;
10997
    virDomainSnapshotDefPtr def = virDomainSnapshotObjGetDef(snapshot);
10998

10999
    if (virDomainSnapshotGetCurrent(vm->snapshots) == snapshot)
11000
        flags |= VIR_DOMAIN_SNAPSHOT_FORMAT_CURRENT;
11001
    virUUIDFormat(vm->def->uuid, uuidstr);
11002
    newxml = virDomainSnapshotDefFormat(uuidstr, def, xmlopt, flags);
11003
    if (newxml == NULL)
11004 11005
        return -1;

11006
    snapDir = g_strdup_printf("%s/%s", snapshotDir, vm->def->name);
11007 11008 11009
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
11010
        return -1;
11011 11012
    }

11013
    snapFile = g_strdup_printf("%s/%s.xml", snapDir, def->parent.name);
11014

11015
    return virXMLSaveFile(snapFile, NULL, "snapshot-edit", newxml);
11016 11017
}

11018

11019 11020
/* The domain is expected to be locked and inactive. Return -1 on normal
 * failure, 1 if we skipped a disk due to try_all.  */
11021
static int
11022
qemuDomainSnapshotForEachQcow2Raw(virQEMUDriverPtr driver,
11023 11024 11025 11026 11027
                                  virDomainDefPtr def,
                                  const char *name,
                                  const char *op,
                                  bool try_all,
                                  int ndisks)
11028 11029
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
11030
    size_t i;
11031 11032 11033 11034 11035 11036 11037 11038 11039
    bool skipped = false;

    qemuimgarg[0] = qemuFindQemuImgBinary(driver);
    if (qemuimgarg[0] == NULL) {
        /* qemuFindQemuImgBinary set the error */
        return -1;
    }

    qemuimgarg[2] = op;
11040
    qemuimgarg[3] = name;
11041

11042
    for (i = 0; i < ndisks; i++) {
11043
        /* FIXME: we also need to handle LVM here */
11044
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
11045 11046 11047
            int format = virDomainDiskGetFormat(def->disks[i]);

            if (format > 0 && format != VIR_STORAGE_FILE_QCOW2) {
11048 11049 11050 11051 11052
                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",
11053
                             def->disks[i]->dst);
11054 11055
                    skipped = true;
                    continue;
11056 11057 11058 11059 11060
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
11061
                }
11062 11063 11064 11065
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("Disk device '%s' does not support"
                                 " snapshotting"),
                               def->disks[i]->dst);
11066 11067 11068
                return -1;
            }

11069
            qemuimgarg[4] = virDomainDiskGetSource(def->disks[i]);
11070 11071 11072 11073

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
11074
                             def->disks[i]->dst);
11075 11076
                    skipped = true;
                    continue;
11077 11078 11079 11080 11081
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
11082 11083 11084 11085 11086 11087 11088 11089 11090
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

11091 11092 11093
/* 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
11094
qemuDomainSnapshotForEachQcow2(virQEMUDriverPtr driver,
11095
                               virDomainObjPtr vm,
11096
                               virDomainMomentObjPtr snap,
11097 11098 11099 11100 11101 11102
                               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.  */
11103
    virDomainDefPtr def = snap->def->dom;
11104 11105 11106

    if (!def)
        def = vm->def;
11107
    return qemuDomainSnapshotForEachQcow2Raw(driver, def, snap->def->name,
11108 11109 11110
                                             op, try_all, def->ndisks);
}

11111 11112
/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
11113
qemuDomainSnapshotDiscard(virQEMUDriverPtr driver,
11114
                          virDomainObjPtr vm,
11115
                          virDomainMomentObjPtr snap,
11116
                          bool update_parent,
11117 11118
                          bool metadata_only)
{
11119
    g_autofree char *snapFile = NULL;
11120
    qemuDomainObjPrivatePtr priv;
11121
    virDomainMomentObjPtr parentsnap = NULL;
11122
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
11123 11124 11125 11126 11127 11128

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
11129
                return -1;
11130 11131
        } else {
            priv = vm->privateData;
11132
            qemuDomainObjEnterMonitor(driver, vm);
11133
            /* we continue on even in the face of error */
11134
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
11135
            ignore_value(qemuDomainObjExitMonitor(driver, vm));
11136 11137 11138
        }
    }

11139 11140
    snapFile = g_strdup_printf("%s/%s/%s.xml", cfg->snapshotDir, vm->def->name,
                               snap->def->name);
11141

11142 11143
    if (snap == virDomainSnapshotGetCurrent(vm->snapshots)) {
        virDomainSnapshotSetCurrent(vm->snapshots, NULL);
11144
        if (update_parent && snap->def->parent_name) {
11145
            parentsnap = virDomainSnapshotFindByName(vm->snapshots,
11146
                                                     snap->def->parent_name);
11147 11148
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
11149
                         snap->def->parent_name);
11150
            } else {
11151
                virDomainSnapshotSetCurrent(vm->snapshots, parentsnap);
11152
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap,
11153
                                                    driver->xmlopt,
11154
                                                    cfg->snapshotDir) < 0) {
11155
                    VIR_WARN("failed to set parent snapshot '%s' as current",
11156
                             snap->def->parent_name);
11157
                    virDomainSnapshotSetCurrent(vm->snapshots, NULL);
11158 11159 11160 11161 11162 11163 11164
                }
            }
        }
    }

    if (unlink(snapFile) < 0)
        VIR_WARN("Failed to unlink %s", snapFile);
11165
    if (update_parent)
11166
        virDomainMomentDropParent(snap);
11167
    virDomainSnapshotObjListRemove(vm->snapshots, snap);
11168

11169
    return 0;
11170 11171 11172
}

/* Hash iterator callback to discard multiple snapshots.  */
11173
int qemuDomainMomentDiscardAll(void *payload,
J
Ján Tomko 已提交
11174
                               const void *name G_GNUC_UNUSED,
11175
                               void *data)
11176
{
11177 11178
    virDomainMomentObjPtr moment = payload;
    virQEMUMomentRemovePtr curr = data;
11179 11180
    int err;

11181 11182 11183 11184
    if (!curr->found && curr->current == moment)
        curr->found = true;
    err = curr->momentDiscard(curr->driver, curr->vm, moment, false,
                              curr->metadata_only);
11185 11186
    if (err && !curr->err)
        curr->err = err;
11187
    return 0;
11188 11189 11190
}

int
11191
qemuDomainSnapshotDiscardAllMetadata(virQEMUDriverPtr driver,
11192 11193
                                     virDomainObjPtr vm)
{
11194 11195 11196
    virQEMUMomentRemove rem = {
        .driver = driver,
        .vm = vm,
11197 11198
        .metadata_only = true,
        .momentDiscard = qemuDomainSnapshotDiscard,
11199
    };
11200

11201
    virDomainSnapshotForEach(vm->snapshots, qemuDomainMomentDiscardAll, &rem);
11202
    virDomainSnapshotObjListRemoveAll(vm->snapshots);
11203 11204 11205 11206

    return rem.err;
}

11207

11208 11209 11210
static void
qemuDomainRemoveInactiveCommon(virQEMUDriverPtr driver,
                               virDomainObjPtr vm)
11211
{
11212
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
11213 11214
    g_autofree char *snapDir = NULL;
    g_autofree char *chkDir = NULL;
11215

11216 11217 11218 11219
    /* 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);
11220 11221 11222 11223 11224
    } 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);
11225
    }
11226
    /* Remove any checkpoint metadata prior to removing the domain */
11227
    if (qemuCheckpointDiscardAllMetadata(driver, vm) < 0) {
11228 11229
        VIR_WARN("unable to remove all checkpoints for domain %s",
                 vm->def->name);
11230 11231 11232 11233 11234
    } 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);
11235
    }
11236
    qemuExtDevicesCleanupHost(driver, vm->def);
11237 11238 11239
}


11240 11241 11242 11243 11244 11245 11246 11247 11248 11249 11250 11251 11252 11253 11254 11255 11256 11257 11258 11259
/**
 * 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);
}


11260 11261 11262 11263 11264 11265 11266 11267 11268 11269 11270 11271 11272 11273 11274 11275 11276 11277 11278 11279 11280 11281
/**
 * 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);
}


11282 11283 11284 11285 11286 11287 11288 11289 11290 11291 11292 11293 11294 11295 11296 11297 11298
/**
 * 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);
11299 11300

    if (haveJob)
11301
        qemuDomainObjEndJob(driver, vm);
11302
}
11303

11304

11305 11306 11307 11308 11309 11310 11311 11312 11313 11314 11315 11316 11317 11318 11319 11320 11321 11322 11323 11324 11325
/**
 * 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);
}


11326
void
11327
qemuDomainSetFakeReboot(virQEMUDriverPtr driver,
11328 11329 11330 11331
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
11332
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
11333 11334

    if (priv->fakeReboot == value)
11335
        return;
11336 11337 11338

    priv->fakeReboot = value;

11339
    if (virDomainObjSave(vm, driver->xmlopt, cfg->stateDir) < 0)
11340 11341
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
}
M
Michal Privoznik 已提交
11342

11343
static void
11344 11345
qemuDomainCheckRemoveOptionalDisk(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
11346
                                  size_t diskIndex)
11347 11348
{
    char uuid[VIR_UUID_STRING_BUFLEN];
11349
    virObjectEventPtr event = NULL;
11350
    virDomainDiskDefPtr disk = vm->def->disks[diskIndex];
11351
    const char *src = virDomainDiskGetSource(disk);
11352 11353 11354 11355 11356

    virUUIDFormat(vm->def->uuid, uuid);

    VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
              "due to inaccessible source '%s'",
11357
              disk->dst, vm->def->name, uuid, src);
11358 11359 11360 11361

    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM ||
        disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY) {

11362
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
11363 11364
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
11365
        virDomainDiskEmptySource(disk);
11366 11367
        /* keeping the old startup policy would be invalid for new images */
        disk->startupPolicy = VIR_DOMAIN_STARTUP_POLICY_DEFAULT;
11368
    } else {
11369
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
11370 11371
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_DROP_MISSING_ON_START);
11372 11373
        virDomainDiskRemove(vm->def, diskIndex);
        virDomainDiskDefFree(disk);
11374 11375
    }

11376
    virObjectEventStateQueue(driver->domainEventState, event);
11377 11378
}

11379 11380 11381 11382 11383 11384 11385 11386 11387 11388 11389 11390 11391 11392 11393 11394 11395

/**
 * 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
11396 11397
qemuDomainCheckDiskStartupPolicy(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
11398
                                 size_t diskIndex,
11399 11400
                                 bool cold_boot)
{
11401
    int startupPolicy = vm->def->disks[diskIndex]->startupPolicy;
11402
    int device = vm->def->disks[diskIndex]->device;
11403

11404
    switch ((virDomainStartupPolicy) startupPolicy) {
11405
        case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
11406 11407 11408 11409 11410 11411
            /* 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)
11412
                return -1;
11413 11414
            break;

11415
        case VIR_DOMAIN_STARTUP_POLICY_DEFAULT:
11416
        case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
11417
            return -1;
11418 11419

        case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
11420
            if (cold_boot)
11421
                return -1;
11422 11423 11424 11425 11426 11427 11428
            break;

        case VIR_DOMAIN_STARTUP_POLICY_LAST:
            /* this should never happen */
            break;
    }

11429
    qemuDomainCheckRemoveOptionalDisk(driver, vm, diskIndex);
11430
    virResetLastError();
11431 11432 11433
    return 0;
}

11434

11435 11436 11437 11438 11439 11440 11441 11442 11443 11444

/*
 * The vm must be locked when any of the following cleanup functions is
 * called.
 */
int
qemuDomainCleanupAdd(virDomainObjPtr vm,
                     qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
11445
    size_t i;
11446 11447 11448 11449 11450 11451 11452 11453 11454 11455

    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,
11456
                     priv->ncleanupCallbacks, 1) < 0)
11457 11458 11459 11460 11461 11462 11463 11464 11465 11466 11467
        return -1;

    priv->cleanupCallbacks[priv->ncleanupCallbacks++] = cb;
    return 0;
}

void
qemuDomainCleanupRemove(virDomainObjPtr vm,
                        qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
11468
    size_t i;
11469 11470 11471 11472

    VIR_DEBUG("vm=%s, cb=%p", vm->def->name, cb);

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
11473 11474 11475
        if (priv->cleanupCallbacks[i] == cb)
            VIR_DELETE_ELEMENT_INPLACE(priv->cleanupCallbacks,
                                       i, priv->ncleanupCallbacks);
11476 11477 11478 11479 11480 11481 11482 11483
    }

    VIR_SHRINK_N(priv->cleanupCallbacks,
                 priv->ncleanupCallbacks_max,
                 priv->ncleanupCallbacks_max - priv->ncleanupCallbacks);
}

void
11484
qemuDomainCleanupRun(virQEMUDriverPtr driver,
11485 11486 11487
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
11488
    size_t i;
11489 11490 11491 11492

    VIR_DEBUG("driver=%p, vm=%s", driver, vm->def->name);

    /* run cleanup callbacks in reverse order */
11493 11494
    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[priv->ncleanupCallbacks - (i + 1)])
11495 11496 11497 11498 11499 11500 11501
            priv->cleanupCallbacks[i](driver, vm);
    }

    VIR_FREE(priv->cleanupCallbacks);
    priv->ncleanupCallbacks = 0;
    priv->ncleanupCallbacks_max = 0;
}
11502

11503
void
11504 11505
qemuDomainGetImageIds(virQEMUDriverConfigPtr cfg,
                      virDomainObjPtr vm,
11506
                      virStorageSourcePtr src,
11507
                      virStorageSourcePtr parentSrc,
11508 11509 11510 11511 11512 11513 11514 11515 11516 11517 11518 11519 11520 11521 11522 11523 11524 11525
                      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;
    }

11526 11527
    if (vm && (vmlabel = virDomainDefGetSecurityLabelDef(vm->def, "dac")) &&
        vmlabel->label)
11528 11529
        virParseOwnershipIds(vmlabel->label, uid, gid);

11530 11531 11532 11533 11534
    if (parentSrc &&
        (disklabel = virStorageSourceGetSecurityLabelDef(parentSrc, "dac")) &&
        disklabel->label)
        virParseOwnershipIds(disklabel->label, uid, gid);

11535
    if ((disklabel = virStorageSourceGetSecurityLabelDef(src, "dac")) &&
11536
        disklabel->label)
11537 11538 11539 11540
        virParseOwnershipIds(disklabel->label, uid, gid);
}


11541 11542 11543
int
qemuDomainStorageFileInit(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
11544 11545
                          virStorageSourcePtr src,
                          virStorageSourcePtr parent)
11546
{
11547
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
11548 11549 11550
    uid_t uid;
    gid_t gid;

11551
    qemuDomainGetImageIds(cfg, vm, src, parent, &uid, &gid);
11552 11553

    if (virStorageFileInitAs(src, uid, gid) < 0)
11554
        return -1;
11555

11556
    return 0;
11557 11558 11559
}


11560 11561 11562 11563 11564
char *
qemuDomainStorageAlias(const char *device, int depth)
{
    char *alias;

11565
    device = qemuAliasDiskDriveSkipPrefix(device);
11566 11567

    if (!depth)
11568
        alias = g_strdup(device);
11569
    else
11570
        alias = g_strdup_printf("%s.%d", device, depth);
11571 11572 11573 11574
    return alias;
}


11575 11576 11577 11578 11579 11580 11581 11582 11583 11584 11585 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 11612 11613 11614 11615 11616 11617 11618 11619 11620 11621
/**
 * 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;
}


11622 11623 11624 11625 11626 11627 11628 11629 11630 11631 11632 11633 11634 11635 11636 11637 11638 11639 11640 11641 11642 11643 11644 11645 11646 11647
/**
 * 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;
    }
}


11648 11649 11650 11651 11652 11653 11654 11655 11656 11657 11658 11659 11660
/**
 * 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.
 */
11661
int
11662
qemuDomainDetermineDiskChain(virQEMUDriverPtr driver,
11663
                             virDomainObjPtr vm,
11664
                             virDomainDiskDefPtr disk,
11665
                             virStorageSourcePtr disksrc,
11666
                             bool report_broken)
11667
{
11668
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
11669 11670
    virStorageSourcePtr src; /* iterator for the backing chain declared in XML */
    virStorageSourcePtr n; /* iterator for the backing chain detected from disk */
11671
    qemuDomainObjPrivatePtr priv = vm->privateData;
11672 11673
    uid_t uid;
    gid_t gid;
11674

11675 11676 11677
    if (!disksrc)
        disksrc = disk->src;

11678 11679
    if (virStorageSourceIsEmpty(disksrc))
        return 0;
11680

11681 11682
    /* There is no need to check the backing chain for disks without backing
     * support */
11683 11684 11685
    if (virStorageSourceIsLocalStorage(disksrc) &&
        disksrc->format > VIR_STORAGE_FILE_NONE &&
        disksrc->format < VIR_STORAGE_FILE_BACKING) {
11686

11687
        if (!virFileExists(disksrc->path)) {
11688
            if (report_broken)
11689
                virStorageFileReportBrokenChain(errno, disksrc, disksrc);
11690

11691
            return -1;
11692 11693
        }

11694
        /* terminate the chain for such images as the code below would do */
11695
        if (!disksrc->backingStore &&
11696
            !(disksrc->backingStore = virStorageSourceNew()))
11697
            return -1;
11698

11699 11700 11701
        /* 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 &&
11702 11703 11704 11705
            disksrc->format == VIR_STORAGE_FILE_RAW &&
            virStorageSourceIsBlockLocal(disksrc) &&
            virFileIsCDROM(disksrc->path) == 1)
            disksrc->hostcdrom = true;
11706

11707
        return 0;
11708 11709
    }

11710
    src = disksrc;
11711 11712
    /* skip to the end of the chain if there is any */
    while (virStorageSourceHasBacking(src)) {
11713 11714
        if (report_broken) {
            int rv = virStorageFileSupportsAccess(src);
11715

11716
            if (rv < 0)
11717
                return -1;
11718

11719
            if (rv > 0) {
11720
                if (qemuDomainStorageFileInit(driver, vm, src, disksrc) < 0)
11721
                    return -1;
11722 11723

                if (virStorageFileAccess(src, F_OK) < 0) {
11724
                    virStorageFileReportBrokenChain(errno, src, disksrc);
11725
                    virStorageFileDeinit(src);
11726
                    return -1;
11727 11728
                }

11729
                virStorageFileDeinit(src);
11730 11731
            }
        }
11732
        src = src->backingStore;
11733 11734 11735 11736
    }

    /* We skipped to the end of the chain. Skip detection if there's the
     * terminator. (An allocated but empty backingStore) */
11737 11738 11739 11740
    if (src->backingStore) {
        if (qemuDomainStorageSourceValidateDepth(disksrc, 0, disk->dst) < 0)
            return -1;

11741
        return 0;
11742
    }
11743

11744
    qemuDomainGetImageIds(cfg, vm, src, disksrc, &uid, &gid);
11745

11746
    if (virStorageFileGetMetadata(src, uid, gid, report_broken) < 0)
11747
        return -1;
11748

