qemu_domain.c 508.2 KB
Newer Older
1
/*
2
 * qemu_domain.c: QEMU domain private state
3
 *
4
 * Copyright (C) 2006-2019 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24
 */

#include <config.h>

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

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

80
#include "dosname.h"
81

82 83
#define QEMU_QXL_VGAMEM_DEFAULT 16 * 1024

84 85
#define VIR_FROM_THIS VIR_FROM_QEMU

86 87
VIR_LOG_INIT("qemu.qemu_domain");

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

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

108 109
VIR_ENUM_IMPL(qemuDomainAsyncJob,
              QEMU_ASYNC_JOB_LAST,
110 111 112 113 114
              "none",
              "migration out",
              "migration in",
              "save",
              "dump",
115
              "snapshot",
116
              "start",
117
              "backup",
118 119
);

120 121
VIR_ENUM_IMPL(qemuDomainNamespace,
              QEMU_DOMAIN_NS_LAST,
122 123 124
              "mount",
);

125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155

/**
 * 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;
}


156
struct _qemuDomainLogContext {
157 158
    virObject parent;

159
    int writefd;
160
    int readfd; /* Only used if manager == NULL */
161
    off_t pos;
162
    ino_t inode; /* Only used if manager != NULL */
163
    char *path;
164
    virLogManagerPtr manager;
165 166
};

167
static virClassPtr qemuDomainLogContextClass;
168
static virClassPtr qemuDomainSaveCookieClass;
169 170

static void qemuDomainLogContextDispose(void *obj);
171
static void qemuDomainSaveCookieDispose(void *obj);
172

173

174
static int
175
qemuDomainOnceInit(void)
176
{
177
    if (!VIR_CLASS_NEW(qemuDomainLogContext, virClassForObject()))
178 179
        return -1;

180
    if (!VIR_CLASS_NEW(qemuDomainSaveCookie, virClassForObject()))
181 182
        return -1;

183 184 185
    return 0;
}

186
VIR_ONCE_GLOBAL_INIT(qemuDomain);
187 188 189 190 191 192 193 194 195 196 197 198 199

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

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

J
Jiri Denemark 已提交
200
const char *
201
qemuDomainAsyncJobPhaseToString(qemuDomainAsyncJob job,
J
Ján Tomko 已提交
202
                                int phase G_GNUC_UNUSED)
J
Jiri Denemark 已提交
203 204 205 206
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
207 208
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
209 210
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
211
    case QEMU_ASYNC_JOB_SNAPSHOT:
212
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
213
    case QEMU_ASYNC_JOB_NONE:
214
    case QEMU_ASYNC_JOB_BACKUP:
215
        G_GNUC_FALLTHROUGH;
216 217
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
218 219 220 221 222 223
    }

    return "none";
}

int
224
qemuDomainAsyncJobPhaseFromString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
225 226 227 228 229 230 231 232
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
233 234
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
235 236
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
237
    case QEMU_ASYNC_JOB_SNAPSHOT:
238
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
239
    case QEMU_ASYNC_JOB_NONE:
240
    case QEMU_ASYNC_JOB_BACKUP:
241
        G_GNUC_FALLTHROUGH;
242 243
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
244 245 246 247 248 249 250 251
    }

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

252

253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
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;
}


285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
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;
        }
    }
}


301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320
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);
321
    virObjectEventStateQueue(driver->domainEventState, event);
322 323 324
}


325 326 327 328 329 330 331 332
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

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

333
    if (virCondInit(&priv->job.asyncCond) < 0) {
334
        virCondDestroy(&priv->job.cond);
335 336 337
        return -1;
    }

338 339 340 341 342 343
    return 0;
}

static void
qemuDomainObjResetJob(qemuDomainObjPrivatePtr priv)
{
344
    qemuDomainJobObjPtr job = &priv->job;
345 346

    job->active = QEMU_JOB_NONE;
347
    job->owner = 0;
348
    job->ownerAPI = NULL;
J
Jiri Denemark 已提交
349
    job->started = 0;
350 351
}

352 353 354 355 356 357 358 359 360 361 362 363 364

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

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


365 366 367
static void
qemuDomainObjResetAsyncJob(qemuDomainObjPrivatePtr priv)
{
368
    qemuDomainJobObjPtr job = &priv->job;
369 370

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
371
    job->asyncOwner = 0;
372
    job->asyncOwnerAPI = NULL;
J
Jiri Denemark 已提交
373
    job->asyncStarted = 0;
J
Jiri Denemark 已提交
374
    job->phase = 0;
375
    job->mask = QEMU_JOB_DEFAULT_MASK;
376
    job->abortJob = false;
377
    job->spiceMigration = false;
378
    job->spiceMigrated = false;
379 380
    job->dumpCompleted = false;
    VIR_FREE(job->error);
J
Jiri Denemark 已提交
381
    VIR_FREE(job->current);
382 383
    qemuMigrationParamsFree(job->migParams);
    job->migParams = NULL;
384
    job->apiFlags = 0;
385 386
}

387 388
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
389
                        qemuDomainJobObjPtr job)
390 391 392 393 394
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
395
    job->owner = priv->job.owner;
396
    job->asyncJob = priv->job.asyncJob;
397
    job->asyncOwner = priv->job.asyncOwner;
J
Jiri Denemark 已提交
398
    job->phase = priv->job.phase;
399
    job->migParams = g_steal_pointer(&priv->job.migParams);
400
    job->apiFlags = priv->job.apiFlags;
401 402 403 404 405

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

406 407 408
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
409 410
    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
J
Jiri Denemark 已提交
411
    VIR_FREE(priv->job.current);
412
    VIR_FREE(priv->job.completed);
413 414
    virCondDestroy(&priv->job.cond);
    virCondDestroy(&priv->job.asyncCond);
415 416
}

417
static bool
418
qemuDomainTrackJob(qemuDomainJob job)
419 420 421 422
{
    return (QEMU_DOMAIN_TRACK_JOBS & JOB_MASK(job)) != 0;
}

423

J
Jiri Denemark 已提交
424 425 426 427 428 429 430 431 432 433 434
int
qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->started)
        return 0;

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

435 436 437 438 439 440
    if (now < jobInfo->started) {
        VIR_WARN("Async job starts in the future");
        jobInfo->started = 0;
        return 0;
    }

J
Jiri Denemark 已提交
441 442 443 444
    jobInfo->timeElapsed = now - jobInfo->started;
    return 0;
}

445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461
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;
    }

462 463
    jobInfo->stats.mig.downtime = now - jobInfo->stopped;
    jobInfo->stats.mig.downtime_set = true;
464 465 466
    return 0;
}

467 468 469 470 471 472 473 474
static virDomainJobType
qemuDomainJobStatusToType(qemuDomainJobStatus status)
{
    switch (status) {
    case QEMU_DOMAIN_JOB_STATUS_NONE:
        break;

    case QEMU_DOMAIN_JOB_STATUS_ACTIVE:
475
    case QEMU_DOMAIN_JOB_STATUS_MIGRATING:
476
    case QEMU_DOMAIN_JOB_STATUS_QEMU_COMPLETED:
477
    case QEMU_DOMAIN_JOB_STATUS_POSTCOPY:
478
    case QEMU_DOMAIN_JOB_STATUS_PAUSED:
479 480 481 482 483 484 485 486 487 488 489 490 491 492 493
        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
494 495 496 497
int
qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                        virDomainJobInfoPtr info)
{
498
    info->type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
499 500
    info->timeElapsed = jobInfo->timeElapsed;

501 502 503 504 505 506 507 508 509 510 511 512 513
    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
514

515 516 517 518 519 520
    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;

521
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
522 523 524 525 526
        info->memTotal = jobInfo->stats.dump.total;
        info->memProcessed = jobInfo->stats.dump.completed;
        info->memRemaining = info->memTotal - info->memProcessed;
        break;

527 528 529 530 531 532
    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;

533 534 535
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
        break;
    }
J
Jiri Denemark 已提交
536 537 538 539 540 541 542 543

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

    return 0;
}

544 545 546 547 548 549

static int
qemuDomainMigrationJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                                   int *type,
                                   virTypedParameterPtr *params,
                                   int *nparams)
J
Jiri Denemark 已提交
550
{
551
    qemuMonitorMigrationStats *stats = &jobInfo->stats.mig;
552
    qemuDomainMirrorStatsPtr mirrorStats = &jobInfo->mirrorStats;
J
Jiri Denemark 已提交
553 554 555
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;
556 557
    unsigned long long mirrorRemaining = mirrorStats->total -
                                         mirrorStats->transferred;
J
Jiri Denemark 已提交
558

559 560 561 562 563
    if (virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_OPERATION,
                             jobInfo->operation) < 0)
        goto error;

J
Jiri Denemark 已提交
564 565 566 567 568
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED,
                                jobInfo->timeElapsed) < 0)
        goto error;

569 570 571 572 573 574 575
    if (jobInfo->timeDeltaSet &&
        jobInfo->timeElapsed > jobInfo->timeDelta &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED_NET,
                                jobInfo->timeElapsed - jobInfo->timeDelta) < 0)
        goto error;

576
    if (stats->downtime_set &&
J
Jiri Denemark 已提交
577 578
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME,
579
                                stats->downtime) < 0)
J
Jiri Denemark 已提交
580 581
        goto error;

582
    if (stats->downtime_set &&
583
        jobInfo->timeDeltaSet &&
584
        stats->downtime > jobInfo->timeDelta &&
585 586
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME_NET,
587
                                stats->downtime - jobInfo->timeDelta) < 0)
588 589
        goto error;

590
    if (stats->setup_time_set &&
591 592
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_SETUP_TIME,
593
                                stats->setup_time) < 0)
594 595
        goto error;

J
Jiri Denemark 已提交
596 597
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_TOTAL,
598
                                stats->ram_total +
599 600
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
601 602
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_PROCESSED,
603
                                stats->ram_transferred +
604 605
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
606 607
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_REMAINING,
608
                                stats->ram_remaining +
609 610
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
611 612 613 614
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
615
                                stats->ram_total) < 0 ||
J
Jiri Denemark 已提交
616 617
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
618
                                stats->ram_transferred) < 0 ||
J
Jiri Denemark 已提交
619 620
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
621
                                stats->ram_remaining) < 0)
J
Jiri Denemark 已提交
622 623
        goto error;

624
    if (stats->ram_bps &&
625 626
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_BPS,
627
                                stats->ram_bps) < 0)
628 629
        goto error;

630
    if (stats->ram_duplicate_set) {
J
Jiri Denemark 已提交
631 632
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_CONSTANT,
633
                                    stats->ram_duplicate) < 0 ||
J
Jiri Denemark 已提交
634 635
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL,
636
                                    stats->ram_normal) < 0 ||
J
Jiri Denemark 已提交
637 638
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL_BYTES,
639
                                    stats->ram_normal_bytes) < 0)
J
Jiri Denemark 已提交
640 641 642
            goto error;
    }

643 644 645 646 647
    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,
648 649 650 651
                                stats->ram_iteration) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_POSTCOPY_REQS,
                                stats->ram_postcopy_reqs) < 0)
652 653
        goto error;

654 655 656 657 658 659
    if (stats->ram_page_size > 0 &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PAGE_SIZE,
                                stats->ram_page_size) < 0)
        goto error;

660 661 662 663 664
    /* 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
665 666
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_TOTAL,
667 668
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
669 670
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_PROCESSED,
671 672
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
673 674
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_REMAINING,
675 676
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
677 678
        goto error;

679
    if (stats->disk_bps &&
680 681
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_BPS,
682
                                stats->disk_bps) < 0)
683 684
        goto error;

685
    if (stats->xbzrle_set) {
J
Jiri Denemark 已提交
686 687
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE,
688
                                    stats->xbzrle_cache_size) < 0 ||
J
Jiri Denemark 已提交
689 690
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_BYTES,
691
                                    stats->xbzrle_bytes) < 0 ||
J
Jiri Denemark 已提交
692 693
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_PAGES,
694
                                    stats->xbzrle_pages) < 0 ||
J
Jiri Denemark 已提交
695 696
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE_MISSES,
697
                                    stats->xbzrle_cache_miss) < 0 ||
J
Jiri Denemark 已提交
698 699
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_OVERFLOW,
700
                                    stats->xbzrle_overflow) < 0)
J
Jiri Denemark 已提交
701 702 703
            goto error;
    }

704 705 706 707 708 709
    if (stats->cpu_throttle_percentage &&
        virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_AUTO_CONVERGE_THROTTLE,
                             stats->cpu_throttle_percentage) < 0)
        goto error;

710
 done:
711
    *type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
712 713 714 715 716 717 718 719 720 721
    *params = par;
    *nparams = npar;
    return 0;

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


722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764
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;
}


765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817
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;
}


818 819 820 821 822 823 824 825
int
qemuDomainJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                          int *type,
                          virTypedParameterPtr *params,
                          int *nparams)
{
    switch (jobInfo->statsType) {
    case QEMU_DOMAIN_JOB_STATS_TYPE_MIGRATION:
826
    case QEMU_DOMAIN_JOB_STATS_TYPE_SAVEDUMP:
827 828
        return qemuDomainMigrationJobInfoToParams(jobInfo, type, params, nparams);

829
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
830 831
        return qemuDomainDumpJobInfoToParams(jobInfo, type, params, nparams);

832 833 834
    case QEMU_DOMAIN_JOB_STATS_TYPE_BACKUP:
        return qemuDomainBackupJobInfoToParams(jobInfo, type, params, nparams);

835
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
836 837 838 839 840 841
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid job statistics type"));
        break;

    default:
        virReportEnumRangeError(qemuDomainJobStatsType, jobInfo->statsType);
842 843 844 845 846 847 848
        break;
    }

    return -1;
}


J
John Ferlan 已提交
849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
/* 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:
871 872
 * @driver: qemu driver data
 * @vm: Pointer to the vm object
J
John Ferlan 已提交
873 874 875 876 877
 *
 * 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
 */
878
int
M
Martin Kletzander 已提交
879 880
qemuDomainWriteMasterKeyFile(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
J
John Ferlan 已提交
881 882 883 884
{
    char *path;
    int fd = -1;
    int ret = -1;
M
Martin Kletzander 已提交
885
    qemuDomainObjPrivatePtr priv = vm->privateData;
J
John Ferlan 已提交
886

887 888 889 890
    /* Only gets filled in if we have the capability */
    if (!priv->masterKey)
        return 0;

J
John Ferlan 已提交
891 892 893 894 895 896 897 898 899 900 901 902 903 904 905
    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;
    }

906
    if (qemuSecurityDomainSetPathLabel(driver, vm, path, false) < 0)
M
Martin Kletzander 已提交
907 908
        goto cleanup;

J
John Ferlan 已提交
909 910 911 912 913 914 915 916 917 918
    ret = 0;

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

    return ret;
}


919 920 921 922 923 924 925 926 927
static void
qemuDomainMasterKeyFree(qemuDomainObjPrivatePtr priv)
{
    if (!priv->masterKey)
        return;

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

J
John Ferlan 已提交
928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025
/* qemuDomainMasterKeyReadFile:
 * @priv: pointer to domain private object
 *
 * Expected to be called during qemuProcessReconnect once the domain
 * libDir has been generated through qemuStateInitialize calling
 * virDomainObjListLoadAllConfigs which will restore the libDir path
 * to the domain private object.
 *
 * This function will get the path to the master key file and if it
 * exists, it will read the contents of the file saving it in priv->masterKey.
 *
 * Once the file exists, the validity checks may cause failures; however,
 * if the file doesn't exist or the capability doesn't exist, we just
 * return (mostly) quietly.
 *
 * Returns 0 on success or lack of capability
 *        -1 on failure with error message indicating failure
 */
int
qemuDomainMasterKeyReadFile(qemuDomainObjPrivatePtr priv)
{
    char *path;
    int fd = -1;
    uint8_t *masterKey = NULL;
    ssize_t masterKeyLen = 0;

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

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

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

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

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

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

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

    ignore_value(VIR_REALLOC_N_QUIET(masterKey, masterKeyLen));

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

    VIR_FORCE_CLOSE(fd);
    VIR_FREE(path);

    return 0;

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

    VIR_FORCE_CLOSE(fd);
    VIR_FREE(path);

    return -1;
}


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

    if (!priv->masterKey)
        return;

    /* Clear the contents */
1026
    qemuDomainMasterKeyFree(priv);
J
John Ferlan 已提交
1027 1028 1029 1030 1031 1032 1033 1034 1035 1036

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

    VIR_FREE(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 1111 1112 1113 1114 1115 1116
}


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

    qemuDomainSecretInfoClear(*secinfo, false);
J
John Ferlan 已提交
1117

1118 1119 1120 1121
    VIR_FREE(*secinfo);
}


1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
/**
 * qemuDomainSecretInfoDestroy:
 * @secinfo: object to destroy
 *
 * Removes any data unnecessary for further use, but keeps alias allocated.
 */
void
qemuDomainSecretInfoDestroy(qemuDomainSecretInfoPtr secinfo)
{
    qemuDomainSecretInfoClear(secinfo, true);
}


1135
static virClassPtr qemuDomainDiskPrivateClass;
1136
static void qemuDomainDiskPrivateDispose(void *obj);
1137 1138 1139 1140

static int
qemuDomainDiskPrivateOnceInit(void)
{
1141
    if (!VIR_CLASS_NEW(qemuDomainDiskPrivate, virClassForObject()))
1142
        return -1;
1143 1144

    return 0;
1145 1146
}

1147
VIR_ONCE_GLOBAL_INIT(qemuDomainDiskPrivate);
1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162

static virObjectPtr
qemuDomainDiskPrivateNew(void)
{
    qemuDomainDiskPrivatePtr priv;

    if (qemuDomainDiskPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}

1163 1164 1165 1166 1167
static void
qemuDomainDiskPrivateDispose(void *obj)
{
    qemuDomainDiskPrivatePtr priv = obj;

1168
    virObjectUnref(priv->migrSource);
1169
    VIR_FREE(priv->qomName);
1170
    VIR_FREE(priv->nodeCopyOnRead);
1171
    virObjectUnref(priv->blockjob);
1172
}
1173

1174 1175 1176 1177 1178 1179
static virClassPtr qemuDomainStorageSourcePrivateClass;
static void qemuDomainStorageSourcePrivateDispose(void *obj);

static int
qemuDomainStorageSourcePrivateOnceInit(void)
{
1180
    if (!VIR_CLASS_NEW(qemuDomainStorageSourcePrivate, virClassForObject()))
1181
        return -1;
1182 1183

    return 0;
1184 1185
}

1186
VIR_ONCE_GLOBAL_INIT(qemuDomainStorageSourcePrivate);
1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212

virObjectPtr
qemuDomainStorageSourcePrivateNew(void)
{
    qemuDomainStorageSourcePrivatePtr priv;

    if (qemuDomainStorageSourcePrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}


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

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


1213 1214 1215 1216 1217 1218
static virClassPtr qemuDomainVcpuPrivateClass;
static void qemuDomainVcpuPrivateDispose(void *obj);

static int
qemuDomainVcpuPrivateOnceInit(void)
{
1219
    if (!VIR_CLASS_NEW(qemuDomainVcpuPrivate, virClassForObject()))
1220
        return -1;
1221 1222

    return 0;
1223 1224
}

1225
VIR_ONCE_GLOBAL_INIT(qemuDomainVcpuPrivate);
1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242

static virObjectPtr
qemuDomainVcpuPrivateNew(void)
{
    qemuDomainVcpuPrivatePtr priv;

    if (qemuDomainVcpuPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}


static void
1243
qemuDomainVcpuPrivateDispose(void *obj)
1244
{
1245 1246 1247 1248
    qemuDomainVcpuPrivatePtr priv = obj;

    VIR_FREE(priv->type);
    VIR_FREE(priv->alias);
1249
    virJSONValueFree(priv->props);
1250 1251 1252 1253
    return;
}


1254 1255
static virClassPtr qemuDomainChrSourcePrivateClass;
static void qemuDomainChrSourcePrivateDispose(void *obj);
1256 1257

static int
1258
qemuDomainChrSourcePrivateOnceInit(void)
1259
{
1260
    if (!VIR_CLASS_NEW(qemuDomainChrSourcePrivate, virClassForObject()))
1261
        return -1;
1262 1263

    return 0;
1264 1265
}

1266
VIR_ONCE_GLOBAL_INIT(qemuDomainChrSourcePrivate);
1267 1268

static virObjectPtr
1269
qemuDomainChrSourcePrivateNew(void)
1270
{
1271
    qemuDomainChrSourcePrivatePtr priv;
1272

1273
    if (qemuDomainChrSourcePrivateInitialize() < 0)
1274 1275
        return NULL;

1276
    if (!(priv = virObjectNew(qemuDomainChrSourcePrivateClass)))
1277 1278 1279 1280 1281 1282 1283
        return NULL;

    return (virObjectPtr) priv;
}


static void
1284
qemuDomainChrSourcePrivateDispose(void *obj)
1285
{
1286
    qemuDomainChrSourcePrivatePtr priv = obj;
1287 1288 1289 1290 1291

    qemuDomainSecretInfoFree(&priv->secinfo);
}


J
Ján Tomko 已提交
1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303
static virClassPtr qemuDomainVsockPrivateClass;
static void qemuDomainVsockPrivateDispose(void *obj);

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

    return 0;
}

1304
VIR_ONCE_GLOBAL_INIT(qemuDomainVsockPrivate);
J
Ján Tomko 已提交
1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316

static virObjectPtr
qemuDomainVsockPrivateNew(void)
{
    qemuDomainVsockPrivatePtr priv;

    if (qemuDomainVsockPrivateInitialize() < 0)
        return NULL;

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

1317 1318
    priv->vhostfd = -1;

J
Ján Tomko 已提交
1319 1320 1321 1322 1323
    return (virObjectPtr) priv;
}


static void
J
Ján Tomko 已提交
1324
qemuDomainVsockPrivateDispose(void *obj G_GNUC_UNUSED)
J
Ján Tomko 已提交
1325
{
1326 1327 1328
    qemuDomainVsockPrivatePtr priv = obj;

    VIR_FORCE_CLOSE(priv->vhostfd);
J
Ján Tomko 已提交
1329 1330 1331
}


1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
static virClassPtr qemuDomainGraphicsPrivateClass;
static void qemuDomainGraphicsPrivateDispose(void *obj);

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

    return 0;
}

1344
VIR_ONCE_GLOBAL_INIT(qemuDomainGraphicsPrivate);
1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366

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);
1367
    qemuDomainSecretInfoFree(&priv->secinfo);
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 1400 1401 1402 1403
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
1404
qemuDomainNetworkPrivateDispose(void *obj G_GNUC_UNUSED)
1405
{
1406 1407 1408
    qemuDomainNetworkPrivatePtr priv = obj;

    qemuSlirpFree(priv->slirp);
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
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;

1439 1440
    priv->vhost_user_fd = -1;

1441 1442 1443 1444 1445
    return (virObjectPtr) priv;
}


static void
1446
qemuDomainVideoPrivateDispose(void *obj)
1447
{
1448 1449 1450
    qemuDomainVideoPrivatePtr priv = obj;

    VIR_FORCE_CLOSE(priv->vhost_user_fd);
1451 1452 1453
}


1454 1455
/* qemuDomainSecretPlainSetup:
 * @secinfo: Pointer to secret info
J
John Ferlan 已提交
1456
 * @usageType: The virSecretUsageType
1457 1458
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1459 1460 1461 1462 1463 1464
 *
 * Taking a secinfo, fill in the plaintext information
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1465
qemuDomainSecretPlainSetup(qemuDomainSecretInfoPtr secinfo,
J
John Ferlan 已提交
1466
                           virSecretUsageType usageType,
1467 1468
                           const char *username,
                           virSecretLookupTypeDefPtr seclookupdef)
1469
{
1470 1471 1472 1473 1474 1475 1476
    virConnectPtr conn;
    int ret = -1;

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

1477
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN;
1478
    secinfo->s.plain.username = g_strdup(username);
1479

1480 1481 1482 1483 1484 1485
    ret = virSecretGetSecretString(conn, seclookupdef, usageType,
                                   &secinfo->s.plain.secret,
                                   &secinfo->s.plain.secretlen);

    virObjectUnref(conn);
    return ret;
1486 1487 1488
}


1489 1490 1491 1492
/* qemuDomainSecretAESSetup:
 * @priv: pointer to domain private object
 * @secinfo: Pointer to secret info
 * @srcalias: Alias of the disk/hostdev used to generate the secret alias
J
John Ferlan 已提交
1493
 * @usageType: The virSecretUsageType
1494 1495
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1496
 * @isLuks: True/False for is for luks (alias generation)
1497 1498 1499 1500 1501 1502
 *
 * Taking a secinfo, fill in the AES specific information using the
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1503
qemuDomainSecretAESSetup(qemuDomainObjPrivatePtr priv,
1504 1505
                         qemuDomainSecretInfoPtr secinfo,
                         const char *srcalias,
J
John Ferlan 已提交
1506
                         virSecretUsageType usageType,
1507
                         const char *username,
1508 1509
                         virSecretLookupTypeDefPtr seclookupdef,
                         bool isLuks)
1510
{
1511
    virConnectPtr conn;
1512 1513 1514 1515 1516 1517 1518 1519
    int ret = -1;
    uint8_t *raw_iv = NULL;
    size_t ivlen = QEMU_DOMAIN_AES_IV_LEN;
    uint8_t *secret = NULL;
    size_t secretlen = 0;
    uint8_t *ciphertext = NULL;
    size_t ciphertextlen = 0;

1520 1521 1522 1523
    conn = virGetConnectSecret();
    if (!conn)
        return -1;

1524
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_AES;
1525
    secinfo->s.aes.username = g_strdup(username);
1526

1527
    if (!(secinfo->s.aes.alias = qemuDomainGetSecretAESAlias(srcalias, isLuks)))
1528
        goto cleanup;
1529

1530 1531 1532
    if (VIR_ALLOC_N(raw_iv, ivlen) < 0)
        goto cleanup;

1533
    /* Create a random initialization vector */
1534
    if (virRandomBytes(raw_iv, ivlen) < 0)
1535
        goto cleanup;
1536 1537

    /* Encode the IV and save that since qemu will need it */
1538
    secinfo->s.aes.iv = g_base64_encode(raw_iv, ivlen);
1539 1540

    /* Grab the unencoded secret */
J
John Ferlan 已提交
1541
    if (virSecretGetSecretString(conn, seclookupdef, usageType,
1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554
                                 &secret, &secretlen) < 0)
        goto cleanup;

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

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

    /* Now encode the ciphertext and store to be passed to qemu */
1555 1556
    secinfo->s.aes.ciphertext = g_base64_encode(ciphertext,
                                                ciphertextlen);
1557 1558 1559 1560 1561 1562 1563

    ret = 0;

 cleanup:
    VIR_DISPOSE_N(raw_iv, ivlen);
    VIR_DISPOSE_N(secret, secretlen);
    VIR_DISPOSE_N(ciphertext, ciphertextlen);
1564
    virObjectUnref(conn);
1565 1566 1567 1568
    return ret;
}


1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584
/**
 * 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;
}


1585
/* qemuDomainSecretInfoNewPlain:
1586
 * @usageType: Secret usage type
1587 1588
 * @username: username
 * @lookupDef: lookup def describing secret
1589
 *
1590
 * Helper function to create a secinfo to be used for secinfo consumers. This
1591
 * sets up a 'plain' (unencrypted) secret for legacy consumers.
1592 1593 1594 1595 1596
 *
 * Returns @secinfo on success, NULL on failure. Caller is responsible
 * to eventually free @secinfo.
 */
static qemuDomainSecretInfoPtr
1597
qemuDomainSecretInfoNewPlain(virSecretUsageType usageType,
1598
                             const char *username,
1599
                             virSecretLookupTypeDefPtr lookupDef)
1600 1601 1602 1603 1604 1605
{
    qemuDomainSecretInfoPtr secinfo = NULL;

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

1606 1607 1608
    if (qemuDomainSecretPlainSetup(secinfo, usageType, username, lookupDef) < 0) {
        qemuDomainSecretInfoFree(&secinfo);
        return NULL;
1609 1610 1611 1612 1613 1614
    }

    return secinfo;
}


1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657
/* qemuDomainSecretInfoNew:
 * @priv: pointer to domain private object
 * @srcAlias: Alias base to use for TLS object
 * @usageType: Secret usage type
 * @username: username
 * @looupDef: lookup def describing secret
 * @isLuks: boolean for luks lookup
 *
 * Helper function to create a secinfo to be used for secinfo consumers. This
 * sets up encrypted data to be used with qemu's 'secret' object.
 *
 * Returns @secinfo on success, NULL on failure. Caller is responsible
 * to eventually free @secinfo.
 */
static qemuDomainSecretInfoPtr
qemuDomainSecretInfoNew(qemuDomainObjPrivatePtr priv,
                        const char *srcAlias,
                        virSecretUsageType usageType,
                        const char *username,
                        virSecretLookupTypeDefPtr lookupDef,
                        bool isLuks)
{
    qemuDomainSecretInfoPtr secinfo = NULL;

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

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

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

    return secinfo;
}


1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668
/**
 * 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.
 */
1669
qemuDomainSecretInfoPtr
1670
qemuDomainSecretInfoTLSNew(qemuDomainObjPrivatePtr priv,
1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683
                           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;

1684 1685 1686
    return qemuDomainSecretInfoNew(priv, srcAlias,
                                   VIR_SECRET_USAGE_TYPE_TLS, NULL,
                                   &seclookupdef, false);
1687 1688 1689
}