11749
    for (n = src->backingStore; virStorageSourceIsBacking(n); n = n->backingStore) {
11750 11751 11752 11753
        /* convert detected ISO format to 'raw' as qemu would not understand it */
        if (n->format == VIR_STORAGE_FILE_ISO)
            n->format = VIR_STORAGE_FILE_RAW;

11754
        if (qemuDomainValidateStorageSource(n, priv->qemuCaps) < 0)
11755
            return -1;
11756

11757
        qemuDomainPrepareStorageSourceConfig(n, cfg, priv->qemuCaps);
11758
        qemuDomainPrepareDiskSourceData(disk, n);
11759 11760 11761

        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV) &&
            qemuDomainPrepareStorageSourceBlockdev(disk, n, priv, cfg) < 0)
11762
            return -1;
11763
    }
11764

11765 11766 11767
    if (qemuDomainStorageSourceValidateDepth(disksrc, 0, disk->dst) < 0)
        return -1;

11768
    return 0;
11769
}
11770

11771

11772 11773 11774 11775 11776 11777 11778 11779 11780 11781 11782 11783 11784 11785 11786 11787 11788 11789 11790 11791 11792 11793 11794 11795 11796
/**
 * qemuDomainDiskGetTopNodename:
 *
 * @disk: disk definition object
 *
 * Returns the pointer to the node-name of the topmost layer used by @disk as
 * backend. Currently returns the nodename of the copy-on-read filter if enabled
 * or the nodename of the top image's format driver. Empty disks return NULL.
 * This must be used only when VIR_QEMU_CAPS_BLOCKDEV is enabled.
 */
const char *
qemuDomainDiskGetTopNodename(virDomainDiskDefPtr disk)
{
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);

    if (virStorageSourceIsEmpty(disk->src))
        return NULL;

    if (disk->copy_on_read == VIR_TRISTATE_SWITCH_ON)
        return priv->nodeCopyOnRead;

    return disk->src->nodeformat;
}


11797 11798 11799 11800 11801 11802 11803 11804 11805 11806 11807 11808 11809 11810 11811 11812
/**
 * 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,
11813
                              virQEMUCapsPtr qemuCaps,
11814 11815 11816 11817
                              char **backendAlias)
{
    *backendAlias = NULL;

11818 11819 11820 11821 11822 11823 11824
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if (!(*backendAlias = qemuAliasDiskDriveFromDisk(disk)))
            return -1;

        return 0;
    }

11825
    *backendAlias = g_strdup(qemuDomainDiskGetTopNodename(disk));
11826 11827 11828 11829
    return 0;
}


11830 11831 11832
typedef enum {
    /* revoke access to the image instead of allowing it */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE = 1 << 0,
11833 11834
    /* operate on full backing chain rather than single image */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN = 1 << 1,
11835 11836 11837 11838
    /* 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,
11839
    /* don't revoke permissions when modification has failed */
11840
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE = 1 << 4,
11841
    /* VM already has access to the source and we are just modifying it */
11842
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS = 1 << 5,
11843 11844
    /* whether the image is the top image of the backing chain (e.g. disk source) */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN_TOP = 1 << 6,
11845 11846 11847
} qemuDomainStorageSourceAccessFlags;


11848 11849 11850 11851 11852 11853 11854 11855 11856 11857 11858 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 11894 11895
static int
qemuDomainStorageSourceAccessModifyNVMe(virQEMUDriverPtr driver,
                                        virDomainObjPtr vm,
                                        virStorageSourcePtr src,
                                        bool revoke)
{
    bool revoke_maxmemlock = false;
    bool revoke_hostdev = false;
    int ret = -1;

    if (!virStorageSourceChainHasNVMe(src))
        return 0;

    VIR_DEBUG("Modifying access for a NVMe disk src=%p revoke=%d",
              src, revoke);

    if (revoke) {
        revoke_maxmemlock = true;
        revoke_hostdev = true;
        ret = 0;
        goto revoke;
    }

    if (qemuDomainAdjustMaxMemLock(vm, true) < 0)
        goto revoke;

    revoke_maxmemlock = true;

    if (qemuHostdevPrepareOneNVMeDisk(driver, vm->def->name, src) < 0)
        goto revoke;

    revoke_hostdev = true;

    return 0;

 revoke:
    if (revoke_maxmemlock) {
        if (qemuDomainAdjustMaxMemLock(vm, false) < 0)
            VIR_WARN("Unable to change max memlock limit");
    }

    if (revoke_hostdev)
        qemuHostdevReAttachOneNVMeDisk(driver, vm->def->name, src);

    return ret;
}


11896
/**
11897
 * qemuDomainStorageSourceAccessModify:
11898 11899 11900
 * @driver: qemu driver struct
 * @vm: domain object
 * @src: Source to prepare
11901
 * @flags: bitwise or of qemuDomainStorageSourceAccessFlags
11902 11903
 *
 * Setup the locks, cgroups and security permissions on a disk source and its
11904
 * backing chain.
11905 11906 11907
 *
 * Returns 0 on success and -1 on error. Reports libvirt error.
 */
11908
static int
11909 11910 11911
qemuDomainStorageSourceAccessModify(virQEMUDriverPtr driver,
                                    virDomainObjPtr vm,
                                    virStorageSourcePtr src,
11912
                                    qemuDomainStorageSourceAccessFlags flags)
11913
{
11914
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
11915 11916 11917
    const char *srcstr = NULLSTR(src->path);
    int ret = -1;
    virErrorPtr orig_err = NULL;
11918
    bool chain = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
11919 11920 11921
    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;
11922
    bool chain_top = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN_TOP;
11923
    int rc;
11924
    bool was_readonly = src->readonly;
11925 11926 11927
    bool revoke_cgroup = false;
    bool revoke_label = false;
    bool revoke_namespace = false;
11928
    bool revoke_nvme = false;
11929
    bool revoke_lockspace = false;
11930

11931 11932 11933 11934
    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)
11935
        src->readonly = true;
11936

11937
    if (force_rw)
11938 11939
        src->readonly = false;

11940
    /* just tear down the disk access */
11941
    if (revoke) {
11942
        virErrorPreserveLast(&orig_err);
11943 11944 11945
        revoke_cgroup = true;
        revoke_label = true;
        revoke_namespace = true;
11946
        revoke_nvme = true;
11947
        revoke_lockspace = true;
11948
        ret = 0;
11949
        goto revoke;
11950 11951 11952
    }

    if (virDomainLockImageAttach(driver->lockManager, cfg->uri, vm, src) < 0)
11953 11954 11955
        goto revoke;

    revoke_lockspace = true;
11956

11957 11958 11959
    if (!(flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS)) {
        if (qemuDomainStorageSourceAccessModifyNVMe(driver, vm, src, false) < 0)
            goto revoke;
11960

11961
        revoke_nvme = true;
11962

11963 11964
        if (qemuDomainNamespaceSetupDisk(vm, src) < 0)
            goto revoke;
11965

11966 11967
        revoke_namespace = true;
    }
11968

11969
    if (qemuSecuritySetImageLabel(driver, vm, src, chain, chain_top) < 0)
11970 11971 11972
        goto revoke;

    revoke_label = true;
11973

11974 11975 11976 11977 11978 11979
    if (chain)
        rc = qemuSetupImageChainCgroup(vm, src);
    else
        rc = qemuSetupImageCgroup(vm, src);

    if (rc < 0)
11980 11981 11982
        goto revoke;

    revoke_cgroup = true;
11983 11984 11985 11986

    ret = 0;
    goto cleanup;

11987
 revoke:
11988 11989 11990
    if (flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE)
        goto cleanup;

11991 11992 11993 11994 11995
    if (revoke_cgroup) {
        if (chain)
            rc = qemuTeardownImageChainCgroup(vm, src);
        else
            rc = qemuTeardownImageCgroup(vm, src);
11996

11997 11998 11999 12000 12001 12002 12003 12004 12005 12006 12007 12008 12009 12010
        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);
    }

12011 12012 12013
    if (revoke_nvme)
        qemuDomainStorageSourceAccessModifyNVMe(driver, vm, src, true);

12014 12015 12016 12017
    if (revoke_lockspace) {
        if (virDomainLockImageDetach(driver->lockManager, vm, src) < 0)
            VIR_WARN("Unable to release lock on %s", srcstr);
    }
12018 12019

 cleanup:
12020
    src->readonly = was_readonly;
12021 12022 12023 12024 12025 12026
    virErrorRestore(&orig_err);

    return ret;
}


12027 12028 12029 12030 12031
int
qemuDomainStorageSourceChainAccessAllow(virQEMUDriverPtr driver,
                                        virDomainObjPtr vm,
                                        virStorageSourcePtr src)
{
12032 12033
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN |
                                               QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN_TOP;
12034 12035

    return qemuDomainStorageSourceAccessModify(driver, vm, src, flags);
12036 12037 12038 12039 12040 12041 12042 12043
}


int
qemuDomainStorageSourceChainAccessRevoke(virQEMUDriverPtr driver,
                                         virDomainObjPtr vm,
                                         virStorageSourcePtr src)
{
12044
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE |
12045 12046
                                               QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN |
                                               QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN_TOP;
12047 12048

    return qemuDomainStorageSourceAccessModify(driver, vm, src, flags);
12049 12050 12051
}


12052
/**
12053
 * qemuDomainStorageSourceAccessRevoke:
12054 12055 12056 12057 12058
 *
 * Revoke access to a single backing chain element. This restores the labels,
 * removes cgroup ACLs for devices and removes locks.
 */
void
12059 12060 12061
qemuDomainStorageSourceAccessRevoke(virQEMUDriverPtr driver,
                                    virDomainObjPtr vm,
                                    virStorageSourcePtr elem)
12062
{
12063
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE;
12064

12065
    ignore_value(qemuDomainStorageSourceAccessModify(driver, vm, elem, flags));
12066 12067 12068 12069
}


/**
12070
 * qemuDomainStorageSourceAccessAllow:
12071 12072 12073 12074 12075
 * @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
12076
 * @chainTop: @elem is top parent of backing chain
12077 12078 12079
 *
 * 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
12080 12081 12082 12083
 * 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.
12084 12085 12086 12087 12088 12089
 *
 * The @chainTop flag must be set if the @elem image is the topmost image of a
 * given backing chain or meant to become the topmost image (for e.g.
 * snapshots, or blockcopy or even in the end for active layer block commit,
 * where we discard the top of the backing chain so one of the intermediates
 * (the base) becomes the top of the chain).
12090
 */
12091
int
12092 12093 12094 12095
qemuDomainStorageSourceAccessAllow(virQEMUDriverPtr driver,
                                   virDomainObjPtr vm,
                                   virStorageSourcePtr elem,
                                   bool readonly,
12096 12097
                                   bool newSource,
                                   bool chainTop)
12098
{
12099
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE;
12100

12101
    if (readonly)
12102 12103 12104
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_ONLY;
    else
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_WRITE;
12105

12106
    if (!newSource)
12107
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS;
12108

12109 12110 12111
    if (chainTop)
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN_TOP;

12112
    return qemuDomainStorageSourceAccessModify(driver, vm, elem, flags);
12113 12114 12115
}


12116 12117 12118 12119 12120 12121 12122 12123 12124 12125
/*
 * 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)
{
12126 12127 12128 12129 12130 12131 12132 12133 12134 12135
#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; \
        } \
12136 12137
    } while (0)

12138 12139 12140 12141 12142 12143 12144 12145 12146 12147 12148 12149
#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)

12150
    CHECK_EQ(device, "device", false);
12151 12152 12153 12154
    CHECK_EQ(bus, "bus", false);
    if (STRNEQ(disk->dst, orig_disk->dst)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
12155
                       "target");
12156 12157 12158 12159
        return false;
    }
    CHECK_EQ(tray_status, "tray", true);
    CHECK_EQ(removable, "removable", true);
12160 12161 12162 12163 12164 12165 12166 12167 12168 12169 12170 12171 12172 12173 12174 12175 12176 12177 12178 12179 12180 12181 12182 12183 12184 12185 12186 12187 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

    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);
12214 12215
    CHECK_STREQ_NULLABLE(blkdeviotune.group_name,
                         "blkdeviotune group name");
12216

12217 12218 12219 12220 12221 12222 12223 12224
    CHECK_STREQ_NULLABLE(serial,
                         "serial");
    CHECK_STREQ_NULLABLE(wwn,
                         "wwn");
    CHECK_STREQ_NULLABLE(vendor,
                         "vendor");
    CHECK_STREQ_NULLABLE(product,
                         "product");
12225

12226 12227 12228 12229 12230 12231 12232
    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);
12233
    /* "snapshot" is a libvirt internal field and thus can be changed */
12234
    /* startupPolicy is allowed to be updated. Therefore not checked here. */
12235
    CHECK_EQ(transient, "transient", true);
12236 12237 12238 12239 12240 12241 12242 12243 12244 12245 12246 12247 12248

    /* 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;
    }

12249
    /* device alias is checked already in virDomainDefCompatibleDevice */
12250

12251
    CHECK_EQ(info.bootIndex, "boot order", true);
12252 12253 12254 12255 12256
    CHECK_EQ(rawio, "rawio", true);
    CHECK_EQ(sgio, "sgio", true);
    CHECK_EQ(discard, "discard", true);
    CHECK_EQ(iothread, "iothread", true);

12257 12258
    CHECK_STREQ_NULLABLE(domain_name,
                         "backenddomain");
12259

12260 12261 12262 12263 12264 12265
    /* 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);

12266 12267 12268 12269 12270 12271 12272 12273
    if (!virStoragePRDefIsEqual(disk->src->pr,
                                orig_disk->src->pr)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "reservations");
        return false;
    }

12274
#undef CHECK_EQ
12275
#undef CHECK_STREQ_NULLABLE
12276 12277 12278 12279

    return true;
}

12280

12281 12282 12283
bool
qemuDomainDiskBlockJobIsActive(virDomainDiskDefPtr disk)
{
12284 12285
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

12286 12287 12288 12289 12290 12291 12292 12293
    if (disk->mirror) {
        virReportError(VIR_ERR_BLOCK_COPY_ACTIVE,
                       _("disk '%s' already in active block job"),
                       disk->dst);

        return true;
    }

12294
    if (diskPriv->blockjob &&
12295
        qemuBlockJobIsRunning(diskPriv->blockjob)) {
12296 12297 12298 12299 12300 12301 12302 12303 12304 12305
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("disk '%s' already in active block job"),
                       disk->dst);
        return true;
    }

    return false;
}


12306 12307 12308 12309 12310 12311 12312 12313 12314 12315 12316 12317 12318 12319 12320
/**
 * 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++) {
12321 12322 12323
        virDomainDiskDefPtr disk = vm->def->disks[i];
        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

12324 12325
        if (!copy_only && diskPriv->blockjob &&
            qemuBlockJobIsRunning(diskPriv->blockjob))
12326 12327
            return true;

12328
        if (disk->mirror && disk->mirrorJob == VIR_DOMAIN_BLOCK_JOB_TYPE_COPY)
12329 12330 12331 12332 12333 12334 12335
            return true;
    }

    return false;
}


12336 12337
int
qemuDomainUpdateDeviceList(virQEMUDriverPtr driver,
12338 12339
                           virDomainObjPtr vm,
                           int asyncJob)
12340 12341 12342
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char **aliases;
12343
    int rc;
12344

12345 12346
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
12347 12348 12349 12350
    rc = qemuMonitorGetDeviceAliases(priv->mon, &aliases);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
    if (rc < 0)
12351 12352
        return -1;

12353
    virStringListFree(priv->qemuDevices);
12354 12355 12356
    priv->qemuDevices = aliases;
    return 0;
}
12357

12358 12359 12360 12361 12362 12363 12364 12365 12366 12367 12368 12369 12370 12371 12372 12373 12374 12375 12376

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);

12377 12378
    if (qemuDomainObjExitMonitor(driver, vm) < 0) {
        virHashFree(meminfo);
12379
        return -1;
12380
    }
12381 12382 12383 12384 12385 12386 12387 12388 12389 12390 12391 12392 12393 12394 12395 12396 12397 12398 12399 12400 12401 12402 12403 12404 12405 12406 12407 12408

    /* 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;
}


12409 12410 12411 12412
static bool
qemuDomainABIStabilityCheck(const virDomainDef *src,
                            const virDomainDef *dst)
{
12413 12414
    size_t i;

12415 12416 12417 12418 12419 12420 12421 12422 12423
    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;
    }

12424 12425 12426 12427 12428 12429 12430 12431 12432 12433 12434 12435 12436
    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;
        }
    }

12437 12438 12439 12440 12441 12442 12443 12444 12445
    return true;
}


virDomainABIStability virQEMUDriverDomainABIStability = {
    .domain = qemuDomainABIStabilityCheck,
};


12446 12447 12448 12449 12450 12451 12452 12453 12454 12455 12456 12457 12458 12459 12460 12461 12462 12463 12464 12465
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;
}


12466 12467 12468
#define COPY_FLAGS (VIR_DOMAIN_XML_SECURE | \
                    VIR_DOMAIN_XML_MIGRATABLE)

12469 12470
bool
qemuDomainDefCheckABIStability(virQEMUDriverPtr driver,
12471
                               virQEMUCapsPtr qemuCaps,
12472 12473 12474 12475 12476 12477 12478
                               virDomainDefPtr src,
                               virDomainDefPtr dst)
{
    virDomainDefPtr migratableDefSrc = NULL;
    virDomainDefPtr migratableDefDst = NULL;
    bool ret = false;

12479 12480
    if (!(migratableDefSrc = qemuDomainDefCopy(driver, qemuCaps, src, COPY_FLAGS)) ||
        !(migratableDefDst = qemuDomainDefCopy(driver, qemuCaps, dst, COPY_FLAGS)))
12481 12482
        goto cleanup;

12483 12484 12485
    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   src, migratableDefSrc,
                                                   dst, migratableDefDst);
12486

12487
 cleanup:
12488 12489 12490 12491
    virDomainDefFree(migratableDefSrc);
    virDomainDefFree(migratableDefDst);
    return ret;
}
12492

12493 12494 12495 12496 12497 12498

bool
qemuDomainCheckABIStability(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
                            virDomainDefPtr dst)
{
12499
    qemuDomainObjPrivatePtr priv = vm->privateData;
12500 12501
    virDomainDefPtr migratableSrc = NULL;
    virDomainDefPtr migratableDst = NULL;
12502
    g_autofree char *xml = NULL;
12503 12504 12505
    bool ret = false;

    if (!(xml = qemuDomainFormatXML(driver, vm, COPY_FLAGS)) ||
12506 12507
        !(migratableSrc = qemuDomainDefFromXML(driver, priv->qemuCaps, xml)) ||
        !(migratableDst = qemuDomainDefCopy(driver, priv->qemuCaps, dst, COPY_FLAGS)))
12508 12509 12510 12511 12512 12513 12514 12515 12516 12517 12518 12519 12520 12521 12522
        goto cleanup;

    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   vm->def, migratableSrc,
                                                   dst, migratableDst);

 cleanup:
    virDomainDefFree(migratableSrc);
    virDomainDefFree(migratableDst);
    return ret;
}

#undef COPY_FLAGS


12523
bool
12524
qemuDomainAgentAvailable(virDomainObjPtr vm,
12525 12526
                         bool reportError)
{
12527 12528
    qemuDomainObjPrivatePtr priv = vm->privateData;

12529 12530 12531 12532 12533 12534 12535
    if (virDomainObjGetState(vm, NULL) != VIR_DOMAIN_RUNNING) {
        if (reportError) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
        }
        return false;
    }
12536 12537 12538 12539 12540 12541 12542 12543 12544
    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) {
12545 12546 12547 12548 12549 12550 12551 12552 12553 12554 12555 12556
        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;
12557 12558 12559 12560
        }
    }
    return true;
}
12561

12562

12563
static unsigned long long
12564
qemuDomainGetMemorySizeAlignment(virDomainDefPtr def)
12565
{
12566 12567 12568 12569 12570
    /* 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;

12571
    /* Align memory size. QEMU requires rounding to next 4KiB block.
J
Ján Tomko 已提交
12572
     * We'll take the "traditional" path and round it to 1MiB */
12573 12574 12575 12576 12577

    return 1024;
}


12578 12579
static unsigned long long
qemuDomainGetMemoryModuleSizeAlignment(const virDomainDef *def,
J
Ján Tomko 已提交
12580
                                       const virDomainMemoryDef *mem G_GNUC_UNUSED)
12581 12582 12583 12584 12585 12586 12587 12588 12589 12590 12591 12592
{
    /* 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;
}


12593 12594 12595 12596 12597 12598 12599 12600 12601 12602 12603 12604 12605 12606 12607 12608 12609 12610 12611 12612 12613 12614 12615 12616 12617 12618 12619 12620 12621 12622 12623 12624 12625 12626 12627 12628 12629 12630 12631 12632
static int
qemuDomainNVDimmAlignSizePseries(virDomainDefPtr def,
                                 virDomainMemoryDefPtr mem)
{
    /* For NVDIMMs in ppc64 in we want to align down the guest
     * visible space, instead of align up, to avoid writing
     * beyond the end of file by adding a potential 256MiB
     * to the user specified size.
     *
     * The label-size is mandatory for ppc64 as well, meaning that
     * the guest visible space will be target_size-label_size.
     *
     * Finally, target_size must include label_size.
     *
     * The above can be summed up as follows:
     *
     * target_size = AlignDown(target_size - label_size) + label_size
     */
    unsigned long long ppc64AlignSize = qemuDomainGetMemorySizeAlignment(def);
    unsigned long long guestArea = mem->size - mem->labelsize;

    /* Align down guest_area. 256MiB is the minimum size. Error
     * out if target_size is smaller than 256MiB + label_size,
     * since aligning it up will cause QEMU errors. */
    if (mem->size < (ppc64AlignSize + mem->labelsize)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("minimum target size for the NVDIMM "
                         "must be 256MB plus the label size"));
        return -1;
    }

    guestArea = (guestArea/ppc64AlignSize) * ppc64AlignSize;
    guestArea = MAX(guestArea, ppc64AlignSize);

    mem->size = guestArea + mem->labelsize;

    return 0;
}


12633 12634 12635
int
qemuDomainAlignMemorySizes(virDomainDefPtr def)
{
12636 12637
    unsigned long long maxmemkb = virMemoryMaxValue(false) >> 10;
    unsigned long long maxmemcapped = virMemoryMaxValue(true) >> 10;
12638
    unsigned long long initialmem = 0;
12639
    unsigned long long hotplugmem = 0;
12640
    unsigned long long mem;
12641
    unsigned long long align = qemuDomainGetMemorySizeAlignment(def);
12642 12643 12644 12645 12646
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    /* align NUMA cell sizes if relevant */
    for (i = 0; i < ncells; i++) {
12647 12648
        mem = VIR_ROUND_UP(virDomainNumaGetNodeMemorySize(def->numa, i), align);
        initialmem += mem;
12649 12650 12651 12652 12653 12654 12655

        if (mem > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("memory size of NUMA node '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
12656
        virDomainNumaSetNodeMemorySize(def->numa, i, mem);
12657 12658
    }

12659 12660 12661 12662 12663
    /* 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);

12664 12665 12666 12667 12668 12669
    if (initialmem > maxmemcapped) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("initial memory size overflowed after alignment"));
        return -1;
    }

12670
    def->mem.max_memory = VIR_ROUND_UP(def->mem.max_memory, align);
12671 12672 12673 12674 12675
    if (def->mem.max_memory > maxmemkb) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("maximum memory size overflowed after alignment"));
        return -1;
    }
12676

12677
    /* Align memory module sizes */
12678
    for (i = 0; i < def->nmems; i++) {
12679 12680 12681 12682 12683 12684 12685 12686 12687
        if (def->mems[i]->model == VIR_DOMAIN_MEMORY_MODEL_NVDIMM &&
            ARCH_IS_PPC64(def->os.arch)) {
            if (qemuDomainNVDimmAlignSizePseries(def, def->mems[i]) < 0)
                return -1;
        } else {
            align = qemuDomainGetMemoryModuleSizeAlignment(def, def->mems[i]);
            def->mems[i]->size = VIR_ROUND_UP(def->mems[i]->size, align);
        }

12688
        hotplugmem += def->mems[i]->size;
12689 12690 12691 12692 12693 12694 12695

        if (def->mems[i]->size > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("size of memory module '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
12696
    }
12697

12698 12699
    virDomainDefSetMemoryTotal(def, initialmem + hotplugmem);

12700 12701
    return 0;
}
12702 12703 12704 12705 12706 12707 12708 12709 12710 12711


/**
 * 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).
 */
12712
int
12713 12714
qemuDomainMemoryDeviceAlignSize(virDomainDefPtr def,
                                virDomainMemoryDefPtr mem)
12715
{
12716 12717 12718 12719 12720 12721 12722 12723 12724
    if (mem->model == VIR_DOMAIN_MEMORY_MODEL_NVDIMM &&
        ARCH_IS_PPC64(def->os.arch)) {
        return qemuDomainNVDimmAlignSizePseries(def, mem);
    } else {
        mem->size = VIR_ROUND_UP(mem->size,
                                 qemuDomainGetMemorySizeAlignment(def));
    }

    return 0;
12725
}
12726 12727 12728 12729 12730 12731 12732 12733 12734 12735 12736 12737 12738


/**
 * qemuDomainGetMonitor:
 * @vm: domain object
 *
 * Returns the monitor pointer corresponding to the domain object @vm.
 */
qemuMonitorPtr
qemuDomainGetMonitor(virDomainObjPtr vm)
{
    return ((qemuDomainObjPrivatePtr) vm->privateData)->mon;
}
12739 12740


12741 12742 12743 12744 12745 12746 12747
/**
 * 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.
 */
12748
virDomainChrDefPtr
12749 12750 12751 12752 12753 12754 12755 12756 12757 12758
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;

12759 12760
        if (STREQ_NULLABLE(channel->target.name, "org.qemu.guest_agent.0"))
            return channel;
12761 12762
    }

12763
    return NULL;
12764
}
12765 12766


12767
static bool
12768
qemuDomainMachineIsQ35(const char *machine,
12769
                       const virArch arch)
12770
{
12771 12772 12773
    if (!ARCH_IS_X86(arch))
        return false;

12774 12775 12776 12777 12778 12779
    if (STREQ(machine, "q35") ||
        STRPREFIX(machine, "pc-q35-")) {
        return true;
    }

    return false;
12780
}
12781 12782


12783
static bool
12784
qemuDomainMachineIsI440FX(const char *machine,
12785
                          const virArch arch)
12786
{
12787 12788 12789
    if (!ARCH_IS_X86(arch))
        return false;

12790 12791 12792 12793 12794 12795 12796 12797 12798
    if (STREQ(machine, "pc") ||
        STRPREFIX(machine, "pc-0.") ||
        STRPREFIX(machine, "pc-1.") ||
        STRPREFIX(machine, "pc-i440fx-") ||
        STRPREFIX(machine, "rhel")) {
        return true;
    }

    return false;
12799 12800 12801
}


12802
static bool
12803
qemuDomainMachineIsS390CCW(const char *machine,
12804
                           const virArch arch)
12805
{
12806 12807 12808
    if (!ARCH_IS_S390(arch))
        return false;

12809 12810 12811 12812
    if (STRPREFIX(machine, "s390-ccw"))
        return true;

    return false;
A
Andrea Bolognani 已提交
12813
}
12814

A
Andrea Bolognani 已提交
12815

12816 12817
/* You should normally avoid this function and use
 * qemuDomainIsARMVirt() instead. */
A
Andrea Bolognani 已提交
12818 12819 12820 12821 12822 12823
bool
qemuDomainMachineIsARMVirt(const char *machine,
                           const virArch arch)
{
    if (arch != VIR_ARCH_ARMV6L &&
        arch != VIR_ARCH_ARMV7L &&
12824
        arch != VIR_ARCH_AARCH64) {
12825
        return false;
12826
    }
12827

12828 12829 12830 12831
    if (STREQ(machine, "virt") ||
        STRPREFIX(machine, "virt-")) {
        return true;
    }
12832

12833
    return false;
12834 12835 12836
}


12837
static bool
A
Andrea Bolognani 已提交
12838 12839
qemuDomainMachineIsRISCVVirt(const char *machine,
                             const virArch arch)
12840
{
A
Andrea Bolognani 已提交
12841 12842
    if (!ARCH_IS_RISCV(arch))
        return false;
12843

12844 12845 12846 12847
    if (STREQ(machine, "virt") ||
        STRPREFIX(machine, "virt-")) {
        return true;
    }
12848

12849
    return false;
A
Andrea Bolognani 已提交
12850 12851 12852
}


12853 12854
/* You should normally avoid this function and use
 * qemuDomainIsPSeries() instead. */
A
Andrea Bolognani 已提交
12855 12856 12857 12858 12859 12860 12861
bool
qemuDomainMachineIsPSeries(const char *machine,
                           const virArch arch)
{
    if (!ARCH_IS_PPC64(arch))
        return false;

12862 12863 12864 12865
    if (STREQ(machine, "pseries") ||
        STRPREFIX(machine, "pseries-")) {
        return true;
    }
12866

12867
    return false;
12868 12869 12870
}


12871 12872
/* You should normally avoid this function and use
 * qemuDomainHasBuiltinIDE() instead. */
12873
bool
12874 12875
qemuDomainMachineHasBuiltinIDE(const char *machine,
                               const virArch arch)
12876
{
12877
    return qemuDomainMachineIsI440FX(machine, arch) ||
A
Andrea Bolognani 已提交
12878 12879 12880
        STREQ(machine, "malta") ||
        STREQ(machine, "sun4u") ||
        STREQ(machine, "g3beige");
12881 12882 12883
}


12884
static bool
12885
qemuDomainMachineNeedsFDC(const char *machine,
12886
                          const virArch arch)
12887 12888
{
    const char *p = STRSKIP(machine, "pc-q35-");
12889

12890 12891 12892
    if (!ARCH_IS_X86(arch))
        return false;

12893 12894 12895 12896 12897 12898 12899 12900 12901
    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;
12902
    }
12903 12904

    return true;
12905 12906 12907
}


12908
bool
A
Andrea Bolognani 已提交
12909
qemuDomainIsQ35(const virDomainDef *def)
12910
{
12911
    return qemuDomainMachineIsQ35(def->os.machine, def->os.arch);
12912 12913
}

12914

J
Ján Tomko 已提交
12915
bool
A
Andrea Bolognani 已提交
12916
qemuDomainIsI440FX(const virDomainDef *def)
J
Ján Tomko 已提交
12917
{
12918
    return qemuDomainMachineIsI440FX(def->os.machine, def->os.arch);
12919 12920 12921 12922
}


bool
A
Andrea Bolognani 已提交
12923
qemuDomainIsS390CCW(const virDomainDef *def)
12924
{
12925
    return qemuDomainMachineIsS390CCW(def->os.machine, def->os.arch);
12926 12927 12928 12929
}


bool
A
Andrea Bolognani 已提交
12930
qemuDomainIsARMVirt(const virDomainDef *def)
12931
{
A
Andrea Bolognani 已提交
12932
    return qemuDomainMachineIsARMVirt(def->os.machine, def->os.arch);
J
Ján Tomko 已提交
12933 12934 12935
}


12936 12937 12938 12939 12940 12941 12942 12943
bool
qemuDomainIsRISCVVirt(const virDomainDef *def)
{
    return qemuDomainMachineIsRISCVVirt(def->os.machine, def->os.arch);
}


bool
A
Andrea Bolognani 已提交
12944
qemuDomainIsPSeries(const virDomainDef *def)
12945
{
A
Andrea Bolognani 已提交
12946 12947 12948 12949 12950 12951 12952 12953 12954 12955
    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)
12956 12957
        return false;

A
Andrea Bolognani 已提交
12958
    if (def->controllers[root]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)
12959 12960 12961 12962 12963 12964
        return false;

    return true;
}


12965
bool
A
Andrea Bolognani 已提交
12966
qemuDomainHasPCIeRoot(const virDomainDef *def)
12967
{
A
Andrea Bolognani 已提交
12968 12969 12970 12971 12972 12973 12974 12975 12976
    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;
12977 12978 12979 12980
}


bool
A
Andrea Bolognani 已提交
12981
qemuDomainHasBuiltinIDE(const virDomainDef *def)
12982
{
12983
    return qemuDomainMachineHasBuiltinIDE(def->os.machine, def->os.arch);
A
Andrea Bolognani 已提交
12984
}
12985 12986


A
Andrea Bolognani 已提交
12987 12988 12989
bool
qemuDomainNeedsFDC(const virDomainDef *def)
{
12990
    return qemuDomainMachineNeedsFDC(def->os.machine, def->os.arch);
12991 12992 12993
}


12994 12995 12996 12997
bool
qemuDomainSupportsPCI(virDomainDefPtr def,
                      virQEMUCapsPtr qemuCaps)
{
12998 12999 13000 13001
    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)) {
13002
        return true;
13003
    }
13004 13005 13006 13007 13008 13009

    if (STREQ(def->os.machine, "versatilepb"))
        return true;

    if ((qemuDomainIsARMVirt(def) ||
         qemuDomainIsRISCVVirt(def)) &&
13010
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX)) {
13011
        return true;
13012
    }
13013 13014 13015 13016 13017

    return false;
}


13018 13019 13020 13021 13022 13023 13024 13025 13026 13027 13028 13029 13030 13031 13032 13033 13034 13035 13036 13037 13038 13039 13040 13041 13042 13043 13044 13045 13046 13047 13048 13049 13050 13051 13052 13053 13054 13055 13056
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
13057
    case VIR_DOMAIN_MEMORY_MODEL_NVDIMM:
13058 13059 13060 13061 13062 13063 13064 13065
        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;
        }

13066 13067 13068
        if (virDomainNumaGetNodeCount(def->numa) != 0) {
            if (mem->targetNode == -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
Y
Yuri Chornoivan 已提交
13069
                               _("target NUMA node needs to be specified for "
13070 13071 13072
                                 "memory device"));
                return -1;
            }
13073 13074
        }

13075 13076 13077 13078 13079 13080 13081 13082 13083 13084 13085 13086 13087 13088 13089 13090 13091 13092 13093 13094 13095 13096 13097 13098
        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;
}


13099 13100 13101 13102 13103 13104 13105 13106 13107 13108 13109 13110 13111 13112 13113 13114 13115 13116 13117 13118 13119
/**
 * 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
13120 13121
    bool needPCDimmCap = false;
    bool needNvdimmCap = false;
13122 13123 13124 13125 13126 13127 13128
    size_t i;

    hotplugSpace = def->mem.max_memory - virDomainDefGetMemoryInitial(def);

    if (mem) {
        nmems++;
        hotplugMemory = mem->size;
13129 13130 13131

        if (qemuDomainDefValidateMemoryHotplugDevice(mem, def) < 0)
            return -1;
13132 13133 13134 13135 13136 13137 13138 13139 13140 13141 13142 13143 13144
    }

    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;
    }

13145 13146 13147 13148 13149 13150 13151 13152 13153 13154 13155
    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;
        }
13156 13157 13158 13159 13160 13161 13162 13163 13164
    }

    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;
    }

13165
    for (i = 0; i < def->nmems; i++) {
13166 13167
        hotplugMemory += def->mems[i]->size;

M
Michal Privoznik 已提交
13168 13169 13170 13171 13172 13173 13174 13175 13176 13177 13178 13179 13180 13181
        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;
        }

13182 13183 13184 13185 13186 13187
        /* 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
13188 13189 13190 13191 13192 13193 13194 13195 13196 13197 13198 13199 13200 13201
    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;
    }

13202 13203 13204 13205 13206 13207 13208 13209 13210 13211
    if (hotplugMemory > hotplugSpace) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory device total size exceeds hotplug space"));
        return -1;
    }

    return 0;
}


13212 13213 13214
/**
 * qemuDomainUpdateCurrentMemorySize:
 *
13215 13216
 * In case when the balloon is not present for the domain, the function
 * recalculates the maximum size to reflect possible changes.
13217
 */
13218 13219
void
qemuDomainUpdateCurrentMemorySize(virDomainObjPtr vm)
13220 13221 13222
{
    /* inactive domain doesn't need size update */
    if (!virDomainObjIsActive(vm))
13223
        return;
13224 13225 13226

    /* if no balloning is available, the current size equals to the current
     * full memory size */
13227
    if (!virDomainDefHasMemballoon(vm->def))
13228
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
13229
}
13230 13231


13232 13233 13234 13235 13236 13237 13238 13239
/**
 * 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.
 */
13240
static bool
13241 13242 13243 13244 13245 13246
ppc64VFIODeviceIsNV2Bridge(const char *device)
{
    const char *nvlink2Files[] = {"ibm,gpu", "ibm,nvlink",
                                  "ibm,nvlink-speed", "memory-region"};
    size_t i;

13247
    for (i = 0; i < G_N_ELEMENTS(nvlink2Files); i++) {
13248
        g_autofree char *file = NULL;
13249

13250 13251
        file = g_strdup_printf("/sys/bus/pci/devices/%s/of_node/%s",
                               device, nvlink2Files[i]);
13252 13253 13254 13255 13256 13257 13258 13259 13260

        if (!virFileExists(file))
            return false;
    }

    return true;
}


13261 13262 13263
/**
 * getPPC64MemLockLimitBytes:
 * @def: domain definition
13264
 * @forceVFIO: force VFIO usage
13265 13266 13267 13268 13269
 *
 * A PPC64 helper that calculates the memory locking limit in order for
 * the guest to operate properly.
 */
static unsigned long long
13270 13271
getPPC64MemLockLimitBytes(virDomainDefPtr def,
                          bool forceVFIO)
13272 13273 13274 13275 13276 13277 13278
{
    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;
13279
    virPCIDeviceAddressPtr pciAddr;
13280
    bool usesVFIO = false;
13281
    bool nvlink2Capable = false;
13282 13283 13284 13285 13286 13287 13288 13289 13290 13291 13292 13293 13294