1690 1691 1692
void
qemuDomainSecretDiskDestroy(virDomainDiskDefPtr disk)
{
1693 1694
    qemuDomainStorageSourcePrivatePtr srcPriv;
    virStorageSourcePtr n;
1695

1696 1697 1698 1699 1700 1701
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if ((srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(n))) {
            qemuDomainSecretInfoDestroy(srcPriv->secinfo);
            qemuDomainSecretInfoDestroy(srcPriv->encinfo);
        }
    }
1702 1703 1704
}


1705
bool
1706
qemuDomainStorageSourceHasAuth(virStorageSourcePtr src)
1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718
{
    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;
}


1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730
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;
}


1731 1732 1733 1734 1735 1736
/**
 * 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
1737
 *
1738 1739 1740 1741
 * 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.
1742
 *
1743
 * Returns 0 on success; -1 on error while reporting an libvirt error.
1744
 */
1745
static int
1746
qemuDomainSecretStorageSourcePrepare(qemuDomainObjPrivatePtr priv,
1747 1748 1749
                                     virStorageSourcePtr src,
                                     const char *authalias,
                                     const char *encalias)
1750
{
1751
    qemuDomainStorageSourcePrivatePtr srcPriv;
1752
    bool iscsiHasPS = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET);
1753
    bool hasAuth = qemuDomainStorageSourceHasAuth(src);
1754 1755 1756 1757
    bool hasEnc = qemuDomainDiskHasEncryptionSecret(src);

    if (!hasAuth && !hasEnc)
        return 0;
1758

1759
    if (!(src->privateData = qemuDomainStorageSourcePrivateNew()))
1760 1761
        return -1;

1762
    srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
1763

1764
    if (hasAuth) {
J
John Ferlan 已提交
1765
        virSecretUsageType usageType = VIR_SECRET_USAGE_TYPE_ISCSI;
1766

1767
        if (src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD)
J
John Ferlan 已提交
1768
            usageType = VIR_SECRET_USAGE_TYPE_CEPH;
1769

1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784
        if (!qemuDomainSupportsEncryptedSecret(priv) ||
            (src->protocol == VIR_STORAGE_NET_PROTOCOL_ISCSI && !iscsiHasPS)) {
            srcPriv->secinfo = qemuDomainSecretInfoNewPlain(usageType,
                                                            src->auth->username,
                                                            &src->auth->seclookupdef);
        } else {
            srcPriv->secinfo = qemuDomainSecretInfoNew(priv, authalias,
                                                       usageType,
                                                       src->auth->username,
                                                       &src->auth->seclookupdef,
                                                       false);
        }

        if (!srcPriv->secinfo)
            return -1;
1785 1786
    }

1787
    if (hasEnc) {
1788
        if (!(srcPriv->encinfo =
1789 1790 1791 1792
              qemuDomainSecretInfoNew(priv, encalias,
                                      VIR_SECRET_USAGE_TYPE_VOLUME, NULL,
                                      &src->encryption->secrets[0]->seclookupdef,
                                      true)))
1793
              return -1;
1794 1795
    }

1796 1797 1798 1799
    return 0;
}


1800 1801 1802
void
qemuDomainSecretHostdevDestroy(virDomainHostdevDefPtr hostdev)
{
1803
    qemuDomainStorageSourcePrivatePtr srcPriv;
1804

1805 1806 1807
    if (virHostdevIsSCSIDevice(hostdev)) {
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1808

1809 1810 1811 1812 1813 1814
        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI) {
            srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(iscsisrc->src);
            if (srcPriv && srcPriv->secinfo)
                qemuDomainSecretInfoFree(&srcPriv->secinfo);
        }
    }
1815 1816 1817 1818
}


/* qemuDomainSecretHostdevPrepare:
J
John Ferlan 已提交
1819
 * @priv: pointer to domain private object
1820 1821 1822 1823 1824 1825 1826
 * @hostdev: Pointer to a hostdev definition
 *
 * For the right host device, generate the qemuDomainSecretInfo structure.
 *
 * Returns 0 on success, -1 on failure
 */
int
1827
qemuDomainSecretHostdevPrepare(qemuDomainObjPrivatePtr priv,
1828 1829
                               virDomainHostdevDefPtr hostdev)
{
1830
    if (virHostdevIsSCSIDevice(hostdev)) {
1831 1832
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1833
        virStorageSourcePtr src = iscsisrc->src;
1834 1835

        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI &&
1836
            src->auth) {
1837 1838
            if (qemuDomainSecretStorageSourcePrepare(priv, src,
                                                     hostdev->info->alias, NULL) < 0)
1839 1840 1841 1842 1843 1844 1845 1846
                return -1;
        }
    }

    return 0;
}


1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
void
qemuDomainSecretChardevDestroy(virDomainChrSourceDefPtr dev)
{
    qemuDomainChrSourcePrivatePtr chrSourcePriv =
        QEMU_DOMAIN_CHR_SOURCE_PRIVATE(dev);

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

    qemuDomainSecretInfoFree(&chrSourcePriv->secinfo);
}


/* qemuDomainSecretChardevPrepare:
 * @cfg: Pointer to driver config object
 * @priv: pointer to domain private object
 * @chrAlias: Alias of the chr device
 * @dev: Pointer to a char source definition
 *
 * For a TCP character device, generate a qemuDomainSecretInfo to be used
 * by the command line code to generate the secret for the tls-creds to use.
 *
 * Returns 0 on success, -1 on failure
 */
int
1872
qemuDomainSecretChardevPrepare(virQEMUDriverConfigPtr cfg,
1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887
                               qemuDomainObjPrivatePtr priv,
                               const char *chrAlias,
                               virDomainChrSourceDefPtr dev)
{
    char *charAlias = NULL;

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

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

        if (!(charAlias = qemuAliasChardevFromDevAlias(chrAlias)))
1888
            return -1;
1889

1890
        chrSourcePriv->secinfo =
1891
            qemuDomainSecretInfoTLSNew(priv, charAlias,
1892
                                       cfg->chardevTLSx509secretUUID);
1893
        VIR_FREE(charAlias);
1894 1895 1896

        if (!chrSourcePriv->secinfo)
            return -1;
1897 1898 1899 1900 1901 1902
    }

    return 0;
}


1903 1904 1905 1906 1907 1908 1909 1910 1911
static void
qemuDomainSecretGraphicsDestroy(virDomainGraphicsDefPtr graphics)
{
    qemuDomainGraphicsPrivatePtr gfxPriv = QEMU_DOMAIN_GRAPHICS_PRIVATE(graphics);

    if (!gfxPriv)
        return;

    VIR_FREE(gfxPriv->tlsAlias);
1912
    qemuDomainSecretInfoFree(&gfxPriv->secinfo);
1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932
}


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;

1933
    gfxPriv->tlsAlias = g_strdup("vnc-tls-creds0");
1934

1935 1936 1937 1938 1939 1940 1941
    if (cfg->vncTLSx509secretUUID) {
        gfxPriv->secinfo = qemuDomainSecretInfoTLSNew(priv, gfxPriv->tlsAlias,
                                                      cfg->vncTLSx509secretUUID);
        if (!gfxPriv->secinfo)
            return -1;
    }

1942 1943 1944 1945
    return 0;
}


1946 1947 1948
/* qemuDomainSecretDestroy:
 * @vm: Domain object
 *
1949
 * Removes all unnecessary data which was needed to generate 'secret' objects.
1950 1951 1952 1953 1954 1955 1956 1957
 */
void
qemuDomainSecretDestroy(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++)
        qemuDomainSecretDiskDestroy(vm->def->disks[i]);
1958 1959 1960

    for (i = 0; i < vm->def->nhostdevs; i++)
        qemuDomainSecretHostdevDestroy(vm->def->hostdevs[i]);
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986

    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);
1987 1988 1989

    for (i = 0; i < vm->def->ngraphics; i++)
        qemuDomainSecretGraphicsDestroy(vm->def->graphics[i]);
1990 1991 1992 1993
}


/* qemuDomainSecretPrepare:
1994
 * @driver: Pointer to driver object
1995 1996 1997
 * @vm: Domain object
 *
 * For any objects that may require an auth/secret setup, create a
1998
 * qemuDomainSecretInfo and save it in the appropriate place within
1999 2000 2001 2002 2003 2004 2005
 * 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
2006
qemuDomainSecretPrepare(virQEMUDriverPtr driver,
2007 2008
                        virDomainObjPtr vm)
{
J
John Ferlan 已提交
2009
    qemuDomainObjPrivatePtr priv = vm->privateData;
2010
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2011
    size_t i;
2012
    int ret = -1;
2013

2014
    /* disk secrets are prepared when preparing disks */
2015

2016
    for (i = 0; i < vm->def->nhostdevs; i++) {
2017
        if (qemuDomainSecretHostdevPrepare(priv,
J
John Ferlan 已提交
2018
                                           vm->def->hostdevs[i]) < 0)
2019
            goto cleanup;
2020 2021
    }

2022
    for (i = 0; i < vm->def->nserials; i++) {
2023
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2024 2025 2026 2027 2028 2029
                                           vm->def->serials[i]->info.alias,
                                           vm->def->serials[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nparallels; i++) {
2030
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2031 2032 2033 2034 2035 2036
                                           vm->def->parallels[i]->info.alias,
                                           vm->def->parallels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nchannels; i++) {
2037
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2038 2039 2040 2041 2042 2043
                                           vm->def->channels[i]->info.alias,
                                           vm->def->channels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nconsoles; i++) {
2044
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2045 2046 2047 2048 2049 2050 2051 2052
                                           vm->def->consoles[i]->info.alias,
                                           vm->def->consoles[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nsmartcards; i++)
        if (vm->def->smartcards[i]->type ==
            VIR_DOMAIN_SMARTCARD_TYPE_PASSTHROUGH &&
2053
            qemuDomainSecretChardevPrepare(cfg, priv,
2054 2055 2056 2057 2058 2059
                                           vm->def->smartcards[i]->info.alias,
                                           vm->def->smartcards[i]->data.passthru) < 0)
            goto cleanup;

    for (i = 0; i < vm->def->nrngs; i++) {
        if (vm->def->rngs[i]->backend == VIR_DOMAIN_RNG_BACKEND_EGD &&
2060
            qemuDomainSecretChardevPrepare(cfg, priv,
2061 2062 2063 2064 2065 2066
                                           vm->def->rngs[i]->info.alias,
                                           vm->def->rngs[i]->source.chardev) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nredirdevs; i++) {
2067
        if (qemuDomainSecretChardevPrepare(cfg, priv,
2068 2069 2070 2071 2072
                                           vm->def->redirdevs[i]->info.alias,
                                           vm->def->redirdevs[i]->source) < 0)
            goto cleanup;
    }

2073 2074 2075 2076 2077
    for (i = 0; i < vm->def->ngraphics; i++) {
        if (qemuDomainSecretGraphicsPrepare(cfg, priv, vm->def->graphics[i]) < 0)
            goto cleanup;
    }

2078 2079 2080 2081 2082
    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
2083 2084 2085
}


2086
/* This is the old way of setting up per-domain directories */
2087
static void
2088 2089 2090 2091 2092 2093
qemuDomainSetPrivatePathsOld(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

2094 2095
    if (!priv->libDir)
        priv->libDir = g_strdup_printf("%s/domain-%s", cfg->libDir, vm->def->name);
2096

2097 2098 2099
    if (!priv->channelTargetDir)
        priv->channelTargetDir = g_strdup_printf("%s/domain-%s",
                                                 cfg->channelTargetDir, vm->def->name);
2100 2101 2102 2103 2104 2105

    virObjectUnref(cfg);
}


int
2106 2107
qemuDomainSetPrivatePaths(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
2108
{
2109 2110
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainObjPrivatePtr priv = vm->privateData;
2111
    char *domname = virDomainDefGetShortName(vm->def);
2112
    int ret = -1;
2113

M
Martin Kletzander 已提交
2114 2115 2116
    if (!domname)
        goto cleanup;

2117 2118
    if (!priv->libDir)
        priv->libDir = g_strdup_printf("%s/domain-%s", cfg->libDir, domname);
2119

2120 2121 2122
    if (!priv->channelTargetDir)
        priv->channelTargetDir = g_strdup_printf("%s/domain-%s",
                                                 cfg->channelTargetDir, domname);
2123

2124 2125 2126
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
M
Martin Kletzander 已提交
2127
    VIR_FREE(domname);
2128
    return ret;
2129 2130 2131
}


M
Marc-André Lureau 已提交
2132
static void
2133
dbusVMStateHashFree(void *opaque)
M
Marc-André Lureau 已提交
2134 2135 2136 2137 2138
{
    qemuDBusVMStateFree(opaque);
}


2139
static void *
2140
qemuDomainObjPrivateAlloc(void *opaque)
2141 2142 2143 2144 2145 2146
{
    qemuDomainObjPrivatePtr priv;

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

2147 2148 2149
    if (qemuDomainObjInitJob(priv) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to init qemu driver mutexes"));
2150
        goto error;
2151
    }
2152

2153
    if (!(priv->devs = virChrdevAlloc()))
2154 2155
        goto error;

2156 2157 2158
    if (!(priv->blockjobs = virHashCreate(5, virObjectFreeHashData)))
        goto error;

M
Marc-André Lureau 已提交
2159 2160 2161
    if (!(priv->dbusVMStates = virHashCreate(5, dbusVMStateHashFree)))
        goto error;

2162 2163
    /* agent commands block by default, user can choose different behavior */
    priv->agentTimeout = VIR_DOMAIN_AGENT_RESPONSE_TIMEOUT_BLOCK;
2164
    priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
2165
    priv->driver = opaque;
2166

2167
    return priv;
2168

2169
 error:
2170 2171
    VIR_FREE(priv);
    return NULL;
2172 2173
}

2174 2175 2176 2177 2178 2179 2180 2181 2182
/**
 * 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)
2183
{
2184 2185 2186
    virStringListFree(priv->qemuDevices);
    priv->qemuDevices = NULL;

2187
    virCgroupFree(&priv->cgroup);
2188 2189 2190 2191 2192

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

    VIR_FREE(priv->machineName);
2193

2194
    virObjectUnref(priv->qemuCaps);
2195
    priv->qemuCaps = NULL;
2196

2197
    VIR_FREE(priv->pidfile);
2198

2199 2200 2201
    VIR_FREE(priv->libDir);
    VIR_FREE(priv->channelTargetDir);

2202 2203
    priv->memPrealloc = false;

2204 2205 2206 2207 2208 2209 2210
    /* remove automatic pinning data */
    virBitmapFree(priv->autoNodeset);
    priv->autoNodeset = NULL;
    virBitmapFree(priv->autoCpuset);
    priv->autoCpuset = NULL;

    /* remove address data */
2211
    virDomainPCIAddressSetFree(priv->pciaddrs);
2212
    priv->pciaddrs = NULL;
J
Ján Tomko 已提交
2213
    virDomainUSBAddressSetFree(priv->usbaddrs);
2214 2215 2216 2217 2218 2219 2220 2221
    priv->usbaddrs = NULL;

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

    /* clear previously used namespaces */
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
2222

2223 2224
    priv->rememberOwner = false;

2225
    priv->reconnectBlockjobs = VIR_TRISTATE_BOOL_ABSENT;
2226
    priv->allowReboot = VIR_TRISTATE_BOOL_ABSENT;
2227 2228 2229

    virBitmapFree(priv->migrationCaps);
    priv->migrationCaps = NULL;
2230

2231
    virHashRemoveAll(priv->blockjobs);
M
Marc-André Lureau 已提交
2232
    virHashRemoveAll(priv->dbusVMStates);
2233 2234 2235 2236 2237

    virObjectUnref(priv->pflash0);
    priv->pflash0 = NULL;
    virObjectUnref(priv->pflash1);
    priv->pflash1 = NULL;
2238 2239 2240

    virDomainBackupDefFree(priv->backup);
    priv->backup = NULL;
2241 2242 2243 2244 2245 2246 2247 2248 2249 2250
}


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

    qemuDomainObjPrivateDataClear(priv);

2251
    virObjectUnref(priv->monConfig);
2252
    qemuDomainObjFreeJob(priv);
2253
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
2254
    VIR_FREE(priv->origname);
2255

2256
    virChrdevFree(priv->devs);
2257

2258 2259
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
2260
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
2261 2262
        qemuMonitorClose(priv->mon);
    }
D
Daniel P. Berrange 已提交
2263 2264 2265 2266
    if (priv->agent) {
        VIR_ERROR(_("Unexpected QEMU agent still active during domain deletion"));
        qemuAgentClose(priv->agent);
    }
2267
    VIR_FREE(priv->cleanupCallbacks);
2268 2269

    qemuDomainSecretInfoFree(&priv->migSecinfo);
2270
    qemuDomainMasterKeyFree(priv);
2271

2272
    virHashFree(priv->blockjobs);
M
Marc-André Lureau 已提交
2273
    virHashFree(priv->dbusVMStates);
2274

2275 2276 2277 2278
    VIR_FREE(priv);
}


2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293
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)
2294
        (*secinfo)->s.aes.alias = g_steal_pointer(&*alias);
2295 2296 2297 2298 2299

    return 0;
}


2300 2301 2302 2303
static int
qemuStorageSourcePrivateDataParse(xmlXPathContextPtr ctxt,
                                  virStorageSourcePtr src)
{
2304 2305 2306 2307 2308
    qemuDomainStorageSourcePrivatePtr priv;
    char *authalias = NULL;
    char *encalias = NULL;
    int ret = -1;

2309 2310
    src->nodestorage = virXPathString("string(./nodenames/nodename[@type='storage']/@name)", ctxt);
    src->nodeformat = virXPathString("string(./nodenames/nodename[@type='format']/@name)", ctxt);
2311
    src->tlsAlias = virXPathString("string(./objects/TLSx509/@alias)", ctxt);
2312

2313 2314 2315
    if (src->pr)
        src->pr->mgralias = virXPathString("string(./reservations/@mgralias)", ctxt);

2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332
    authalias = virXPathString("string(./objects/secret[@type='auth']/@alias)", ctxt);
    encalias = virXPathString("string(./objects/secret[@type='encryption']/@alias)", ctxt);

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

        priv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);

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

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

2333
    if (virStorageSourcePrivateDataParseRelPath(ctxt, src) < 0)
2334
        goto cleanup;
2335

2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357
    ret = 0;

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

    return ret;
}


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

    virBufferAsprintf(buf, "<secret type='%s' alias='%s'/>\n",
                      type, secinfo->s.aes.alias);
2358 2359 2360 2361 2362 2363 2364
}


static int
qemuStorageSourcePrivateDataFormat(virStorageSourcePtr src,
                                   virBufferPtr buf)
{
2365
    g_auto(virBuffer) tmp = VIR_BUFFER_INIT_CHILD(buf);
2366 2367
    qemuDomainStorageSourcePrivatePtr srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);

2368 2369 2370 2371 2372 2373 2374 2375 2376
    if (src->nodestorage || src->nodeformat) {
        virBufferAddLit(buf, "<nodenames>\n");
        virBufferAdjustIndent(buf, 2);
        virBufferEscapeString(buf, "<nodename type='storage' name='%s'/>\n", src->nodestorage);
        virBufferEscapeString(buf, "<nodename type='format' name='%s'/>\n", src->nodeformat);
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</nodenames>\n");
    }

2377 2378 2379
    if (src->pr)
        virBufferAsprintf(buf, "<reservations mgralias='%s'/>\n", src->pr->mgralias);

2380
    if (virStorageSourcePrivateDataFormatRelPath(src, buf) < 0)
2381
        return -1;
2382

2383 2384 2385 2386 2387
    if (srcPriv) {
        qemuStorageSourcePrivateDataFormatSecinfo(&tmp, srcPriv->secinfo, "auth");
        qemuStorageSourcePrivateDataFormatSecinfo(&tmp, srcPriv->encinfo, "encryption");
    }

2388 2389 2390
    if (src->tlsAlias)
        virBufferAsprintf(&tmp, "<TLSx509 alias='%s'/>\n", src->tlsAlias);

2391
    virXMLFormatElement(buf, "objects", NULL, &tmp);
2392

2393
    return 0;
2394 2395 2396
}


2397 2398 2399 2400 2401 2402 2403
static int
qemuDomainDiskPrivateParse(xmlXPathContextPtr ctxt,
                           virDomainDiskDefPtr disk)
{
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);

    priv->qomName = virXPathString("string(./qom/@name)", ctxt);
2404
    priv->nodeCopyOnRead = virXPathString("string(./nodenames/nodename[@type='copyOnRead']/@name)", ctxt);
2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417

    return 0;
}


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

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

2418 2419 2420 2421 2422 2423 2424 2425 2426
    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");
    }

2427 2428 2429 2430
    return 0;
}


2431 2432
static void
qemuDomainObjPrivateXMLFormatVcpus(virBufferPtr buf,
2433
                                   virDomainDefPtr def)
2434 2435
{
    size_t i;
2436 2437 2438
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    pid_t tid;
2439 2440 2441 2442

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

2443 2444 2445 2446 2447 2448 2449 2450 2451
    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);
    }
2452 2453 2454 2455 2456 2457

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


2458
static int
2459
qemuDomainObjPrivateXMLFormatAutomaticPlacement(virBufferPtr buf,
2460
                                                qemuDomainObjPrivatePtr priv)
2461 2462
{
    char *nodeset = NULL;
2463
    char *cpuset = NULL;
2464 2465
    int ret = -1;

2466
    if (!priv->autoNodeset && !priv->autoCpuset)
2467 2468
        return 0;

2469 2470
    if (priv->autoNodeset &&
        !((nodeset = virBitmapFormat(priv->autoNodeset))))
2471 2472
        goto cleanup;

2473 2474 2475 2476 2477 2478 2479 2480
    if (priv->autoCpuset &&
        !((cpuset = virBitmapFormat(priv->autoCpuset))))
        goto cleanup;

    virBufferAddLit(buf, "<numad");
    virBufferEscapeString(buf, " nodeset='%s'", nodeset);
    virBufferEscapeString(buf, " cpuset='%s'", cpuset);
    virBufferAddLit(buf, "/>\n");
2481 2482 2483 2484 2485

    ret = 0;

 cleanup:
    VIR_FREE(nodeset);
2486
    VIR_FREE(cpuset);
2487 2488 2489 2490
    return ret;
}


2491 2492 2493 2494 2495 2496 2497
typedef struct qemuDomainPrivateBlockJobFormatData {
    virDomainXMLOptionPtr xmlopt;
    virBufferPtr buf;
} qemuDomainPrivateBlockJobFormatData;


static int
2498 2499 2500 2501 2502
qemuDomainObjPrivateXMLFormatBlockjobFormatSource(virBufferPtr buf,
                                                  const char *element,
                                                  virStorageSourcePtr src,
                                                  virDomainXMLOptionPtr xmlopt,
                                                  bool chain)
2503
{
2504
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2505
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2506 2507 2508 2509 2510 2511 2512 2513 2514
    unsigned int xmlflags = VIR_DOMAIN_DEF_FORMAT_STATUS;

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

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

2515 2516
    if (chain &&
        virDomainDiskBackingStoreFormat(&childBuf, src, xmlopt, xmlflags) < 0)
2517 2518
        return -1;

2519
    virXMLFormatElement(buf, element, &attrBuf, &childBuf);
2520 2521 2522 2523 2524

    return 0;
}


2525 2526
static int
qemuDomainObjPrivateXMLFormatBlockjobIterator(void *payload,
J
Ján Tomko 已提交
2527
                                              const void *name G_GNUC_UNUSED,
2528
                                              void *opaque)
2529
{
2530
    struct qemuDomainPrivateBlockJobFormatData *data = opaque;
2531
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2532 2533
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(data->buf);
    g_auto(virBuffer) chainsBuf = VIR_BUFFER_INIT_CHILD(&childBuf);
2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544
    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);
2545 2546
    if (job->brokentype != QEMU_BLOCKJOB_TYPE_NONE)
        virBufferEscapeString(&attrBuf, " brokentype='%s'", qemuBlockjobTypeToString(job->brokentype));
2547 2548
    virBufferEscapeString(&childBuf, "<errmsg>%s</errmsg>", job->errmsg);

2549
    if (job->disk) {
2550 2551 2552 2553
        virBufferEscapeString(&childBuf, "<disk dst='%s'", job->disk->dst);
        if (job->mirrorChain)
            virBufferAddLit(&childBuf, " mirror='yes'");
        virBufferAddLit(&childBuf, "/>\n");
2554 2555
    } else {
        if (job->chain &&
2556 2557 2558 2559 2560
            qemuDomainObjPrivateXMLFormatBlockjobFormatSource(&chainsBuf,
                                                              "disk",
                                                              job->chain,
                                                              data->xmlopt,
                                                              true) < 0)
2561
            return -1;
2562

2563
        if (job->mirrorChain &&
2564 2565 2566 2567 2568
            qemuDomainObjPrivateXMLFormatBlockjobFormatSource(&chainsBuf,
                                                              "mirror",
                                                              job->mirrorChain,
                                                              data->xmlopt,
                                                              true) < 0)
2569 2570
            return -1;

2571
        virXMLFormatElement(&childBuf, "chains", NULL, &chainsBuf);
2572 2573
    }

2574 2575 2576 2577 2578 2579 2580 2581
    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:
2582 2583 2584 2585 2586 2587
            if (job->data.commit.base)
                virBufferAsprintf(&childBuf, "<base node='%s'/>\n", job->data.commit.base->nodeformat);
            if (job->data.commit.top)
                virBufferAsprintf(&childBuf, "<top node='%s'/>\n", job->data.commit.top->nodeformat);
            if (job->data.commit.topparent)
                virBufferAsprintf(&childBuf, "<topparent node='%s'/>\n", job->data.commit.topparent->nodeformat);
2588 2589
            if (job->data.commit.deleteCommittedImages)
                virBufferAddLit(&childBuf, "<deleteCommittedImages/>\n");
2590 2591
            break;

2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604
        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;

2605
        case QEMU_BLOCKJOB_TYPE_COPY:
2606 2607 2608 2609
            if (job->data.copy.shallownew)
                virBufferAddLit(&attrBuf, " shallownew='yes'");
            break;

P
Peter Krempa 已提交
2610
        case QEMU_BLOCKJOB_TYPE_BACKUP:
2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622
            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;

                if (job->data.backup.deleteStore)
                    virBufferAddLit(&childBuf, "<deleteStore/>\n");
            }
P
Peter Krempa 已提交
2623
            break;
2624 2625

        case QEMU_BLOCKJOB_TYPE_BROKEN:
2626 2627 2628 2629 2630 2631
        case QEMU_BLOCKJOB_TYPE_NONE:
        case QEMU_BLOCKJOB_TYPE_INTERNAL:
        case QEMU_BLOCKJOB_TYPE_LAST:
            break;
    }

2632 2633
    virXMLFormatElement(data->buf, "blockjob", &attrBuf, &childBuf);
    return 0;
2634 2635 2636
}


2637 2638 2639 2640
static int
qemuDomainObjPrivateXMLFormatBlockjobs(virBufferPtr buf,
                                       virDomainObjPtr vm)
{
2641
    qemuDomainObjPrivatePtr priv = vm->privateData;
2642
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2643
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2644
    bool bj = qemuDomainHasBlockjob(vm, false);
2645 2646
    struct qemuDomainPrivateBlockJobFormatData iterdata = { priv->driver->xmlopt,
                                                            &childBuf };
2647 2648 2649 2650

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

2651 2652 2653
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV) &&
        virHashForEach(priv->blockjobs,
                       qemuDomainObjPrivateXMLFormatBlockjobIterator,
2654
                       &iterdata) < 0)
2655 2656
        return -1;

2657 2658
    virXMLFormatElement(buf, "blockjobs", &attrBuf, &childBuf);
    return 0;
2659 2660 2661
}


2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681
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;
}


2682
void
2683 2684 2685 2686 2687 2688 2689 2690 2691
qemuDomainObjPrivateXMLFormatAllowReboot(virBufferPtr buf,
                                         virTristateBool allowReboot)
{
    virBufferAsprintf(buf, "<allowReboot value='%s'/>\n",
                      virTristateBoolTypeToString(allowReboot));

}


2692 2693 2694 2695 2696 2697 2698 2699 2700
static void
qemuDomainObjPrivateXMLFormatPR(virBufferPtr buf,
                                qemuDomainObjPrivatePtr priv)
{
    if (priv->prDaemonRunning)
        virBufferAddLit(buf, "<prDaemon/>\n");
}


2701 2702
static int
qemuDomainObjPrivateXMLFormatNBDMigrationSource(virBufferPtr buf,
2703 2704
                                                virStorageSourcePtr src,
                                                virDomainXMLOptionPtr xmlopt)
2705
{
2706
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2707
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2708 2709 2710 2711 2712

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

2713
    if (virDomainDiskSourceFormat(&childBuf, src, "source", 0, false,
2714
                                  VIR_DOMAIN_DEF_FORMAT_STATUS, xmlopt) < 0)
2715
        return -1;
2716