    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];

13295
        if (virHostdevIsVFIODevice(dev)) {
13296
            usesVFIO = true;
13297 13298 13299

            pciAddr = &dev->source.subsys.u.pci.addr;
            if (virPCIDeviceAddressIsValid(pciAddr, false)) {
13300
                g_autofree char *pciAddrStr = NULL;
13301 13302 13303 13304 13305 13306 13307

                pciAddrStr = virPCIDeviceAddressAsString(pciAddr);
                if (ppc64VFIODeviceIsNV2Bridge(pciAddrStr)) {
                    nvlink2Capable = true;
                    break;
                }
            }
13308 13309 13310
        }
    }

13311 13312 13313
    if (virDomainDefHasNVMeDisk(def))
        usesVFIO = true;

13314 13315 13316 13317 13318 13319 13320 13321 13322 13323 13324 13325 13326 13327 13328 13329 13330 13331 13332 13333 13334 13335 13336
    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;

13337 13338 13339 13340 13341
    /* 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.
13342
     *
13343 13344 13345
     * 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.
13346
     *
13347 13348 13349 13350
     * 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.
13351
     *
13352 13353 13354 13355 13356 13357 13358 13359 13360 13361
     * 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;
13362
    } else if (usesVFIO || forceVFIO) {
13363 13364 13365 13366 13367 13368 13369 13370 13371 13372 13373 13374 13375 13376 13377 13378 13379 13380 13381 13382 13383 13384 13385
        /* 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) */
13386 13387 13388
        passthroughLimit = MAX(2 * 1024 * 1024 * nPCIHostBridges,
                               memory +
                               memory / 512 * nPCIHostBridges + 8192);
13389
    }
13390 13391 13392 13393 13394 13395 13396

    memKB = baseLimit + passthroughLimit;

    return memKB << 10;
}


13397
/**
13398
 * qemuDomainGetMemLockLimitBytes:
13399
 * @def: domain definition
13400
 * @forceVFIO: force VFIO calculation
13401
 *
13402 13403 13404 13405
 * 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.
13406 13407
 * The @forceVFIO argument can be used to tell this function will use VFIO even
 * though @def doesn't indicates so right now.
13408 13409
 *
 * Returns: the memory locking limit, or 0 if setting the limit is not needed
13410 13411
 */
unsigned long long
13412 13413
qemuDomainGetMemLockLimitBytes(virDomainDefPtr def,
                               bool forceVFIO)
13414
{
13415
    unsigned long long memKB = 0;
13416
    bool usesVFIO = false;
13417
    size_t i;
13418

13419 13420 13421
    /* prefer the hard limit */
    if (virMemoryLimitIsSet(def->mem.hard_limit)) {
        memKB = def->mem.hard_limit;
13422
        return memKB << 10;
13423 13424
    }

13425 13426 13427 13428 13429 13430 13431
    /* 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;
13432

13433
    if (ARCH_IS_PPC64(def->os.arch) && def->virtType == VIR_DOMAIN_VIRT_KVM)
13434
        return getPPC64MemLockLimitBytes(def, forceVFIO);
13435

13436 13437 13438 13439 13440 13441 13442 13443 13444 13445 13446 13447 13448 13449 13450 13451 13452 13453
    /* 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.
     */
13454 13455 13456 13457 13458 13459 13460
    if (!forceVFIO) {
        for (i = 0; i < def->nhostdevs; i++) {
            if (virHostdevIsVFIODevice(def->hostdevs[i]) ||
                virHostdevIsMdevDevice(def->hostdevs[i])) {
                usesVFIO = true;
                break;
            }
13461
        }
13462

13463 13464 13465
        if (virDomainDefHasNVMeDisk(def))
            usesVFIO = true;
    }
13466

13467
    if (usesVFIO || forceVFIO)
13468 13469
        memKB = virDomainDefGetMemoryTotal(def) + 1024 * 1024;

13470
    return memKB << 10;
13471
}
13472

13473

13474 13475 13476
/**
 * qemuDomainAdjustMaxMemLock:
 * @vm: domain
13477
 * @forceVFIO: apply VFIO requirements even if vm's def doesn't require it
13478 13479
 *
 * Adjust the memory locking limit for the QEMU process associated to @vm, in
13480 13481 13482
 * order to comply with VFIO or architecture requirements. If @forceVFIO is
 * true then the limit is changed even if nothing in @vm's definition indicates
 * so.
13483
 *
13484 13485 13486 13487
 * 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.
13488 13489 13490 13491
 *
 * Returns: 0 on success, <0 on failure
 */
int
13492 13493
qemuDomainAdjustMaxMemLock(virDomainObjPtr vm,
                           bool forceVFIO)
13494 13495 13496
{
    unsigned long long bytes = 0;

13497
    bytes = qemuDomainGetMemLockLimitBytes(vm->def, forceVFIO);
13498 13499

    if (bytes) {
13500 13501 13502 13503 13504 13505 13506 13507 13508 13509 13510 13511 13512 13513
        /* 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;
    }
13514 13515 13516

    /* Trying to set the memory locking limit to zero is a no-op */
    if (virProcessSetMaxMemLock(vm->pid, bytes) < 0)
13517
        return -1;
13518

13519
    return 0;
13520
}
13521

13522 13523 13524 13525 13526 13527 13528 13529 13530 13531 13532 13533 13534 13535 13536 13537 13538 13539 13540 13541 13542

/**
 * 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;
13543
    if (qemuDomainAdjustMaxMemLock(vm, false) < 0)
13544 13545 13546 13547 13548 13549 13550 13551
        ret = -1;

    vm->def->hostdevs[--(vm->def->nhostdevs)] = NULL;

    return ret;
}


13552 13553 13554 13555 13556 13557 13558 13559 13560
/**
 * qemuDomainHasVcpuPids:
 * @vm: Domain object
 *
 * Returns true if we were able to successfully detect vCPU pids for the VM.
 */
bool
qemuDomainHasVcpuPids(virDomainObjPtr vm)
{
13561 13562 13563 13564 13565 13566
    size_t i;
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    virDomainVcpuDefPtr vcpu;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
13567

13568 13569 13570 13571 13572
        if (QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid > 0)
            return true;
    }

    return false;
13573
}
13574 13575 13576 13577 13578 13579 13580 13581 13582 13583 13584


/**
 * 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,
13585
                     unsigned int vcpuid)
13586
{
13587 13588
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
    return QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid;
13589
}
13590 13591


13592 13593 13594 13595 13596 13597 13598 13599 13600 13601 13602 13603 13604 13605 13606 13607 13608 13609 13610 13611 13612 13613 13614 13615 13616 13617 13618 13619 13620 13621 13622 13623 13624 13625 13626 13627 13628 13629 13630 13631 13632 13633
/**
 * 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;
}


13634 13635 13636 13637 13638 13639 13640 13641 13642
bool
qemuDomainSupportsNewVcpuHotplug(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
}


13643
/**
13644
 * qemuDomainRefreshVcpuInfo:
13645 13646 13647
 * @driver: qemu driver data
 * @vm: domain object
 * @asyncJob: current asynchronous job type
13648
 * @state: refresh vcpu state
13649
 *
13650 13651
 * Updates vCPU information private data of @vm. Due to historical reasons this
 * function returns success even if some data were not reported by qemu.
13652
 *
13653 13654
 * If @state is true, the vcpu state is refreshed as reported by the monitor.
 *
13655
 * Returns 0 on success and -1 on fatal error.
13656 13657
 */
int
13658 13659
qemuDomainRefreshVcpuInfo(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
13660 13661
                          int asyncJob,
                          bool state)
13662
{
13663
    virDomainVcpuDefPtr vcpu;
13664 13665
    qemuDomainVcpuPrivatePtr vcpupriv;
    qemuMonitorCPUInfoPtr info = NULL;
13666
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
13667
    size_t i, j;
13668
    bool hotplug;
13669
    bool fast;
13670
    bool validTIDs = true;
13671
    int rc;
13672
    int ret = -1;
13673

13674
    hotplug = qemuDomainSupportsNewVcpuHotplug(vm);
13675 13676
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
13677

13678 13679
    VIR_DEBUG("Maxvcpus %zu hotplug %d fast query %d", maxvcpus, hotplug, fast);

13680 13681
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
13682

13683 13684
    rc = qemuMonitorGetCPUInfo(qemuDomainGetMonitor(vm), &info, maxvcpus,
                               hotplug, fast);
13685

13686
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
13687 13688
        goto cleanup;

13689
    if (rc < 0)
13690 13691
        goto cleanup;

13692 13693 13694 13695 13696 13697 13698 13699 13700 13701 13702 13703 13704 13705 13706 13707 13708 13709 13710 13711 13712 13713 13714 13715 13716 13717 13718 13719 13720 13721 13722 13723 13724
    /*
     * 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++) {
13725
            if (info[i].tid != 0 && info[i].tid == info[j].tid) {
13726 13727 13728 13729 13730 13731 13732
                VIR_DEBUG("vCPU[%zu] PID %llu duplicates vCPU[%zu]",
                          i, (unsigned long long)info[i].tid, j);
                validTIDs = false;
            }
        }

        if (validTIDs)
13733 13734 13735 13736 13737 13738 13739 13740
            VIR_DEBUG("vCPU[%zu] PID %llu is valid "
                      "(node=%d socket=%d die=%d core=%d thread=%d)",
                      i, (unsigned long long)info[i].tid,
                      info[i].node_id,
                      info[i].socket_id,
                      info[i].die_id,
                      info[i].core_id,
                      info[i].thread_id);
13741 13742 13743
    }

    VIR_DEBUG("Extracting vCPU information validTIDs=%d", validTIDs);
13744 13745
    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
13746
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
13747

13748
        if (validTIDs)
13749 13750 13751 13752 13753
            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;
13754
        vcpupriv->node_id = info[i].node_id;
13755 13756
        vcpupriv->vcpus = info[i].vcpus;
        VIR_FREE(vcpupriv->type);
13757
        vcpupriv->type = g_steal_pointer(&info[i].type);
13758
        VIR_FREE(vcpupriv->alias);
13759
        vcpupriv->alias = g_steal_pointer(&info[i].alias);
13760
        virJSONValueFree(vcpupriv->props);
13761
        vcpupriv->props = g_steal_pointer(&info[i].props);
13762
        vcpupriv->enable_id = info[i].id;
13763
        vcpupriv->qemu_id = info[i].qemu_id;
13764

13765
        if (hotplug && state) {
13766 13767 13768 13769 13770
            vcpu->online = info[i].online;
            if (info[i].hotpluggable)
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_YES;
            else
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_NO;
13771
        }
13772 13773
    }

13774
    ret = 0;
13775 13776

 cleanup:
13777
    qemuMonitorCPUInfoFree(info, maxvcpus);
13778
    return ret;
13779
}
13780

13781 13782 13783 13784 13785 13786 13787 13788 13789 13790 13791 13792 13793 13794 13795 13796 13797 13798 13799 13800 13801 13802 13803 13804 13805 13806 13807 13808 13809 13810 13811
/**
 * 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;
13812
    qemuDomainVcpuPrivatePtr vcpupriv;
13813
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
13814
    virBitmapPtr haltedmap = NULL;
13815 13816
    size_t i;
    int ret = -1;
13817
    bool fast;
13818 13819 13820 13821 13822

    /* Not supported currently for TCG, see qemuDomainRefreshVcpuInfo */
    if (vm->def->virtType == VIR_DOMAIN_VIRT_QEMU)
        return 0;

13823
    /* The halted state is interresting only on s390(x). On other platforms
13824 13825 13826 13827 13828 13829 13830
     * 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))
13831 13832
        return 0;

13833 13834 13835
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;

13836 13837 13838 13839
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
    haltedmap = qemuMonitorGetCpuHalted(qemuDomainGetMonitor(vm), maxvcpus,
                                        fast);
13840
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || !haltedmap)
13841 13842 13843 13844
        goto cleanup;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
13845
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
13846 13847
        vcpupriv->halted = virTristateBoolFromBool(virBitmapIsBitSet(haltedmap,
                                                                     vcpupriv->qemu_id));
13848 13849 13850 13851 13852
    }

    ret = 0;

 cleanup:
13853
    virBitmapFree(haltedmap);
13854 13855
    return ret;
}
13856 13857 13858 13859 13860 13861

bool
qemuDomainSupportsNicdev(virDomainDefPtr def,
                         virDomainNetDefPtr net)
{
    /* non-virtio ARM nics require legacy -net nic */
S
Stefan Schallenberg 已提交
13862 13863
    if (((def->os.arch == VIR_ARCH_ARMV6L) ||
        (def->os.arch == VIR_ARCH_ARMV7L) ||
13864 13865 13866 13867 13868 13869 13870 13871
        (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;
}

13872 13873 13874 13875
bool
qemuDomainNetSupportsMTU(virDomainNetType type)
{
    switch (type) {
13876 13877
    case VIR_DOMAIN_NET_TYPE_NETWORK:
    case VIR_DOMAIN_NET_TYPE_BRIDGE:
13878 13879
    case VIR_DOMAIN_NET_TYPE_ETHERNET:
    case VIR_DOMAIN_NET_TYPE_VHOSTUSER:
13880 13881
        return true;
    case VIR_DOMAIN_NET_TYPE_USER:
13882 13883 13884 13885 13886 13887 13888 13889 13890 13891 13892 13893
    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
13894

P
Peter Krempa 已提交
13895 13896 13897 13898 13899 13900 13901 13902

virDomainDiskDefPtr
qemuDomainDiskByName(virDomainDefPtr def,
                     const char *name)
{
    virDomainDiskDefPtr ret;

    if (!(ret = virDomainDiskByName(def, name, true))) {
13903 13904
        virReportError(VIR_ERR_INVALID_ARG,
                       _("disk '%s' not found in domain"), name);
P
Peter Krempa 已提交
13905 13906 13907 13908 13909
        return NULL;
    }

    return ret;
}
13910 13911 13912 13913 13914 13915 13916 13917 13918 13919 13920 13921 13922 13923 13924 13925 13926 13927 13928 13929 13930 13931 13932 13933 13934 13935 13936 13937 13938 13939 13940


/**
 * 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;
}
13941 13942 13943 13944 13945 13946


int
qemuDomainPrepareChannel(virDomainChrDefPtr channel,
                         const char *domainChannelTargetDir)
{
S
Scott Garfinkle 已提交
13947 13948 13949 13950 13951 13952
    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) {
13953 13954 13955
        channel->source->data.nix.path = g_strdup_printf("%s/%s",
                                                         domainChannelTargetDir,
                                                         channel->target.name);
13956 13957
    } else {
        /* Generate a unique name */
13958 13959 13960 13961 13962
        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);
13963 13964 13965 13966
    }

    return 0;
}
13967 13968


13969
/* qemuDomainPrepareChardevSourceTLS:
13970 13971 13972 13973 13974 13975 13976 13977 13978 13979 13980 13981 13982 13983 13984 13985 13986
 * @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;
13987
            source->data.tcp.tlsFromConfig = true;
13988 13989 13990 13991 13992
        }
    }
}


13993
/* qemuDomainPrepareChardevSource:
13994
 * @def: live domain definition
13995
 * @cfg: driver configuration
13996 13997 13998 13999 14000 14001
 *
 * Iterate through all devices that use virDomainChrSourceDefPtr as host
 * interface part.
 */
void
qemuDomainPrepareChardevSource(virDomainDefPtr def,
14002
                               virQEMUDriverConfigPtr cfg)
14003 14004 14005 14006 14007 14008 14009 14010 14011 14012 14013 14014 14015 14016 14017 14018 14019 14020 14021 14022 14023 14024 14025 14026 14027 14028 14029 14030 14031
{
    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);
}


14032 14033 14034 14035 14036 14037 14038 14039 14040 14041 14042 14043 14044 14045 14046
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;
    }

14047 14048
    if (src->haveTLS == VIR_TRISTATE_BOOL_YES)
        src->tlsCertdir = g_strdup(cfg->vxhsTLSx509certdir);
14049 14050 14051 14052 14053

    return 0;
}


14054 14055 14056 14057 14058 14059 14060 14061 14062 14063 14064 14065 14066 14067 14068 14069 14070 14071 14072 14073
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;
        }

14074
        src->tlsCertdir = g_strdup(cfg->nbdTLSx509certdir);
14075 14076 14077 14078 14079 14080
    }

    return 0;
}


14081
/* qemuProcessPrepareStorageSourceTLS:
14082
 * @source: source for a disk
14083
 * @cfg: driver configuration
14084
 * @parentAlias: alias of the parent device
14085 14086 14087 14088 14089 14090 14091
 *
 * 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
 */
14092
static int
14093
qemuDomainPrepareStorageSourceTLS(virStorageSourcePtr src,
14094
                                  virQEMUDriverConfigPtr cfg,
14095 14096
                                  const char *parentAlias,
                                  virQEMUCapsPtr qemuCaps)
14097
{
14098 14099 14100 14101 14102 14103 14104 14105
    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;
14106

14107
    case VIR_STORAGE_NET_PROTOCOL_NBD:
14108 14109 14110 14111
        if (qemuProcessPrepareStorageSourceTLSNBD(src, cfg, qemuCaps) < 0)
            return -1;
        break;

14112 14113 14114 14115 14116 14117 14118 14119 14120 14121
    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:
14122 14123 14124 14125 14126 14127
        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;
        }
14128 14129 14130 14131 14132 14133
        break;

    case VIR_STORAGE_NET_PROTOCOL_NONE:
    case VIR_STORAGE_NET_PROTOCOL_LAST:
    default:
        virReportEnumRangeError(virStorageNetProtocol, src->protocol);
14134
        return -1;
14135
    }
14136

14137 14138 14139 14140
    if (src->haveTLS == VIR_TRISTATE_BOOL_YES &&
        !(src->tlsAlias = qemuAliasTLSObjFromSrcAlias(parentAlias)))
        return -1;

14141 14142 14143 14144
    return 0;
}


14145
void
14146 14147 14148 14149
qemuDomainPrepareShmemChardev(virDomainShmemDefPtr shmem)
{
    if (!shmem->server.enabled ||
        shmem->server.chr.data.nix.path)
14150
        return;
14151

14152 14153
    shmem->server.chr.data.nix.path = g_strdup_printf("/var/lib/libvirt/shmem-%s-sock",
                                                      shmem->name);
14154 14155 14156
}


14157 14158 14159 14160 14161 14162 14163 14164 14165 14166 14167 14168 14169 14170 14171 14172 14173 14174 14175 14176 14177 14178 14179 14180 14181 14182 14183 14184 14185 14186 14187 14188 14189
/**
 * 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);
}
14190 14191 14192 14193 14194 14195 14196 14197 14198 14199 14200 14201 14202 14203 14204 14205 14206 14207 14208 14209 14210 14211 14212 14213 14214 14215 14216 14217 14218 14219 14220 14221 14222 14223


/**
 * 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;
        }
    }
}
14224 14225 14226 14227 14228 14229 14230 14231 14232 14233 14234 14235 14236 14237 14238 14239 14240 14241 14242 14243


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;
}
14244 14245 14246 14247 14248 14249


bool
qemuDomainSupportsVideoVga(virDomainVideoDefPtr video,
                           virQEMUCapsPtr qemuCaps)
{
14250 14251 14252 14253 14254 14255 14256 14257
    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;
        }
    }
14258 14259 14260

    return true;
}
14261 14262


14263 14264 14265 14266
bool
qemuDomainNeedsVFIO(const virDomainDef *def)
{
    return virDomainDefHasVFIOHostdev(def) ||
14267 14268
        virDomainDefHasMdevHostdev(def) ||
        virDomainDefHasNVMeDisk(def);
14269 14270 14271
}


14272 14273 14274 14275 14276 14277
/**
 * qemuDomainGetHostdevPath:
 * @dev: host device definition
 * @path: resulting path to @dev
 * @perms: Optional pointer to VIR_CGROUP_DEVICE_* perms
 *
14278
 * For given device @dev fetch its host path and store it at
14279
 * @path. Optionally, caller can get @perms on the path (e.g.
14280 14281
 * rw/ro). When called on a missing device, the function will return success
 * and store NULL at @path.
14282
 *
14283 14284
 * The caller is responsible for freeing the @path when no longer
 * needed.
14285 14286 14287 14288
 *
 * Returns 0 on success, -1 otherwise.
 */
int
14289 14290 14291
qemuDomainGetHostdevPath(virDomainHostdevDefPtr dev,
                         char **path,
                         int *perms)
14292 14293 14294 14295 14296
{
    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;
14297
    virDomainHostdevSubsysMediatedDevPtr mdevsrc = &dev->source.subsys.u.mdev;
14298 14299 14300 14301
    g_autoptr(virUSBDevice) usb = NULL;
    g_autoptr(virSCSIDevice) scsi = NULL;
    g_autoptr(virSCSIVHostDevice) host = NULL;
    g_autofree char *tmpPath = NULL;
14302
    int perm = 0;
14303 14304 14305

    switch ((virDomainHostdevMode) dev->mode) {
    case VIR_DOMAIN_HOSTDEV_MODE_SUBSYS:
14306
        switch ((virDomainHostdevSubsysType)dev->source.subsys.type) {
14307 14308
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
            if (pcisrc->backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO) {
14309
                if (!(tmpPath = virPCIDeviceAddressGetIOMMUGroupDev(&pcisrc->addr)))
14310
                    return -1;
14311 14312

                perm = VIR_CGROUP_DEVICE_RW;
14313 14314 14315 14316 14317 14318 14319 14320 14321 14322
            }
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
            if (dev->missing)
                break;
            usb = virUSBDeviceNew(usbsrc->bus,
                                  usbsrc->device,
                                  NULL);
            if (!usb)
14323
                return -1;
14324

J
Ján Tomko 已提交
14325
            tmpPath = g_strdup(virUSBDeviceGetPath(usb));
14326
            perm = VIR_CGROUP_DEVICE_RW;
14327 14328 14329 14330 14331
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI:
            if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI) {
                virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
14332
                VIR_DEBUG("Not updating /dev for hostdev iSCSI path '%s'", iscsisrc->src->path);
14333 14334 14335 14336 14337 14338 14339 14340 14341 14342 14343
            } else {
                virDomainHostdevSubsysSCSIHostPtr scsihostsrc = &scsisrc->u.host;
                scsi = virSCSIDeviceNew(NULL,
                                        scsihostsrc->adapter,
                                        scsihostsrc->bus,
                                        scsihostsrc->target,
                                        scsihostsrc->unit,
                                        dev->readonly,
                                        dev->shareable);

                if (!scsi)
14344
                    return -1;
14345

J
Ján Tomko 已提交
14346
                tmpPath = g_strdup(virSCSIDeviceGetPath(scsi));
14347 14348
                perm = virSCSIDeviceGetReadonly(scsi) ?
                    VIR_CGROUP_DEVICE_READ : VIR_CGROUP_DEVICE_RW;
14349 14350 14351 14352 14353 14354 14355
            }
            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)))
14356
                    return -1;
14357

J
Ján Tomko 已提交
14358
                tmpPath = g_strdup(virSCSIVHostDeviceGetPath(host));
14359
                perm = VIR_CGROUP_DEVICE_RW;
14360 14361 14362 14363
            }
            break;
        }

14364
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
14365
            if (!(tmpPath = virMediatedDeviceGetIOMMUGroupDev(mdevsrc->uuidstr)))
14366
                return -1;
14367 14368 14369

            perm = VIR_CGROUP_DEVICE_RW;
            break;
14370 14371 14372 14373 14374 14375 14376 14377 14378 14379 14380
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
            break;
        }
        break;

    case VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES:
    case VIR_DOMAIN_HOSTDEV_MODE_LAST:
        /* nada */
        break;
    }

14381 14382 14383
    *path = g_steal_pointer(&tmpPath);
    if (perms)
        *perms = perm;
14384
    return 0;
14385 14386 14387
}


14388 14389 14390 14391
/**
 * qemuDomainGetPreservedMountPath:
 * @cfg: driver configuration data
 * @vm: domain object
14392
 * @mountpoint: mount point path to convert
14393
 *
14394
 * For given @mountpoint return new path where the mount point
14395 14396 14397 14398 14399 14400 14401 14402
 * 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,
14403
                                const char *mountpoint)
14404 14405 14406
{
    char *path = NULL;
    char *tmp;
14407
    const char *suffix = mountpoint + strlen(QEMU_DEVPREFIX);
14408
    g_autofree char *domname = virDomainDefGetShortName(vm->def);
14409 14410
    size_t off;

14411 14412 14413
    if (!domname)
        return NULL;

14414
    if (STREQ(mountpoint, "/dev"))
14415 14416
        suffix = "dev";

14417
    path = g_strdup_printf("%s/%s.%s", cfg->stateDir, domname, suffix);
14418

14419
    /* Now consider that @mountpoint is "/dev/blah/blah2".
14420 14421 14422 14423 14424 14425 14426 14427 14428 14429 14430 14431 14432 14433 14434 14435 14436
     * @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++;
    }

    return path;
}


14437 14438 14439 14440 14441 14442 14443 14444 14445 14446 14447 14448
/**
 * 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
14449
qemuDomainGetPreservedMounts(virQEMUDriverConfigPtr cfg,
14450 14451 14452 14453 14454 14455
                             virDomainObjPtr vm,
                             char ***devPath,
                             char ***devSavePath,
                             size_t *ndevPath)
{
    char **paths = NULL, **mounts = NULL;
14456
    size_t i, j, nmounts;
14457

14458
    if (virFileGetMountSubtree(QEMU_PROC_MOUNTS, "/dev",
14459 14460 14461 14462 14463 14464 14465 14466 14467
                               &mounts, &nmounts) < 0)
        goto error;

    if (!nmounts) {
        if (ndevPath)
            *ndevPath = 0;
        return 0;
    }

14468 14469 14470 14471 14472 14473 14474 14475 14476 14477 14478 14479
    /* 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) {
14480 14481
            char *c = STRSKIP(mounts[j], mounts[i]);

14482
            if (c && (*c == '/' || *c == '\0')) {
14483 14484 14485 14486 14487 14488 14489 14490
                VIR_DEBUG("Dropping path %s because of %s", mounts[j], mounts[i]);
                VIR_DELETE_ELEMENT(mounts, j, nmounts);
            } else {
                j++;
            }
        }
    }

14491 14492 14493 14494
    if (VIR_ALLOC_N(paths, nmounts) < 0)
        goto error;

    for (i = 0; i < nmounts; i++) {
14495
        if (!(paths[i] = qemuDomainGetPreservedMountPath(cfg, vm, mounts[i])))
14496 14497 14498 14499 14500 14501 14502 14503 14504 14505 14506 14507 14508 14509 14510 14511 14512 14513 14514 14515 14516 14517 14518 14519 14520
            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;
}


14521 14522
struct qemuDomainCreateDeviceData {
    const char *path;     /* Path to temp new /dev location */
14523 14524
    char * const *devMountsPath;
    size_t ndevMountsPath;
14525 14526 14527
};


14528
static int
14529
qemuDomainCreateDeviceRecursive(const char *device,
14530
                                const struct qemuDomainCreateDeviceData *data,
14531 14532
                                bool allow_noent,
                                unsigned int ttl)
14533
{
14534 14535
    g_autofree char *devicePath = NULL;
    g_autofree char *target = NULL;
14536
    GStatBuf sb;
14537
    int ret = -1;
14538
    bool isLink = false;
14539
    bool isDev = false;
14540
    bool isReg = false;
14541
    bool isDir = false;
14542
    bool create = false;
14543 14544 14545
#ifdef WITH_SELINUX
    char *tcon = NULL;
#endif
14546

14547 14548 14549 14550 14551 14552 14553
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             device);
        return ret;
    }

14554
    if (g_lstat(device, &sb) < 0) {
14555 14556
        if (errno == ENOENT && allow_noent) {
            /* Ignore non-existent device. */
14557
            return 0;
14558
        }
14559 14560
        virReportSystemError(errno, _("Unable to stat %s"), device);
        return ret;
14561 14562
    }

14563
    isLink = S_ISLNK(sb.st_mode);
14564
    isDev = S_ISCHR(sb.st_mode) || S_ISBLK(sb.st_mode);
14565
    isReg = S_ISREG(sb.st_mode) || S_ISFIFO(sb.st_mode) || S_ISSOCK(sb.st_mode);
14566
    isDir = S_ISDIR(sb.st_mode);
14567 14568 14569 14570 14571 14572 14573 14574 14575 14576 14577 14578 14579 14580 14581 14582

    /* 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.
     */
14583
    if (STRPREFIX(device, QEMU_DEVPREFIX)) {
14584
        size_t i;
14585

14586 14587 14588 14589 14590 14591 14592 14593 14594 14595
        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. */
14596 14597
            devicePath = g_strdup_printf("%s/%s", data->path,
                                         device + strlen(QEMU_DEVPREFIX));
14598 14599 14600 14601 14602 14603 14604 14605 14606 14607 14608 14609

            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]);
14610
        }
14611 14612
    }

14613
    if (isLink) {
14614 14615
        g_autoptr(GError) gerr = NULL;

14616 14617
        /* We are dealing with a symlink. Create a dangling symlink and descend
         * down one level which hopefully creates the symlink's target. */
14618 14619 14620
        if (!(target = g_file_read_link(device, &gerr))) {
            virReportError(VIR_ERR_SYSTEM_ERROR,
                           _("failed to resolve symlink %s: %s"), device, gerr->message);
14621 14622
            goto cleanup;
        }
14623

14624 14625 14626 14627 14628 14629 14630 14631 14632
        if (create &&
            symlink(target, devicePath) < 0) {
            if (errno == EEXIST) {
                ret = 0;
            } else {
                virReportSystemError(errno,
                                     _("unable to create symlink %s"),
                                     devicePath);
            }
14633 14634 14635
            goto cleanup;
        }

14636 14637 14638 14639 14640 14641 14642 14643
        /* 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)
         */
14644
        if (!g_path_is_absolute(target)) {
14645 14646
            g_autofree char *devTmp = g_strdup(device);
            char *c = NULL, *tmp = NULL;
14647

14648 14649 14650
            if ((c = strrchr(devTmp, '/')))
                *(c + 1) = '\0';

14651
            tmp = g_strdup_printf("%s%s", devTmp, target);
14652
            VIR_FREE(target);
14653
            target = g_steal_pointer(&tmp);
14654 14655
        }

14656
        if (qemuDomainCreateDeviceRecursive(target, data,
14657
                                            allow_noent, ttl - 1) < 0)
14658
            goto cleanup;
14659
    } else if (isDev) {
14660 14661 14662
        if (create) {
            unlink(devicePath);
            if (mknod(devicePath, sb.st_mode, sb.st_rdev) < 0) {
14663 14664 14665
                virReportSystemError(errno,
                                     _("Failed to make device %s"),
                                     devicePath);
14666
                goto cleanup;
14667
            }
14668
        }
14669
    } else if (isReg) {
14670
        if (create &&
14671
            virFileTouch(devicePath, sb.st_mode) < 0)
14672
            goto cleanup;
14673 14674
        /* Just create the file here so that code below sets
         * proper owner and mode. Bind mount only after that. */
14675 14676
    } else if (isDir) {
        if (create &&
14677 14678 14679 14680
            virFileMakePathWithMode(devicePath, sb.st_mode) < 0) {
            virReportSystemError(errno,
                                 _("Unable to make dir %s"),
                                 devicePath);
14681
            goto cleanup;
14682
        }
14683 14684 14685 14686 14687
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       device, sb.st_mode);
        goto cleanup;
14688 14689 14690 14691
    }

    if (!create) {
        ret = 0;
14692 14693 14694
        goto cleanup;
    }

14695
    if (lchown(devicePath, sb.st_uid, sb.st_gid) < 0) {
14696 14697 14698 14699 14700 14701
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             devicePath);
        goto cleanup;
    }

14702 14703 14704 14705 14706 14707 14708 14709 14710
    /* 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;
    }

14711 14712 14713
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileCopyACLs(device, devicePath) < 0 &&
14714 14715 14716 14717 14718 14719 14720
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Failed to copy ACLs on device %s"),
                             devicePath);
        goto cleanup;
    }

14721
#ifdef WITH_SELINUX
14722
    if (lgetfilecon_raw(device, &tcon) < 0 &&
14723 14724 14725
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"),
14726
                             device);
14727 14728 14729 14730
        goto cleanup;
    }

    if (tcon &&
14731
        lsetfilecon_raw(devicePath, (const char *)tcon) < 0) {
14732 14733 14734 14735 14736 14737 14738 14739 14740 14741 14742
        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

14743
    /* Finish mount process started earlier. */
14744
    if ((isReg || isDir) &&
14745 14746 14747
        virFileBindMountDevice(device, devicePath) < 0)
        goto cleanup;

14748 14749
    ret = 0;
 cleanup:
14750 14751 14752
#ifdef WITH_SELINUX
    freecon(tcon);
#endif
14753 14754 14755 14756
    return ret;
}


14757 14758
static int
qemuDomainCreateDevice(const char *device,
14759
                       const struct qemuDomainCreateDeviceData *data,
14760 14761 14762 14763
                       bool allow_noent)
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

14764
    return qemuDomainCreateDeviceRecursive(device, data,
14765 14766 14767
                                           allow_noent, symloop_max);
}

14768 14769

static int
14770
qemuDomainPopulateDevices(virQEMUDriverConfigPtr cfg,
J
Ján Tomko 已提交
14771
                          virDomainObjPtr vm G_GNUC_UNUSED,
14772
                          const struct qemuDomainCreateDeviceData *data)
14773 14774 14775 14776 14777 14778 14779 14780
{
    const char *const *devices = (const char *const *) cfg->cgroupDeviceACL;
    size_t i;

    if (!devices)
        devices = defaultDeviceACL;

    for (i = 0; devices[i]; i++) {
14781
        if (qemuDomainCreateDevice(devices[i], data, true) < 0)
14782
            return -1;
14783 14784
    }

14785
    return 0;
14786 14787 14788 14789
}


static int
14790 14791
qemuDomainSetupDev(virQEMUDriverConfigPtr cfg,
                   virSecurityManagerPtr mgr,
14792
                   virDomainObjPtr vm,
14793
                   const struct qemuDomainCreateDeviceData *data)
14794
{
14795 14796
    g_autofree char *mount_options = NULL;
    g_autofree char *opts = NULL;
14797 14798 14799

    VIR_DEBUG("Setting up /dev/ for domain %s", vm->def->name);

14800
    mount_options = qemuSecurityGetMountOptions(mgr, vm->def);
14801

14802 14803
    if (!mount_options)
        mount_options = g_strdup("");
14804 14805 14806 14807 14808

    /*
     * tmpfs is limited to 64kb, since we only have device nodes in there
     * and don't want to DOS the entire OS RAM usage
     */
14809
    opts = g_strdup_printf("mode=755,size=65536%s", mount_options);
14810

14811
    if (virFileSetupDev(data->path, opts) < 0)
14812
        return -1;
14813

14814
    if (qemuDomainPopulateDevices(cfg, vm, data) < 0)
14815
        return -1;
14816

14817
    return 0;
14818 14819 14820
}


14821
static int
J
Ján Tomko 已提交
14822
qemuDomainSetupDisk(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14823
                    virDomainDiskDefPtr disk,
14824
                    const struct qemuDomainCreateDeviceData *data)
14825 14826
{
    virStorageSourcePtr next;
14827
    bool hasNVMe = false;
14828

14829
    for (next = disk->src; virStorageSourceIsBacking(next); next = next->backingStore) {
14830 14831 14832
        VIR_AUTOSTRINGLIST targetPaths = NULL;
        size_t i;

14833 14834
        if (next->type == VIR_STORAGE_TYPE_NVME) {
            g_autofree char *nvmePath = NULL;
14835

14836 14837 14838
            hasNVMe = true;

            if (!(nvmePath = virPCIDeviceAddressGetIOMMUGroupDev(&next->nvme->pciAddr)))
14839
                return -1;
14840 14841

            if (qemuDomainCreateDevice(nvmePath, data, false) < 0)
14842
                return -1;
14843 14844 14845 14846 14847 14848 14849
        } else {
            if (!next->path || !virStorageSourceIsLocalStorage(next)) {
                /* Not creating device. Just continue. */
                continue;
            }

            if (qemuDomainCreateDevice(next->path, data, false) < 0)
14850
                return -1;
14851 14852 14853 14854 14855 14856 14857 14858 14859 14860 14861 14862 14863

            if (virDevMapperGetTargets(next->path, &targetPaths) < 0 &&
                errno != ENOSYS && errno != EBADF) {
                virReportSystemError(errno,
                                     _("Unable to get devmapper targets for %s"),
                                     next->path);
                return -1;
            }

            for (i = 0; targetPaths && targetPaths[i]; i++) {
                if (qemuDomainCreateDevice(targetPaths[i], data, false) < 0)
                    return -1;
            }
14864
        }
14865 14866
    }

14867
    /* qemu-pr-helper might require access to /dev/mapper/control. */
14868
    if (disk->src->pr &&
14869
        qemuDomainCreateDevice(QEMU_DEVICE_MAPPER_CONTROL_PATH, data, true) < 0)
14870
        return -1;
14871

14872 14873
    if (hasNVMe &&
        qemuDomainCreateDevice(QEMU_DEV_VFIO, data, false) < 0)
14874
        return -1;
14875

14876
    return 0;
14877 14878 14879 14880
}


static int
14881
qemuDomainSetupAllDisks(virQEMUDriverConfigPtr cfg,
14882
                        virDomainObjPtr vm,
14883
                        const struct qemuDomainCreateDeviceData *data)
14884 14885 14886 14887 14888
{
    size_t i;
    VIR_DEBUG("Setting up disks");

    for (i = 0; i < vm->def->ndisks; i++) {
14889
        if (qemuDomainSetupDisk(cfg,
14890
                                vm->def->disks[i],
14891
                                data) < 0)
14892 14893 14894 14895 14896 14897 14898 14899
            return -1;
    }