2717
    virXMLFormatElement(buf, "migrationSource", &attrBuf, &childBuf);
2718

2719
    return 0;
2720 2721 2722 2723
}


static int
2724 2725 2726
qemuDomainObjPrivateXMLFormatNBDMigration(virBufferPtr buf,
                                          virDomainObjPtr vm)
{
2727
    qemuDomainObjPrivatePtr priv = vm->privateData;
2728 2729 2730 2731 2732
    size_t i;
    virDomainDiskDefPtr disk;
    qemuDomainDiskPrivatePtr diskPriv;

    for (i = 0; i < vm->def->ndisks; i++) {
2733 2734
        g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
        g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2735 2736
        disk = vm->def->disks[i];
        diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
2737 2738

        virBufferAsprintf(&attrBuf, " dev='%s' migrating='%s'",
2739
                          disk->dst, diskPriv->migrating ? "yes" : "no");
2740 2741 2742

        if (diskPriv->migrSource &&
            qemuDomainObjPrivateXMLFormatNBDMigrationSource(&childBuf,
2743 2744
                                                            diskPriv->migrSource,
                                                            priv->driver->xmlopt) < 0)
2745
            return -1;
2746

2747
        virXMLFormatElement(buf, "disk", &attrBuf, &childBuf);
2748
    }
2749

2750
    return 0;
2751 2752 2753
}


2754
static int
2755 2756 2757 2758
qemuDomainObjPrivateXMLFormatJob(virBufferPtr buf,
                                 virDomainObjPtr vm,
                                 qemuDomainObjPrivatePtr priv)
{
2759
    g_auto(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
2760
    g_auto(virBuffer) childBuf = VIR_BUFFER_INIT_CHILD(buf);
2761
    qemuDomainJob job = priv->job.active;
2762 2763

    if (!qemuDomainTrackJob(job))
2764
        job = QEMU_JOB_NONE;
2765

2766 2767
    if (job == QEMU_JOB_NONE &&
        priv->job.asyncJob == QEMU_ASYNC_JOB_NONE)
2768 2769 2770
        return 0;

    virBufferAsprintf(&attrBuf, " type='%s' async='%s'",
2771 2772
                      qemuDomainJobTypeToString(job),
                      qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
2773

2774
    if (priv->job.phase) {
2775
        virBufferAsprintf(&attrBuf, " phase='%s'",
2776 2777 2778
                          qemuDomainAsyncJobPhaseToString(priv->job.asyncJob,
                                                          priv->job.phase));
    }
2779

2780 2781 2782
    if (priv->job.asyncJob != QEMU_ASYNC_JOB_NONE)
        virBufferAsprintf(&attrBuf, " flags='0x%lx'", priv->job.apiFlags);

2783 2784
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_OUT &&
        qemuDomainObjPrivateXMLFormatNBDMigration(&childBuf, vm) < 0)
2785
        return -1;
2786

2787 2788 2789
    if (priv->job.migParams)
        qemuMigrationParamsFormat(&childBuf, priv->job.migParams);

2790
    virXMLFormatElement(buf, "job", &attrBuf, &childBuf);
2791

2792
    return 0;
2793 2794 2795
}


2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811
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;
}


2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829
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;
}


2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870
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;
}

2871
static int
2872 2873
qemuDomainObjPrivateXMLFormat(virBufferPtr buf,
                              virDomainObjPtr vm)
2874
{
2875
    qemuDomainObjPrivatePtr priv = vm->privateData;
2876 2877 2878 2879
    const char *monitorpath;

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
2880
        switch (priv->monConfig->type) {
2881
        case VIR_DOMAIN_CHR_TYPE_UNIX:
2882
            monitorpath = priv->monConfig->data.nix.path;
2883 2884 2885
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
2886
            monitorpath = priv->monConfig->data.file.path;
2887 2888 2889
            break;
        }

2890
        virBufferEscapeString(buf, "<monitor path='%s'", monitorpath);
2891
        virBufferAsprintf(buf, " type='%s'/>\n",
2892
                          virDomainChrTypeToString(priv->monConfig->type));
2893 2894
    }

2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905
    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");
    }

2906
    qemuDomainObjPrivateXMLFormatVcpus(buf, vm->def);
2907

2908
    if (priv->qemuCaps) {
2909
        size_t i;
2910 2911
        virBufferAddLit(buf, "<qemuCaps>\n");
        virBufferAdjustIndent(buf, 2);
2912
        for (i = 0; i < QEMU_CAPS_LAST; i++) {
2913
            if (virQEMUCapsGet(priv->qemuCaps, i)) {
2914
                virBufferAsprintf(buf, "<flag name='%s'/>\n",
2915
                                  virQEMUCapsTypeToString(i));
2916 2917
            }
        }
2918 2919
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</qemuCaps>\n");
2920 2921
    }

2922
    if (priv->lockState)
2923
        virBufferAsprintf(buf, "<lockstate>%s</lockstate>\n", priv->lockState);
2924

2925 2926
    if (qemuDomainObjPrivateXMLFormatJob(buf, vm, priv) < 0)
        return -1;
2927

2928
    if (priv->fakeReboot)
2929
        virBufferAddLit(buf, "<fakereboot/>\n");
2930

2931 2932
    if (priv->qemuDevices && *priv->qemuDevices) {
        char **tmp = priv->qemuDevices;
2933 2934
        virBufferAddLit(buf, "<devices>\n");
        virBufferAdjustIndent(buf, 2);
2935
        while (*tmp) {
2936
            virBufferAsprintf(buf, "<device alias='%s'/>\n", *tmp);
2937 2938
            tmp++;
        }
2939 2940
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</devices>\n");
2941 2942
    }

2943
    if (qemuDomainObjPrivateXMLFormatAutomaticPlacement(buf, priv) < 0)
2944
        return -1;
2945

2946 2947 2948 2949 2950
    /* Various per-domain paths */
    virBufferEscapeString(buf, "<libDir path='%s'/>\n", priv->libDir);
    virBufferEscapeString(buf, "<channelTargetDir path='%s'/>\n",
                          priv->channelTargetDir);

2951
    virCPUDefFormatBufFull(buf, priv->origCPU, NULL);
2952

2953
    if (priv->chardevStdioLogd)
2954
        virBufferAddLit(buf, "<chardevStdioLogd/>\n");
2955

2956 2957 2958
    if (priv->rememberOwner)
        virBufferAddLit(buf, "<rememberOwner/>\n");

2959 2960
    qemuDomainObjPrivateXMLFormatAllowReboot(buf, priv->allowReboot);

2961 2962
    qemuDomainObjPrivateXMLFormatPR(buf, priv);

2963 2964 2965
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV))
        virBufferAsprintf(buf, "<nodename index='%llu'/>\n", priv->nodenameindex);

2966 2967 2968
    if (priv->memPrealloc)
        virBufferAddLit(buf, "<memPrealloc/>\n");

2969 2970 2971
    if (qemuDomainObjPrivateXMLFormatBlockjobs(buf, vm) < 0)
        return -1;

2972 2973 2974
    if (qemuDomainObjPrivateXMLFormatSlirp(buf, vm) < 0)
        return -1;

2975 2976
    virBufferAsprintf(buf, "<agentTimeout>%i</agentTimeout>\n", priv->agentTimeout);

2977 2978 2979
    if (qemuDomainObjPrivateXMLFormatBackups(buf, vm) < 0)
        return -1;

2980 2981 2982
    return 0;
}

2983 2984 2985 2986

static int
qemuDomainObjPrivateXMLParseVcpu(xmlNodePtr node,
                                 unsigned int idx,
2987
                                 virDomainDefPtr def)
2988
{
2989
    virDomainVcpuDefPtr vcpu;
2990
    char *idstr;
2991
    char *pidstr;
2992
    unsigned int tmp;
2993 2994
    int ret = -1;

2995 2996
    idstr = virXMLPropString(node, "id");

2997 2998
    if (idstr &&
        (virStrToLong_uip(idstr, NULL, 10, &idx) < 0)) {
2999
        virReportError(VIR_ERR_INTERNAL_ERROR,
3000 3001 3002 3003 3004 3005
                       _("cannot parse vcpu index '%s'"), idstr);
        goto cleanup;
    }
    if (!(vcpu = virDomainDefGetVcpu(def, idx))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("invalid vcpu index '%u'"), idx);
3006
        goto cleanup;
3007 3008
    }

3009 3010 3011
    if (!(pidstr = virXMLPropString(node, "pid")))
        goto cleanup;

3012
    if (virStrToLong_uip(pidstr, NULL, 10, &tmp) < 0)
3013 3014
        goto cleanup;

3015 3016
    QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid = tmp;

3017 3018 3019
    ret = 0;

 cleanup:
3020
    VIR_FREE(idstr);
3021 3022 3023 3024 3025
    VIR_FREE(pidstr);
    return ret;
}


3026 3027 3028 3029 3030
static int
qemuDomainObjPrivateXMLParseAutomaticPlacement(xmlXPathContextPtr ctxt,
                                               qemuDomainObjPrivatePtr priv,
                                               virQEMUDriverPtr driver)
{
3031
    g_autoptr(virCapsHostNUMA) caps = NULL;
3032
    char *nodeset;
3033
    char *cpuset;
3034 3035
    int nodesetSize = 0;
    size_t i;
3036 3037 3038
    int ret = -1;

    nodeset = virXPathString("string(./numad/@nodeset)", ctxt);
3039
    cpuset = virXPathString("string(./numad/@cpuset)", ctxt);
3040

3041
    if (!nodeset && !cpuset)
3042 3043
        return 0;

3044
    if (!(caps = virQEMUDriverGetHostNUMACaps(driver)))
3045 3046
        goto cleanup;

3047 3048 3049
    /* 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 */
3050
    for (i = 0; i < caps->cells->len; i++) {
3051
        virCapsHostNUMACellPtr cell =
3052
            g_ptr_array_index(caps->cells, i);
3053 3054
        nodesetSize = MAX(nodesetSize, cell->num + 1);
    }
3055

3056
    if (nodeset &&
3057
        virBitmapParse(nodeset, &priv->autoNodeset, nodesetSize) < 0)
3058 3059
        goto cleanup;

3060 3061 3062 3063 3064 3065
    if (cpuset) {
        if (virBitmapParse(cpuset, &priv->autoCpuset, VIR_DOMAIN_CPUMASK_LEN) < 0)
            goto cleanup;
    } else {
        /* autoNodeset is present in this case, since otherwise we wouldn't
         * reach this code */
3066
        if (!(priv->autoCpuset = virCapabilitiesHostNUMAGetCpus(caps,
3067
                                                                priv->autoNodeset)))
3068 3069
            goto cleanup;
    }
3070 3071 3072 3073 3074 3075

    ret = 0;

 cleanup:
    virObjectUnref(caps);
    VIR_FREE(nodeset);
3076
    VIR_FREE(cpuset);
3077 3078 3079 3080 3081

    return ret;
}


3082 3083 3084 3085 3086 3087 3088
static virStorageSourcePtr
qemuDomainObjPrivateXMLParseBlockjobChain(xmlNodePtr node,
                                          xmlXPathContextPtr ctxt,
                                          virDomainXMLOptionPtr xmlopt)

{
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3089 3090 3091
    g_autofree char *format = NULL;
    g_autofree char *type = NULL;
    g_autofree char *index = NULL;
3092
    g_autoptr(virStorageSource) src = NULL;
3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115
    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
3116
    return g_steal_pointer(&src);
3117 3118 3119
}


3120 3121 3122 3123 3124 3125
static void
qemuDomainObjPrivateXMLParseBlockjobNodename(qemuBlockJobDataPtr job,
                                             const char *xpath,
                                             virStorageSourcePtr *src,
                                             xmlXPathContextPtr ctxt)
{
3126
    g_autofree char *nodename = NULL;
3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153

    *src = NULL;

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

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

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

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

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


static void
qemuDomainObjPrivateXMLParseBlockjobDataSpecific(qemuBlockJobDataPtr job,
3154 3155
                                                 xmlXPathContextPtr ctxt,
                                                 virDomainXMLOptionPtr xmlopt)
3156
{
3157 3158
    g_autofree char *createmode = NULL;
    g_autofree char *shallownew = NULL;
3159 3160
    xmlNodePtr tmp;

3161 3162 3163 3164 3165 3166 3167 3168 3169 3170
    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:
3171 3172 3173 3174 3175 3176 3177 3178
            qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                         "string(./topparent/@node)",
                                                         &job->data.commit.topparent,
                                                         ctxt);

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

3179
            G_GNUC_FALLTHROUGH;
3180
        case QEMU_BLOCKJOB_TYPE_ACTIVE_COMMIT:
3181 3182 3183 3184 3185 3186 3187 3188
            qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                         "string(./top/@node)",
                                                         &job->data.commit.top,
                                                         ctxt);
            qemuDomainObjPrivateXMLParseBlockjobNodename(job,
                                                         "string(./base/@node)",
                                                         &job->data.commit.base,
                                                         ctxt);
3189 3190
            if (virXPathNode("./deleteCommittedImages", ctxt))
                job->data.commit.deleteCommittedImages = true;
3191 3192 3193 3194 3195
            if (!job->data.commit.top ||
                !job->data.commit.base)
                goto broken;
            break;

3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208
        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;

3209
        case QEMU_BLOCKJOB_TYPE_COPY:
3210 3211 3212 3213 3214 3215 3216 3217
            if ((shallownew =  virXPathString("string(./@shallownew)", ctxt))) {
                if (STRNEQ(shallownew, "yes"))
                    goto broken;

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

P
Peter Krempa 已提交
3218
        case QEMU_BLOCKJOB_TYPE_BACKUP:
3219 3220 3221 3222 3223 3224 3225 3226 3227
            job->data.backup.bitmap =  virXPathString("string(./bitmap/@name)", ctxt);

            if (!(tmp = virXPathNode("./store", ctxt)) ||
                !(job->data.backup.store = qemuDomainObjPrivateXMLParseBlockjobChain(tmp, ctxt, xmlopt)))
                goto broken;

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

P
Peter Krempa 已提交
3228
            break;
3229 3230

        case QEMU_BLOCKJOB_TYPE_BROKEN:
3231 3232 3233 3234 3235 3236 3237
        case QEMU_BLOCKJOB_TYPE_NONE:
        case QEMU_BLOCKJOB_TYPE_INTERNAL:
        case QEMU_BLOCKJOB_TYPE_LAST:
            break;
    }

    return;
3238 3239 3240 3241

 broken:
    VIR_DEBUG("marking block job '%s' as invalid: malformed job data", job->name);
    job->invalidData = true;
3242 3243 3244
}


3245
static int
3246 3247
qemuDomainObjPrivateXMLParseBlockjobData(virDomainObjPtr vm,
                                         xmlNodePtr node,
3248 3249
                                         xmlXPathContextPtr ctxt,
                                         virDomainXMLOptionPtr xmlopt)
3250 3251 3252
{
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
    virDomainDiskDefPtr disk = NULL;
3253
    g_autoptr(qemuBlockJobData) job = NULL;
3254 3255
    g_autofree char *name = NULL;
    g_autofree char *typestr = NULL;
3256
    g_autofree char *brokentypestr = NULL;
3257
    int type;
3258
    g_autofree char *statestr = NULL;
3259
    int state = QEMU_BLOCKJOB_STATE_FAILED;
3260 3261 3262
    g_autofree char *diskdst = NULL;
    g_autofree char *newstatestr = NULL;
    g_autofree char *mirror = NULL;
3263 3264
    int newstate = -1;
    bool invalidData = false;
3265
    xmlNodePtr tmp;
3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277

    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) {
3278
        type = QEMU_BLOCKJOB_TYPE_BROKEN;
3279 3280 3281 3282 3283 3284
        invalidData = true;
    }

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

3285 3286 3287 3288
    if ((brokentypestr = virXPathString("string(./@brokentype)", ctxt)) &&
        (job->brokentype = qemuBlockjobTypeFromString(brokentypestr)) < 0)
        job->brokentype = QEMU_BLOCKJOB_TYPE_NONE;

3289 3290 3291 3292 3293 3294 3295 3296 3297
    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)) &&
3298
        !(disk = virDomainDiskByTarget(vm->def, diskdst)))
3299 3300
        invalidData = true;

3301 3302 3303 3304
    if ((mirror = virXPathString("string(./disk/@mirror)", ctxt)) &&
        STRNEQ(mirror, "yes"))
        invalidData = true;

3305 3306 3307 3308 3309 3310 3311 3312 3313 3314
    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;
    }

3315 3316 3317 3318 3319 3320 3321
    if (mirror) {
        if (disk)
            job->mirrorChain = virObjectRef(disk->mirror);
        else
            invalidData = true;
    }

3322 3323 3324 3325
    job->state = state;
    job->newstate = newstate;
    job->errmsg = virXPathString("string(./errmsg)", ctxt);
    job->invalidData = invalidData;
3326 3327
    job->disk = disk;

3328
    qemuDomainObjPrivateXMLParseBlockjobDataSpecific(job, ctxt, xmlopt);
3329

3330
    if (qemuBlockJobRegister(job, vm, disk, false) < 0)
3331 3332 3333 3334 3335 3336 3337 3338 3339
        return -1;

    return 0;
}


static int
qemuDomainObjPrivateXMLParseBlockjobs(virDomainObjPtr vm,
                                      qemuDomainObjPrivatePtr priv,
3340 3341
                                      xmlXPathContextPtr ctxt)
{
3342
    g_autofree xmlNodePtr *nodes = NULL;
3343
    ssize_t nnodes = 0;
3344
    g_autofree char *active = NULL;
3345
    int tmp;
3346
    size_t i;
3347 3348 3349 3350 3351

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

3352 3353 3354 3355 3356
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if ((nnodes = virXPathNodeSet("./blockjobs/blockjob", ctxt, &nodes)) < 0)
            return -1;

        for (i = 0; i < nnodes; i++) {
3357 3358
            if (qemuDomainObjPrivateXMLParseBlockjobData(vm, nodes[i], ctxt,
                                                         priv->driver->xmlopt) < 0)
3359 3360 3361 3362
                return -1;
        }
    }

3363 3364 3365 3366
    return 0;
}


3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394
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;
}


3395
int
3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419
qemuDomainObjPrivateXMLParseAllowReboot(xmlXPathContextPtr ctxt,
                                        virTristateBool *allowReboot)
{
    int ret = -1;
    int val;
    char *valStr;

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

    ret = 0;

 cleanup:
    VIR_FREE(valStr);
    return ret;
}


3420 3421 3422 3423 3424 3425 3426 3427
static void
qemuDomainObjPrivateXMLParsePR(xmlXPathContextPtr ctxt,
                               bool *prDaemonRunning)
{
    *prDaemonRunning = virXPathBoolean("boolean(./prDaemon)", ctxt) > 0;
}


3428 3429 3430
static int
qemuDomainObjPrivateXMLParseJobNBDSource(xmlNodePtr node,
                                         xmlXPathContextPtr ctxt,
3431 3432
                                         virDomainDiskDefPtr disk,
                                         virDomainXMLOptionPtr xmlopt)
3433
{
3434
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3435
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
3436 3437
    g_autofree char *format = NULL;
    g_autofree char *type = NULL;
3438
    g_autoptr(virStorageSource) migrSource = NULL;
3439
    xmlNodePtr sourceNode;
3440 3441 3442

    ctxt->node = node;

3443 3444
    if (!(ctxt->node = virXPathNode("./migrationSource", ctxt)))
        return 0;
3445 3446 3447 3448

    if (!(type = virXMLPropString(ctxt->node, "type"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing storage source type"));
3449
        return -1;
3450 3451 3452 3453
    }

    if (!(format = virXMLPropString(ctxt->node, "format"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
3454
                       _("missing storage source format"));
3455
        return -1;
3456 3457
    }

3458
    if (!(migrSource = virDomainStorageSourceParseBase(type, format, NULL)))
3459
        return -1;
3460

3461 3462 3463 3464 3465
    /* newer libvirt uses the <source> subelement instead of formatting the
     * source directly into <migrationSource> */
    if ((sourceNode = virXPathNode("./source", ctxt)))
        ctxt->node = sourceNode;

3466
    if (virDomainStorageSourceParse(ctxt->node, ctxt, migrSource,
3467
                                    VIR_DOMAIN_DEF_PARSE_STATUS, xmlopt) < 0)
3468
        return -1;
3469

3470
    diskPriv->migrSource = g_steal_pointer(&migrSource);
3471
    return 0;
3472 3473 3474
}


3475 3476 3477 3478 3479 3480
static int
qemuDomainObjPrivateXMLParseJobNBD(virDomainObjPtr vm,
                                   qemuDomainObjPrivatePtr priv,
                                   xmlXPathContextPtr ctxt)
{
    xmlNodePtr *nodes = NULL;
3481
    char *dst = NULL;
3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497
    size_t i;
    int n;
    int ret = -1;

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

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

3498
            if ((dst = virXMLPropString(nodes[i], "dev")) &&
3499
                (disk = virDomainDiskByTarget(vm->def, dst))) {
3500
                QEMU_DOMAIN_DISK_PRIVATE(disk)->migrating = true;
3501 3502

                if (qemuDomainObjPrivateXMLParseJobNBDSource(nodes[i], ctxt,
3503 3504
                                                             disk,
                                                             priv->driver->xmlopt) < 0)
3505 3506 3507
                    goto cleanup;
            }

3508 3509 3510 3511 3512 3513 3514 3515
            VIR_FREE(dst);
        }
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
3516
    VIR_FREE(dst);
3517 3518 3519 3520
    return ret;
}


3521 3522 3523 3524 3525
static int
qemuDomainObjPrivateXMLParseJob(virDomainObjPtr vm,
                                qemuDomainObjPrivatePtr priv,
                                xmlXPathContextPtr ctxt)
{
3526
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3527 3528 3529
    char *tmp = NULL;
    int ret = -1;

3530 3531 3532 3533 3534 3535
    if (!(ctxt->node = virXPathNode("./job[1]", ctxt))) {
        ret = 0;
        goto cleanup;
    }

    if ((tmp = virXPathString("string(@type)", ctxt))) {
3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546
        int type;

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

3547
    if ((tmp = virXPathString("string(@async)", ctxt))) {
3548 3549 3550 3551 3552 3553 3554 3555 3556 3557
        int async;

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

3558
        if ((tmp = virXPathString("string(@phase)", ctxt))) {
3559 3560 3561 3562 3563 3564 3565 3566 3567 3568
            priv->job.phase = qemuDomainAsyncJobPhaseFromString(async, tmp);
            if (priv->job.phase < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Unknown job phase %s"), tmp);
                goto cleanup;
            }
            VIR_FREE(tmp);
        }
    }

3569 3570 3571 3572 3573
    if (virXPathULongHex("string(@flags)", ctxt, &priv->job.apiFlags) == -2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Invalid job flags"));
        goto cleanup;
    }

3574
    if (qemuDomainObjPrivateXMLParseJobNBD(vm, priv, ctxt) < 0)
3575 3576
        goto cleanup;

3577 3578 3579
    if (qemuMigrationParamsParse(ctxt, &priv->job.migParams) < 0)
        goto cleanup;

3580 3581 3582 3583 3584 3585 3586 3587
    ret = 0;

 cleanup:
    VIR_FREE(tmp);
    return ret;
}


3588 3589 3590 3591 3592 3593
static int
qemuDomainObjPrivateXMLParseSlirpFeatures(xmlNodePtr featuresNode,
                                          xmlXPathContextPtr ctxt,
                                          qemuSlirpPtr slirp)
{
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
3594
    g_autofree xmlNodePtr *nodes = NULL;
3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606
    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++) {
3607
        g_autofree char *str = virXMLPropString(nodes[i], "name");
3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626
        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;
}


3627
static int
3628
qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt,
3629
                             virDomainObjPtr vm,
3630
                             virDomainDefParserConfigPtr config)
3631
{
3632
    qemuDomainObjPrivatePtr priv = vm->privateData;
3633
    virQEMUDriverPtr driver = config->priv;
3634
    char *monitorpath;
3635
    char *tmp = NULL;
3636 3637
    int n;
    size_t i;
3638
    xmlNodePtr *nodes = NULL;
3639
    xmlNodePtr node = NULL;
3640
    virQEMUCapsPtr qemuCaps = NULL;
3641

3642
    if (!(priv->monConfig = virDomainChrSourceDefNew(NULL)))
3643 3644 3645 3646
        goto error;

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
3647 3648
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("no monitor path"));
3649 3650 3651 3652 3653
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
3654
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
3655
    else
3656
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
3657 3658
    VIR_FREE(tmp);

3659
    switch (priv->monConfig->type) {
3660
    case VIR_DOMAIN_CHR_TYPE_PTY:
3661
        priv->monConfig->data.file.path = monitorpath;
3662 3663
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
3664
        priv->monConfig->data.nix.path = monitorpath;
3665 3666 3667
        break;
    default:
        VIR_FREE(monitorpath);
3668 3669 3670
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported monitor type '%s'"),
                       virDomainChrTypeToString(priv->monConfig->type));
3671 3672 3673
        goto error;
    }

3674 3675 3676 3677 3678 3679
    if (virXPathInt("string(./agentTimeout)", ctxt, &priv->agentTimeout) == -2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse agent timeout"));
        goto error;
    }

3680 3681 3682 3683
    if ((node = virXPathNode("./namespaces", ctxt))) {
        xmlNodePtr next;

        for (next = node->children; next; next = next->next) {
3684
            int ns = qemuDomainNamespaceTypeFromString((const char *)next->name);
3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703

            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;
    }

3704 3705
    priv->rememberOwner = virXPathBoolean("count(./rememberOwner) > 0", ctxt);

3706
    if ((n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes)) < 0)
3707 3708
        goto error;

3709
    for (i = 0; i < n; i++) {
3710
        if (qemuDomainObjPrivateXMLParseVcpu(nodes[i], i, vm->def) < 0)
3711
            goto error;
3712
    }
3713
    VIR_FREE(nodes);
3714

3715
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
3716 3717
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse qemu capabilities flags"));
3718 3719 3720
        goto error;
    }
    if (n > 0) {
3721
        if (!(qemuCaps = virQEMUCapsNew()))
3722 3723
            goto error;

3724
        for (i = 0; i < n; i++) {
3725 3726
            char *str = virXMLPropString(nodes[i], "name");
            if (str) {
3727
                int flag = virQEMUCapsTypeFromString(str);
3728
                if (flag < 0) {
3729 3730
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Unknown qemu capabilities flag %s"), str);
3731
                    VIR_FREE(str);
3732 3733
                    goto error;
                }
3734
                VIR_FREE(str);
3735
                virQEMUCapsSet(qemuCaps, flag);
3736 3737 3738
            }
        }

3739
        priv->qemuCaps = g_steal_pointer(&qemuCaps);
3740 3741 3742
    }
    VIR_FREE(nodes);

3743
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
3744

3745
    if (qemuDomainObjPrivateXMLParseJob(vm, priv, ctxt) < 0)
3746 3747
        goto error;

3748 3749
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770
    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);

3771 3772 3773 3774 3775 3776
    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++) {
3777 3778
        g_autofree char *alias = virXMLPropString(nodes[i], "alias");
        g_autofree char *pid = virXMLPropString(nodes[i], "pid");
J
Ján Tomko 已提交
3779
        g_autoptr(qemuSlirp) slirp = qemuSlirpNew();
3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795
        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;

3796
        QEMU_DOMAIN_NETWORK_PRIVATE(dev.data.net)->slirp = g_steal_pointer(&slirp);
3797 3798 3799
    }
    VIR_FREE(nodes);

3800
    if (qemuDomainObjPrivateXMLParseAutomaticPlacement(ctxt, priv, driver) < 0)
3801 3802
        goto error;

3803 3804 3805 3806 3807 3808
    if ((tmp = virXPathString("string(./libDir/@path)", ctxt)))
        priv->libDir = tmp;
    if ((tmp = virXPathString("string(./channelTargetDir/@path)", ctxt)))
        priv->channelTargetDir = tmp;
    tmp = NULL;

3809
    qemuDomainSetPrivatePathsOld(driver, vm);
3810

3811 3812 3813
    if (virCPUDefParseXML(ctxt, "./cpu", VIR_CPU_TYPE_GUEST, &priv->origCPU) < 0)
        goto error;

3814 3815 3816
    priv->chardevStdioLogd = virXPathBoolean("boolean(./chardevStdioLogd)",
                                             ctxt) == 1;

3817 3818
    qemuDomainObjPrivateXMLParseAllowReboot(ctxt, &priv->allowReboot);

3819 3820
    qemuDomainObjPrivateXMLParsePR(ctxt, &priv->prDaemonRunning);

3821
    if (qemuDomainObjPrivateXMLParseBlockjobs(vm, priv, ctxt) < 0)
3822 3823
        goto error;

3824 3825 3826
    if (qemuDomainObjPrivateXMLParseBackups(priv, ctxt) < 0)
        goto error;

3827 3828 3829 3830 3831 3832 3833 3834
    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;
    }

3835 3836
    priv->memPrealloc = virXPathBoolean("boolean(./memPrealloc)", ctxt) == 1;

3837 3838
    return 0;

3839
 error:
3840
    VIR_FREE(nodes);
3841
    VIR_FREE(tmp);
3842 3843
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
3844
    virObjectUnref(priv->monConfig);
3845
    priv->monConfig = NULL;
3846
    virStringListFree(priv->qemuDevices);
3847
    priv->qemuDevices = NULL;
3848
    virObjectUnref(qemuCaps);
3849 3850 3851 3852
    return -1;
}


3853 3854 3855 3856 3857 3858 3859 3860 3861
static void *
qemuDomainObjPrivateXMLGetParseOpaque(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return priv->qemuCaps;
}