    VIR_DEBUG("Setup all disks");
    return 0;
}


14900
static int
J
Ján Tomko 已提交
14901
qemuDomainSetupHostdev(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14902
                       virDomainHostdevDefPtr dev,
14903
                       const struct qemuDomainCreateDeviceData *data)
14904
{
14905
    g_autofree char *path = NULL;
14906

14907
    if (qemuDomainGetHostdevPath(dev, &path, NULL) < 0)
14908
        return -1;
14909

14910
    if (path && qemuDomainCreateDevice(path, data, false) < 0)
14911
        return -1;
14912

14913 14914
    if (qemuHostdevNeedsVFIO(dev) &&
        qemuDomainCreateDevice(QEMU_DEV_VFIO, data, false) < 0)
14915
        return -1;
14916

14917
    return 0;
14918 14919 14920 14921
}


static int
14922
qemuDomainSetupAllHostdevs(virQEMUDriverConfigPtr cfg,
14923
                           virDomainObjPtr vm,
14924
                           const struct qemuDomainCreateDeviceData *data)
14925 14926 14927 14928 14929
{
    size_t i;

    VIR_DEBUG("Setting up hostdevs");
    for (i = 0; i < vm->def->nhostdevs; i++) {
14930
        if (qemuDomainSetupHostdev(cfg,
14931
                                   vm->def->hostdevs[i],
14932
                                   data) < 0)
14933 14934 14935 14936 14937 14938 14939
            return -1;
    }
    VIR_DEBUG("Setup all hostdevs");
    return 0;
}


M
Michal Privoznik 已提交
14940
static int
J
Ján Tomko 已提交
14941
qemuDomainSetupMemory(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
M
Michal Privoznik 已提交
14942
                      virDomainMemoryDefPtr mem,
14943
                      const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
14944 14945 14946 14947
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

14948
    return qemuDomainCreateDevice(mem->nvdimmPath, data, false);
M
Michal Privoznik 已提交
14949 14950 14951 14952 14953 14954
}


static int
qemuDomainSetupAllMemories(virQEMUDriverConfigPtr cfg,
                           virDomainObjPtr vm,
14955
                           const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
14956 14957 14958 14959 14960 14961 14962
{
    size_t i;

    VIR_DEBUG("Setting up memories");
    for (i = 0; i < vm->def->nmems; i++) {
        if (qemuDomainSetupMemory(cfg,
                                  vm->def->mems[i],
14963
                                  data) < 0)
M
Michal Privoznik 已提交
14964 14965 14966 14967 14968 14969 14970
            return -1;
    }
    VIR_DEBUG("Setup all memories");
    return 0;
}


14971
static int
J
Ján Tomko 已提交
14972
qemuDomainSetupChardev(virDomainDefPtr def G_GNUC_UNUSED,
14973 14974 14975
                       virDomainChrDefPtr dev,
                       void *opaque)
{
14976
    const struct qemuDomainCreateDeviceData *data = opaque;
14977
    const char *path = NULL;
14978

14979
    if (!(path = virDomainChrSourceDefGetPath(dev->source)))
14980 14981
        return 0;

14982 14983 14984 14985 14986 14987
    /* 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);
14988 14989 14990 14991
}


static int
J
Ján Tomko 已提交
14992
qemuDomainSetupAllChardevs(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14993
                           virDomainObjPtr vm,
14994
                           const struct qemuDomainCreateDeviceData *data)
14995 14996 14997 14998 14999 15000
{
    VIR_DEBUG("Setting up chardevs");

    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuDomainSetupChardev,
15001
                               (void *)data) < 0)
15002 15003 15004 15005 15006 15007 15008
        return -1;

    VIR_DEBUG("Setup all chardevs");
    return 0;
}


15009
static int
J
Ján Tomko 已提交
15010
qemuDomainSetupTPM(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
15011
                   virDomainObjPtr vm,
15012
                   const struct qemuDomainCreateDeviceData *data)
15013 15014 15015 15016 15017 15018 15019 15020 15021 15022 15023
{
    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,
15024
                                   data, false) < 0)
15025 15026 15027
            return -1;
        break;

15028
    case VIR_DOMAIN_TPM_TYPE_EMULATOR:
15029 15030 15031 15032 15033 15034 15035 15036 15037 15038
    case VIR_DOMAIN_TPM_TYPE_LAST:
        /* nada */
        break;
    }

    VIR_DEBUG("Setup TPM");
    return 0;
}


15039
static int
J
Ján Tomko 已提交
15040
qemuDomainSetupGraphics(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
15041
                        virDomainGraphicsDefPtr gfx,
15042
                        const struct qemuDomainCreateDeviceData *data)
15043
{
15044
    const char *rendernode = virDomainGraphicsGetRenderNode(gfx);
15045

15046
    if (!rendernode)
15047 15048
        return 0;

15049
    return qemuDomainCreateDevice(rendernode, data, false);
15050 15051 15052 15053
}


static int
15054
qemuDomainSetupAllGraphics(virQEMUDriverConfigPtr cfg,
15055
                           virDomainObjPtr vm,
15056
                           const struct qemuDomainCreateDeviceData *data)
15057 15058 15059 15060 15061
{
    size_t i;

    VIR_DEBUG("Setting up graphics");
    for (i = 0; i < vm->def->ngraphics; i++) {
15062
        if (qemuDomainSetupGraphics(cfg,
15063
                                    vm->def->graphics[i],
15064
                                    data) < 0)
15065 15066 15067 15068 15069 15070 15071 15072
            return -1;
    }

    VIR_DEBUG("Setup all graphics");
    return 0;
}


15073
static int
J
Ján Tomko 已提交
15074
qemuDomainSetupInput(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
15075
                     virDomainInputDefPtr input,
15076
                     const struct qemuDomainCreateDeviceData *data)
15077
{
J
Ján Tomko 已提交
15078
    const char *path = virDomainInputDefGetPath(input);
15079

J
Ján Tomko 已提交
15080 15081
    if (path && qemuDomainCreateDevice(path, data, false) < 0)
        return -1;
15082

J
Ján Tomko 已提交
15083
    return 0;
15084 15085 15086 15087
}


static int
15088
qemuDomainSetupAllInputs(virQEMUDriverConfigPtr cfg,
15089
                         virDomainObjPtr vm,
15090
                         const struct qemuDomainCreateDeviceData *data)
15091 15092 15093
{
    size_t i;

15094
    VIR_DEBUG("Setting up inputs");
15095
    for (i = 0; i < vm->def->ninputs; i++) {
15096
        if (qemuDomainSetupInput(cfg,
15097
                                 vm->def->inputs[i],
15098
                                 data) < 0)
15099 15100
            return -1;
    }
15101
    VIR_DEBUG("Setup all inputs");
15102 15103 15104 15105
    return 0;
}


15106
static int
J
Ján Tomko 已提交
15107
qemuDomainSetupRNG(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
15108
                   virDomainRNGDefPtr rng,
15109
                   const struct qemuDomainCreateDeviceData *data)
15110 15111 15112
{
    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
15113
        if (qemuDomainCreateDevice(rng->source.file, data, false) < 0)
15114
            return -1;
H
Han Han 已提交
15115
        break;
15116 15117

    case VIR_DOMAIN_RNG_BACKEND_EGD:
H
Han Han 已提交
15118
    case VIR_DOMAIN_RNG_BACKEND_BUILTIN:
15119 15120 15121 15122 15123 15124 15125 15126 15127 15128
    case VIR_DOMAIN_RNG_BACKEND_LAST:
        /* nada */
        break;
    }

    return 0;
}


static int
15129
qemuDomainSetupAllRNGs(virQEMUDriverConfigPtr cfg,
15130
                       virDomainObjPtr vm,
15131
                       const struct qemuDomainCreateDeviceData *data)
15132 15133 15134 15135 15136
{
    size_t i;

    VIR_DEBUG("Setting up RNGs");
    for (i = 0; i < vm->def->nrngs; i++) {
15137
        if (qemuDomainSetupRNG(cfg,
15138
                               vm->def->rngs[i],
15139
                               data) < 0)
15140 15141 15142 15143 15144 15145 15146 15147
            return -1;
    }

    VIR_DEBUG("Setup all RNGs");
    return 0;
}


15148
static int
J
Ján Tomko 已提交
15149
qemuDomainSetupLoader(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
15150 15151 15152 15153 15154 15155 15156 15157 15158 15159 15160
                      virDomainObjPtr vm,
                      const struct qemuDomainCreateDeviceData *data)
{
    virDomainLoaderDefPtr loader = vm->def->os.loader;

    VIR_DEBUG("Setting up loader");

    if (loader) {
        switch ((virDomainLoader) loader->type) {
        case VIR_DOMAIN_LOADER_TYPE_ROM:
            if (qemuDomainCreateDevice(loader->path, data, false) < 0)
15161
                return -1;
15162 15163 15164 15165
            break;

        case VIR_DOMAIN_LOADER_TYPE_PFLASH:
            if (qemuDomainCreateDevice(loader->path, data, false) < 0)
15166
                return -1;
15167 15168 15169

            if (loader->nvram &&
                qemuDomainCreateDevice(loader->nvram, data, false) < 0)
15170
                return -1;
15171 15172
            break;

15173
        case VIR_DOMAIN_LOADER_TYPE_NONE:
15174 15175 15176 15177 15178 15179
        case VIR_DOMAIN_LOADER_TYPE_LAST:
            break;
        }
    }

    VIR_DEBUG("Setup loader");
15180
    return 0;
15181 15182 15183
}


15184
static int
J
Ján Tomko 已提交
15185
qemuDomainSetupLaunchSecurity(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
15186 15187 15188 15189 15190 15191 15192 15193 15194 15195
                              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");

15196
    if (qemuDomainCreateDevice(QEMU_DEV_SEV, data, false) < 0)
15197 15198 15199 15200 15201 15202 15203
        return -1;

    VIR_DEBUG("Set up launch security");
    return 0;
}


15204
int
15205 15206
qemuDomainBuildNamespace(virQEMUDriverConfigPtr cfg,
                         virSecurityManagerPtr mgr,
15207 15208
                         virDomainObjPtr vm)
{
15209
    struct qemuDomainCreateDeviceData data;
15210
    const char *devPath = NULL;
15211
    char **devMountsPath = NULL, **devMountsSavePath = NULL;
15212 15213 15214 15215 15216 15217 15218 15219
    size_t ndevMountsPath = 0, i;
    int ret = -1;

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT)) {
        ret = 0;
        goto cleanup;
    }

15220
    if (qemuDomainGetPreservedMounts(cfg, vm,
15221 15222
                                     &devMountsPath, &devMountsSavePath,
                                     &ndevMountsPath) < 0)
15223 15224
        goto cleanup;

15225 15226 15227 15228 15229 15230 15231 15232 15233 15234 15235 15236 15237
    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;
    }

15238
    data.path = devPath;
15239 15240
    data.devMountsPath = devMountsPath;
    data.ndevMountsPath = ndevMountsPath;
15241

15242 15243 15244
    if (virProcessSetupPrivateMountNS() < 0)
        goto cleanup;

15245
    if (qemuDomainSetupDev(cfg, mgr, vm, &data) < 0)
15246 15247
        goto cleanup;

15248
    if (qemuDomainSetupAllDisks(cfg, vm, &data) < 0)
15249 15250
        goto cleanup;

15251
    if (qemuDomainSetupAllHostdevs(cfg, vm, &data) < 0)
15252 15253
        goto cleanup;

15254
    if (qemuDomainSetupAllMemories(cfg, vm, &data) < 0)
15255 15256
        goto cleanup;

15257
    if (qemuDomainSetupAllChardevs(cfg, vm, &data) < 0)
15258 15259
        goto cleanup;

15260
    if (qemuDomainSetupTPM(cfg, vm, &data) < 0)
15261 15262
        goto cleanup;

15263
    if (qemuDomainSetupAllGraphics(cfg, vm, &data) < 0)
15264 15265
        goto cleanup;

15266
    if (qemuDomainSetupAllInputs(cfg, vm, &data) < 0)
15267 15268
        goto cleanup;

15269
    if (qemuDomainSetupAllRNGs(cfg, vm, &data) < 0)
15270 15271
        goto cleanup;

15272 15273 15274
    if (qemuDomainSetupLoader(cfg, vm, &data) < 0)
        goto cleanup;

15275 15276 15277
    if (qemuDomainSetupLaunchSecurity(cfg, vm, &data) < 0)
        goto cleanup;

15278 15279
    /* Save some mount points because we want to share them with the host */
    for (i = 0; i < ndevMountsPath; i++) {
15280 15281
        struct stat sb;

15282 15283 15284
        if (devMountsSavePath[i] == devPath)
            continue;

15285 15286 15287 15288 15289 15290 15291
        if (stat(devMountsPath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsPath[i]);
            goto cleanup;
        }

15292 15293 15294
        /* At this point, devMountsPath is either:
         * a file (regular or special), or
         * a directory. */
15295
        if ((S_ISDIR(sb.st_mode) && virFileMakePath(devMountsSavePath[i]) < 0) ||
15296
            (!S_ISDIR(sb.st_mode) && virFileTouch(devMountsSavePath[i], sb.st_mode) < 0)) {
15297 15298 15299 15300 15301 15302
            virReportSystemError(errno,
                                 _("Failed to create %s"),
                                 devMountsSavePath[i]);
            goto cleanup;
        }

15303
        if (virFileMoveMount(devMountsPath[i], devMountsSavePath[i]) < 0)
15304 15305 15306
            goto cleanup;
    }

15307
    if (virFileMoveMount(devPath, "/dev") < 0)
15308 15309 15310
        goto cleanup;

    for (i = 0; i < ndevMountsPath; i++) {
15311 15312
        struct stat sb;

15313 15314 15315
        if (devMountsSavePath[i] == devPath)
            continue;

15316 15317 15318 15319
        if (stat(devMountsSavePath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsSavePath[i]);
15320 15321 15322
            goto cleanup;
        }

15323 15324 15325 15326 15327 15328 15329 15330 15331 15332 15333 15334 15335 15336 15337
        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;
            }
        }

15338
        if (virFileMoveMount(devMountsSavePath[i], devMountsPath[i]) < 0)
15339 15340 15341 15342 15343
            goto cleanup;
    }

    ret = 0;
 cleanup:
15344
    for (i = 0; i < ndevMountsPath; i++) {
15345 15346 15347
#if defined(__linux__)
        umount(devMountsSavePath[i]);
#endif /* defined(__linux__) */
15348 15349
        /* The path can be either a regular file or a dir. */
        if (virFileIsDir(devMountsSavePath[i]))
15350
            virFileDeleteTree(devMountsSavePath[i]);
15351 15352 15353
        else
            unlink(devMountsSavePath[i]);
    }
15354
    virStringListFreeCount(devMountsPath, ndevMountsPath);
15355
    virStringListFreeCount(devMountsSavePath, ndevMountsPath);
15356 15357 15358 15359 15360 15361 15362 15363
    return ret;
}


int
qemuDomainCreateNamespace(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
{
15364
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
15365

15366 15367
    if (virBitmapIsBitSet(cfg->namespaces, QEMU_DOMAIN_NS_MOUNT) &&
        qemuDomainEnableNamespace(vm, QEMU_DOMAIN_NS_MOUNT) < 0)
15368
        return -1;
15369

15370
    return 0;
15371 15372 15373
}


15374
void
J
Ján Tomko 已提交
15375
qemuDomainDestroyNamespace(virQEMUDriverPtr driver G_GNUC_UNUSED,
15376 15377 15378 15379 15380 15381 15382
                           virDomainObjPtr vm)
{
    if (qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        qemuDomainDisableNamespace(vm, QEMU_DOMAIN_NS_MOUNT);
}


15383
bool
J
Ján Tomko 已提交
15384
qemuDomainNamespaceAvailable(qemuDomainNamespace ns G_GNUC_UNUSED)
15385 15386 15387 15388 15389 15390 15391 15392 15393 15394 15395 15396 15397 15398 15399 15400 15401 15402 15403 15404 15405 15406 15407 15408 15409 15410 15411
{
#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__) */
}


15412 15413 15414 15415
struct qemuDomainAttachDeviceMknodData {
    virQEMUDriverPtr driver;
    virDomainObjPtr vm;
    const char *file;
15416
    const char *target;
15417
    GStatBuf sb;
15418
    void *acl;
15419 15420 15421
#ifdef WITH_SELINUX
    char *tcon;
#endif
15422 15423 15424
};


15425 15426
/* Our way of creating devices is highly linux specific */
#if defined(__linux__)
15427
static int
J
Ján Tomko 已提交
15428
qemuDomainAttachDeviceMknodHelper(pid_t pid G_GNUC_UNUSED,
15429 15430 15431 15432
                                  void *opaque)
{
    struct qemuDomainAttachDeviceMknodData *data = opaque;
    int ret = -1;
15433
    bool delDevice = false;
15434
    bool isLink = S_ISLNK(data->sb.st_mode);
15435
    bool isDev = S_ISCHR(data->sb.st_mode) || S_ISBLK(data->sb.st_mode);
15436
    bool isReg = S_ISREG(data->sb.st_mode) || S_ISFIFO(data->sb.st_mode) || S_ISSOCK(data->sb.st_mode);
15437
    bool isDir = S_ISDIR(data->sb.st_mode);
15438

15439
    qemuSecurityPostFork(data->driver->securityManager);
15440 15441 15442 15443 15444 15445 15446

    if (virFileMakeParentPath(data->file) < 0) {
        virReportSystemError(errno,
                             _("Unable to create %s"), data->file);
        goto cleanup;
    }

15447 15448
    if (isLink) {
        VIR_DEBUG("Creating symlink %s -> %s", data->file, data->target);
15449 15450 15451 15452 15453 15454 15455 15456 15457 15458 15459 15460

        /* 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;
        }

15461
        if (symlink(data->target, data->file) < 0) {
15462 15463 15464 15465
            virReportSystemError(errno,
                                 _("Unable to create symlink %s (pointing to %s)"),
                                 data->file, data->target);
            goto cleanup;
15466 15467
        } else {
            delDevice = true;
15468
        }
15469
    } else if (isDev) {
15470 15471
        VIR_DEBUG("Creating dev %s (%d,%d)",
                  data->file, major(data->sb.st_rdev), minor(data->sb.st_rdev));
15472
        unlink(data->file);
15473
        if (mknod(data->file, data->sb.st_mode, data->sb.st_rdev) < 0) {
15474 15475 15476 15477
            virReportSystemError(errno,
                                 _("Unable to create device %s"),
                                 data->file);
            goto cleanup;
15478 15479 15480
        } else {
            delDevice = true;
        }
15481
    } else if (isReg || isDir) {
15482 15483 15484 15485 15486
        /* 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 &&
15487
            errno != ENOENT && errno != EINVAL) {
15488 15489 15490 15491 15492
            virReportSystemError(errno,
                                 _("Unable to umount %s"),
                                 data->file);
            goto cleanup;
        }
15493 15494
        if ((isReg && virFileTouch(data->file, data->sb.st_mode) < 0) ||
            (isDir && virFileMakePathWithMode(data->file, data->sb.st_mode) < 0))
15495 15496 15497 15498
            goto cleanup;
        delDevice = true;
        /* Just create the file here so that code below sets
         * proper owner and mode. Move the mount only after that. */
15499 15500 15501 15502 15503
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       data->file, data->sb.st_mode);
        goto cleanup;
15504 15505
    }

15506 15507 15508 15509 15510 15511 15512
    if (lchown(data->file, data->sb.st_uid, data->sb.st_gid) < 0) {
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             data->file);
        goto cleanup;
    }

15513 15514 15515 15516 15517 15518 15519 15520 15521
    /* 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;
    }

15522 15523 15524
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileSetACLs(data->file, data->acl) < 0 &&
15525 15526 15527 15528 15529 15530
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to set ACLs on %s"), data->file);
        goto cleanup;
    }

15531
# ifdef WITH_SELINUX
15532
    if (data->tcon &&
15533
        lsetfilecon_raw(data->file, (const char *)data->tcon) < 0) {
15534 15535 15536 15537 15538 15539 15540 15541 15542
        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;
        }
    }
15543
# endif
15544

15545
    /* Finish mount process started earlier. */
15546
    if ((isReg || isDir) &&
15547 15548 15549
        virFileMoveMount(data->target, data->file) < 0)
        goto cleanup;

15550 15551
    ret = 0;
 cleanup:
15552 15553 15554 15555 15556 15557
    if (ret < 0 && delDevice) {
        if (isDir)
            virFileDeleteTree(data->file);
        else
            unlink(data->file);
    }
15558
# ifdef WITH_SELINUX
15559
    freecon(data->tcon);
15560
# endif
15561 15562 15563 15564 15565 15566
    virFileFreeACLs(&data->acl);
    return ret;
}


static int
15567 15568 15569
qemuDomainAttachDeviceMknodRecursive(virQEMUDriverPtr driver,
                                     virDomainObjPtr vm,
                                     const char *file,
15570 15571
                                     char * const *devMountsPath,
                                     size_t ndevMountsPath,
15572
                                     unsigned int ttl)
15573
{
15574
    g_autoptr(virQEMUDriverConfig) cfg = NULL;
15575 15576
    struct qemuDomainAttachDeviceMknodData data;
    int ret = -1;
15577
    g_autofree char *target = NULL;
15578
    bool isLink;
15579
    bool isReg;
15580
    bool isDir;
15581

15582 15583 15584 15585 15586 15587 15588
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             file);
        return ret;
    }

15589 15590 15591 15592 15593 15594
    memset(&data, 0, sizeof(data));

    data.driver = driver;
    data.vm = vm;
    data.file = file;

15595
    if (g_lstat(file, &data.sb) < 0) {
15596 15597 15598 15599 15600
        virReportSystemError(errno,
                             _("Unable to access %s"), file);
        return ret;
    }

15601
    isLink = S_ISLNK(data.sb.st_mode);
15602
    isReg = S_ISREG(data.sb.st_mode) || S_ISFIFO(data.sb.st_mode) || S_ISSOCK(data.sb.st_mode);
15603
    isDir = S_ISDIR(data.sb.st_mode);
15604

15605
    if ((isReg || isDir) && STRPREFIX(file, QEMU_DEVPREFIX)) {
15606 15607 15608 15609 15610 15611 15612 15613 15614
        cfg = virQEMUDriverGetConfig(driver);
        if (!(target = qemuDomainGetPreservedMountPath(cfg, vm, file)))
            goto cleanup;

        if (virFileBindMountDevice(file, target) < 0)
            goto cleanup;

        data.target = target;
    } else if (isLink) {
15615 15616 15617 15618 15619
        g_autoptr(GError) gerr = NULL;

        if (!(target = g_file_read_link(file, &gerr))) {
            virReportError(VIR_ERR_SYSTEM_ERROR,
                           _("failed to resolve symlink %s: %s"), file, gerr->message);
15620 15621 15622
            return ret;
        }

15623
        if (!g_path_is_absolute(target)) {
15624 15625
            g_autofree char *fileTmp = g_strdup(file);
            char *c = NULL, *tmp = NULL;
15626 15627 15628 15629

            if ((c = strrchr(fileTmp, '/')))
                *(c + 1) = '\0';

15630
            tmp = g_strdup_printf("%s%s", fileTmp, target);
15631
            VIR_FREE(target);
15632
            target = g_steal_pointer(&tmp);
15633 15634 15635 15636 15637 15638 15639 15640
        }

        data.target = target;
    }

    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileGetACLs(file, &data.acl) < 0 &&
15641 15642 15643
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to get ACLs on %s"), file);
15644
        goto cleanup;
15645 15646
    }

15647
# ifdef WITH_SELINUX
15648
    if (lgetfilecon_raw(file, &data.tcon) < 0 &&
15649 15650 15651 15652 15653
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"), file);
        goto cleanup;
    }
15654
# endif
15655

15656
    if (STRPREFIX(file, QEMU_DEVPREFIX)) {
15657 15658 15659 15660 15661 15662 15663 15664 15665 15666 15667 15668
        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;
15669

15670 15671 15672 15673 15674 15675
            if (virProcessRunInMountNamespace(vm->pid,
                                              qemuDomainAttachDeviceMknodHelper,
                                              &data) < 0) {
                qemuSecurityPostFork(driver->securityManager);
                goto cleanup;
            }
15676
            qemuSecurityPostFork(driver->securityManager);
15677 15678 15679
        } else {
            VIR_DEBUG("Skipping dev %s because of %s mount point",
                      file, devMountsPath[i]);
15680
        }
15681 15682
    }

15683
    if (isLink &&
15684 15685 15686
        qemuDomainAttachDeviceMknodRecursive(driver, vm, target,
                                             devMountsPath, ndevMountsPath,
                                             ttl -1) < 0)
15687
        goto cleanup;
15688 15689 15690

    ret = 0;
 cleanup:
15691
# ifdef WITH_SELINUX
15692
    freecon(data.tcon);
15693
# endif
15694
    virFileFreeACLs(&data.acl);
15695 15696
    if (isReg && target)
        umount(target);
15697
    return ret;
15698 15699 15700
}


15701 15702 15703 15704
#else /* !defined(__linux__) */


static int
J
Ján Tomko 已提交
15705 15706 15707 15708 15709 15710
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)
15711 15712 15713 15714 15715 15716 15717 15718 15719 15720
{
    virReportSystemError(ENOSYS, "%s",
                         _("Namespaces are not supported on this platform."));
    return -1;
}


#endif /* !defined(__linux__) */


15721 15722 15723
static int
qemuDomainAttachDeviceMknod(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
15724 15725 15726
                            const char *file,
                            char * const *devMountsPath,
                            size_t ndevMountsPath)
15727 15728 15729
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

15730 15731 15732
    return qemuDomainAttachDeviceMknodRecursive(driver, vm, file,
                                                devMountsPath, ndevMountsPath,
                                                symloop_max);
15733 15734 15735
}


15736
static int
J
Ján Tomko 已提交
15737
qemuDomainDetachDeviceUnlinkHelper(pid_t pid G_GNUC_UNUSED,
15738 15739 15740 15741 15742 15743 15744 15745 15746 15747 15748 15749 15750 15751 15752 15753
                                   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
15754
qemuDomainDetachDeviceUnlink(virQEMUDriverPtr driver G_GNUC_UNUSED,
15755
                             virDomainObjPtr vm,
15756 15757 15758
                             const char *file,
                             char * const *devMountsPath,
                             size_t ndevMountsPath)
15759
{
15760
    size_t i;
15761

15762
    if (STRPREFIX(file, QEMU_DEVPREFIX)) {
15763 15764 15765 15766 15767 15768 15769 15770 15771 15772 15773
        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)
15774
                return -1;
15775 15776 15777
        }
    }

15778
    return 0;
15779 15780 15781
}


15782 15783 15784 15785
static int
qemuDomainNamespaceMknodPaths(virDomainObjPtr vm,
                              const char **paths,
                              size_t npaths)
15786
{
15787 15788
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
15789
    g_autoptr(virQEMUDriverConfig) cfg = NULL;
15790 15791
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0;
15792
    int ret = -1;
15793
    size_t i;
15794

15795 15796
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
15797 15798
        return 0;

15799 15800 15801 15802 15803 15804
    cfg = virQEMUDriverGetConfig(driver);
    if (qemuDomainGetPreservedMounts(cfg, vm,
                                     &devMountsPath, NULL,
                                     &ndevMountsPath) < 0)
        goto cleanup;

15805 15806 15807 15808 15809 15810 15811 15812 15813 15814 15815 15816 15817 15818 15819
    for (i = 0; i < npaths; i++) {
        if (qemuDomainAttachDeviceMknod(driver,
                                        vm,
                                        paths[i],
                                        devMountsPath, ndevMountsPath) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    return ret;
}


15820 15821 15822 15823 15824 15825 15826 15827 15828 15829
static int
qemuDomainNamespaceMknodPath(virDomainObjPtr vm,
                             const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceMknodPaths(vm, paths, 1);
}


15830 15831 15832 15833 15834 15835 15836
static int
qemuDomainNamespaceUnlinkPaths(virDomainObjPtr vm,
                               const char **paths,
                               size_t npaths)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
15837
    g_autoptr(virQEMUDriverConfig) cfg = NULL;
15838 15839 15840 15841 15842
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0;
    size_t i;
    int ret = -1;

15843 15844
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
15845 15846 15847 15848 15849 15850 15851 15852 15853 15854 15855 15856 15857 15858 15859
        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;
    }

15860
    ret = 0;
15861 15862 15863 15864 15865 15866
 cleanup:
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    return ret;
}


15867 15868 15869 15870 15871 15872 15873 15874 15875 15876
static int
qemuDomainNamespaceUnlinkPath(virDomainObjPtr vm,
                              const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceUnlinkPaths(vm, paths, 1);
}


15877
int
15878
qemuDomainNamespaceSetupDisk(virDomainObjPtr vm,
15879 15880 15881
                             virStorageSourcePtr src)
{
    virStorageSourcePtr next;
15882
    VIR_AUTOSTRINGLIST paths = NULL;
15883
    size_t npaths = 0;
15884
    bool hasNVMe = false;
15885

15886
    for (next = src; virStorageSourceIsBacking(next); next = next->backingStore) {
15887 15888 15889 15890 15891 15892
        g_autofree char *tmpPath = NULL;

        if (next->type == VIR_STORAGE_TYPE_NVME) {
            hasNVMe = true;

            if (!(tmpPath = virPCIDeviceAddressGetIOMMUGroupDev(&next->nvme->pciAddr)))
15893
                return -1;
15894
        } else {
15895 15896
            VIR_AUTOSTRINGLIST targetPaths = NULL;

15897 15898 15899 15900 15901 15902 15903
            if (virStorageSourceIsEmpty(next) ||
                !virStorageSourceIsLocalStorage(next)) {
                /* Not creating device. Just continue. */
                continue;
            }

            tmpPath = g_strdup(next->path);
15904

15905 15906 15907 15908 15909 15910 15911
            if (virDevMapperGetTargets(next->path, &targetPaths) < 0 &&
                errno != ENOSYS && errno != EBADF) {
                virReportSystemError(errno,
                                     _("Unable to get devmapper targets for %s"),
                                     next->path);
                return -1;
            }
15912

15913 15914
            if (virStringListMerge(&paths, &targetPaths) < 0)
                return -1;
15915 15916
        }

15917
        if (virStringListAdd(&paths, tmpPath) < 0)
15918
            return -1;
15919 15920
    }

15921
    /* qemu-pr-helper might require access to /dev/mapper/control. */
15922 15923 15924
    if (src->pr &&
        virStringListAdd(&paths, QEMU_DEVICE_MAPPER_CONTROL_PATH) < 0)
        return -1;
15925

15926 15927 15928
    if (hasNVMe &&
        virStringListAdd(&paths, QEMU_DEV_VFIO) < 0)
        return -1;
15929

15930
    npaths = virStringListLength((const char **) paths);
15931
    if (qemuDomainNamespaceMknodPaths(vm, (const char **) paths, npaths) < 0)
15932
        return -1;
15933

15934
    return 0;
15935 15936 15937 15938
}


int
J
Ján Tomko 已提交
15939 15940
qemuDomainNamespaceTeardownDisk(virDomainObjPtr vm G_GNUC_UNUSED,
                                virStorageSourcePtr src G_GNUC_UNUSED)
15941 15942 15943 15944 15945 15946 15947 15948 15949
{
    /* 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;
}
15950 15951


15952 15953 15954 15955 15956 15957 15958 15959 15960 15961 15962
/**
 * qemuDomainNamespaceSetupHostdev:
 * @vm: domain object
 * @hostdev: hostdev to create in @vm's namespace
 *
 * For given @hostdev, create its devfs representation (if it has one) in
 * domain namespace. Note, @hostdev must not be in @vm's definition.
 *
 * Returns: 0 on success,
 *         -1 otherwise.
 */
15963
int
15964
qemuDomainNamespaceSetupHostdev(virDomainObjPtr vm,
15965 15966
                                virDomainHostdevDefPtr hostdev)
{
15967
    g_autofree char *path = NULL;
15968

15969
    if (qemuDomainGetHostdevPath(hostdev, &path, NULL) < 0)
15970
        return -1;
15971

15972
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
15973
        return -1;
15974

15975 15976 15977
    if (qemuHostdevNeedsVFIO(hostdev) &&
        !qemuDomainNeedsVFIO(vm->def) &&
        qemuDomainNamespaceMknodPath(vm, QEMU_DEV_VFIO) < 0)
15978
        return -1;
15979

15980
    return 0;
15981 15982 15983
}


15984 15985 15986 15987 15988 15989 15990 15991 15992 15993 15994
/**
 * qemuDomainNamespaceTeardownHostdev:
 * @vm: domain object
 * @hostdev: hostdev to remove in @vm's namespace
 *
 * For given @hostdev, remove its devfs representation (if it has one) in
 * domain namespace. Note, @hostdev must not be in @vm's definition.
 *
 * Returns: 0 on success,
 *         -1 otherwise.
 */
15995
int
15996
qemuDomainNamespaceTeardownHostdev(virDomainObjPtr vm,
15997 15998
                                   virDomainHostdevDefPtr hostdev)
{
15999
    g_autofree char *path = NULL;
16000

16001
    if (qemuDomainGetHostdevPath(hostdev, &path, NULL) < 0)
16002
        return -1;
16003

16004
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
16005
        return -1;
16006

16007 16008 16009
    if (qemuHostdevNeedsVFIO(hostdev) &&
        !qemuDomainNeedsVFIO(vm->def) &&
        qemuDomainNamespaceUnlinkPath(vm, QEMU_DEV_VFIO) < 0)
16010
        return -1;
16011

16012
    return 0;
16013
}
16014 16015


M
Michal Privoznik 已提交
16016
int
16017
qemuDomainNamespaceSetupMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
16018 16019 16020 16021 16022
                               virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

16023 16024
    if (qemuDomainNamespaceMknodPath(vm, mem->nvdimmPath) < 0)
        return -1;
16025

16026
    return 0;
M
Michal Privoznik 已提交
16027 16028 16029 16030
}


int
16031
qemuDomainNamespaceTeardownMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
16032 16033 16034 16035 16036
                                  virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

16037 16038
    if (qemuDomainNamespaceUnlinkPath(vm, mem->nvdimmPath) < 0)
        return -1;
16039

16040
    return 0;
M
Michal Privoznik 已提交
16041 16042 16043
}


16044
int
16045
qemuDomainNamespaceSetupChardev(virDomainObjPtr vm,
16046 16047 16048 16049
                                virDomainChrDefPtr chr)
{
    const char *path;

16050
    if (!(path = virDomainChrSourceDefGetPath(chr->source)))
16051 16052
        return 0;

16053 16054 16055 16056
    /* 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;
16057

16058 16059
    if (qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
16060

16061
    return 0;
16062 16063 16064 16065
}


int
16066
qemuDomainNamespaceTeardownChardev(virDomainObjPtr vm,
16067 16068 16069 16070 16071 16072 16073 16074 16075
                                   virDomainChrDefPtr chr)
{
    const char *path = NULL;

    if (chr->source->type != VIR_DOMAIN_CHR_TYPE_DEV)
        return 0;

    path = chr->source->data.file.path;

16076 16077
    if (qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
16078

16079
    return 0;
16080
}
16081 16082 16083


int
16084
qemuDomainNamespaceSetupRNG(virDomainObjPtr vm,
16085 16086 16087 16088 16089 16090 16091 16092 16093 16094
                            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:
H
Han Han 已提交
16095
    case VIR_DOMAIN_RNG_BACKEND_BUILTIN:
16096
    case VIR_DOMAIN_RNG_BACKEND_LAST:
16097
        break;
16098 16099
    }

16100 16101
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
16102

16103
    return 0;
16104 16105 16106 16107
}


int
16108
qemuDomainNamespaceTeardownRNG(virDomainObjPtr vm,
16109 16110 16111 16112 16113 16114 16115 16116 16117 16118
                               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:
H
Han Han 已提交
16119
    case VIR_DOMAIN_RNG_BACKEND_BUILTIN:
16120
    case VIR_DOMAIN_RNG_BACKEND_LAST:
16121
        break;
16122 16123
    }

16124 16125
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
16126

16127
    return 0;
16128 16129 16130 16131 16132 16133 16134 16135 16136 16137 16138 16139
}


int
qemuDomainNamespaceSetupInput(virDomainObjPtr vm,
                              virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

16140 16141 16142
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
    return 0;
16143 16144 16145 16146 16147 16148 16149 16150 16151 16152 16153 16154
}


int
qemuDomainNamespaceTeardownInput(virDomainObjPtr vm,
                                 virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

16155 16156
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
16157

16158
    return 0;
16159
}
16160 16161 16162 16163 16164 16165 16166 16167 16168 16169 16170 16171 16172 16173 16174 16175 16176 16177 16178 16179 16180 16181 16182 16183 16184 16185 16186 16187 16188 16189 16190 16191 16192 16193 16194 16195 16196 16197 16198 16199 16200 16201 16202


/**
 * 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;
}
16203 16204 16205 16206 16207 16208 16209 16210 16211 16212


/**
 * 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
16213
                                  virStorageSourcePtr src G_GNUC_UNUSED,
16214 16215 16216 16217 16218
                                  unsigned int idx)
{
    char *ret = NULL;

    if (idx)
16219
        ret = g_strdup_printf("%s[%d]", disk->dst, idx);
16220
    else
16221
        ret = g_strdup(disk->dst);
16222 16223 16224

    return ret;
}
16225 16226 16227 16228 16229 16230 16231 16232


virStorageSourcePtr
qemuDomainGetStorageSourceByDevstr(const char *devstr,
                                   virDomainDefPtr def)
{
    virDomainDiskDefPtr disk = NULL;
    virStorageSourcePtr src = NULL;
16233
    g_autofree char *target = NULL;
16234 16235 16236 16237 16238 16239 16240 16241 16242 16243 16244 16245 16246 16247 16248 16249 16250 16251
    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
16252
                       _("failed to find disk '%s'"), target);
16253
        return NULL;
16254 16255
    }