3862 3863 3864
virDomainXMLPrivateDataCallbacks virQEMUDriverPrivateDataCallbacks = {
    .alloc = qemuDomainObjPrivateAlloc,
    .free = qemuDomainObjPrivateFree,
3865
    .diskNew = qemuDomainDiskPrivateNew,
3866 3867
    .diskParse = qemuDomainDiskPrivateParse,
    .diskFormat = qemuDomainDiskPrivateFormat,
3868
    .vcpuNew = qemuDomainVcpuPrivateNew,
3869
    .chrSourceNew = qemuDomainChrSourcePrivateNew,
J
Ján Tomko 已提交
3870
    .vsockNew = qemuDomainVsockPrivateNew,
3871
    .graphicsNew = qemuDomainGraphicsPrivateNew,
3872
    .networkNew = qemuDomainNetworkPrivateNew,
3873
    .videoNew = qemuDomainVideoPrivateNew,
3874 3875
    .parse = qemuDomainObjPrivateXMLParse,
    .format = qemuDomainObjPrivateXMLFormat,
3876
    .getParseOpaque = qemuDomainObjPrivateXMLGetParseOpaque,
3877 3878
    .storageParse = qemuStorageSourcePrivateDataParse,
    .storageFormat = qemuStorageSourcePrivateDataFormat,
3879 3880 3881
};


3882 3883 3884 3885 3886 3887
static void
qemuDomainXmlNsDefFree(qemuDomainXmlNsDefPtr def)
{
    if (!def)
        return;

3888 3889 3890
    virStringListFreeCount(def->args, def->num_args);
    virStringListFreeCount(def->env_name, def->num_env);
    virStringListFreeCount(def->env_value, def->num_env);
3891 3892
    virStringListFreeCount(def->capsadd, def->ncapsadd);
    virStringListFreeCount(def->capsdel, def->ncapsdel);
3893

3894 3895 3896 3897
    VIR_FREE(def);
}


3898 3899 3900
static void
qemuDomainDefNamespaceFree(void *nsdata)
{
3901
    qemuDomainXmlNsDefPtr cmd = nsdata;
3902

3903
    qemuDomainXmlNsDefFree(cmd);
3904 3905
}

3906 3907 3908 3909 3910

static int
qemuDomainDefNamespaceParseCommandlineArgs(qemuDomainXmlNsDefPtr nsdef,
                                           xmlXPathContextPtr ctxt)
{
3911
    g_autofree xmlNodePtr *nodes = NULL;
3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935
    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;
}


3936 3937 3938
static int
qemuDomainDefNamespaceParseCommandlineEnvNameValidate(const char *envname)
{
3939
    if (!g_ascii_isalpha(envname[0]) && envname[0] != '_') {
3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958
        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)
{
3959
    g_autofree xmlNodePtr *nodes = NULL;
3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991
    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;
}


3992 3993 3994 3995
static int
qemuDomainDefNamespaceParseCaps(qemuDomainXmlNsDefPtr nsdef,
                                xmlXPathContextPtr ctxt)
{
3996
    g_autofree xmlNodePtr *nodesadd = NULL;
3997
    ssize_t nnodesadd;
3998
    g_autofree xmlNodePtr *nodesdel = NULL;
3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035
    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;
}


4036
static int
4037
qemuDomainDefNamespaceParse(xmlXPathContextPtr ctxt,
4038 4039
                            void **data)
{
4040 4041
    qemuDomainXmlNsDefPtr nsdata = NULL;
    int ret = -1;
4042

4043
    if (VIR_ALLOC(nsdata) < 0)
4044 4045
        return -1;

4046
    if (qemuDomainDefNamespaceParseCommandlineArgs(nsdata, ctxt) < 0 ||
4047 4048
        qemuDomainDefNamespaceParseCommandlineEnv(nsdata, ctxt) < 0 ||
        qemuDomainDefNamespaceParseCaps(nsdata, ctxt) < 0)
4049
        goto cleanup;
4050

4051 4052
    if (nsdata->num_args > 0 || nsdata->num_env > 0 ||
        nsdata->ncapsadd > 0 || nsdata->ncapsdel > 0)
4053
        *data = g_steal_pointer(&nsdata);
4054

4055
    ret = 0;
4056

4057 4058 4059
 cleanup:
    qemuDomainDefNamespaceFree(nsdata);
    return ret;
4060 4061
}

4062 4063 4064 4065

static void
qemuDomainDefNamespaceFormatXMLCommandline(virBufferPtr buf,
                                           qemuDomainXmlNsDefPtr cmd)
4066
{
4067
    size_t i;
4068 4069

    if (!cmd->num_args && !cmd->num_env)
4070
        return;
4071

4072 4073 4074
    virBufferAddLit(buf, "<qemu:commandline>\n");
    virBufferAdjustIndent(buf, 2);

4075
    for (i = 0; i < cmd->num_args; i++)
4076
        virBufferEscapeString(buf, "<qemu:arg value='%s'/>\n",
4077 4078
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
4079
        virBufferAsprintf(buf, "<qemu:env name='%s'", cmd->env_name[i]);
4080 4081 4082 4083 4084
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }

4085 4086
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:commandline>\n");
4087 4088 4089
}


4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112
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");
}


4113 4114 4115 4116 4117 4118 4119
static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainXmlNsDefPtr cmd = nsdata;

    qemuDomainDefNamespaceFormatXMLCommandline(buf, cmd);
4120
    qemuDomainDefNamespaceFormatXMLCaps(buf, cmd);
4121

4122 4123 4124 4125
    return 0;
}


4126
virXMLNamespace virQEMUDriverDomainXMLNamespace = {
4127 4128 4129
    .parse = qemuDomainDefNamespaceParse,
    .free = qemuDomainDefNamespaceFree,
    .format = qemuDomainDefNamespaceFormatXML,
4130
    .prefix = "qemu",
4131
    .uri = "http://libvirt.org/schemas/domain/qemu/1.0",
4132
};
4133

4134

P
Pavel Hrdina 已提交
4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153
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;
}


4154
static int
4155 4156
qemuDomainDefAddDefaultDevices(virDomainDefPtr def,
                               virQEMUCapsPtr qemuCaps)
4157
{
4158
    bool addDefaultUSB = true;
4159
    int usbModel = -1; /* "default for machinetype" */
4160
    int pciRoot;       /* index within def->controllers */
4161
    bool addImplicitSATA = false;
4162
    bool addPCIRoot = false;
L
Laine Stump 已提交
4163
    bool addPCIeRoot = false;
4164
    bool addDefaultMemballoon = true;
4165 4166
    bool addDefaultUSBKBD = false;
    bool addDefaultUSBMouse = false;
4167
    bool addPanicDevice = false;
4168

P
Pavel Hrdina 已提交
4169 4170
    /* add implicit input devices */
    if (qemuDomainDefAddImplicitInputDevice(def) < 0)
4171
        return -1;
P
Pavel Hrdina 已提交
4172

4173 4174 4175 4176
    /* 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
4177
        if (STREQ(def->os.machine, "isapc")) {
4178
            addDefaultUSB = false;
4179
            break;
4180
        }
4181
        if (qemuDomainIsQ35(def)) {
4182 4183
            addPCIeRoot = true;
            addImplicitSATA = true;
4184

4185 4186 4187
            /* 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.
4188
             */
4189 4190 4191
            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))
4192 4193
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
            else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_USB_EHCI1))
4194 4195 4196
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_ICH9_EHCI1;
            else
                addDefaultUSB = false;
4197
            break;
L
Laine Stump 已提交
4198
        }
4199
        if (qemuDomainIsI440FX(def))
4200
            addPCIRoot = true;
4201 4202
        break;

4203 4204 4205 4206 4207 4208 4209
    case VIR_ARCH_ARMV6L:
        addDefaultUSB = false;
        addDefaultMemballoon = false;
        if (STREQ(def->os.machine, "versatilepb"))
            addPCIRoot = true;
        break;

4210
    case VIR_ARCH_ARMV7L:
4211
    case VIR_ARCH_AARCH64:
4212 4213
        addDefaultUSB = false;
        addDefaultMemballoon = false;
4214
        if (qemuDomainIsARMVirt(def))
4215 4216
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
        break;
4217

4218
    case VIR_ARCH_PPC64:
4219
    case VIR_ARCH_PPC64LE:
4220 4221 4222
        addPCIRoot = true;
        addDefaultUSBKBD = true;
        addDefaultUSBMouse = true;
4223 4224 4225
        /* For pSeries guests, the firmware provides the same
         * functionality as the pvpanic device, so automatically
         * add the definition if not already present */
4226
        if (qemuDomainIsPSeries(def))
4227
            addPanicDevice = true;
4228 4229
        break;

4230 4231 4232 4233 4234 4235 4236
    case VIR_ARCH_ALPHA:
    case VIR_ARCH_PPC:
    case VIR_ARCH_PPCEMB:
    case VIR_ARCH_SH4:
    case VIR_ARCH_SH4EB:
        addPCIRoot = true;
        break;
4237

4238 4239 4240
    case VIR_ARCH_RISCV32:
    case VIR_ARCH_RISCV64:
        addDefaultUSB = false;
4241 4242
        if (qemuDomainIsRISCVVirt(def))
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
4243 4244
        break;

4245 4246 4247
    case VIR_ARCH_S390:
    case VIR_ARCH_S390X:
        addDefaultUSB = false;
4248
        addPanicDevice = true;
4249
        addPCIRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_ZPCI);
4250
        break;
4251 4252 4253 4254 4255 4256

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

4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276
    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:
4277 4278 4279 4280
    default:
        break;
    }

4281
    if (addDefaultUSB &&
4282 4283
        virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_USB, 0) < 0 &&
        virDomainDefAddUSBController(def, 0, usbModel) < 0)
4284
        return -1;
4285

4286 4287 4288
    if (addImplicitSATA &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_SATA, 0, -1) < 0)
4289
        return -1;
4290

4291 4292
    pciRoot = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

4293 4294 4295
    /* NB: any machine that sets addPCIRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
     */
4296 4297 4298 4299 4300 4301 4302 4303
    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));
4304
                return -1;
4305 4306 4307
            }
        } else if (!virDomainDefAddController(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
                                              VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)) {
4308
            return -1;
4309 4310
        }
    }
4311

4312 4313 4314
    /* 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
4315 4316 4317
     *
     * NB: any machine that sets addPCIeRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
4318 4319
     */
    if (addPCIeRoot) {
4320 4321 4322 4323 4324 4325 4326
        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));
4327
                return -1;
4328 4329 4330
            }
        } else if (!virDomainDefAddController(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
                                             VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT)) {
4331
            return -1;
4332
        }
4333
    }
4334

4335
    if (addDefaultMemballoon && !def->memballoon) {
4336 4337
        virDomainMemballoonDefPtr memballoon;
        if (VIR_ALLOC(memballoon) < 0)
4338
            return -1;
4339 4340 4341 4342 4343

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

4344 4345 4346 4347
    if (STRPREFIX(def->os.machine, "s390-virtio") &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_S390) && def->memballoon)
        def->memballoon->model = VIR_DOMAIN_MEMBALLOON_MODEL_NONE;

4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367
    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;
    }

4368 4369 4370 4371 4372
    if (addDefaultUSBKBD &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_KBD,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
4373
        return -1;
4374 4375 4376 4377 4378 4379

    if (addDefaultUSBMouse &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
4380
        return -1;
4381

D
Dmitry Andreev 已提交
4382 4383 4384 4385
    if (addPanicDevice) {
        size_t j;
        for (j = 0; j < def->npanics; j++) {
            if (def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT ||
4386 4387 4388 4389
                (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
4390 4391
                break;
        }
4392

D
Dmitry Andreev 已提交
4393 4394 4395 4396 4397 4398
        if (j == def->npanics) {
            virDomainPanicDefPtr panic;
            if (VIR_ALLOC(panic) < 0 ||
                VIR_APPEND_ELEMENT_COPY(def->panics,
                                        def->npanics, panic) < 0) {
                VIR_FREE(panic);
4399
                return -1;
D
Dmitry Andreev 已提交
4400 4401
            }
        }
4402 4403
    }

4404
    return 0;
4405 4406 4407
}


A
Andrea Bolognani 已提交
4408 4409 4410
/**
 * qemuDomainDefEnableDefaultFeatures:
 * @def: domain definition
4411
 * @qemuCaps: QEMU capabilities
A
Andrea Bolognani 已提交
4412 4413 4414 4415 4416
 *
 * Make sure that features that should be enabled by default are actually
 * enabled and configure default values related to those features.
 */
static void
4417 4418
qemuDomainDefEnableDefaultFeatures(virDomainDefPtr def,
                                   virQEMUCapsPtr qemuCaps)
A
Andrea Bolognani 已提交
4419
{
4420
    /* The virt machine type always uses GIC: if the relevant information
4421 4422
     * was not included in the domain XML, we need to choose a suitable
     * GIC version ourselves */
4423
    if ((def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ABSENT &&
4424
         qemuDomainIsARMVirt(def)) ||
4425 4426 4427
        (def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ON &&
         def->gic_version == VIR_GIC_VERSION_NONE)) {
        virGICVersion version;
4428 4429 4430 4431 4432

        VIR_DEBUG("Looking for usable GIC version in domain capabilities");
        for (version = VIR_GIC_VERSION_LAST - 1;
             version > VIR_GIC_VERSION_NONE;
             version--) {
4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447

            /* 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;
            }

4448 4449 4450 4451 4452 4453 4454 4455
            if (virQEMUCapsSupportsGICVersion(qemuCaps,
                                              def->virtType,
                                              version)) {
                VIR_DEBUG("Using GIC version %s",
                          virGICVersionTypeToString(version));
                def->gic_version = version;
                break;
            }
4456 4457
        }

4458 4459 4460 4461 4462 4463 4464
        /* 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;
        }

4465 4466 4467
        /* 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;
4468
    }
A
Andrea Bolognani 已提交
4469 4470 4471
}


4472 4473 4474 4475 4476
static int
qemuCanonicalizeMachine(virDomainDefPtr def, virQEMUCapsPtr qemuCaps)
{
    const char *canon;

4477 4478
    if (!(canon = virQEMUCapsGetCanonicalMachine(qemuCaps, def->virtType,
                                                 def->os.machine)))
4479 4480 4481 4482
        return 0;

    if (STRNEQ(canon, def->os.machine)) {
        char *tmp;
4483
        tmp = g_strdup(canon);
4484 4485 4486 4487 4488 4489 4490 4491
        VIR_FREE(def->os.machine);
        def->os.machine = tmp;
    }

    return 0;
}


4492
static int
4493 4494 4495 4496 4497
qemuDomainRecheckInternalPaths(virDomainDefPtr def,
                               virQEMUDriverConfigPtr cfg,
                               unsigned int flags)
{
    size_t i = 0;
4498
    size_t j = 0;
4499 4500 4501 4502

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

4503 4504 4505 4506 4507 4508
        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
4509
             * the auto-generated socket based on config option from qemu.conf
4510 4511 4512 4513
             * 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 &&
4514
                !glisten->autoGenerated &&
4515 4516 4517 4518 4519 4520 4521
                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;
                }
4522
            }
4523 4524
        }
    }
4525 4526

    return 0;
4527 4528 4529
}


4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599
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;
}


4600 4601
static int
qemuDomainDefSetDefaultCPU(virDomainDefPtr def,
4602
                           virArch hostarch,
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
                           virQEMUCapsPtr qemuCaps)
{
    const char *model;

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

    /* 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)
4634
        def->cpu = virCPUDefNew();
4635 4636 4637 4638

    def->cpu->type = VIR_CPU_TYPE_GUEST;

    if (STREQ(model, "host")) {
4639
        if (ARCH_IS_S390(def->os.arch) &&
4640
            virQEMUCapsIsCPUModeSupported(qemuCaps, hostarch, def->virtType,
4641 4642 4643 4644 4645 4646 4647 4648
                                          VIR_CPU_MODE_HOST_MODEL)) {
            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));
4649
    } else {
4650 4651 4652 4653 4654 4655
        /* 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;
4656 4657 4658 4659
        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);
4660 4661 4662

        VIR_DEBUG("Setting default CPU model for domain '%s' to %s",
                  def->name, model);
4663 4664 4665 4666 4667 4668
    }

    return 0;
}


4669 4670 4671
static int
qemuDomainDefCPUPostParse(virDomainDefPtr def)
{
A
Andrea Bolognani 已提交
4672 4673 4674 4675
    virCPUFeatureDefPtr sveFeature = NULL;
    bool sveVectorLengthsProvided = false;
    size_t i;

4676 4677 4678
    if (!def->cpu)
        return 0;

4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732
    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
4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765
    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++;
        }
    }

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
    /* 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;
}


4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830
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;
}


4831 4832
static int
qemuDomainDefPostParseBasic(virDomainDefPtr def,
J
Ján Tomko 已提交
4833
                            void *opaque G_GNUC_UNUSED)
4834
{
4835 4836
    virQEMUDriverPtr driver = opaque;

4837
    /* check for emulator and create a default one if needed */
4838 4839
    if (!def->emulator) {
        if (!(def->emulator = virQEMUCapsGetDefaultEmulator(
4840 4841 4842 4843
                  driver->hostarch, def->os.arch))) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("No emulator found for arch '%s'"),
                           virArchToString(def->os.arch));
4844
            return 1;
4845
        }
4846
    }
4847 4848 4849 4850 4851

    return 0;
}


4852 4853
static int
qemuDomainDefPostParse(virDomainDefPtr def,
4854
                       unsigned int parseFlags,
4855
                       void *opaque,
4856
                       void *parseOpaque)
4857 4858
{
    virQEMUDriverPtr driver = opaque;
4859
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
4860
    virQEMUCapsPtr qemuCaps = parseOpaque;
4861

4862 4863 4864 4865 4866
    /* 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)
4867 4868
        return 1;

4869 4870 4871
    if (def->os.bootloader || def->os.bootloaderArgs) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bootloader is not supported by QEMU"));
4872
        return -1;
4873 4874
    }

4875
    if (!def->os.machine) {
4876 4877 4878
        const char *machine = virQEMUCapsGetPreferredMachine(qemuCaps,
                                                             def->virtType);
        def->os.machine = g_strdup(machine);
4879 4880
    }

4881
    qemuDomainNVRAMPathGenerate(cfg, def);
4882

4883
    if (qemuDomainDefAddDefaultDevices(def, qemuCaps) < 0)
4884
        return -1;
4885

4886
    if (qemuCanonicalizeMachine(def, qemuCaps) < 0)
4887
        return -1;
4888

4889
    if (qemuDomainDefSetDefaultCPU(def, driver->hostarch, qemuCaps) < 0)
4890 4891
        return -1;

4892
    qemuDomainDefEnableDefaultFeatures(def, qemuCaps);
A
Andrea Bolognani 已提交
4893

4894
    if (qemuDomainRecheckInternalPaths(def, cfg, parseFlags) < 0)
4895
        return -1;
4896

4897
    if (qemuSecurityVerify(driver->securityManager, def) < 0)
4898
        return -1;
4899

4900
    if (qemuDomainDefVcpusPostParse(def) < 0)
4901
        return -1;
4902

4903
    if (qemuDomainDefCPUPostParse(def) < 0)
4904
        return -1;
4905

4906
    if (qemuDomainDefTsegPostParse(def, qemuCaps) < 0)
4907
        return -1;
4908

4909
    return 0;
4910 4911
}

4912

4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942
/**
 * 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;
}


4943 4944 4945
#define QEMU_MAX_VCPUS_WITHOUT_EIM 255


4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 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 5022 5023 5024 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
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;
}

5054
static int
5055 5056
qemuDomainDefValidateFeatures(const virDomainDef *def,
                              virQEMUCapsPtr qemuCaps)
5057
{
5058
    size_t i;
5059

5060 5061 5062 5063 5064
    for (i = 0; i < VIR_DOMAIN_FEATURE_LAST; i++) {
        const char *featureName = virDomainFeatureTypeToString(i);

        switch ((virDomainFeature) i) {
        case VIR_DOMAIN_FEATURE_IOAPIC:
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
            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;
                }
5098 5099 5100 5101
            }
            break;

        case VIR_DOMAIN_FEATURE_HPT:
5102
        case VIR_DOMAIN_FEATURE_HTM:
5103
        case VIR_DOMAIN_FEATURE_NESTED_HV:
5104
        case VIR_DOMAIN_FEATURE_CCF_ASSIST:
5105
            if (qemuDomainDefValidatePSeriesFeature(def, qemuCaps, i) < 0)
5106 5107 5108
                return -1;
            break;

5109 5110
        case VIR_DOMAIN_FEATURE_GIC:
            if (def->features[i] == VIR_TRISTATE_SWITCH_ON &&
5111
                !qemuDomainIsARMVirt(def)) {
5112 5113 5114 5115 5116 5117 5118 5119 5120 5121
                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;

5122 5123
        case VIR_DOMAIN_FEATURE_SMM:
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
5124
                !virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT)) {
5125 5126 5127 5128 5129 5130
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("smm is not available with this QEMU binary"));
                return -1;
            }
            break;

5131 5132 5133 5134 5135 5136 5137 5138 5139 5140
        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;

5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151
        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;

5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162
        case VIR_DOMAIN_FEATURE_ACPI:
        case VIR_DOMAIN_FEATURE_APIC:
        case VIR_DOMAIN_FEATURE_PAE:
        case VIR_DOMAIN_FEATURE_HAP:
        case VIR_DOMAIN_FEATURE_VIRIDIAN:
        case VIR_DOMAIN_FEATURE_PRIVNET:
        case VIR_DOMAIN_FEATURE_HYPERV:
        case VIR_DOMAIN_FEATURE_PVSPINLOCK:
        case VIR_DOMAIN_FEATURE_CAPABILITIES:
        case VIR_DOMAIN_FEATURE_PMU:
        case VIR_DOMAIN_FEATURE_VMCOREINFO:
5163
        case VIR_DOMAIN_FEATURE_MSRS:
5164 5165 5166
        case VIR_DOMAIN_FEATURE_LAST:
            break;
        }
5167 5168 5169 5170 5171 5172
    }

    return 0;
}


5173
static int
M
Marc-André Lureau 已提交
5174 5175
qemuDomainDefValidateMemory(const virDomainDef *def,
                            virQEMUCapsPtr qemuCaps)
5176 5177
{
    const long system_page_size = virGetSystemPageSizeKB();
5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195
    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;
    }
5196

M
Marc-André Lureau 已提交
5197 5198 5199
    if (mem->source == VIR_DOMAIN_MEMORY_SOURCE_MEMFD &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
5200
                       _("hugepages is not supported with memfd memory source"));
M
Marc-André Lureau 已提交
5201 5202 5203
        return -1;
    }

5204 5205
    /* We can't guarantee any other mem.access
     * if no guest NUMA nodes are defined. */
5206
    if (mem->hugepages[0].size != system_page_size &&
5207
        virDomainNumaGetNodeCount(def->numa) == 0 &&
5208 5209
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_DEFAULT &&
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_PRIVATE) {
5210 5211 5212
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("memory access mode '%s' not supported "
                         "without guest numa node"),
5213
                       virDomainMemoryAccessTypeToString(mem->access));
5214 5215 5216
        return -1;
    }

5217 5218 5219 5220 5221 5222 5223
    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;
    }

5224 5225 5226 5227
    return 0;
}


5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290
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;
}


5291 5292 5293 5294
static int
qemuDomainValidateCpuCount(const virDomainDef *def,
                            virQEMUCapsPtr qemuCaps)
{
5295 5296
    unsigned int maxCpus = virQEMUCapsGetMachineMaxCpus(qemuCaps, def->virtType,
                                                        def->os.machine);
5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314

    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;
}


5315 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
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;
}


5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368
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;
}


5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465
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;
        }
    }

    return 0;
}


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
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;
}


5498 5499
static int
qemuDomainDefValidate(const virDomainDef *def,
5500
                      void *opaque)
5501
{
5502
    virQEMUDriverPtr driver = opaque;
5503
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
5504 5505
    virQEMUCapsPtr qemuCaps = NULL;
    int ret = -1;
5506
    size_t i;
5507

5508
    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
5509 5510 5511
                                            def->emulator)))
        goto cleanup;

5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532
    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));
        goto cleanup;
    }

    if (!virQEMUCapsIsArchSupported(qemuCaps, def->os.arch)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Emulator '%s' does not support arch '%s'"),
                       def->emulator, virArchToString(def->os.arch));
        goto cleanup;
    }

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

5533 5534 5535
    if (def->mem.min_guarantee) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Parameter 'min_guarantee' not supported by QEMU."));
5536
        goto cleanup;
5537 5538 5539
    }

    /* On x86, UEFI requires ACPI */
5540 5541 5542
    if ((def->os.firmware == VIR_DOMAIN_OS_DEF_FIRMWARE_EFI ||
         (def->os.loader &&
          def->os.loader->type == VIR_DOMAIN_LOADER_TYPE_PFLASH)) &&
5543 5544 5545 5546 5547 5548 5549 5550 5551 5552
        ARCH_IS_X86(def->os.arch) &&
        def->features[VIR_DOMAIN_FEATURE_ACPI] != VIR_TRISTATE_SWITCH_ON) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("UEFI requires ACPI on this architecture"));
        goto cleanup;
    }

    /* On aarch64, ACPI requires UEFI */
    if (def->features[VIR_DOMAIN_FEATURE_ACPI] == VIR_TRISTATE_SWITCH_ON &&
        def->os.arch == VIR_ARCH_AARCH64 &&
5553 5554 5555
        (def->os.firmware != VIR_DOMAIN_OS_DEF_FIRMWARE_EFI &&
         (!def->os.loader ||
          def->os.loader->type != VIR_DOMAIN_LOADER_TYPE_PFLASH))) {
5556 5557 5558
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("ACPI requires UEFI on this architecture"));
        goto cleanup;
5559 5560
    }

M
Michal Privoznik 已提交
5561 5562 5563 5564 5565
    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. */

5566
        if (!qemuDomainIsQ35(def)) {
M
Michal Privoznik 已提交
5567 5568
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot is supported with q35 machine types only"));
5569
            goto cleanup;
M
Michal Privoznik 已提交
5570 5571 5572 5573 5574 5575 5576 5577
        }

        /* 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"));
5578
            goto cleanup;
M
Michal Privoznik 已提交
5579 5580
        }

5581 5582 5583
        /* 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
5584 5585
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot requires SMM feature enabled"));
5586
            goto cleanup;
M
Michal Privoznik 已提交
5587 5588 5589
        }
    }

5590 5591 5592 5593 5594 5595 5596
    if (def->genidRequested &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMGENID)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("this QEMU does not support the 'genid' capability"));
        goto cleanup;
    }

5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610
    /* 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"));
            goto cleanup;
        }
        if (!def->nserials) {
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("need at least one serial port to use SGA"));
            goto cleanup;
        }
    }

5611 5612 5613
    if (qemuDomainDefValidateClockTimers(def, qemuCaps) < 0)
        goto cleanup;

5614 5615 5616
    if (qemuDomainDefValidatePM(def, qemuCaps) < 0)
        goto cleanup;

5617 5618 5619 5620 5621
    /* 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;
5622
        unsigned int granularity;
5623
        unsigned int numacpus;
5624 5625 5626 5627

        /* 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 */
5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640
        if (virDomainDefGetVcpusTopology(def, &topologycpus) == 0) {
            if (topologycpus != virDomainDefGetVcpusMax(def)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("CPU topology doesn't match maximum vcpu count"));
                goto cleanup;
            }

            numacpus = virDomainNumaGetCPUCountTotal(def->numa);
            if ((numacpus != 0) && (topologycpus != numacpus)) {
                VIR_WARN("CPU topology doesn't match numa CPU count; "
                         "partial NUMA mapping is obsoleted and will "
                         "be removed in future");
            }
5641
        }
5642 5643 5644 5645 5646 5647 5648 5649 5650 5651

        /* vCPU hotplug granularity must be respected */
        granularity = qemuDomainDefGetVcpuHotplugGranularity(def);
        if ((virDomainDefGetVcpus(def) % granularity) != 0) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("vCPUs count must be a multiple of the vCPU "
                             "hotplug granularity (%u)"),
                           granularity);
            goto cleanup;
        }
5652 5653
    }

5654
    if (qemuDomainValidateCpuCount(def, qemuCaps) < 0)
5655 5656
        goto cleanup;

5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674
    if (ARCH_IS_X86(def->os.arch) &&
        virDomainDefGetVcpusMax(def) > QEMU_MAX_VCPUS_WITHOUT_EIM) {
        if (!qemuDomainIsQ35(def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("more than %d vCPUs are only supported on "
                             "q35-based machine types"),
                           QEMU_MAX_VCPUS_WITHOUT_EIM);
            goto cleanup;
        }
        if (!def->iommu || def->iommu->eim != VIR_TRISTATE_SWITCH_ON) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("more than %d vCPUs require extended interrupt "
                             "mode enabled on the iommu device"),
                           QEMU_MAX_VCPUS_WITHOUT_EIM);
            goto cleanup;
        }
    }

B
Bing Niu 已提交
5675
    if (def->nresctrls &&
5676 5677 5678 5679 5680 5681
        def->virtType != VIR_DOMAIN_VIRT_KVM) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("cachetune is only supported for KVM domains"));
        goto cleanup;
    }

5682
    if (qemuDomainDefValidateFeatures(def, qemuCaps) < 0)
5683 5684
        goto cleanup;

M
Marc-André Lureau 已提交
5685
    if (qemuDomainDefValidateMemory(def, qemuCaps) < 0)
5686 5687
        goto cleanup;

5688 5689 5690
    if (qemuDomainDefValidateNuma(def, qemuCaps) < 0)
        goto cleanup;

5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702
    if (cfg->vncTLS && cfg->vncTLSx509secretUUID &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_TLS_CREDS_X509)) {
        for (i = 0; i < def->ngraphics; i++) {
            if (def->graphics[i]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("encrypted VNC TLS keys are not supported with "
                                 "this QEMU binary"));
                goto cleanup;
            }
        }
    }

5703 5704 5705 5706
    ret = 0;

 cleanup:
    virObjectUnref(qemuCaps);
5707
    virObjectUnref(cfg);
5708
    return ret;
5709 5710 5711
}


5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735
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;
}


5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750
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
5751 5752
qemuDomainChrSourceDefValidate(const virDomainChrSourceDef *def,
                               virQEMUCapsPtr qemuCaps)
5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764
{
    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;

5765 5766 5767 5768 5769 5770 5771 5772 5773
    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;

5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787
    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;
    }

5788 5789 5790 5791 5792 5793 5794 5795
    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;
        }
    }

5796 5797 5798 5799
    return 0;
}


5800 5801 5802
static int
qemuDomainChrSerialTargetTypeToAddressType(int targetType)
{
5803
    switch ((virDomainChrSerialTargetType)targetType) {
5804 5805 5806 5807 5808 5809
    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;
5810 5811
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
        return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO;
5812
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
5813
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
5814 5815 5816 5817 5818 5819 5820 5821 5822
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        break;
    }

    return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE;
}


5823 5824 5825 5826 5827 5828 5829 5830 5831 5832
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;
5833 5834
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
5835
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
5836
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
5837
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
5838 5839 5840
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
5841 5842 5843 5844 5845 5846 5847 5848 5849
    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;
}


5850
static int
5851
qemuDomainChrTargetDefValidate(const virDomainChrDef *chr)
5852
{
5853 5854
    int expected;

5855
    switch ((virDomainChrDeviceType)chr->deviceType) {
5856 5857 5858
    case VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL:

        /* Validate target type */
5859
        switch ((virDomainChrSerialTargetType)chr->targetType) {
5860
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA:
5861 5862
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI:
5863
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
5864

5865
            expected = qemuDomainChrSerialTargetTypeToAddressType(chr->targetType);
5866 5867

            if (chr->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
5868 5869 5870 5871 5872
                chr->info.type != expected) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Target type '%s' requires address type '%s'"),
                               virDomainChrSerialTargetTypeToString(chr->targetType),
                               virDomainDeviceAddressTypeToString(expected));
5873 5874 5875 5876
                return -1;
            }
            break;

5877
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
5878
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
5879 5880 5881 5882 5883 5884 5885 5886 5887
            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;

5888 5889 5890 5891
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            break;
        }
5892 5893 5894 5895 5896 5897

        /* 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:
5898
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
5899
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
5900 5901
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
5902
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915 5916 5917 5918

            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;
        }
5919 5920 5921 5922 5923 5924 5925 5926 5927 5928 5929 5930 5931 5932
        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;
}


5933
static int
5934
qemuDomainChrDefValidate(const virDomainChrDef *dev,
5935 5936
                         const virDomainDef *def,
                         virQEMUCapsPtr qemuCaps)
5937
{
5938
    if (qemuDomainChrSourceDefValidate(dev->source, qemuCaps) < 0)
5939 5940
        return -1;

5941
    if (qemuDomainChrTargetDefValidate(dev) < 0)
5942 5943
        return -1;

5944
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL &&
5945
        (ARCH_IS_S390(def->os.arch) || qemuDomainIsPSeries(def))) {
5946 5947 5948 5949 5950
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("parallel ports are not supported"));
            return -1;
    }

5951 5952 5953
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL) {
        bool isCompatible = true;

5954 5955 5956 5957 5958 5959 5960 5961 5962 5963 5964
        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;
            }
        }

5965 5966 5967 5968 5969 5970
        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;
        }

5971 5972 5973 5974 5975 5976 5977
        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;
        }

5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988
        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;
        }
    }

5989 5990 5991 5992 5993
    return 0;
}


static int
5994 5995
qemuDomainSmartcardDefValidate(const virDomainSmartcardDef *def,
                               virQEMUCapsPtr qemuCaps)
5996 5997
{
    if (def->type == VIR_DOMAIN_SMARTCARD_TYPE_PASSTHROUGH &&
5998
        qemuDomainChrSourceDefValidate(def->data.passthru, qemuCaps) < 0)
5999 6000 6001 6002 6003 6004 6005
        return -1;

    return 0;
}


static int
6006
qemuDomainRNGDefValidate(const virDomainRNGDef *def,
J
Ján Tomko 已提交
6007
                         virQEMUCapsPtr qemuCaps G_GNUC_UNUSED)
6008 6009
{
    if (def->backend == VIR_DOMAIN_RNG_BACKEND_EGD &&
6010
        qemuDomainChrSourceDefValidate(def->source.chardev, qemuCaps) < 0)
6011 6012 6013 6014 6015 6016 6017
        return -1;

    return 0;
}


static int
6018 6019
qemuDomainRedirdevDefValidate(const virDomainRedirdevDef *def,
                              virQEMUCapsPtr qemuCaps)
6020
{
6021
    if (qemuDomainChrSourceDefValidate(def->source, qemuCaps) < 0)
6022 6023 6024 6025 6026 6027
        return -1;

    return 0;
}


6028 6029 6030 6031 6032 6033 6034 6035 6036
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
6037
                           _("%s model of watchdog can go only on PCI bus"),
6038 6039 6040 6041 6042 6043 6044 6045 6046
                           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
6047
                           _("%s model of watchdog can go only on ISA bus"),
6048 6049 6050 6051 6052 6053 6054 6055
                           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
6056
                           _("%s model of watchdog is virtual and cannot go on any bus."),
6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075
                           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;
}


6076 6077 6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089
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'.)
     */
6090
    char macstr[VIR_MAC_STRING_BUFLEN];
6091 6092
    virDomainNetType actualType = virDomainNetGetActualType(net);

6093 6094
    virMacAddrFormat(&net->mac, macstr);

6095 6096 6097 6098 6099 6100
    /* hypervisor-agnostic validation */
    if (virDomainActualNetDefValidate(net) < 0)
        return -1;

    /* QEMU-specific validation */

6101 6102 6103 6104 6105 6106 6107 6108 6109
    /* 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,
6110 6111
                           _("interface %s - multiqueue is not supported for network interfaces of type %s"),
                           macstr, virDomainNetTypeToString(actualType));
6112 6113 6114 6115 6116 6117
            return -1;
        }

        if (net->driver.virtio.queues > 1 &&
            actualType == VIR_DOMAIN_NET_TYPE_VHOSTUSER &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_VHOSTUSER_MULTIQUEUE)) {
6118 6119 6120
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("interface %s - multiqueue is not supported for network interfaces of type vhost-user with this QEMU binary"),
                           macstr);
6121 6122 6123 6124 6125 6126 6127 6128 6129 6130
            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) {
6131
        const virNetDevVPortProfile *vport = virDomainNetGetActualVirtPortProfile(net);
6132

6133 6134 6135 6136
        if (!(actualType == VIR_DOMAIN_NET_TYPE_NETWORK ||
              actualType == VIR_DOMAIN_NET_TYPE_BRIDGE ||
              actualType == VIR_DOMAIN_NET_TYPE_ETHERNET)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
6137 6138
                           _("interface %s - filterref is not supported for network interfaces of type %s"),
                           macstr, virDomainNetTypeToString(actualType));
6139 6140 6141 6142 6143
            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,
6144 6145
                           _("interface %s - filterref is not supported for network interfaces with virtualport type %s"),
                           macstr, virNetDevVPortTypeToString(vport->virtPortType));
6146 6147 6148 6149 6150 6151 6152 6153 6154
            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,
6155 6156
                       _("interface %s - custom tap device path is not supported for network interfaces of type %s"),
                       macstr, virDomainNetTypeToString(actualType));
6157 6158 6159 6160 6161 6162 6163
        return -1;
    }

    return 0;
}


6164
static int
6165
qemuDomainDeviceDefValidateNetwork(const virDomainNetDef *net)
6166
{
6167 6168
    bool hasIPv4 = false;
    bool hasIPv6 = false;
6169
    size_t i;
6170

6171 6172 6173 6174 6175 6176 6177 6178 6179 6180
    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];
6181

6182
            if (VIR_SOCKET_ADDR_VALID(&net->guestIP.ips[i]->peer)) {
6183
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
6184 6185
                               _("Invalid attempt to set peer IP for guest"));
                return -1;
6186
            }
6187

6188 6189
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ip->address, AF_INET)) {
                if (hasIPv4) {
6190
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
6191 6192 6193
                                   _("Only one IPv4 address per "
                                     "interface is allowed"));
                    return -1;
6194
                }
6195
                hasIPv4 = true;
6196

6197 6198
                if (ip->prefix > 0 &&
                    (ip->prefix < 4 || ip->prefix > 27)) {
6199
                    virReportError(VIR_ERR_XML_ERROR, "%s",
6200
                                   _("invalid prefix, must be in range of 4-27"));
6201
                    return -1;
6202
                }
6203
            }
6204

6205 6206 6207 6208 6209 6210
            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;
6211
                }
6212
                hasIPv6 = true;
6213

6214 6215 6216 6217 6218
                if (ip->prefix > 120) {
                    virReportError(VIR_ERR_XML_ERROR, "%s",
                                   _("prefix too long"));
                    return -1;
                }
6219
            }
6220
        }
6221 6222 6223 6224 6225 6226 6227
    } 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;
    }
6228

6229
    if (virDomainNetIsVirtioModel(net)) {
6230 6231 6232 6233 6234 6235 6236 6237 6238
        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;
6239
        }
6240
    }
6241

6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260
    if (net->mtu &&
        !qemuDomainNetSupportsMTU(net->type)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("setting MTU on interface type %s is not supported yet"),
                       virDomainNetTypeToString(net->type));
        return -1;
    }

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

    return 0;
}


6261
static int
6262
qemuDomainMdevDefVFIOPCIValidate(const virDomainHostdevDef *hostdev,
6263 6264
                                 const virDomainDef *def,
                                 virQEMUCapsPtr qemuCaps)
6265
{
6266 6267
    const virDomainHostdevSubsysMediatedDev *dev;

6268 6269 6270 6271 6272 6273 6274
    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;
    }

6275 6276 6277 6278 6279 6280 6281 6282 6283
    /* 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;
    }

6284
    dev = &hostdev->source.subsys.u.mdev;
6285
    if (dev->display == VIR_TRISTATE_SWITCH_ABSENT)
6286 6287 6288 6289 6290 6291 6292 6293 6294
        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;
    }

6295
    if (dev->model != VIR_MDEV_MODEL_TYPE_VFIO_PCI) {
6296 6297 6298 6299 6300 6301 6302
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("<hostdev> attribute 'display' is only supported"
                         " with model='vfio-pci'"));

        return -1;
    }

6303
    if (dev->display == VIR_TRISTATE_SWITCH_ON) {
6304 6305 6306 6307 6308 6309 6310 6311 6312 6313 6314 6315
        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;
}


6316
static int
6317
qemuDomainMdevDefVFIOAPValidate(const virDomainHostdevDef *hostdev,
6318 6319
                                const virDomainDef *def,
                                virQEMUCapsPtr qemuCaps)
6320 6321 6322 6323
{
    size_t i;
    bool vfioap_found = false;

6324 6325 6326 6327 6328 6329 6330
    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;
    }

6331 6332 6333 6334 6335 6336 6337 6338 6339
    /* 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;
    }

6340 6341
    /* VFIO-AP is restricted to a single mediated device only */
    for (i = 0; i < def->nhostdevs; i++) {
6342
        virDomainHostdevDefPtr hdev = def->hostdevs[i];
6343

6344 6345
        if (virHostdevIsMdevDevice(hdev) &&
            hdev->source.subsys.u.mdev.model == VIR_MDEV_MODEL_TYPE_VFIO_AP) {
6346 6347 6348 6349 6350 6351 6352 6353 6354 6355 6356 6357 6358 6359
            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;
}


6360
static int
6361
qemuDomainMdevDefValidate(const virDomainHostdevDef *hostdev,
6362 6363 6364
                          const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
{
6365 6366 6367
    const virDomainHostdevSubsysMediatedDev *mdevsrc;

    mdevsrc = &hostdev->source.subsys.u.mdev;
6368 6369
    switch ((virMediatedDeviceModelType) mdevsrc->model) {
    case VIR_MDEV_MODEL_TYPE_VFIO_PCI:
6370
        return qemuDomainMdevDefVFIOPCIValidate(hostdev, def, qemuCaps);
6371
    case VIR_MDEV_MODEL_TYPE_VFIO_AP:
6372
        return qemuDomainMdevDefVFIOAPValidate(hostdev, def, qemuCaps);
6373
    case VIR_MDEV_MODEL_TYPE_VFIO_CCW:
6374 6375 6376 6377 6378 6379
        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;
        }
6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391
        break;
    case VIR_MDEV_MODEL_TYPE_LAST:
    default:
        virReportEnumRangeError(virMediatedDeviceModelType,
                                mdevsrc->model);
        return -1;
    }

    return 0;
}


6392 6393
static int
qemuDomainDeviceDefValidateHostdev(const virDomainHostdevDef *hostdev,
6394 6395
                                   const virDomainDef *def,
                                   virQEMUCapsPtr qemuCaps)
6396
{
6397 6398
    int backend;

6399 6400 6401 6402 6403 6404 6405 6406 6407
    /* 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;
    }

6408 6409 6410 6411
    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:
6412
            break;
6413 6414 6415 6416 6417 6418 6419 6420 6421 6422 6423 6424 6425 6426

        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;

6427
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI_HOST:
6428 6429
            if (hostdev->info->bootIndex) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
6430 6431
                               _("booting from assigned devices is not "
                                 "supported by vhost SCSI devices"));
6432 6433
                return -1;
            }
6434 6435
            break;
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
6436
            return qemuDomainMdevDefValidate(hostdev, def, qemuCaps);
6437 6438 6439 6440 6441 6442 6443 6444
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
        default:
            virReportEnumRangeError(virDomainHostdevSubsysType,
                                    hostdev->source.subsys.type);
            return -1;
        }
    }

6445 6446 6447 6448
    return 0;
}


6449
static int
6450 6451
qemuDomainDeviceDefValidateVideo(const virDomainVideoDef *video,
                                 virQEMUCapsPtr qemuCaps)
6452
{
6453 6454
    /* there's no properties to validate for NONE video devices */
    if (video->type == VIR_DOMAIN_VIDEO_TYPE_NONE)
6455
        return 0;
6456 6457 6458 6459 6460 6461 6462 6463 6464 6465 6466 6467 6468 6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479 6480 6481 6482 6483

    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);
6484 6485
            return -1;
        }
6486 6487 6488 6489 6490 6491 6492
        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;
            }
6493

6494 6495 6496 6497 6498 6499 6500 6501
            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;
            }
        }
    }

6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512
    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;
        }
    }

6513 6514 6515 6516 6517 6518
    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)"));
6519 6520 6521 6522
            return -1;
        }
    }

6523 6524 6525 6526 6527 6528 6529
    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;
        }
6530 6531 6532 6533 6534 6535 6536 6537 6538
    } 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;
        }
6539 6540
    }

6541 6542 6543 6544
    return 0;
}


6545
int
6546 6547
qemuDomainValidateStorageSource(virStorageSourcePtr src,
                                virQEMUCapsPtr qemuCaps)
6548
{
6549 6550
    int actualType = virStorageSourceGetActualType(src);

6551 6552 6553 6554 6555 6556
    if (src->format == VIR_STORAGE_FILE_COW) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                      _("'cow' storage format is not supported"));
        return -1;
    }

6557 6558 6559 6560 6561 6562 6563
    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;
    }

6564 6565 6566 6567 6568 6569 6570
    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;
    }

6571 6572 6573 6574 6575 6576 6577 6578 6579 6580
    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;
    }

6581 6582 6583 6584 6585
    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",
6586
                       _("LUKS encrypted QCOW2 images are not supported by this QEMU"));
6587 6588 6589
        return -1;
    }

6590
    if (src->format == VIR_STORAGE_FILE_FAT &&
6591
        actualType != VIR_STORAGE_TYPE_VOLUME &&
6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608 6609 6610 6611 6612 6613
        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;
        }
    }

6614 6615 6616 6617 6618
    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;
6619 6620
    }

6621 6622 6623 6624 6625 6626 6627 6628 6629
    /* 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;
    }

6630 6631 6632 6633
    return 0;
}


6634
int
6635 6636
qemuDomainDeviceDefValidateDisk(const virDomainDiskDef *disk,
                                virQEMUCapsPtr qemuCaps)
6637
{
6638
    const char *driverName = virDomainDiskGetDriver(disk);
6639
    virStorageSourcePtr n;
6640 6641
    int idx;
    int partition;
6642

6643 6644 6645 6646 6647 6648
    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;
6649 6650
    }

6651 6652 6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664 6665
    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;
        }
6666 6667
    }

6668 6669 6670 6671 6672 6673 6674 6675 6676 6677 6678 6679 6680 6681 6682 6683 6684 6685 6686 6687 6688 6689 6690 6691 6692 6693 6694
    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;
    }

6695 6696 6697 6698 6699 6700 6701
    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
6702 6703 6704 6705 6706 6707 6708 6709
    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;
    }

6710 6711 6712 6713 6714 6715 6716 6717 6718 6719 6720 6721 6722
    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;
    }

6723 6724 6725 6726 6727
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if (qemuDomainValidateStorageSource(n, qemuCaps) < 0)
            return -1;
    }

6728 6729 6730 6731
    return 0;
}


6732 6733 6734 6735
static int
qemuDomainDeviceDefValidateControllerAttributes(const virDomainControllerDef *controller)
{
    if (!(controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
6736 6737 6738
          (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))) {
6739 6740 6741 6742 6743 6744 6745 6746 6747 6748 6749 6750 6751 6752 6753 6754 6755 6756 6757 6758
        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;
        }
6759 6760 6761 6762 6763
        if (controller->iothread) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'iothread' is only supported for virtio-scsi controller"));
            return -1;
        }
6764 6765 6766 6767 6768 6769
    }

    return 0;
}


6770 6771 6772 6773 6774 6775
/**
 * @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
6776
 * Returns true if acceptable, false otherwise with error message set.
6777 6778 6779 6780 6781 6782 6783 6784 6785 6786 6787 6788 6789 6790 6791
 */
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:
6792 6793
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL:
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL:
6794 6795 6796 6797 6798 6799 6800 6801 6802 6803 6804 6805 6806 6807 6808 6809 6810 6811 6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824 6825 6826
        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;
6827 6828 6829 6830 6831 6832
    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;
6833 6834 6835 6836 6837 6838
    }

    return true;
}


6839 6840 6841 6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857 6858 6859 6860 6861 6862 6863 6864
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;
}


6865 6866 6867 6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881
/* 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;

6882 6883
    if (controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI &&
6884 6885
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
       virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
6886 6887
                       _("virtio-scsi IOThreads only available for virtio "
                         "pci and virtio ccw controllers"));
6888 6889 6890 6891 6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906 6907 6908
       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:
6909 6910
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL:
6911 6912 6913 6914 6915 6916 6917 6918 6919 6920 6921
            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:
6922
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT:
6923 6924 6925 6926 6927 6928 6929 6930
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LAST:
            break;
    }

    return 0;
}


6931 6932 6933 6934 6935 6936 6937 6938 6939 6940 6941 6942 6943 6944 6945 6946 6947 6948 6949 6950 6951 6952 6953 6954 6955 6956 6957 6958 6959 6960 6961 6962
/**
 * 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;
6963 6964
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCIE_PCI_BRIDGE:
        return QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE;
6965 6966 6967 6968 6969
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE:
        return 0;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_LAST:
    default:
        return -1;
6970 6971
    }

6972
    return -1;
6973 6974 6975
}


6976 6977 6978 6979 6980 6981 6982 6983 6984 6985 6986 6987 6988 6989 6990 6991 6992
#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));


6993 6994 6995 6996 6997 6998 6999 7000 7001
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);
7002
    int cap = virDomainControllerPCIModelNameToQEMUCaps(pciopts->modelName);
7003 7004 7005 7006 7007 7008 7009 7010 7011 7012

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

7013 7014 7015 7016 7017 7018 7019 7020 7021
    /* 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:
7022
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
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 7071 7072 7073 7074 7075 7076 7077 7078 7079 7080 7081 7082 7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093 7094 7095 7096 7097 7098 7099 7100 7101 7102 7103 7104 7105 7106 7107 7108 7109 7110 7111 7112 7113 7114 7115 7116 7117 7118 7119 7120 7121 7122 7123 7124 7125 7126
        /* 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;

7127 7128 7129 7130 7131 7132 7133
    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;

7134 7135 7136 7137 7138 7139 7140
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

7141 7142 7143 7144 7145 7146 7147 7148 7149
    /* 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:
7150
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7151 7152 7153 7154 7155 7156 7157 7158 7159 7160 7161 7162 7163 7164 7165 7166 7167 7168 7169 7170 7171 7172 7173 7174 7175 7176 7177 7178 7179 7180 7181 7182 7183 7184
        if (cont->idx == 0) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Index for '%s' controllers must be > 0"),
                           model);
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
        /* pSeries guests can have multiple PHBs, so it's expected that
         * the index will not be zero for some of them */
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT &&
            pciopts->modelName == VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_SPAPR_PCI_HOST_BRIDGE) {
            break;
        }

        /* For all other pci-root and pcie-root controllers, though,
         * the index must be zero*/
        if (cont->idx != 0) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Index for '%s' controllers must be 0"),
                           model);
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

7185 7186 7187 7188 7189 7190 7191 7192 7193 7194 7195 7196 7197 7198 7199 7200 7201 7202 7203 7204 7205 7206 7207 7208 7209 7210 7211
    /* 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:
7212
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7213 7214 7215 7216 7217 7218 7219 7220 7221 7222 7223 7224 7225
        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;
    }

7226 7227 7228 7229 7230 7231 7232 7233 7234 7235 7236 7237 7238 7239 7240 7241 7242 7243 7244 7245
    /* pcihole64 */
    switch ((virDomainControllerModelPCI) cont->model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT:
        /* The pcihole64 option only applies to x86 guests */
        if ((pciopts->pcihole64 ||
             pciopts->pcihole64size != 0) &&
            !ARCH_IS_X86(def->os.arch)) {
            virReportControllerInvalidOption(cont, model, modelName, "pcihole64");
            return -1;
        }
        break;

    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_UPSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_SWITCH_DOWNSTREAM_PORT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS:
7246
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7247 7248 7249 7250 7251 7252 7253 7254 7255 7256 7257 7258 7259 7260
        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;
    }

7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271 7272 7273 7274 7275 7276 7277
    /* 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:
7278
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7279 7280 7281 7282 7283 7284 7285 7286 7287 7288 7289 7290 7291
        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;
    }

7292 7293 7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320 7321 7322
    /* 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:
7323
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7324 7325 7326 7327 7328 7329 7330 7331 7332 7333 7334 7335 7336
        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;
    }

7337 7338 7339 7340 7341 7342 7343 7344 7345 7346 7347 7348 7349 7350 7351 7352 7353
    /* 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:
7354
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7355 7356 7357 7358 7359 7360 7361 7362 7363 7364 7365 7366 7367
        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;
    }

7368 7369 7370 7371 7372 7373 7374 7375 7376 7377 7378 7379 7380 7381 7382 7383 7384 7385 7386 7387 7388
    /* 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:
7389
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
7390 7391 7392 7393 7394 7395 7396 7397 7398 7399 7400 7401 7402 7403 7404 7405
        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);
    }

7406 7407 7408 7409 7410 7411 7412 7413 7414 7415 7416 7417 7418 7419 7420 7421 7422 7423 7424 7425 7426 7427 7428 7429 7430 7431
    /* 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;
    }

7432
    return 0;
7433 7434 7435
}


7436 7437 7438 7439 7440
#undef virReportControllerInvalidValue
#undef virReportControllerInvalidOption
#undef virReportControllerMissingOption


7441 7442 7443 7444 7445 7446 7447 7448 7449 7450 7451 7452 7453 7454 7455 7456 7457 7458
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;
}


7459
static int
7460 7461 7462
qemuDomainDeviceDefValidateController(const virDomainControllerDef *controller,
                                      const virDomainDef *def,
                                      virQEMUCapsPtr qemuCaps)
7463
{
7464 7465
    int ret = 0;

7466
    if (!qemuDomainCheckCCWS390AddressSupport(def, &controller->info, qemuCaps,
7467 7468 7469
                                              "controller"))
        return -1;

7470 7471 7472 7473
    if (controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
        !qemuDomainCheckSCSIControllerModel(qemuCaps, controller->model))
        return -1;

7474 7475 7476
    if (qemuDomainDeviceDefValidateControllerAttributes(controller) < 0)
        return -1;

7477
    switch ((virDomainControllerType)controller->type) {
7478
    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
7479 7480 7481
        ret = qemuDomainDeviceDefValidateControllerIDE(controller, def);
        break;

7482
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
7483 7484 7485
        ret = qemuDomainDeviceDefValidateControllerSCSI(controller, def);
        break;

7486
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
7487 7488
        ret = qemuDomainDeviceDefValidateControllerPCI(controller, def,
                                                       qemuCaps);
7489 7490
        break;

7491
    case VIR_DOMAIN_CONTROLLER_TYPE_SATA:
7492 7493 7494 7495 7496
        ret = qemuDomainDeviceDefValidateControllerSATA(controller, def,
                                                        qemuCaps);
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
7497 7498 7499
    case VIR_DOMAIN_CONTROLLER_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_CONTROLLER_TYPE_CCID:
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
7500
    case VIR_DOMAIN_CONTROLLER_TYPE_XENBUS:
7501 7502 7503 7504
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
    }

7505
    return ret;
7506 7507 7508
}


7509
static int
7510 7511
qemuDomainDeviceDefValidateVsock(const virDomainVsockDef *vsock,
                                 const virDomainDef *def,
7512 7513 7514 7515 7516 7517 7518 7519
                                 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;
    }
7520

7521
    if (!qemuDomainCheckCCWS390AddressSupport(def, &vsock->info, qemuCaps,
7522 7523 7524
                                              "vsock"))
        return -1;

7525 7526
    return 0;
}
7527

7528 7529 7530

static int
qemuDomainDeviceDefValidateTPM(virDomainTPMDef *tpm,
7531 7532
                               const virDomainDef *def,
                               virQEMUCapsPtr qemuCaps)
7533 7534 7535 7536 7537 7538 7539 7540 7541 7542 7543 7544 7545 7546 7547 7548 7549 7550 7551 7552 7553
{
    /* TPM 1.2 and 2 are not compatible, so we choose a specific version here */
    if (tpm->version == VIR_DOMAIN_TPM_VERSION_DEFAULT)
        tpm->version = VIR_DOMAIN_TPM_VERSION_1_2;

    switch (tpm->version) {
    case VIR_DOMAIN_TPM_VERSION_1_2:
        /* only TIS available for emulator */
        if (tpm->type == VIR_DOMAIN_TPM_TYPE_EMULATOR &&
            tpm->model != VIR_DOMAIN_TPM_MODEL_TIS) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported interface %s for TPM 1.2"),
                           virDomainTPMModelTypeToString(tpm->model));
            return -1;
        }
        break;
    case VIR_DOMAIN_TPM_VERSION_2_0:
    case VIR_DOMAIN_TPM_VERSION_DEFAULT:
    case VIR_DOMAIN_TPM_VERSION_LAST:
        break;
    }
7554 7555 7556 7557 7558 7559 7560 7561 7562 7563 7564 7565 7566 7567 7568 7569

    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;
    }

7570
    return 0;
7571 7572 7573 7574 7575 7576 7577 7578

 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;
7579 7580 7581
}


7582 7583 7584 7585 7586 7587 7588 7589 7590
static int
qemuDomainDeviceDefValidateGraphics(const virDomainGraphicsDef *graphics,
                                    const virDomainDef *def,
                                    virQEMUCapsPtr qemuCaps)
{
    bool have_egl_headless = false;
    size_t i;

    for (i = 0; i < def->ngraphics; i++) {
7591
        if (def->graphics[i]->type == VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS) {
7592 7593 7594 7595 7596 7597 7598 7599 7600 7601 7602 7603 7604 7605 7606 7607 7608 7609 7610 7611 7612 7613 7614 7615 7616 7617 7618 7619 7620 7621 7622 7623 7624 7625 7626 7627 7628 7629 7630 7631 7632 7633
            have_egl_headless = true;
            break;
        }
    }

    /* Only VNC and SPICE can be paired with egl-headless, the other types
     * either don't make sense to pair with egl-headless or aren't even
     * supported by QEMU.
     */
    if (have_egl_headless) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_EGL_HEADLESS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("egl-headless display is not supported with this "
                             "QEMU binary"));
            return -1;
        }

        if (graphics->type != VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS &&
            graphics->type != VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
            graphics->type != VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("graphics type 'egl-headless' is only supported "
                             "with one of: 'vnc', 'spice' graphics types"));
            return -1;
        }

        /* '-spice gl=on' and '-display egl-headless' are mutually
         * exclusive
         */
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE &&
            graphics->data.spice.gl == VIR_TRISTATE_BOOL_YES) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("multiple OpenGL displays are not supported "
                             "by QEMU"));
            return -1;
        }
    }

    return 0;
}


7634 7635
static int
qemuDomainDeviceDefValidateInput(const virDomainInputDef *input,
J
Ján Tomko 已提交
7636
                                 const virDomainDef *def G_GNUC_UNUSED,
7637 7638
                                 virQEMUCapsPtr qemuCaps)
{
7639 7640 7641 7642
    const char *baseName;
    int cap;
    int ccwCap;