16256 16257 16258 16259
    if (idx == 0)
        src = disk->src;
    else
        src = virStorageFileChainLookup(disk->src, NULL, NULL, idx, NULL);
16260 16261 16262

    return src;
}
16263 16264 16265 16266 16267 16268 16269 16270


static void
qemuDomainSaveCookieDispose(void *obj)
{
    qemuDomainSaveCookiePtr cookie = obj;

    VIR_DEBUG("cookie=%p", cookie);
16271 16272

    virCPUDefFree(cookie->cpu);
16273 16274 16275 16276
}


qemuDomainSaveCookiePtr
16277
qemuDomainSaveCookieNew(virDomainObjPtr vm)
16278
{
16279
    qemuDomainObjPrivatePtr priv = vm->privateData;
16280
    g_autoptr(qemuDomainSaveCookie) cookie = NULL;
16281 16282

    if (qemuDomainInitialize() < 0)
16283
        return NULL;
16284 16285

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
16286
        return NULL;
16287

16288
    if (priv->origCPU && !(cookie->cpu = virCPUDefCopy(vm->def->cpu)))
16289
        return NULL;
16290

16291 16292 16293 16294
    cookie->slirpHelper = qemuDomainGetSlirpHelperOk(vm);

    VIR_DEBUG("Save cookie %p, cpu=%p, slirpHelper=%d",
              cookie, cookie->cpu, cookie->slirpHelper);
16295

16296
    return g_steal_pointer(&cookie);
16297 16298 16299 16300
}


static int
J
Ján Tomko 已提交
16301
qemuDomainSaveCookieParse(xmlXPathContextPtr ctxt G_GNUC_UNUSED,
16302 16303
                          virObjectPtr *obj)
{
16304
    g_autoptr(qemuDomainSaveCookie) cookie = NULL;
16305 16306

    if (qemuDomainInitialize() < 0)
16307
        return -1;
16308 16309

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
16310
        return -1;
16311

16312 16313
    if (virCPUDefParseXML(ctxt, "./cpu[1]", VIR_CPU_TYPE_GUEST,
                          &cookie->cpu) < 0)
16314
        return -1;
16315

16316 16317
    cookie->slirpHelper = virXPathBoolean("boolean(./slirpHelper)", ctxt) > 0;

16318
    *obj = (virObjectPtr) g_steal_pointer(&cookie);
16319 16320 16321 16322 16323
    return 0;
}


static int
16324 16325
qemuDomainSaveCookieFormat(virBufferPtr buf,
                           virObjectPtr obj)
16326
{
16327 16328 16329
    qemuDomainSaveCookiePtr cookie = (qemuDomainSaveCookiePtr) obj;

    if (cookie->cpu &&
16330
        virCPUDefFormatBufFull(buf, cookie->cpu, NULL) < 0)
16331 16332
        return -1;

16333 16334 16335
    if (cookie->slirpHelper)
        virBufferAddLit(buf, "<slirpHelper/>\n");

16336 16337 16338 16339 16340 16341 16342 16343
    return 0;
}


virSaveCookieCallbacks virQEMUDriverDomainSaveCookie = {
    .parse = qemuDomainSaveCookieParse,
    .format = qemuDomainSaveCookieFormat,
};
16344 16345 16346 16347 16348 16349 16350 16351 16352 16353 16354 16355 16356 16357 16358 16359 16360 16361 16362 16363 16364 16365 16366 16367 16368 16369 16370 16371 16372 16373 16374 16375 16376 16377 16378 16379 16380 16381 16382 16383


/**
 * 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;
}
16384

16385 16386 16387 16388 16389 16390 16391 16392 16393 16394 16395 16396 16397 16398 16399 16400 16401 16402 16403 16404 16405 16406 16407 16408 16409 16410 16411 16412

/**
 * 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;
16413
    int ret = -1;
16414 16415 16416 16417 16418 16419 16420 16421 16422 16423 16424 16425 16426 16427 16428 16429 16430 16431 16432 16433 16434 16435 16436 16437 16438 16439 16440 16441 16442 16443

    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);
16444
        vm->def->cpu = g_steal_pointer(&fixedCPU);
16445 16446 16447 16448
    }

    if (fixedOrig) {
        virCPUDefFree(*origCPU);
16449
        *origCPU = g_steal_pointer(&fixedOrig);
16450 16451 16452 16453 16454 16455 16456 16457 16458 16459 16460
    }

    ret = 0;

 cleanup:
    virCPUDefFree(fixedCPU);
    virCPUDefFree(fixedOrig);
    return ret;
}


16461 16462 16463 16464 16465
char *
qemuDomainGetMachineName(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
16466
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
16467 16468
    char *ret = NULL;

16469
    if (vm->pid > 0) {
16470 16471 16472 16473 16474 16475
        ret = virSystemdGetMachineNameByPID(vm->pid);
        if (!ret)
            virResetLastError();
    }

    if (!ret)
16476 16477
        ret = virDomainGenerateMachineName("qemu", cfg->root,
                                           vm->def->id, vm->def->name,
16478 16479 16480 16481
                                           virQEMUDriverIsPrivileged(driver));

    return ret;
}
16482 16483 16484 16485 16486 16487 16488 16489 16490 16491 16492 16493 16494


/* 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,
16495
                                     const virDomainDeviceInfo *info,
16496 16497 16498
                                     virQEMUCapsPtr qemuCaps,
                                     const char *devicename)
{
16499
    if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
16500 16501 16502 16503 16504 16505
        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;
16506
        } else if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW)) {
16507 16508 16509 16510 16511
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CCW address type is not supported by "
                             "this QEMU"));
            return false;
        }
16512
    } else if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390) {
16513 16514 16515 16516 16517 16518 16519 16520 16521
        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;
}
16522 16523


16524
/**
16525
 * qemuDomainPrepareDiskSourceData:
16526 16527 16528 16529
 *
 * @disk: Disk config object
 * @src: source to start from
 *
16530
 * Prepares various aspects of a storage source belonging to a disk backing
16531 16532
 * chain based on the disk configuration. This function should be also called
 * for detected backing chain members.
16533
 */
16534
void
16535
qemuDomainPrepareDiskSourceData(virDomainDiskDefPtr disk,
16536
                                virStorageSourcePtr src)
16537
{
16538 16539 16540
    if (!disk)
        return;

16541
    /* transfer properties valid only for the top level image */
16542 16543
    if (src == disk->src)
        src->detect_zeroes = disk->detect_zeroes;
16544

16545 16546 16547 16548
    /* transfer properties valid for the full chain */
    src->iomode = disk->iomode;
    src->cachemode = disk->cachemode;
    src->discard = disk->discard;
16549

16550 16551
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY)
        src->floppyimg = true;
16552 16553 16554
}


16555 16556 16557 16558 16559 16560 16561 16562 16563
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;
}


16564 16565
static int
qemuDomainPrepareStorageSourcePR(virStorageSourcePtr src,
16566 16567
                                 qemuDomainObjPrivatePtr priv,
                                 const char *parentalias)
16568 16569 16570 16571 16572
{
    if (!src->pr)
        return 0;

    if (virStoragePRDefIsManaged(src->pr)) {
16573
        VIR_FREE(src->pr->path);
16574 16575
        if (!(src->pr->path = qemuDomainGetManagedPRSocketPath(priv)))
            return -1;
16576
        src->pr->mgralias = g_strdup(qemuDomainGetManagedPRAlias());
16577 16578 16579
    } else {
        if (!(src->pr->mgralias = qemuDomainGetUnmanagedPRAlias(parentalias)))
            return -1;
16580 16581 16582 16583 16584 16585
    }

    return 0;
}


16586 16587 16588 16589 16590 16591 16592 16593 16594 16595 16596 16597
/**
 * 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)
16598
{
16599 16600 16601
    if (qemuDomainValidateStorageSource(disk->src, priv->qemuCaps) < 0)
        return -1;

16602
    qemuDomainPrepareStorageSourceConfig(disk->src, cfg, priv->qemuCaps);
16603
    qemuDomainPrepareDiskSourceData(disk, disk->src);
16604

16605 16606 16607
    if (qemuDomainSecretStorageSourcePrepare(priv, disk->src,
                                             disk->info.alias,
                                             disk->info.alias) < 0)
16608 16609
        return -1;

16610
    if (qemuDomainPrepareStorageSourcePR(disk->src, priv, disk->info.alias) < 0)
16611
        return -1;
16612

16613 16614
    if (qemuDomainPrepareStorageSourceTLS(disk->src, cfg, disk->info.alias,
                                          priv->qemuCaps) < 0)
16615 16616
        return -1;

16617 16618 16619 16620
    return 0;
}


16621
int
16622 16623 16624 16625 16626 16627 16628
qemuDomainPrepareStorageSourceBlockdev(virDomainDiskDefPtr disk,
                                       virStorageSourcePtr src,
                                       qemuDomainObjPrivatePtr priv,
                                       virQEMUDriverConfigPtr cfg)
{
    src->id = qemuDomainStorageIdNew(priv);

16629 16630
    src->nodestorage = g_strdup_printf("libvirt-%u-storage", src->id);
    src->nodeformat = g_strdup_printf("libvirt-%u-format", src->id);
16631

16632
    if (qemuBlockStorageSourceNeedsStorageSliceLayer(src))
16633 16634
        src->sliceStorage->nodename = g_strdup_printf("libvirt-%u-slice-sto", src->id);

16635 16636 16637
    if (qemuDomainValidateStorageSource(src, priv->qemuCaps) < 0)
        return -1;

16638
    qemuDomainPrepareStorageSourceConfig(src, cfg, priv->qemuCaps);
16639
    qemuDomainPrepareDiskSourceData(disk, src);
16640 16641 16642 16643 16644 16645

    if (qemuDomainSecretStorageSourcePrepare(priv, src,
                                             src->nodestorage,
                                             src->nodeformat) < 0)
        return -1;

16646
    if (qemuDomainPrepareStorageSourcePR(src, priv, src->nodestorage) < 0)
16647 16648
        return -1;

16649
    if (qemuDomainPrepareStorageSourceTLS(src, cfg, src->nodestorage,
16650 16651 16652 16653 16654 16655 16656 16657 16658 16659 16660 16661 16662 16663 16664 16665
                                          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 &&
16666 16667
        !diskPriv->nodeCopyOnRead)
        diskPriv->nodeCopyOnRead = g_strdup_printf("libvirt-CoR-%s", disk->dst);
16668 16669 16670 16671 16672 16673 16674 16675 16676 16677

    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if (qemuDomainPrepareStorageSourceBlockdev(disk, n, priv, cfg) < 0)
            return -1;
    }

    return 0;
}


16678 16679 16680 16681 16682 16683 16684
int
qemuDomainPrepareDiskSource(virDomainDiskDefPtr disk,
                            qemuDomainObjPrivatePtr priv,
                            virQEMUDriverConfigPtr cfg)
{
    qemuDomainPrepareDiskCachemode(disk);

16685
    /* set default format for storage pool based disks */
16686
    if (disk->src->type == VIR_STORAGE_TYPE_VOLUME &&
16687 16688 16689 16690 16691 16692 16693 16694
        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;
    }
16695

16696 16697 16698 16699 16700 16701 16702
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if (qemuDomainPrepareDiskSourceBlockdev(disk, priv, cfg) < 0)
            return -1;
    } else {
        if (qemuDomainPrepareDiskSourceLegacy(disk, priv, cfg) < 0)
            return -1;
    }
16703

16704 16705
    return 0;
}
16706 16707


16708 16709 16710 16711 16712 16713 16714 16715 16716 16717 16718 16719 16720 16721 16722 16723 16724 16725 16726 16727 16728 16729 16730 16731 16732 16733 16734 16735 16736 16737 16738 16739 16740 16741 16742 16743 16744 16745 16746 16747 16748 16749 16750 16751 16752 16753 16754 16755 16756 16757 16758 16759 16760 16761 16762 16763 16764 16765 16766 16767 16768 16769 16770 16771 16772 16773 16774 16775 16776 16777 16778 16779 16780 16781 16782
/**
 * 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;
}


16783 16784 16785 16786 16787 16788 16789 16790 16791 16792
void
qemuProcessEventFree(struct qemuProcessEvent *event)
{
    if (!event)
        return;

    switch (event->eventType) {
    case QEMU_PROCESS_EVENT_GUESTPANIC:
        qemuMonitorEventPanicInfoFree(event->data);
        break;
16793 16794 16795
    case QEMU_PROCESS_EVENT_RDMA_GID_STATUS_CHANGED:
        qemuMonitorEventRdmaGidStatusFree(event->data);
        break;
16796 16797 16798 16799 16800 16801
    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:
16802
    case QEMU_PROCESS_EVENT_GUEST_CRASHLOADED:
16803 16804
        VIR_FREE(event->data);
        break;
16805 16806 16807
    case QEMU_PROCESS_EVENT_JOB_STATUS_CHANGE:
        virObjectUnref(event->data);
        break;
16808
    case QEMU_PROCESS_EVENT_PR_DISCONNECT:
16809 16810 16811 16812 16813
    case QEMU_PROCESS_EVENT_LAST:
        break;
    }
    VIR_FREE(event);
}
16814 16815 16816


char *
16817
qemuDomainGetManagedPRSocketPath(qemuDomainObjPrivatePtr priv)
16818 16819 16820
{
    char *ret = NULL;

16821
    ret = g_strdup_printf("%s/%s.sock", priv->libDir, qemuDomainGetManagedPRAlias());
16822 16823 16824

    return ret;
}
16825 16826 16827 16828 16829 16830 16831 16832 16833 16834 16835 16836 16837 16838 16839 16840 16841 16842 16843 16844 16845 16846 16847 16848 16849 16850 16851


/**
 * 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;
}
16852 16853 16854 16855 16856 16857 16858 16859 16860 16861 16862 16863 16864 16865 16866 16867 16868 16869 16870 16871 16872 16873 16874 16875 16876 16877 16878 16879 16880


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
16881 16882 16883 16884 16885


/* qemuDomainIsUsingNoShutdown:
 * @priv: Domain private data
 *
J
Ján Tomko 已提交
16886 16887
 * 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
16888 16889 16890 16891 16892 16893 16894
 *
 * 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
16895
    return priv->allowReboot == VIR_TRISTATE_BOOL_YES;
J
John Ferlan 已提交
16896
}
16897 16898 16899 16900 16901 16902 16903 16904 16905


bool
qemuDomainDiskIsMissingLocalOptional(virDomainDiskDefPtr disk)
{
    return disk->startupPolicy == VIR_DOMAIN_STARTUP_POLICY_OPTIONAL &&
           virStorageSourceIsLocalStorage(disk->src) && disk->src->path &&
           !virFileExists(disk->src->path);
}
16906 16907


16908
void
16909 16910 16911 16912
qemuDomainNVRAMPathFormat(virQEMUDriverConfigPtr cfg,
                            virDomainDefPtr def,
                            char **path)
{
16913
    *path = g_strdup_printf("%s/%s_VARS.fd", cfg->nvramDir, def->name);
16914 16915 16916
}


16917
void
16918 16919 16920
qemuDomainNVRAMPathGenerate(virQEMUDriverConfigPtr cfg,
                            virDomainDefPtr def)
{
16921
    if (virDomainDefHasOldStyleROUEFI(def) &&
16922 16923
        !def->os.loader->nvram)
        qemuDomainNVRAMPathFormat(cfg, def, &def->os.loader->nvram);
16924
}
16925 16926 16927 16928 16929 16930 16931 16932 16933 16934 16935 16936 16937 16938 16939 16940 16941 16942 16943 16944 16945 16946 16947 16948 16949 16950 16951 16952 16953 16954 16955 16956 16957 16958


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;
}
16959 16960 16961 16962


static int
qemuDomainDefHasManagedPRBlockjobIterator(void *payload,
J
Ján Tomko 已提交
16963
                                          const void *name G_GNUC_UNUSED,
16964 16965 16966 16967 16968 16969 16970 16971 16972 16973 16974 16975 16976 16977 16978 16979 16980 16981 16982 16983 16984 16985 16986 16987 16988 16989 16990 16991 16992 16993 16994 16995 16996 16997 16998 16999
                                          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;
}
17000 17001 17002 17003 17004 17005 17006 17007 17008 17009 17010 17011 17012


/**
 * 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)
{
17013 17014 17015 17016
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_INCREMENTAL_BACKUP) &&
        virDomainListCheckpoints(vm->checkpoints, NULL, NULL, NULL, 0) > 0) {
17017 17018 17019 17020 17021 17022 17023
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s",
                       _("cannot perform block operations while checkpoint exists"));
        return -1;
    }

    return 0;
}
17024 17025 17026 17027 17028 17029 17030 17031 17032 17033 17034 17035 17036 17037 17038 17039 17040 17041 17042 17043 17044 17045 17046 17047 17048 17049

/**
 * qemuDomainInitializePflashStorageSource:
 *
 * This helper converts the specification of the source of the 'loader' in case
 * PFLASH is required to virStorageSources in case QEMU_CAPS_BLOCKDEV is present.
 *
 * This helper is used in the intermediate state when we don't support full
 * backing chains for pflash drives in the XML.
 *
 * The nodenames used here have a different prefix to allow for a later
 * conversion. The prefixes are 'libvirt-pflash0-storage',
 * 'libvirt-pflash0-format' for pflash0 and 'libvirt-pflash1-storage' and
 * 'libvirt-pflash1-format' for pflash1.
 */
int
qemuDomainInitializePflashStorageSource(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDefPtr def = vm->def;
    g_autoptr(virStorageSource) pflash0 = NULL;
    g_autoptr(virStorageSource) pflash1 = NULL;

    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV))
        return 0;

17050
    if (!virDomainDefHasOldStyleUEFI(def))
17051 17052 17053 17054 17055 17056 17057 17058 17059 17060 17061 17062 17063 17064 17065 17066 17067 17068 17069 17070 17071 17072 17073 17074 17075 17076 17077 17078 17079 17080
        return 0;

    if (!(pflash0 = virStorageSourceNew()))
        return -1;

    pflash0->type = VIR_STORAGE_TYPE_FILE;
    pflash0->format = VIR_STORAGE_FILE_RAW;
    pflash0->path = g_strdup(def->os.loader->path);
    pflash0->readonly = def->os.loader->readonly;
    pflash0->nodeformat = g_strdup("libvirt-pflash0-format");
    pflash0->nodestorage = g_strdup("libvirt-pflash0-storage");


    if (def->os.loader->nvram) {
        if (!(pflash1 = virStorageSourceNew()))
            return -1;

        pflash1->type = VIR_STORAGE_TYPE_FILE;
        pflash1->format = VIR_STORAGE_FILE_RAW;
        pflash1->path = g_strdup(def->os.loader->nvram);
        pflash1->readonly = false;
        pflash1->nodeformat = g_strdup("libvirt-pflash1-format");
        pflash1->nodestorage = g_strdup("libvirt-pflash1-storage");
    }

    priv->pflash0 = g_steal_pointer(&pflash0);
    priv->pflash1 = g_steal_pointer(&pflash1);

    return 0;
}