7643 7644 7645
    if (input->bus != VIR_DOMAIN_INPUT_BUS_VIRTIO)
        return 0;

7646 7647 7648 7649 7650 7651 7652 7653 7654 7655 7656 7657 7658 7659 7660 7661 7662 7663 7664 7665 7666 7667 7668 7669 7670 7671 7672 7673 7674 7675 7676 7677
    /* 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;
    }

7678 7679
    switch ((virDomainInputType)input->type) {
    case VIR_DOMAIN_INPUT_TYPE_MOUSE:
7680 7681 7682
        baseName = "virtio-mouse";
        cap = QEMU_CAPS_VIRTIO_MOUSE;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_MOUSE_CCW;
7683 7684
        break;
    case VIR_DOMAIN_INPUT_TYPE_TABLET:
7685 7686 7687
        baseName = "virtio-tablet";
        cap = QEMU_CAPS_VIRTIO_TABLET;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_TABLET_CCW;
7688 7689
        break;
    case VIR_DOMAIN_INPUT_TYPE_KBD:
7690 7691 7692
        baseName = "virtio-keyboard";
        cap = QEMU_CAPS_VIRTIO_KEYBOARD;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_KEYBOARD_CCW;
7693 7694
        break;
    case VIR_DOMAIN_INPUT_TYPE_PASSTHROUGH:
7695 7696 7697
        baseName = "virtio-input-host";
        cap = QEMU_CAPS_VIRTIO_INPUT_HOST;
        ccwCap = QEMU_CAPS_LAST;
7698 7699 7700 7701 7702 7703 7704 7705
        break;
    case VIR_DOMAIN_INPUT_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainInputType,
                                input->type);
        return -1;
    }

7706 7707 7708 7709 7710 7711 7712 7713 7714
    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;
    }

7715 7716 7717 7718
    return 0;
}


7719 7720 7721 7722 7723 7724 7725 7726 7727
static int
qemuDomainDeviceDefValidateMemballoon(const virDomainMemballoonDef *memballoon,
                                      virQEMUCapsPtr qemuCaps)
{
    if (!memballoon ||
        memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_NONE) {
        return 0;
    }

7728 7729 7730
    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) {
7731 7732 7733 7734 7735 7736 7737 7738 7739 7740 7741 7742 7743 7744 7745 7746 7747
        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;
}


7748 7749 7750 7751 7752 7753 7754 7755 7756 7757 7758 7759 7760 7761 7762 7763 7764 7765 7766 7767 7768 7769 7770 7771
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;

7772
    case VIR_DOMAIN_IOMMU_MODEL_SMMUV3:
7773 7774 7775 7776 7777 7778 7779 7780 7781 7782 7783 7784 7785 7786
        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;
        }
7787 7788
        break;

7789 7790 7791
    case VIR_DOMAIN_IOMMU_MODEL_LAST:
    default:
        virReportEnumRangeError(virDomainIOMMUModel, iommu->model);
7792
        return -1;
7793 7794
    }

7795 7796 7797 7798 7799 7800 7801 7802 7803 7804 7805 7806 7807 7808 7809 7810 7811 7812 7813 7814 7815 7816 7817 7818 7819 7820 7821 7822 7823 7824 7825 7826 7827 7828
    /* 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;
    }

7829 7830 7831 7832
    return 0;
}


7833 7834
static int
qemuDomainDeviceDefValidateFS(virDomainFSDefPtr fs,
J
Ján Tomko 已提交
7835 7836
                              const virDomainDef *def G_GNUC_UNUSED,
                              virQEMUCapsPtr qemuCaps G_GNUC_UNUSED)
7837 7838 7839 7840 7841 7842 7843 7844 7845 7846 7847 7848 7849 7850 7851 7852 7853 7854 7855 7856 7857 7858 7859 7860 7861 7862 7863 7864 7865 7866 7867 7868 7869 7870 7871 7872 7873
{
    if (fs->type != VIR_DOMAIN_FS_TYPE_MOUNT) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("only supports mount filesystem type"));
        return -1;
    }

    switch ((virDomainFSDriverType) fs->fsdriver) {
    case VIR_DOMAIN_FS_DRIVER_TYPE_DEFAULT:
    case VIR_DOMAIN_FS_DRIVER_TYPE_PATH:
        break;

    case VIR_DOMAIN_FS_DRIVER_TYPE_HANDLE:
        if (fs->accessmode != VIR_DOMAIN_FS_ACCESSMODE_PASSTHROUGH) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("only supports passthrough accessmode"));
            return -1;
        }
        break;

    case VIR_DOMAIN_FS_DRIVER_TYPE_LOOP:
    case VIR_DOMAIN_FS_DRIVER_TYPE_NBD:
    case VIR_DOMAIN_FS_DRIVER_TYPE_PLOOP:
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Filesystem driver type not supported"));
        return -1;

    case VIR_DOMAIN_FS_DRIVER_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainFSDriverType, fs->fsdriver);
        return -1;
    }

    return 0;
}


7874 7875 7876 7877 7878 7879 7880 7881 7882 7883 7884 7885 7886 7887 7888 7889 7890 7891 7892 7893 7894 7895 7896 7897 7898
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;

7899 7900
    switch ((virDomainDeviceAddressType) info->type) {
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI:
7901 7902
        return qemuDomainDeviceDefValidateZPCIAddress(info, qemuCaps);

7903 7904 7905 7906 7907 7908
    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;

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

7923 7924 7925 7926 7927 7928 7929 7930 7931 7932 7933 7934 7935 7936 7937 7938 7939 7940
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCID:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_USB:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_MMIO:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_ISA:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DIMM:
        /* No validation for these address types yet */
        break;

    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainDeviceAddressType, info->type);
        return -1;
    }

7941 7942 7943 7944
    return 0;
}


7945 7946 7947
static int
qemuDomainDeviceDefValidate(const virDomainDeviceDef *dev,
                            const virDomainDef *def,
7948
                            void *opaque)
7949
{
7950
    int ret = 0;
7951
    virQEMUDriverPtr driver = opaque;
7952 7953
    g_autoptr(virQEMUCaps) qemuCaps = NULL;
    g_autoptr(virDomainCaps) domCaps = NULL;
7954 7955 7956 7957

    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
                                            def->emulator)))
        return -1;
7958

7959 7960 7961 7962
    if (!(domCaps = virQEMUDriverGetDomainCapabilities(driver, qemuCaps,
                                                       def->os.machine,
                                                       def->os.arch,
                                                       def->virtType)))
7963
        return -1;
7964

7965
    if ((ret = qemuDomainDeviceDefValidateAddress(dev, qemuCaps)) < 0)
7966
        return ret;
7967

7968
    if ((ret = virDomainCapsDeviceDefValidate(domCaps, dev, def)) < 0)
7969
        return ret;
7970

7971
    switch ((virDomainDeviceType)dev->type) {
7972 7973 7974 7975 7976
    case VIR_DOMAIN_DEVICE_NET:
        ret = qemuDomainDeviceDefValidateNetwork(dev->data.net);
        break;

    case VIR_DOMAIN_DEVICE_CHR:
7977
        ret = qemuDomainChrDefValidate(dev->data.chr, def, qemuCaps);
7978 7979 7980
        break;

    case VIR_DOMAIN_DEVICE_SMARTCARD:
7981
        ret = qemuDomainSmartcardDefValidate(dev->data.smartcard, qemuCaps);
7982 7983 7984
        break;

    case VIR_DOMAIN_DEVICE_RNG:
7985
        ret = qemuDomainRNGDefValidate(dev->data.rng, qemuCaps);
7986 7987 7988
        break;

    case VIR_DOMAIN_DEVICE_REDIRDEV:
7989
        ret = qemuDomainRedirdevDefValidate(dev->data.redirdev, qemuCaps);
7990 7991 7992 7993 7994 7995 7996
        break;

    case VIR_DOMAIN_DEVICE_WATCHDOG:
        ret = qemuDomainWatchdogDefValidate(dev->data.watchdog, def);
        break;

    case VIR_DOMAIN_DEVICE_HOSTDEV:
7997 7998
        ret = qemuDomainDeviceDefValidateHostdev(dev->data.hostdev, def,
                                                 qemuCaps);
7999 8000 8001
        break;

    case VIR_DOMAIN_DEVICE_VIDEO:
8002
        ret = qemuDomainDeviceDefValidateVideo(dev->data.video, qemuCaps);
8003 8004 8005
        break;

    case VIR_DOMAIN_DEVICE_DISK:
8006
        ret = qemuDomainDeviceDefValidateDisk(dev->data.disk, qemuCaps);
8007 8008
        break;

8009
    case VIR_DOMAIN_DEVICE_CONTROLLER:
8010 8011
        ret = qemuDomainDeviceDefValidateController(dev->data.controller, def,
                                                    qemuCaps);
8012 8013
        break;

8014
    case VIR_DOMAIN_DEVICE_VSOCK:
8015
        ret = qemuDomainDeviceDefValidateVsock(dev->data.vsock, def, qemuCaps);
8016 8017
        break;

8018
    case VIR_DOMAIN_DEVICE_TPM:
8019
        ret = qemuDomainDeviceDefValidateTPM(dev->data.tpm, def, qemuCaps);
8020 8021
        break;

8022 8023 8024 8025 8026
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainDeviceDefValidateGraphics(dev->data.graphics, def,
                                                  qemuCaps);
        break;

8027 8028 8029 8030
    case VIR_DOMAIN_DEVICE_INPUT:
        ret = qemuDomainDeviceDefValidateInput(dev->data.input, def, qemuCaps);
        break;

8031 8032 8033 8034
    case VIR_DOMAIN_DEVICE_MEMBALLOON:
        ret = qemuDomainDeviceDefValidateMemballoon(dev->data.memballoon, qemuCaps);
        break;

8035 8036 8037 8038
    case VIR_DOMAIN_DEVICE_IOMMU:
        ret = qemuDomainDeviceDefValidateIOMMU(dev->data.iommu, def, qemuCaps);
        break;

8039
    case VIR_DOMAIN_DEVICE_FS:
8040 8041 8042
        ret = qemuDomainDeviceDefValidateFS(dev->data.fs, def, qemuCaps);
        break;

8043 8044 8045 8046
    case VIR_DOMAIN_DEVICE_NVRAM:
        ret = qemuDomainDeviceDefValidateNVRAM(dev->data.nvram, def, qemuCaps);
        break;

8047 8048 8049 8050
    case VIR_DOMAIN_DEVICE_HUB:
        ret = qemuDomainDeviceDefValidateHub(dev->data.hub, qemuCaps);
        break;

8051
    case VIR_DOMAIN_DEVICE_LEASE:
8052 8053
    case VIR_DOMAIN_DEVICE_SOUND:
    case VIR_DOMAIN_DEVICE_SHMEM:
8054
    case VIR_DOMAIN_DEVICE_MEMORY:
8055 8056 8057 8058
    case VIR_DOMAIN_DEVICE_PANIC:
    case VIR_DOMAIN_DEVICE_NONE:
    case VIR_DOMAIN_DEVICE_LAST:
        break;
8059 8060
    }

8061 8062 8063 8064
    return ret;
}


8065 8066 8067 8068 8069 8070 8071 8072 8073
/**
 * 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.
 */
8074
static int
8075 8076
qemuDomainDefaultNetModel(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
8077
{
8078
    if (ARCH_IS_S390(def->os.arch))
8079
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
8080

S
Stefan Schallenberg 已提交
8081 8082
    if (def->os.arch == VIR_ARCH_ARMV6L ||
        def->os.arch == VIR_ARCH_ARMV7L ||
8083
        def->os.arch == VIR_ARCH_AARCH64) {
8084
        if (STREQ(def->os.machine, "versatilepb"))
8085
            return VIR_DOMAIN_NET_MODEL_SMC91C111;
8086

8087
        if (qemuDomainIsARMVirt(def))
8088
            return VIR_DOMAIN_NET_MODEL_VIRTIO;
8089

8090 8091
        /* Incomplete. vexpress (and a few others) use this, but not all
         * arm boards */
8092
        return VIR_DOMAIN_NET_MODEL_LAN9118;
8093 8094
    }

8095 8096
    /* virtio is a sensible default for RISC-V virt guests */
    if (qemuDomainIsRISCVVirt(def))
8097
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
8098

8099 8100 8101
    /* In all other cases the model depends on the capabilities. If they were
     * not provided don't report any default. */
    if (!qemuCaps)
8102
        return VIR_DOMAIN_NET_MODEL_UNKNOWN;
8103

8104 8105 8106 8107
    /* 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))
8108
        return VIR_DOMAIN_NET_MODEL_RTL8139;
8109
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_E1000))
8110
        return VIR_DOMAIN_NET_MODEL_E1000;
8111
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_NET))
8112
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
8113 8114 8115

    /* We've had no luck detecting support for any network device,
     * but we have to return something: might as well be rtl8139 */
8116
    return VIR_DOMAIN_NET_MODEL_RTL8139;
8117
}
8118

8119 8120

/*
8121 8122 8123 8124 8125 8126 8127 8128 8129 8130 8131 8132 8133 8134 8135 8136
 * 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.
8137
 */
8138
static int
8139 8140 8141
qemuDomainChrDefDropDefaultPath(virDomainChrDefPtr chr,
                                virQEMUDriverPtr driver)
{
8142 8143 8144 8145
    virQEMUDriverConfigPtr cfg;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *regexp = NULL;
    int ret = -1;
8146

8147 8148 8149 8150 8151
    if (chr->deviceType != VIR_DOMAIN_CHR_DEVICE_TYPE_CHANNEL ||
        chr->targetType != VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO ||
        chr->source->type != VIR_DOMAIN_CHR_TYPE_UNIX ||
        !chr->source->data.nix.path) {
        return 0;
8152 8153
    }

8154 8155 8156 8157 8158 8159 8160 8161 8162 8163 8164 8165 8166
    cfg = virQEMUDriverGetConfig(driver);

    virBufferEscapeRegex(&buf, "^%s", cfg->channelTargetDir);
    virBufferAddLit(&buf, "/([^/]+\\.)|(domain-[^/]+/)");
    virBufferEscapeRegex(&buf, "%s$", chr->target.name);

    regexp = virBufferContentAndReset(&buf);

    if (virStringMatch(chr->source->data.nix.path, regexp))
        VIR_FREE(chr->source->data.nix.path);

    ret = 0;
    VIR_FREE(regexp);
8167
    virObjectUnref(cfg);
8168
    return ret;
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 8198 8199 8200 8201 8202 8203 8204 8205 8206 8207 8208 8209 8210 8211 8212 8213 8214 8215 8216 8217 8218 8219 8220 8221
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;
}


8222
#define QEMU_USB_XHCI_MAXPORTS 15
8223 8224


8225 8226 8227
static int
qemuDomainControllerDefPostParse(virDomainControllerDefPtr cont,
                                 const virDomainDef *def,
8228 8229
                                 virQEMUCapsPtr qemuCaps,
                                 unsigned int parseFlags)
8230
{
8231 8232
    switch ((virDomainControllerType)cont->type) {
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
8233 8234 8235
        /* Set the default SCSI controller model if not already set */
        if (qemuDomainSetSCSIControllerModel(def, cont, qemuCaps) < 0)
            return -1;
8236
        break;
8237

8238
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
8239
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT && qemuCaps) {
8240
            /* Pick a suitable default model for the USB controller if none
8241 8242
             * has been selected by the user and we have the qemuCaps for
             * figuring out which contollers are supported.
8243 8244 8245 8246 8247 8248 8249
             *
             * 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() */
8250 8251 8252 8253 8254 8255 8256 8257 8258 8259 8260

            /* 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;
                }
8261
            } else if (ARCH_IS_PPC64(def->os.arch)) {
8262 8263
                /* To not break migration we need to set default USB controller
                 * for ppc64 to pci-ohci if we cannot change ABI of the VM.
8264 8265
                 * The nec-usb-xhci or qemu-xhci controller is used as default
                 * only for newly defined domains or devices. */
8266
                if ((parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE) &&
8267 8268 8269
                    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) &&
8270 8271 8272
                    virQEMUCapsGet(qemuCaps, QEMU_CAPS_NEC_USB_XHCI)) {
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
                } else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_OHCI)) {
8273
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_PCI_OHCI;
8274 8275 8276
                } else {
                    /* Explicitly fallback to legacy USB controller for PPC64. */
                    cont->model = -1;
8277
                }
8278
            } else if (def->os.arch == VIR_ARCH_AARCH64) {
8279 8280 8281
                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))
8282
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
8283 8284 8285 8286 8287 8288 8289 8290 8291 8292 8293
            }
        }
        /* 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;
        }
8294 8295 8296
        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) {
8297
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8298 8299 8300
                           _("'%s' controller only supports up to '%u' ports"),
                           virDomainControllerModelUSBTypeToString(cont->model),
                           QEMU_USB_XHCI_MAXPORTS);
8301 8302 8303
            return -1;
        }
        break;
8304

8305
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
8306 8307 8308 8309 8310 8311 8312

        /* 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) {
8313 8314 8315 8316 8317 8318
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("pci-root and pcie-root controllers "
                             "should have index 0"));
            return -1;
        }

8319
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS &&
8320
            !qemuDomainIsI440FX(def)) {
8321 8322 8323 8324 8325 8326
            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 &&
8327
            !qemuDomainIsQ35(def)) {
8328 8329 8330 8331 8332 8333
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("pcie-expander-bus controllers are only supported "
                             "on q35-based machinetypes"));
            return -1;
        }

8334 8335 8336
        /* 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
8337 8338
         * from 0 .. size-1.
         */
8339 8340
        if (cont->opts.pciopts.numaNode >= 0 &&
            cont->opts.pciopts.numaNode >=
8341
            (int)virDomainNumaGetNodeCount(def->numa)) {
8342 8343 8344 8345 8346 8347 8348 8349 8350 8351
            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;
        }
8352 8353 8354 8355 8356 8357 8358
        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:
8359
    case VIR_DOMAIN_CONTROLLER_TYPE_XENBUS:
8360 8361
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
8362 8363 8364 8365 8366
    }

    return 0;
}

8367 8368 8369 8370 8371 8372
static int
qemuDomainChrDefPostParse(virDomainChrDefPtr chr,
                          const virDomainDef *def,
                          virQEMUDriverPtr driver,
                          unsigned int parseFlags)
{
8373 8374 8375 8376 8377 8378 8379 8380 8381 8382 8383 8384 8385 8386
    /* 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)) {
8387
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
8388
        } else if (qemuDomainIsARMVirt(def) || qemuDomainIsRISCVVirt(def)) {
8389
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
8390 8391
        } else if (ARCH_IS_S390(def->os.arch)) {
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
8392 8393 8394
        }
    }

8395 8396 8397
    /* Set the default target model */
    if (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL &&
        chr->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_NONE) {
8398
        switch ((virDomainChrSerialTargetType)chr->targetType) {
8399 8400 8401 8402 8403 8404 8405 8406 8407
        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;
8408 8409 8410
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY;
            break;
8411
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
8412 8413 8414 8415 8416
            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;
            }
8417
            break;
8418 8419 8420
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE;
            break;
8421 8422 8423 8424 8425 8426 8427
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            /* Nothing to do */
            break;
        }
    }

8428 8429 8430 8431 8432 8433 8434 8435 8436 8437 8438 8439 8440 8441 8442 8443
    /* clear auto generated unix socket path for inactive definitions */
    if (parseFlags & VIR_DOMAIN_DEF_PARSE_INACTIVE) {
        if (qemuDomainChrDefDropDefaultPath(chr, driver) < 0)
            return -1;

        /* For UNIX chardev if no path is provided we generate one.
         * This also implies that the mode is 'bind'. */
        if (chr->source &&
            chr->source->type == VIR_DOMAIN_CHR_TYPE_UNIX &&
            !chr->source->data.nix.path) {
            chr->source->data.nix.listen = true;
        }
    }

    return 0;
}
8444

8445

8446 8447 8448 8449 8450 8451 8452 8453 8454 8455 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 8496 8497 8498 8499 8500 8501 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
/**
 * qemuDomainDeviceDiskDefPostParseRestoreSecAlias:
 *
 * Re-generate aliases for objects related to the storage source if they
 * were not stored in the status XML by an older libvirt.
 *
 * Note that qemuCaps should be always present for a status XML.
 */
static int
qemuDomainDeviceDiskDefPostParseRestoreSecAlias(virDomainDiskDefPtr disk,
                                                virQEMUCapsPtr qemuCaps,
                                                unsigned int parseFlags)
{
    qemuDomainStorageSourcePrivatePtr priv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(disk->src);
    bool restoreAuthSecret = false;
    bool restoreEncSecret = false;
    char *authalias = NULL;
    char *encalias = NULL;
    int ret = -1;

    if (!(parseFlags & VIR_DOMAIN_DEF_PARSE_STATUS) ||
        !qemuCaps ||
        virStorageSourceIsEmpty(disk->src) ||
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

    /* network storage authentication secret */
    if (disk->src->auth &&
        (!priv || !priv->secinfo)) {

        /* only RBD and iSCSI (with capability) were supporting authentication
         * using secret object at the time we did not format the alias into the
         * status XML */
        if (virStorageSourceGetActualType(disk->src) == VIR_STORAGE_TYPE_NETWORK &&
            (disk->src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD ||
             (disk->src->protocol == VIR_STORAGE_NET_PROTOCOL_ISCSI &&
              virQEMUCapsGet(qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET))))
            restoreAuthSecret = true;
    }

    /* disk encryption secret */
    if (disk->src->encryption &&
        disk->src->encryption->format == VIR_STORAGE_ENCRYPTION_FORMAT_LUKS &&
        (!priv || !priv->encinfo))
        restoreEncSecret = true;

    if (!restoreAuthSecret && !restoreEncSecret)
        return 0;

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

        priv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(disk->src);
    }

    if (restoreAuthSecret) {
        if (!(authalias = qemuDomainGetSecretAESAlias(disk->info.alias, false)))
            goto cleanup;

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

    if (restoreEncSecret) {
        if (!(encalias = qemuDomainGetSecretAESAlias(disk->info.alias, true)))
            goto cleanup;

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

    ret = 0;

 cleanup:
    VIR_FREE(authalias);
    VIR_FREE(encalias);
    return ret;
}


8527 8528
static int
qemuDomainDeviceDiskDefPostParse(virDomainDiskDefPtr disk,
8529
                                 virQEMUCapsPtr qemuCaps,
8530
                                 unsigned int parseFlags)
8531 8532
{
    /* set default disk types and drivers */
8533 8534 8535
    if (!virDomainDiskGetDriver(disk) &&
        virDomainDiskSetDriver(disk, "qemu") < 0)
        return -1;
8536

8537 8538
    /* default disk format for drives */
    if (virDomainDiskGetFormat(disk) == VIR_STORAGE_FILE_NONE &&
8539
        virDomainDiskGetType(disk) != VIR_STORAGE_TYPE_VOLUME)
8540
        virDomainDiskSetFormat(disk, VIR_STORAGE_FILE_RAW);
8541

8542 8543 8544 8545
    /* default disk format for mirrored drive */
    if (disk->mirror &&
        disk->mirror->format == VIR_STORAGE_FILE_NONE)
        disk->mirror->format = VIR_STORAGE_FILE_RAW;
8546

8547 8548 8549 8550
    if (qemuDomainDeviceDiskDefPostParseRestoreSecAlias(disk, qemuCaps,
                                                        parseFlags) < 0)
        return -1;

8551 8552 8553 8554 8555 8556 8557
    /* 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;

8558 8559 8560 8561
    return 0;
}


8562 8563 8564 8565 8566 8567
static int
qemuDomainDeviceNetDefPostParse(virDomainNetDefPtr net,
                                const virDomainDef *def,
                                virQEMUCapsPtr qemuCaps)
{
    if (net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
8568 8569
        !virDomainNetGetModelString(net))
        net->model = qemuDomainDefaultNetModel(def, qemuCaps);
8570 8571 8572 8573 8574

    return 0;
}


8575
static int
8576 8577
qemuDomainDefaultVideoDevice(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
8578
{
8579
    if (ARCH_IS_PPC64(def->os.arch))
8580
        return VIR_DOMAIN_VIDEO_TYPE_VGA;
8581 8582 8583
    if (qemuDomainIsARMVirt(def) ||
        qemuDomainIsRISCVVirt(def) ||
        ARCH_IS_S390(def->os.arch)) {
8584
        return VIR_DOMAIN_VIDEO_TYPE_VIRTIO;
8585
    }
8586 8587 8588 8589 8590
    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;
8591 8592 8593
}


8594 8595
static int
qemuDomainDeviceVideoDefPostParse(virDomainVideoDefPtr video,
8596 8597
                                  const virDomainDef *def,
                                  virQEMUCapsPtr qemuCaps)
8598
{
8599
    if (video->type == VIR_DOMAIN_VIDEO_TYPE_DEFAULT)
8600
        video->type = qemuDomainDefaultVideoDevice(def, qemuCaps);
8601 8602 8603 8604 8605 8606 8607 8608 8609 8610

    if (video->type == VIR_DOMAIN_VIDEO_TYPE_QXL &&
        !video->vgamem) {
        video->vgamem = QEMU_QXL_VGAMEM_DEFAULT;
    }

    return 0;
}


8611 8612 8613 8614 8615 8616 8617 8618 8619 8620 8621 8622 8623 8624 8625 8626 8627
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
8628 8629 8630 8631 8632 8633 8634 8635 8636 8637
static int
qemuDomainVsockDefPostParse(virDomainVsockDefPtr vsock)
{
    if (vsock->model == VIR_DOMAIN_VSOCK_MODEL_DEFAULT)
        vsock->model = VIR_DOMAIN_VSOCK_MODEL_VIRTIO;

    return 0;
}


8638 8639 8640 8641 8642 8643 8644
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) &&
8645
        mdevsrc->model == VIR_MDEV_MODEL_TYPE_VFIO_PCI &&
8646 8647 8648 8649 8650 8651 8652 8653 8654 8655 8656 8657 8658 8659 8660 8661 8662 8663 8664 8665 8666 8667
        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;
}


8668 8669
static int
qemuDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
8670
                             const virDomainDef *def,
8671
                             unsigned int parseFlags,
8672
                             void *opaque,
8673
                             void *parseOpaque)
8674
{
8675
    virQEMUDriverPtr driver = opaque;
8676 8677 8678
    /* 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. */
8679
    virQEMUCapsPtr qemuCaps = parseOpaque;
8680
    int ret = -1;
8681

8682 8683 8684 8685
    switch ((virDomainDeviceType) dev->type) {
    case VIR_DOMAIN_DEVICE_NET:
        ret = qemuDomainDeviceNetDefPostParse(dev->data.net, def, qemuCaps);
        break;
8686

8687
    case VIR_DOMAIN_DEVICE_DISK:
8688
        ret = qemuDomainDeviceDiskDefPostParse(dev->data.disk, qemuCaps,
8689
                                               parseFlags);
8690
        break;
8691

8692
    case VIR_DOMAIN_DEVICE_VIDEO:
8693
        ret = qemuDomainDeviceVideoDefPostParse(dev->data.video, def, qemuCaps);
8694
        break;
8695

8696 8697 8698
    case VIR_DOMAIN_DEVICE_PANIC:
        ret = qemuDomainDevicePanicDefPostParse(dev->data.panic, def);
        break;
8699

8700 8701 8702 8703
    case VIR_DOMAIN_DEVICE_CONTROLLER:
        ret = qemuDomainControllerDefPostParse(dev->data.controller, def,
                                               qemuCaps, parseFlags);
        break;
8704

8705 8706 8707
    case VIR_DOMAIN_DEVICE_SHMEM:
        ret = qemuDomainShmemDefPostParse(dev->data.shmem);
        break;
8708

8709 8710 8711
    case VIR_DOMAIN_DEVICE_CHR:
        ret = qemuDomainChrDefPostParse(dev->data.chr, def, driver, parseFlags);
        break;
8712

J
Ján Tomko 已提交
8713 8714 8715 8716
    case VIR_DOMAIN_DEVICE_VSOCK:
        ret = qemuDomainVsockDefPostParse(dev->data.vsock);
        break;

8717 8718 8719 8720
    case VIR_DOMAIN_DEVICE_HOSTDEV:
        ret = qemuDomainHostdevDefPostParse(dev->data.hostdev, qemuCaps);
        break;

8721 8722 8723 8724 8725 8726 8727 8728 8729 8730 8731 8732 8733 8734 8735 8736 8737 8738 8739 8740 8741 8742 8743 8744 8745 8746 8747 8748
    case VIR_DOMAIN_DEVICE_LEASE:
    case VIR_DOMAIN_DEVICE_FS:
    case VIR_DOMAIN_DEVICE_INPUT:
    case VIR_DOMAIN_DEVICE_SOUND:
    case VIR_DOMAIN_DEVICE_WATCHDOG:
    case VIR_DOMAIN_DEVICE_GRAPHICS:
    case VIR_DOMAIN_DEVICE_HUB:
    case VIR_DOMAIN_DEVICE_REDIRDEV:
    case VIR_DOMAIN_DEVICE_SMARTCARD:
    case VIR_DOMAIN_DEVICE_MEMBALLOON:
    case VIR_DOMAIN_DEVICE_NVRAM:
    case VIR_DOMAIN_DEVICE_RNG:
    case VIR_DOMAIN_DEVICE_TPM:
    case VIR_DOMAIN_DEVICE_MEMORY:
    case VIR_DOMAIN_DEVICE_IOMMU:
        ret = 0;
        break;

    case VIR_DOMAIN_DEVICE_NONE:
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("unexpected VIR_DOMAIN_DEVICE_NONE"));
        break;

    case VIR_DOMAIN_DEVICE_LAST:
    default:
        virReportEnumRangeError(virDomainDeviceType, dev->type);
        break;
    }
8749 8750

    return ret;
8751 8752 8753
}


8754 8755
static int
qemuDomainDefAssignAddresses(virDomainDef *def,
J
Ján Tomko 已提交
8756
                             unsigned int parseFlags G_GNUC_UNUSED,
8757
                             void *opaque,
8758
                             void *parseOpaque)
8759 8760
{
    virQEMUDriverPtr driver = opaque;
8761 8762 8763
    /* 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. */
8764
    virQEMUCapsPtr qemuCaps = parseOpaque;
8765
    bool newDomain = parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE;
8766

8767 8768 8769 8770 8771 8772 8773
    /* 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;

8774 8775 8776 8777 8778 8779
    return qemuDomainAssignAddresses(def, qemuCaps, driver, NULL, newDomain);
}


static int
qemuDomainPostParseDataAlloc(const virDomainDef *def,
J
Ján Tomko 已提交
8780
                             unsigned int parseFlags G_GNUC_UNUSED,
8781 8782 8783 8784 8785 8786
                             void *opaque,
                             void **parseOpaque)
{
    virQEMUDriverPtr driver = opaque;

    if (!(*parseOpaque = virQEMUCapsCacheLookup(driver->qemuCapsCache,
8787
                                                def->emulator)))
8788
        return 1;
8789

8790 8791 8792 8793 8794 8795 8796 8797
    return 0;
}


static void
qemuDomainPostParseDataFree(void *parseOpaque)
{
    virQEMUCapsPtr qemuCaps = parseOpaque;
8798 8799 8800 8801 8802

    virObjectUnref(qemuCaps);
}


8803
virDomainDefParserConfig virQEMUDriverDomainDefParserConfig = {
8804
    .domainPostParseBasicCallback = qemuDomainDefPostParseBasic,
8805 8806
    .domainPostParseDataAlloc = qemuDomainPostParseDataAlloc,
    .domainPostParseDataFree = qemuDomainPostParseDataFree,
8807
    .devicesPostParseCallback = qemuDomainDeviceDefPostParse,
8808
    .domainPostParseCallback = qemuDomainDefPostParse,
8809
    .assignAddressesCallback = qemuDomainDefAssignAddresses,
8810
    .domainValidateCallback = qemuDomainDefValidate,
8811 8812
    .deviceValidateCallback = qemuDomainDeviceDefValidate,

8813
    .features = VIR_DOMAIN_DEF_FEATURE_MEMORY_HOTPLUG |
8814
                VIR_DOMAIN_DEF_FEATURE_OFFLINE_VCPUPIN |
8815
                VIR_DOMAIN_DEF_FEATURE_INDIVIDUAL_VCPUS |
8816
                VIR_DOMAIN_DEF_FEATURE_USER_ALIAS |
8817 8818
                VIR_DOMAIN_DEF_FEATURE_FW_AUTOSELECT |
                VIR_DOMAIN_DEF_FEATURE_NET_MODEL_STRING,
8819 8820 8821
};


8822
static void
8823 8824
qemuDomainObjSaveStatus(virQEMUDriverPtr driver,
                        virDomainObjPtr obj)
8825
{
8826 8827 8828
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    if (virDomainObjIsActive(obj)) {
8829
        if (virDomainObjSave(obj, driver->xmlopt, cfg->stateDir) < 0)
8830
            VIR_WARN("Failed to save status on vm %s", obj->def->name);
8831
    }
8832

8833
    virObjectUnref(cfg);
8834 8835
}

8836 8837 8838 8839 8840 8841 8842 8843

void
qemuDomainSaveStatus(virDomainObjPtr obj)
{
    qemuDomainObjSaveStatus(QEMU_DOMAIN_PRIVATE(obj)->driver, obj);
}


8844 8845 8846 8847
void
qemuDomainSaveConfig(virDomainObjPtr obj)
{
    virQEMUDriverPtr driver = QEMU_DOMAIN_PRIVATE(obj)->driver;
8848
    g_autoptr(virQEMUDriverConfig) cfg = NULL;
8849 8850 8851 8852 8853 8854 8855 8856 8857 8858 8859 8860
    virDomainDefPtr def = NULL;

    if (virDomainObjIsActive(obj))
        def = obj->newDef;
    else
        def = obj->def;

    if (!def)
        return;

    cfg = virQEMUDriverGetConfig(driver);

8861
    if (virDomainDefSave(def, driver->xmlopt, cfg->configDir) < 0)
8862 8863 8864 8865
        VIR_WARN("Failed to save config of vm %s", obj->def->name);
}


J
Jiri Denemark 已提交
8866
void
8867
qemuDomainObjSetJobPhase(virQEMUDriverPtr driver,
J
Jiri Denemark 已提交
8868 8869 8870 8871
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
8872
    unsigned long long me = virThreadSelfID();
J
Jiri Denemark 已提交
8873 8874 8875 8876

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

8877 8878 8879 8880 8881
    VIR_DEBUG("Setting '%s' phase to '%s'",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob, phase));

    if (priv->job.asyncOwner && me != priv->job.asyncOwner) {
8882
        VIR_WARN("'%s' async job is owned by thread %llu",
8883 8884 8885 8886
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }

J
Jiri Denemark 已提交
8887
    priv->job.phase = phase;
8888
    priv->job.asyncOwner = me;
8889
    qemuDomainObjSaveStatus(driver, obj);
J
Jiri Denemark 已提交
8890 8891
}

8892
void
8893 8894
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
8895 8896 8897
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

8898 8899 8900 8901 8902 8903 8904
    if (!priv->job.asyncJob)
        return;

    priv->job.mask = allowedJobs | JOB_MASK(QEMU_JOB_DESTROY);
}

void
8905
qemuDomainObjDiscardAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
8906 8907 8908 8909 8910 8911
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
8912
    qemuDomainObjSaveStatus(driver, obj);
8913 8914
}

8915 8916 8917 8918 8919 8920 8921 8922 8923
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()) {
8924
        VIR_WARN("'%s' async job is owned by thread %llu",
8925 8926 8927 8928 8929 8930
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }
    priv->job.asyncOwner = 0;
}

8931
static bool
8932
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
8933
{
8934 8935 8936
    return !priv->job.asyncJob ||
           job == QEMU_JOB_NONE ||
           (priv->job.mask & JOB_MASK(job)) != 0;
8937 8938
}

8939
bool
8940
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
8941 8942 8943 8944
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

8945 8946 8947 8948 8949 8950 8951 8952 8953 8954 8955
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));
}

8956 8957 8958
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

8959 8960 8961 8962 8963 8964
/**
 * qemuDomainObjBeginJobInternal:
 * @driver: qemu driver
 * @obj: domain object
 * @job: qemuDomainJob to start
 * @asyncJob: qemuDomainAsyncJob to start
8965
 * @nowait: don't wait trying to acquire @job
8966 8967 8968 8969
 *
 * 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
8970 8971 8972 8973 8974
 * 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.
8975 8976 8977 8978 8979
 *
 * Returns: 0 on success,
 *         -2 if unable to start job because of timeout or
 *            maxQueuedJobs limit,
 *         -1 otherwise.
8980
 */
8981
static int ATTRIBUTE_NONNULL(1)
8982
qemuDomainObjBeginJobInternal(virQEMUDriverPtr driver,
8983
                              virDomainObjPtr obj,
8984
                              qemuDomainJob job,
8985
                              qemuDomainAgentJob agentJob,
8986 8987
                              qemuDomainAsyncJob asyncJob,
                              bool nowait)
8988 8989
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
8990
    unsigned long long now;
8991
    unsigned long long then;
8992
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
J
Jiri Denemark 已提交
8993
    bool async = job == QEMU_JOB_ASYNC;
8994
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8995
    const char *blocker = NULL;
8996
    const char *agentBlocker = NULL;
J
Jiri Denemark 已提交
8997
    int ret = -1;
J
Jiri Denemark 已提交
8998
    unsigned long long duration = 0;
8999
    unsigned long long agentDuration = 0;
J
Jiri Denemark 已提交
9000
    unsigned long long asyncDuration = 0;
9001

9002 9003 9004 9005 9006 9007
    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
9008
              qemuDomainJobTypeToString(priv->job.active),
9009
              qemuDomainAgentJobTypeToString(priv->job.agentActive),
J
Jiri Denemark 已提交
9010
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
9011

9012 9013
    if (virTimeMillisNow(&now) < 0) {
        virObjectUnref(cfg);
9014
        return -1;
9015 9016
    }

9017
    priv->jobs_queued++;
J
Jiri Denemark 已提交
9018
    then = now + QEMU_JOB_WAIT_TIME;
9019

9020
 retry:
9021 9022
    if ((!async && job != QEMU_JOB_DESTROY) &&
        cfg->maxQueuedJobs &&
9023
        priv->jobs_queued > cfg->maxQueuedJobs) {
9024 9025 9026
        goto error;
    }

9027
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
9028 9029 9030
        if (nowait)
            goto cleanup;

9031
        VIR_DEBUG("Waiting for async job (vm=%p name=%s)", obj, obj->def->name);
9032
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->parent.lock, then) < 0)
9033 9034 9035
            goto error;
    }

9036
    while (!qemuDomainObjCanSetJob(priv, job, agentJob)) {
9037 9038 9039
        if (nowait)
            goto cleanup;

9040
        VIR_DEBUG("Waiting for job (vm=%p name=%s)", obj, obj->def->name);
9041
        if (virCondWaitUntil(&priv->job.cond, &obj->parent.lock, then) < 0)
9042
            goto error;
9043
    }
9044 9045 9046

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
9047
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
9048 9049
        goto retry;

J
Jiri Denemark 已提交
9050 9051
    ignore_value(virTimeMillisNow(&now));

9052 9053 9054 9055 9056 9057 9058 9059 9060 9061 9062 9063 9064 9065 9066 9067 9068 9069 9070 9071 9072 9073 9074 9075 9076 9077 9078 9079 9080 9081 9082 9083 9084 9085 9086 9087 9088 9089 9090 9091
    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;
9092
    }
9093

9094
    if (qemuDomainTrackJob(job))
9095
        qemuDomainObjSaveStatus(driver, obj);
9096

9097
    virObjectUnref(cfg);
9098
    return 0;
9099

9100
 error:
J
Jiri Denemark 已提交
9101 9102 9103
    ignore_value(virTimeMillisNow(&now));
    if (priv->job.active && priv->job.started)
        duration = now - priv->job.started;
9104 9105
    if (priv->job.agentActive && priv->job.agentStarted)
        agentDuration = now - priv->job.agentStarted;
J
Jiri Denemark 已提交
9106 9107 9108
    if (priv->job.asyncJob && priv->job.asyncStarted)
        asyncDuration = now - priv->job.asyncStarted;

9109 9110 9111 9112
    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)",
9113
             qemuDomainJobTypeToString(job),
9114
             qemuDomainAgentJobTypeToString(agentJob),
9115 9116 9117
             qemuDomainAsyncJobTypeToString(asyncJob),
             obj->def->name,
             qemuDomainJobTypeToString(priv->job.active),
9118
             qemuDomainAgentJobTypeToString(priv->job.agentActive),
9119
             qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
9120
             priv->job.owner, NULLSTR(priv->job.ownerAPI),
9121
             priv->job.agentOwner, NULLSTR(priv->job.agentOwnerAPI),
J
Jiri Denemark 已提交
9122
             priv->job.asyncOwner, NULLSTR(priv->job.asyncOwnerAPI),
9123
             priv->job.apiFlags,
9124
             duration / 1000, agentDuration / 1000, asyncDuration / 1000);
9125

9126 9127 9128 9129 9130 9131 9132 9133 9134
    if (job) {
        if (nested || qemuDomainNestedJobAllowed(priv, job))
            blocker = priv->job.ownerAPI;
        else
            blocker = priv->job.asyncOwnerAPI;
    }

    if (agentJob)
        agentBlocker = priv->job.agentOwnerAPI;
9135

9136
    if (errno == ETIMEDOUT) {
9137 9138 9139 9140 9141 9142
        if (blocker && agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by monitor=%s agent=%s)"),
                           blocker, agentBlocker);
        } else if (blocker) {
9143
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
9144 9145
                           _("cannot acquire state change "
                             "lock (held by monitor=%s)"),
9146
                           blocker);
9147 9148 9149 9150 9151
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by agent=%s)"),
                           agentBlocker);
9152 9153 9154 9155
        } else {
            virReportError(VIR_ERR_OPERATION_TIMEOUT, "%s",
                           _("cannot acquire state change lock"));
        }
9156 9157 9158
        ret = -2;
    } else if (cfg->maxQueuedJobs &&
               priv->jobs_queued > cfg->maxQueuedJobs) {
9159 9160 9161 9162 9163 9164 9165
        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) {
9166
            virReportError(VIR_ERR_OPERATION_FAILED,
9167 9168
                           _("cannot acquire state change "
                             "lock (held by monitor=%s) "
9169 9170
                             "due to max_queued limit"),
                           blocker);
9171 9172 9173 9174 9175 9176
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_FAILED,
                           _("cannot acquire state change "
                             "lock (held by agent=%s) "
                             "due to max_queued limit"),
                           agentBlocker);
9177 9178 9179 9180 9181
        } else {
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("cannot acquire state change lock "
                             "due to max_queued limit"));
        }
9182 9183
        ret = -2;
    } else {
9184
        virReportSystemError(errno, "%s", _("cannot acquire job mutex"));
9185
    }
J
Jiri Denemark 已提交
9186 9187

 cleanup:
9188
    priv->jobs_queued--;
9189
    virObjectUnref(cfg);
9190
    return ret;
9191 9192 9193
}

/*
9194
 * obj must be locked before calling
9195 9196 9197 9198
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
9199
 * Successful calls must be followed by EndJob eventually
9200
 */
9201
int qemuDomainObjBeginJob(virQEMUDriverPtr driver,
9202
                          virDomainObjPtr obj,
9203
                          qemuDomainJob job)
9204
{
9205
    if (qemuDomainObjBeginJobInternal(driver, obj, job,
9206
                                      QEMU_AGENT_JOB_NONE,
9207
                                      QEMU_ASYNC_JOB_NONE, false) < 0)
9208 9209 9210
        return -1;
    else
        return 0;
9211 9212
}

9213 9214 9215 9216 9217 9218 9219 9220 9221 9222 9223 9224 9225 9226 9227 9228 9229 9230 9231 9232 9233 9234 9235 9236 9237 9238 9239 9240 9241 9242 9243 9244 9245 9246 9247 9248 9249
/**
 * qemuDomainObjBeginAgentJob:
 *
 * Grabs agent type of job. Use if caller talks to guest agent only.
 *
 * To end job call qemuDomainObjEndAgentJob.
 */
int
qemuDomainObjBeginAgentJob(virQEMUDriverPtr driver,
                           virDomainObjPtr obj,
                           qemuDomainAgentJob agentJob)
{
    return qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_NONE,
                                         agentJob,
                                         QEMU_ASYNC_JOB_NONE, false);
}

/**
 * qemuDomainObjBeginJobWithAgent:
 *
 * Grabs both monitor and agent types of job. Use if caller talks to
 * both monitor and guest agent. However, if @job (or @agentJob) is
 * QEMU_JOB_NONE (or QEMU_AGENT_JOB_NONE) only agent job is acquired (or
 * monitor job).
 *
 * To end job call qemuDomainObjEndJobWithAgent.
 */
int
qemuDomainObjBeginJobWithAgent(virQEMUDriverPtr driver,
                               virDomainObjPtr obj,
                               qemuDomainJob job,
                               qemuDomainAgentJob agentJob)
{
    return qemuDomainObjBeginJobInternal(driver, obj, job, agentJob,
                                         QEMU_ASYNC_JOB_NONE, false);
}

9250
int qemuDomainObjBeginAsyncJob(virQEMUDriverPtr driver,
9251
                               virDomainObjPtr obj,
9252
                               qemuDomainAsyncJob asyncJob,
9253 9254
                               virDomainJobOperation operation,
                               unsigned long apiFlags)
9255
{
9256 9257
    qemuDomainObjPrivatePtr priv;

9258
    if (qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_ASYNC,
9259
                                      QEMU_AGENT_JOB_NONE,
9260
                                      asyncJob, false) < 0)
9261
        return -1;
9262 9263 9264

    priv = obj->privateData;
    priv->job.current->operation = operation;
9265
    priv->job.apiFlags = apiFlags;
9266
    return 0;
9267 9268
}

9269
int
9270 9271
qemuDomainObjBeginNestedJob(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
9272
                            qemuDomainAsyncJob asyncJob)
9273 9274 9275 9276 9277
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (asyncJob != priv->job.asyncJob) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
9278 9279
                       _("unexpected async job %d type expected %d"),
                       asyncJob, priv->job.asyncJob);
9280 9281 9282 9283
        return -1;
    }

    if (priv->job.asyncOwner != virThreadSelfID()) {
9284
        VIR_WARN("This thread doesn't seem to be the async job owner: %llu",
9285 9286 9287 9288 9289
                 priv->job.asyncOwner);
    }

    return qemuDomainObjBeginJobInternal(driver, obj,
                                         QEMU_JOB_ASYNC_NESTED,
9290
                                         QEMU_AGENT_JOB_NONE,
9291 9292
                                         QEMU_ASYNC_JOB_NONE,
                                         false);
9293 9294
}

9295 9296 9297 9298 9299 9300 9301 9302 9303 9304 9305 9306 9307 9308 9309 9310 9311 9312 9313
/**
 * 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,
9314
                                         QEMU_AGENT_JOB_NONE,
9315 9316
                                         QEMU_ASYNC_JOB_NONE, true);
}
9317

9318
/*
9319
 * obj must be locked and have a reference before calling
9320 9321 9322 9323
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 */
9324 9325
void
qemuDomainObjEndJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
9326 9327
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
9328
    qemuDomainJob job = priv->job.active;
9329

9330 9331
    priv->jobs_queued--;

9332
    VIR_DEBUG("Stopping job: %s (async=%s vm=%p name=%s)",
9333
              qemuDomainJobTypeToString(job),
9334 9335
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
9336

9337
    qemuDomainObjResetJob(priv);
9338
    if (qemuDomainTrackJob(job))
9339
        qemuDomainObjSaveStatus(driver, obj);
9340 9341 9342 9343 9344 9345 9346 9347 9348 9349 9350 9351 9352 9353 9354 9355 9356 9357 9358 9359 9360 9361 9362 9363 9364 9365 9366 9367 9368 9369 9370 9371 9372 9373 9374 9375 9376 9377 9378 9379 9380 9381 9382
    /* We indeed need to wake up ALL threads waiting because
     * grabbing a job requires checking more variables. */
    virCondBroadcast(&priv->job.cond);
}

void
qemuDomainObjEndAgentJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
    qemuDomainAgentJob agentJob = priv->job.agentActive;

    priv->jobs_queued--;

    VIR_DEBUG("Stopping agent job: %s (async=%s vm=%p name=%s)",
              qemuDomainAgentJobTypeToString(agentJob),
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);

    qemuDomainObjResetAgentJob(priv);
    /* We indeed need to wake up ALL threads waiting because
     * grabbing a job requires checking more variables. */
    virCondBroadcast(&priv->job.cond);
}

void
qemuDomainObjEndJobWithAgent(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
    qemuDomainJob job = priv->job.active;
    qemuDomainAgentJob agentJob = priv->job.agentActive;

    priv->jobs_queued--;

    VIR_DEBUG("Stopping both jobs: %s %s (async=%s vm=%p name=%s)",
              qemuDomainJobTypeToString(job),
              qemuDomainAgentJobTypeToString(agentJob),
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAgentJob(priv);
    if (qemuDomainTrackJob(job))
9383
        qemuDomainObjSaveStatus(driver, obj);
9384 9385 9386
    /* We indeed need to wake up ALL threads waiting because
     * grabbing a job requires checking more variables. */
    virCondBroadcast(&priv->job.cond);
9387 9388
}

9389
void
9390
qemuDomainObjEndAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
9391 9392
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
9393

9394 9395
    priv->jobs_queued--;

9396 9397 9398
    VIR_DEBUG("Stopping async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
9399

9400
    qemuDomainObjResetAsyncJob(priv);
9401
    qemuDomainObjSaveStatus(driver, obj);
9402 9403 9404
    virCondBroadcast(&priv->job.asyncCond);
}

9405 9406 9407 9408 9409
void
qemuDomainObjAbortAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

9410 9411 9412
    VIR_DEBUG("Requesting abort of async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
9413

9414 9415
    priv->job.abortJob = true;
    virDomainObjBroadcast(obj);
9416 9417
}

9418 9419 9420 9421
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU monitor API call
9422 9423 9424
 * Must have already either called qemuDomainObjBeginJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active; may not be used for nested async jobs.
9425 9426 9427
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
9428
static int
9429
qemuDomainObjEnterMonitorInternal(virQEMUDriverPtr driver,
9430
                                  virDomainObjPtr obj,
9431
                                  qemuDomainAsyncJob asyncJob)
9432 9433 9434
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

9435
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
9436 9437 9438
        int ret;
        if ((ret = qemuDomainObjBeginNestedJob(driver, obj, asyncJob)) < 0)
            return ret;
9439
        if (!virDomainObjIsActive(obj)) {
9440 9441
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
9442
            qemuDomainObjEndJob(driver, obj);
9443 9444
            return -1;
        }
9445 9446 9447
    } 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");
9448 9449 9450 9451 9452
    } 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);
9453 9454
    }

9455 9456
    VIR_DEBUG("Entering monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
9457
    virObjectLock(priv->mon);
9458
    virObjectRef(priv->mon);
9459
    ignore_value(virTimeMillisNow(&priv->monStart));
9460
    virObjectUnlock(obj);
9461 9462

    return 0;
9463 9464
}

9465
static void ATTRIBUTE_NONNULL(1)
9466
qemuDomainObjExitMonitorInternal(virQEMUDriverPtr driver,
9467
                                 virDomainObjPtr obj)
9468 9469
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
9470
    bool hasRefs;
9471

9472
    hasRefs = virObjectUnref(priv->mon);
9473

9474
    if (hasRefs)
9475
        virObjectUnlock(priv->mon);
9476

9477
    virObjectLock(obj);
9478 9479
    VIR_DEBUG("Exited monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
9480

9481
    priv->monStart = 0;
9482
    if (!hasRefs)
9483
        priv->mon = NULL;
9484

J
Jiri Denemark 已提交
9485 9486
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjEndJob(driver, obj);
9487 9488
}

9489
void qemuDomainObjEnterMonitor(virQEMUDriverPtr driver,
9490
                               virDomainObjPtr obj)
9491
{
9492
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, obj,
9493
                                                   QEMU_ASYNC_JOB_NONE));
9494 9495
}

9496
/* obj must NOT be locked before calling
9497 9498
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
9499 9500 9501 9502 9503 9504
 *
 * 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.
9505
 */
9506 9507
int qemuDomainObjExitMonitor(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
9508
{
9509
    qemuDomainObjExitMonitorInternal(driver, obj);
9510
    if (!virDomainObjIsActive(obj)) {
9511
        if (virGetLastErrorCode() == VIR_ERR_OK)
9512 9513
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
9514 9515 9516
        return -1;
    }
    return 0;
9517
}
9518 9519

/*
9520
 * obj must be locked before calling
9521 9522
 *
 * To be called immediately before any QEMU monitor API call.
9523
 * Must have already either called qemuDomainObjBeginJob()
9524 9525 9526 9527 9528
 * 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
9529 9530 9531
 * 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).
9532 9533
 */
int
9534
qemuDomainObjEnterMonitorAsync(virQEMUDriverPtr driver,
9535
                               virDomainObjPtr obj,
9536
                               qemuDomainAsyncJob asyncJob)
9537
{
9538
    return qemuDomainObjEnterMonitorInternal(driver, obj, asyncJob);
9539 9540
}

D
Daniel P. Berrange 已提交
9541

9542 9543 9544 9545
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU agent API call.
9546 9547 9548
 * Must have already called qemuDomainObjBeginAgentJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active.
9549 9550 9551
 *
 * To be followed with qemuDomainObjExitAgent() once complete
 */
9552
qemuAgentPtr
9553
qemuDomainObjEnterAgent(virDomainObjPtr obj)
D
Daniel P. Berrange 已提交
9554 9555
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
9556
    qemuAgentPtr agent = priv->agent;
D
Daniel P. Berrange 已提交
9557

9558 9559
    VIR_DEBUG("Entering agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
9560 9561 9562

    virObjectLock(agent);
    virObjectRef(agent);
9563
    virObjectUnlock(obj);
9564 9565

    return agent;
D
Daniel P. Berrange 已提交
9566 9567
}

9568 9569 9570 9571 9572 9573

/* obj must NOT be locked before calling
 *
 * Should be paired with an earlier qemuDomainObjEnterAgent() call
 */
void
9574
qemuDomainObjExitAgent(virDomainObjPtr obj, qemuAgentPtr agent)
D
Daniel P. Berrange 已提交
9575
{
9576 9577
    virObjectUnlock(agent);
    virObjectUnref(agent);
9578
    virObjectLock(obj);
D
Daniel P. Berrange 已提交
9579

9580 9581
    VIR_DEBUG("Exited agent (agent=%p vm=%p name=%s)",
              agent, obj, obj->def->name);
D
Daniel P. Berrange 已提交
9582 9583
}

9584
void qemuDomainObjEnterRemote(virDomainObjPtr obj)
9585
{
9586 9587
    VIR_DEBUG("Entering remote (vm=%p name=%s)",
              obj, obj->def->name);
9588
    virObjectUnlock(obj);
9589 9590
}

9591 9592 9593 9594

int
qemuDomainObjExitRemote(virDomainObjPtr obj,
                        bool checkActive)
9595
{
9596
    virObjectLock(obj);
9597 9598
    VIR_DEBUG("Exited remote (vm=%p name=%s)",
              obj, obj->def->name);
9599 9600 9601 9602 9603 9604 9605 9606 9607

    if (checkActive && !virDomainObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("domain '%s' is not running"),
                       obj->def->name);
        return -1;
    }

    return 0;
9608
}
9609 9610


9611 9612
static virDomainDefPtr
qemuDomainDefFromXML(virQEMUDriverPtr driver,
9613
                     virQEMUCapsPtr qemuCaps,
9614 9615 9616 9617
                     const char *xml)
{
    virDomainDefPtr def;

9618
    def = virDomainDefParseString(xml, driver->xmlopt, qemuCaps,
9619 9620 9621 9622 9623 9624 9625
                                  VIR_DOMAIN_DEF_PARSE_INACTIVE |
                                  VIR_DOMAIN_DEF_PARSE_SKIP_VALIDATE);

    return def;
}


9626 9627
virDomainDefPtr
qemuDomainDefCopy(virQEMUDriverPtr driver,
9628
                  virQEMUCapsPtr qemuCaps,
9629 9630 9631 9632
                  virDomainDefPtr src,
                  unsigned int flags)
{
    virDomainDefPtr ret = NULL;
9633
    char *xml;
9634

9635
    if (!(xml = qemuDomainDefFormatXML(driver, qemuCaps, src, flags)))
9636
        return NULL;
9637

9638
    ret = qemuDomainDefFromXML(driver, qemuCaps, xml);
9639 9640 9641 9642 9643

    VIR_FREE(xml);
    return ret;
}

9644

9645 9646 9647 9648 9649 9650 9651 9652 9653 9654 9655 9656 9657 9658 9659 9660 9661 9662 9663 9664
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;
}


9665 9666
static int
qemuDomainDefFormatBufInternal(virQEMUDriverPtr driver,
9667
                               virQEMUCapsPtr qemuCaps,
9668 9669 9670 9671
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags,
                               virBuffer *buf)
9672
{
9673
    int ret = -1;
9674
    virDomainDefPtr copy = NULL;
9675

9676 9677
    virCheckFlags(VIR_DOMAIN_XML_COMMON_FLAGS | VIR_DOMAIN_XML_UPDATE_CPU, -1);

9678 9679 9680
    if (!(flags & (VIR_DOMAIN_XML_UPDATE_CPU | VIR_DOMAIN_XML_MIGRATABLE)))
        goto format;

9681
    if (!(copy = virDomainDefCopy(def, driver->xmlopt, qemuCaps,
9682 9683 9684 9685 9686
                                  flags & VIR_DOMAIN_XML_MIGRATABLE)))
        goto cleanup;

    def = copy;

9687
    /* Update guest CPU requirements according to host CPU */
9688
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) &&
9689
        def->cpu &&
9690 9691
        (def->cpu->mode != VIR_CPU_MODE_CUSTOM ||
         def->cpu->model)) {
9692
        g_autoptr(virQEMUCaps) qCaps = NULL;
9693 9694 9695 9696 9697

        if (qemuCaps) {
            qCaps = virObjectRef(qemuCaps);
        } else {
            if (!(qCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
9698
                                                     def->virtType,
9699 9700 9701 9702
                                                     def->emulator,
                                                     def->os.machine)))
                goto cleanup;
        }
9703 9704

        if (virCPUUpdate(def->os.arch, def->cpu,
9705
                         virQEMUCapsGetHostModel(qCaps, def->virtType,
9706
                                                 VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE)) < 0)
9707 9708 9709
            goto cleanup;
    }

9710
    if ((flags & VIR_DOMAIN_XML_MIGRATABLE)) {
9711
        size_t i;
9712
        int toremove = 0;
9713
        virDomainControllerDefPtr usb = NULL, pci = NULL;
9714 9715 9716 9717 9718 9719 9720 9721 9722 9723 9724 9725 9726 9727 9728

        /* 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];
            }
        }
9729 9730 9731 9732 9733 9734 9735 9736 9737

        /* 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'/>.
9738
         */
9739
        if (qemuDomainIsI440FX(def) &&
9740
            usb && usb->idx == 0 &&
9741
            (usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT ||
9742 9743
             usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_PIIX3_UHCI) &&
            !virDomainDeviceAliasIsUserAlias(usb->info.alias)) {
9744 9745
            VIR_DEBUG("Removing default USB controller from domain '%s'"
                      " for migration compatibility", def->name);
9746
            toremove++;
9747 9748 9749 9750 9751 9752 9753 9754 9755 9756 9757 9758 9759 9760 9761 9762 9763
        } 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 &&
9764 9765 9766
            pci->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT &&
            !virDomainDeviceAliasIsUserAlias(pci->info.alias) &&
            !pci->opts.pciopts.pcihole64) {
L
Laine Stump 已提交
9767
            VIR_DEBUG("Removing default pci-root from domain '%s'"
9768
                      " for migration compatibility", def->name);
9769
            toremove++;
9770 9771 9772 9773
        } else {
            pci = NULL;
        }

9774
        if (toremove) {
9775 9776 9777
            virDomainControllerDefPtr *controllers = def->controllers;
            int ncontrollers = def->ncontrollers;

9778
            if (VIR_ALLOC_N(def->controllers, ncontrollers - toremove) < 0) {
9779
                def->controllers = controllers;
9780 9781 9782 9783 9784
                goto cleanup;
            }

            def->ncontrollers = 0;
            for (i = 0; i < ncontrollers; i++) {
9785
                if (controllers[i] != usb && controllers[i] != pci)
9786 9787
                    def->controllers[def->ncontrollers++] = controllers[i];
            }
9788 9789 9790 9791

            VIR_FREE(controllers);
            virDomainControllerDefFree(pci);
            virDomainControllerDefFree(usb);
9792
        }
9793

9794 9795 9796 9797 9798 9799 9800 9801 9802
        /* 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;
            }
        }

9803 9804 9805 9806
        for (i = 0; i < def->nchannels; i++) {
            if (qemuDomainChrDefDropDefaultPath(def->channels[i], driver) < 0)
                goto cleanup;
        }
9807

9808 9809 9810 9811 9812 9813 9814 9815 9816
        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) {
9817
                switch ((virDomainChrSerialTargetType)serial->targetType) {
9818
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
9819
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
9820 9821 9822 9823 9824 9825
                    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:
9826
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
9827 9828 9829 9830 9831 9832 9833 9834
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
                    /* Nothing to do */
                    break;
                }
            }
        }

9835 9836 9837 9838 9839 9840 9841 9842 9843
        /* 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;
        }
9844

9845
        if (def->cpu && qemuDomainMakeCPUMigratable(def->cpu) < 0)
9846
            goto cleanup;
9847 9848
    }

9849
 format:
9850
    ret = virDomainDefFormatInternal(def, driver->xmlopt, buf,
9851
                                     virDomainDefFormatConvertXMLFlags(flags));
9852

9853
 cleanup:
9854
    virDomainDefFree(copy);
9855 9856
    return ret;
}
9857

9858 9859 9860

int
qemuDomainDefFormatBuf(virQEMUDriverPtr driver,
9861
                       virQEMUCapsPtr qemuCaps,
9862 9863 9864 9865
                       virDomainDefPtr def,
                       unsigned int flags,
                       virBufferPtr buf)
{
9866
    return qemuDomainDefFormatBufInternal(driver, qemuCaps, def, NULL, flags, buf);
9867 9868 9869 9870 9871
}


static char *
qemuDomainDefFormatXMLInternal(virQEMUDriverPtr driver,
9872
                               virQEMUCapsPtr qemuCaps,
9873 9874 9875
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags)
9876 9877 9878
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

9879
    if (qemuDomainDefFormatBufInternal(driver, qemuCaps, def, origCPU, flags, &buf) < 0)
9880 9881 9882 9883 9884
        return NULL;

    return virBufferContentAndReset(&buf);
}

9885 9886 9887

char *
qemuDomainDefFormatXML(virQEMUDriverPtr driver,
9888
                       virQEMUCapsPtr qemuCaps,
9889 9890 9891
                       virDomainDefPtr def,
                       unsigned int flags)
{
9892
    return qemuDomainDefFormatXMLInternal(driver, qemuCaps, def, NULL, flags);
9893 9894 9895
}


9896
char *qemuDomainFormatXML(virQEMUDriverPtr driver,
9897
                          virDomainObjPtr vm,
9898
                          unsigned int flags)
9899 9900
{
    virDomainDefPtr def;
9901 9902
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCPUDefPtr origCPU = NULL;
9903

9904
    if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef) {
9905
        def = vm->newDef;
9906
    } else {
9907
        def = vm->def;
9908
        origCPU = priv->origCPU;
9909
    }
9910

9911
    return qemuDomainDefFormatXMLInternal(driver, priv->qemuCaps, def, origCPU, flags);
9912 9913
}

9914
char *
9915
qemuDomainDefFormatLive(virQEMUDriverPtr driver,
9916
                        virQEMUCapsPtr qemuCaps,
9917
                        virDomainDefPtr def,
9918
                        virCPUDefPtr origCPU,
9919 9920
                        bool inactive,
                        bool compatible)
9921 9922 9923 9924 9925
{
    unsigned int flags = QEMU_DOMAIN_FORMAT_LIVE_FLAGS;

    if (inactive)
        flags |= VIR_DOMAIN_XML_INACTIVE;
9926 9927
    if (compatible)
        flags |= VIR_DOMAIN_XML_MIGRATABLE;
9928

9929
    return qemuDomainDefFormatXMLInternal(driver, qemuCaps, def, origCPU, flags);
9930 9931
}

9932

9933
void qemuDomainObjTaint(virQEMUDriverPtr driver,
9934
                        virDomainObjPtr obj,
9935
                        virDomainTaintFlags taint,
9936
                        qemuDomainLogContextPtr logCtxt)
9937
{
9938
    virErrorPtr orig_err = NULL;
9939 9940
    char *timestamp = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
9941
    int rc;
9942

9943 9944 9945 9946 9947 9948 9949 9950 9951 9952 9953 9954 9955 9956
    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 */
9957
    virErrorPreserveLast(&orig_err);
9958

9959 9960 9961
    if (!(timestamp = virTimeStringNow()))
        goto cleanup;

9962 9963 9964 9965 9966 9967 9968 9969 9970 9971 9972 9973
    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));
9974 9975
    }

9976
    if (rc < 0)
9977 9978 9979 9980
        virResetLastError();

 cleanup:
    VIR_FREE(timestamp);
9981
    virErrorRestore(&orig_err);
9982 9983 9984
}


9985
void qemuDomainObjCheckTaint(virQEMUDriverPtr driver,
9986
                             virDomainObjPtr obj,
9987
                             qemuDomainLogContextPtr logCtxt)
9988
{
9989
    size_t i;
9990
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
9991
    qemuDomainObjPrivatePtr priv = obj->privateData;
9992
    bool custom_hypervisor_feat = false;
9993

9994
    if (virQEMUDriverIsPrivileged(driver) &&
9995
        (cfg->user == 0 ||
9996
         cfg->group == 0))
9997
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
9998

9999
    if (priv->hookRun)
10000
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOOK, logCtxt);
10001

10002
    if (obj->def->namespaceData) {
10003 10004
        qemuDomainXmlNsDefPtr qemuxmlns = obj->def->namespaceData;
        if (qemuxmlns->num_args || qemuxmlns->num_env)
10005
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logCtxt);
10006 10007 10008 10009
        if (qemuxmlns->ncapsadd > 0 || qemuxmlns->ncapsdel > 0)
            custom_hypervisor_feat = true;
    }

10010 10011
    if (custom_hypervisor_feat ||
        (cfg->capabilityfilters && *cfg->capabilityfilters)) {
10012 10013
        qemuDomainObjTaint(driver, obj,
                           VIR_DOMAIN_TAINT_CUSTOM_HYPERVISOR_FEATURE, logCtxt);
10014 10015
    }

10016
    if (obj->def->cpu && obj->def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
10017
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOST_CPU, logCtxt);
10018

10019
    for (i = 0; i < obj->def->ndisks; i++)
10020
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logCtxt);
10021

10022 10023
    for (i = 0; i < obj->def->nhostdevs; i++)
        qemuDomainObjCheckHostdevTaint(driver, obj, obj->def->hostdevs[i],
10024
                                       logCtxt);
10025

10026
    for (i = 0; i < obj->def->nnets; i++)
10027
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logCtxt);
10028

10029
    if (obj->def->os.dtb)
10030
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_DTB, logCtxt);
10031

10032
    virObjectUnref(cfg);
10033 10034 10035
}


10036
void qemuDomainObjCheckDiskTaint(virQEMUDriverPtr driver,
10037
                                 virDomainObjPtr obj,
10038
                                 virDomainDiskDefPtr disk,
10039
                                 qemuDomainLogContextPtr logCtxt)
10040
{
10041
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
10042

10043 10044
    if (disk->rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
10045
                           logCtxt);
10046

10047 10048
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
        virStorageSourceGetActualType(disk->src) == VIR_STORAGE_TYPE_BLOCK &&
10049
        disk->src->path && virFileIsCDROM(disk->src->path) == 1)
10050
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CDROM_PASSTHROUGH,
10051
                           logCtxt);
10052

10053
    virObjectUnref(cfg);
10054 10055 10056
}


10057 10058 10059
void qemuDomainObjCheckHostdevTaint(virQEMUDriverPtr driver,
                                    virDomainObjPtr obj,
                                    virDomainHostdevDefPtr hostdev,
10060
                                    qemuDomainLogContextPtr logCtxt)
10061
{
10062
    if (!virHostdevIsSCSIDevice(hostdev))
10063
        return;
10064

10065 10066
    if (hostdev->source.subsys.u.scsi.rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
10067 10068 10069
}


10070
void qemuDomainObjCheckNetTaint(virQEMUDriverPtr driver,
10071
                                virDomainObjPtr obj,
10072
                                virDomainNetDefPtr net,
10073
                                qemuDomainLogContextPtr logCtxt)
10074
{
10075 10076 10077 10078 10079 10080
    /* 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)
10081
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logCtxt);
10082
}
10083 10084


10085 10086 10087 10088 10089 10090 10091
qemuDomainLogContextPtr qemuDomainLogContextNew(virQEMUDriverPtr driver,
                                                virDomainObjPtr vm,
                                                qemuDomainLogContextMode mode)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainLogContextPtr ctxt = NULL;

10092
    if (qemuDomainInitialize() < 0)
10093 10094 10095 10096
        goto cleanup;

    if (!(ctxt = virObjectNew(qemuDomainLogContextClass)))
        goto cleanup;
10097

10098
    VIR_DEBUG("Context new %p stdioLogD=%d", ctxt, cfg->stdioLogD);
10099 10100 10101
    ctxt->writefd = -1;
    ctxt->readfd = -1;

10102
    ctxt->path = g_strdup_printf("%s/%s.log", cfg->logDir, vm->def->name);
10103

10104 10105 10106 10107
    if (cfg->stdioLogD) {
        ctxt->manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver));
        if (!ctxt->manager)
            goto error;
10108

10109 10110 10111 10112
        ctxt->writefd = virLogManagerDomainOpenLogFile(ctxt->manager,
                                                       "qemu",
                                                       vm->def->uuid,
                                                       vm->def->name,
10113
                                                       ctxt->path,
10114 10115 10116 10117 10118 10119
                                                       0,
                                                       &ctxt->inode,
                                                       &ctxt->pos);
        if (ctxt->writefd < 0)
            goto error;
    } else {
10120
        if ((ctxt->writefd = open(ctxt->path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
10121
            virReportSystemError(errno, _("failed to create logfile %s"),
10122
                                 ctxt->path);
10123 10124
            goto error;
        }
10125
        if (virSetCloseExec(ctxt->writefd) < 0) {
10126
            virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
10127
                                 ctxt->path);
10128 10129 10130
            goto error;
        }

10131 10132 10133 10134 10135 10136 10137
        /* 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"),
10138
                                 ctxt->path);
10139 10140 10141 10142
            goto error;
        }

        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START) {
10143
            if ((ctxt->readfd = open(ctxt->path, O_RDONLY, S_IRUSR | S_IWUSR)) < 0) {
10144
                virReportSystemError(errno, _("failed to open logfile %s"),
10145
                                     ctxt->path);
10146 10147 10148 10149
                goto error;
            }
            if (virSetCloseExec(ctxt->readfd) < 0) {
                virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
10150
                                     ctxt->path);
10151 10152 10153 10154
                goto error;
            }
        }

10155 10156
        if ((ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END)) < 0) {
            virReportSystemError(errno, _("failed to seek in log file %s"),
10157
                                 ctxt->path);
10158 10159 10160 10161
            goto error;
        }
    }

10162
 cleanup:
10163 10164 10165 10166
    virObjectUnref(cfg);
    return ctxt;

 error:
10167
    virObjectUnref(ctxt);
10168 10169
    ctxt = NULL;
    goto cleanup;
10170 10171 10172 10173 10174 10175 10176 10177 10178 10179 10180 10181
}


int qemuDomainLogContextWrite(qemuDomainLogContextPtr ctxt,
                              const char *fmt, ...)
{
    va_list argptr;
    char *message = NULL;
    int ret = -1;

    va_start(argptr, fmt);

10182
    message = g_strdup_vprintf(fmt, argptr);
10183 10184
    if (!ctxt->manager &&
        lseek(ctxt->writefd, 0, SEEK_END) < 0) {
10185
        virReportSystemError(errno, "%s",
10186
                             _("Unable to seek to end of domain logfile"));
10187 10188 10189 10190 10191 10192 10193 10194 10195 10196 10197 10198 10199 10200 10201 10202 10203 10204 10205 10206
        goto cleanup;
    }
    if (safewrite(ctxt->writefd, message, strlen(message)) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to write to domain logfile"));
        goto cleanup;
    }

    ret = 0;

 cleanup:
    va_end(argptr);
    VIR_FREE(message);
    return ret;
}


ssize_t qemuDomainLogContextRead(qemuDomainLogContextPtr ctxt,
                                 char **msg)
{
10207 10208 10209 10210
    VIR_DEBUG("Context read %p manager=%p inode=%llu pos=%llu",
              ctxt, ctxt->manager,
              (unsigned long long)ctxt->inode,
              (unsigned long long)ctxt->pos);
10211
    char *buf;
10212 10213 10214
    size_t buflen;
    if (ctxt->manager) {
        buf = virLogManagerDomainReadLogFile(ctxt->manager,
10215
                                             ctxt->path,
10216 10217 10218 10219 10220 10221 10222 10223 10224
                                             ctxt->inode,
                                             ctxt->pos,
                                             1024 * 128,
                                             0);
        if (!buf)
            return -1;
        buflen = strlen(buf);
    } else {
        ssize_t got;
10225

10226
        buflen = 1024 * 128;
10227

10228 10229
        /* Best effort jump to start of messages */
        ignore_value(lseek(ctxt->readfd, ctxt->pos, SEEK_SET));
10230

10231 10232
        if (VIR_ALLOC_N(buf, buflen) < 0)
            return -1;
10233

10234 10235 10236 10237 10238 10239 10240 10241 10242 10243 10244 10245 10246
        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;
    }
10247 10248 10249

    *msg = buf;

10250
    return buflen;
10251 10252 10253
}


10254 10255 10256 10257 10258 10259 10260 10261 10262 10263 10264 10265 10266 10267 10268 10269 10270 10271 10272 10273 10274
/**
 * qemuDomainLogAppendMessage:
 *
 * This is a best-effort attempt to add a log message to the qemu log file
 * either by using virtlogd or the legacy approach */
int
qemuDomainLogAppendMessage(virQEMUDriverPtr driver,
                           virDomainObjPtr vm,
                           const char *fmt,
                           ...)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    virLogManagerPtr manager = NULL;
    va_list ap;
    char *path = NULL;
    int writefd = -1;
    char *message = NULL;
    int ret = -1;

    va_start(ap, fmt);

10275
    message = g_strdup_vprintf(fmt, ap);
10276 10277 10278 10279

    VIR_DEBUG("Append log message (vm='%s' message='%s) stdioLogD=%d",
              vm->def->name, message, cfg->stdioLogD);

10280
    path = g_strdup_printf("%s/%s.log", cfg->logDir, vm->def->name);
10281 10282 10283 10284 10285 10286 10287 10288 10289 10290 10291 10292 10293 10294 10295 10296 10297 10298 10299 10300 10301 10302 10303 10304 10305 10306 10307 10308 10309 10310 10311 10312 10313

    if (cfg->stdioLogD) {
        if (!(manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver))))
            goto cleanup;

        if (virLogManagerDomainAppendMessage(manager, "qemu", vm->def->uuid,
                                             vm->def->name, path, message, 0) < 0)
            goto cleanup;
    } else {
        if ((writefd = open(path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
            virReportSystemError(errno, _("failed to create logfile %s"),
                                 path);
            goto cleanup;
        }

        if (safewrite(writefd, message, strlen(message)) < 0)
            goto cleanup;
    }

    ret = 0;

 cleanup:
    va_end(ap);
    VIR_FREE(message);
    VIR_FORCE_CLOSE(writefd);
    virLogManagerFree(manager);
    virObjectUnref(cfg);
    VIR_FREE(path);

    return ret;
}


10314 10315 10316 10317 10318 10319 10320 10321
int qemuDomainLogContextGetWriteFD(qemuDomainLogContextPtr ctxt)
{
    return ctxt->writefd;
}


void qemuDomainLogContextMarkPosition(qemuDomainLogContextPtr ctxt)
{
10322 10323
    if (ctxt->manager)
        virLogManagerDomainGetLogFilePosition(ctxt->manager,
10324
                                              ctxt->path,
10325 10326 10327 10328 10329
                                              0,
                                              &ctxt->inode,
                                              &ctxt->pos);
    else
        ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END);
10330 10331 10332
}


10333 10334 10335 10336 10337 10338
virLogManagerPtr qemuDomainLogContextGetManager(qemuDomainLogContextPtr ctxt)
{
    return ctxt->manager;
}


10339 10340
/* Locate an appropriate 'qemu-img' binary.  */
const char *
10341
qemuFindQemuImgBinary(virQEMUDriverPtr driver)
10342
{
10343 10344
    if (!driver->qemuImgBinary)
        virReportError(VIR_ERR_INTERNAL_ERROR,
10345
                       "%s", _("unable to find qemu-img"));
10346 10347 10348 10349 10350 10351

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
10352
                                virDomainMomentObjPtr snapshot,
10353
                                virDomainXMLOptionPtr xmlopt,
10354
                                const char *snapshotDir)
10355 10356 10357 10358 10359 10360
{
    char *newxml = NULL;
    int ret = -1;
    char *snapDir = NULL;
    char *snapFile = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
10361 10362
    unsigned int flags = VIR_DOMAIN_SNAPSHOT_FORMAT_SECURE |
        VIR_DOMAIN_SNAPSHOT_FORMAT_INTERNAL;
10363
    virDomainSnapshotDefPtr def = virDomainSnapshotObjGetDef(snapshot);
10364

10365
    if (virDomainSnapshotGetCurrent(vm->snapshots) == snapshot)
10366
        flags |= VIR_DOMAIN_SNAPSHOT_FORMAT_CURRENT;
10367
    virUUIDFormat(vm->def->uuid, uuidstr);
10368
    newxml = virDomainSnapshotDefFormat(uuidstr, def, xmlopt, flags);
10369
    if (newxml == NULL)
10370 10371
        return -1;

10372
    snapDir = g_strdup_printf("%s/%s", snapshotDir, vm->def->name);
10373 10374 10375 10376 10377 10378
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
        goto cleanup;
    }

10379
    snapFile = g_strdup_printf("%s/%s.xml", snapDir, def->parent.name);
10380

J
Ján Tomko 已提交
10381
    ret = virXMLSaveFile(snapFile, NULL, "snapshot-edit", newxml);
10382

10383
 cleanup:
10384 10385 10386 10387 10388 10389
    VIR_FREE(snapFile);
    VIR_FREE(snapDir);
    VIR_FREE(newxml);
    return ret;
}

10390

10391 10392
/* The domain is expected to be locked and inactive. Return -1 on normal
 * failure, 1 if we skipped a disk due to try_all.  */
10393
static int
10394
qemuDomainSnapshotForEachQcow2Raw(virQEMUDriverPtr driver,
10395 10396 10397 10398 10399
                                  virDomainDefPtr def,
                                  const char *name,
                                  const char *op,
                                  bool try_all,
                                  int ndisks)
10400 10401
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
10402
    size_t i;
10403 10404 10405 10406 10407 10408 10409 10410 10411
    bool skipped = false;

    qemuimgarg[0] = qemuFindQemuImgBinary(driver);
    if (qemuimgarg[0] == NULL) {
        /* qemuFindQemuImgBinary set the error */
        return -1;
    }

    qemuimgarg[2] = op;
10412
    qemuimgarg[3] = name;
10413

10414
    for (i = 0; i < ndisks; i++) {
10415
        /* FIXME: we also need to handle LVM here */
10416
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
10417 10418 10419
            int format = virDomainDiskGetFormat(def->disks[i]);

            if (format > 0 && format != VIR_STORAGE_FILE_QCOW2) {
10420 10421 10422 10423 10424
                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",
10425
                             def->disks[i]->dst);
10426 10427
                    skipped = true;
                    continue;
10428 10429 10430 10431 10432
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
10433
                }
10434 10435 10436 10437
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("Disk device '%s' does not support"
                                 " snapshotting"),
                               def->disks[i]->dst);
10438 10439 10440
                return -1;
            }

10441
            qemuimgarg[4] = virDomainDiskGetSource(def->disks[i]);
10442 10443 10444 10445

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
10446
                             def->disks[i]->dst);
10447 10448
                    skipped = true;
                    continue;
10449 10450 10451 10452 10453
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
10454 10455 10456 10457 10458 10459 10460 10461 10462
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

10463 10464 10465
/* 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
10466
qemuDomainSnapshotForEachQcow2(virQEMUDriverPtr driver,
10467
                               virDomainObjPtr vm,
10468
                               virDomainMomentObjPtr snap,
10469 10470 10471 10472 10473 10474
                               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.  */
10475
    virDomainDefPtr def = snap->def->dom;
10476 10477 10478

    if (!def)
        def = vm->def;
10479
    return qemuDomainSnapshotForEachQcow2Raw(driver, def, snap->def->name,
10480 10481 10482
                                             op, try_all, def->ndisks);
}

10483 10484
/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
10485
qemuDomainSnapshotDiscard(virQEMUDriverPtr driver,
10486
                          virDomainObjPtr vm,
10487
                          virDomainMomentObjPtr snap,
10488
                          bool update_parent,
10489 10490 10491 10492 10493
                          bool metadata_only)
{
    char *snapFile = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;
10494
    virDomainMomentObjPtr parentsnap = NULL;
10495
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
10496 10497 10498 10499 10500 10501 10502 10503 10504

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
                goto cleanup;
        } else {
            priv = vm->privateData;
10505
            qemuDomainObjEnterMonitor(driver, vm);
10506
            /* we continue on even in the face of error */
10507
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
10508
            ignore_value(qemuDomainObjExitMonitor(driver, vm));
10509 10510 10511
        }
    }

10512 10513
    snapFile = g_strdup_printf("%s/%s/%s.xml", cfg->snapshotDir, vm->def->name,
                               snap->def->name);
10514

10515 10516
    if (snap == virDomainSnapshotGetCurrent(vm->snapshots)) {
        virDomainSnapshotSetCurrent(vm->snapshots, NULL);
10517
        if (update_parent && snap->def->parent_name) {
10518
            parentsnap = virDomainSnapshotFindByName(vm->snapshots,
10519
                                                     snap->def->parent_name);
10520 10521
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
10522
                         snap->def->parent_name);
10523
            } else {
10524
                virDomainSnapshotSetCurrent(vm->snapshots, parentsnap);
10525
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap,
10526
                                                    driver->xmlopt,
10527
                                                    cfg->snapshotDir) < 0) {
10528
                    VIR_WARN("failed to set parent snapshot '%s' as current",
10529
                             snap->def->parent_name);
10530
                    virDomainSnapshotSetCurrent(vm->snapshots, NULL);
10531 10532 10533 10534 10535 10536 10537
                }
            }
        }
    }

    if (unlink(snapFile) < 0)
        VIR_WARN("Failed to unlink %s", snapFile);
10538
    if (update_parent)
10539
        virDomainMomentDropParent(snap);
10540
    virDomainSnapshotObjListRemove(vm->snapshots, snap);
10541 10542 10543

    ret = 0;

10544
 cleanup:
10545
    VIR_FREE(snapFile);
10546
    virObjectUnref(cfg);
10547 10548 10549 10550
    return ret;
}

/* Hash iterator callback to discard multiple snapshots.  */
10551
int qemuDomainMomentDiscardAll(void *payload,
J
Ján Tomko 已提交
10552
                               const void *name G_GNUC_UNUSED,
10553
                               void *data)
10554
{
10555 10556
    virDomainMomentObjPtr moment = payload;
    virQEMUMomentRemovePtr curr = data;
10557 10558
    int err;

10559 10560 10561 10562
    if (!curr->found && curr->current == moment)
        curr->found = true;
    err = curr->momentDiscard(curr->driver, curr->vm, moment, false,
                              curr->metadata_only);
10563 10564
    if (err && !curr->err)
        curr->err = err;
10565
    return 0;
10566 10567 10568
}

int
10569
qemuDomainSnapshotDiscardAllMetadata(virQEMUDriverPtr driver,
10570 10571
                                     virDomainObjPtr vm)
{
10572 10573 10574
    virQEMUMomentRemove rem = {
        .driver = driver,
        .vm = vm,
10575 10576
        .metadata_only = true,
        .momentDiscard = qemuDomainSnapshotDiscard,
10577
    };
10578

10579
    virDomainSnapshotForEach(vm->snapshots, qemuDomainMomentDiscardAll, &rem);
10580
    virDomainSnapshotObjListRemoveAll(vm->snapshots);
10581 10582 10583 10584

    return rem.err;
}

10585

10586 10587 10588
static void
qemuDomainRemoveInactiveCommon(virQEMUDriverPtr driver,
                               virDomainObjPtr vm)
10589
{
10590
    virQEMUDriverConfigPtr cfg;
10591 10592
    g_autofree char *snapDir = NULL;
    g_autofree char *chkDir = NULL;
10593 10594

    cfg = virQEMUDriverGetConfig(driver);
10595

10596 10597 10598 10599
    /* 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);
10600 10601 10602 10603 10604
    } 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);
10605
    }
10606
    /* Remove any checkpoint metadata prior to removing the domain */
10607
    if (qemuCheckpointDiscardAllMetadata(driver, vm) < 0) {
10608 10609
        VIR_WARN("unable to remove all checkpoints for domain %s",
                 vm->def->name);
10610 10611 10612 10613 10614
    } 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);
10615
    }
10616
    qemuExtDevicesCleanupHost(driver, vm->def);
10617

10618
    virObjectUnref(cfg);
10619 10620 10621
}


10622 10623 10624 10625 10626 10627 10628 10629 10630 10631 10632 10633 10634 10635 10636 10637 10638 10639 10640 10641
/**
 * 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);
}


10642 10643 10644 10645 10646 10647 10648 10649 10650 10651 10652 10653 10654 10655 10656 10657 10658 10659 10660 10661 10662 10663
/**
 * 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);
}


10664 10665 10666 10667 10668 10669 10670 10671 10672 10673 10674 10675 10676 10677 10678 10679 10680
/**
 * 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);
10681 10682

    if (haveJob)
10683
        qemuDomainObjEndJob(driver, vm);
10684
}
10685

10686

10687 10688 10689 10690 10691 10692 10693 10694 10695 10696 10697 10698 10699 10700 10701 10702 10703 10704 10705 10706 10707
/**
 * 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);
}


10708
void
10709
qemuDomainSetFakeReboot(virQEMUDriverPtr driver,
10710 10711 10712 10713
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
10714
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
10715 10716

    if (priv->fakeReboot == value)
10717
        goto cleanup;
10718 10719 10720

    priv->fakeReboot = value;

10721
    if (virDomainObjSave(vm, driver->xmlopt, cfg->stateDir) < 0)
10722
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
10723

10724
 cleanup:
10725
    virObjectUnref(cfg);
10726
}
M
Michal Privoznik 已提交
10727

10728
static void
10729 10730
qemuDomainCheckRemoveOptionalDisk(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
10731
                                  size_t diskIndex)
10732 10733
{
    char uuid[VIR_UUID_STRING_BUFLEN];
10734
    virObjectEventPtr event = NULL;
10735
    virDomainDiskDefPtr disk = vm->def->disks[diskIndex];
10736
    const char *src = virDomainDiskGetSource(disk);
10737 10738 10739 10740 10741

    virUUIDFormat(vm->def->uuid, uuid);

    VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
              "due to inaccessible source '%s'",
10742
              disk->dst, vm->def->name, uuid, src);
10743 10744 10745 10746

    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM ||
        disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY) {

10747
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
10748 10749
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
10750
        virDomainDiskEmptySource(disk);
10751 10752
        /* keeping the old startup policy would be invalid for new images */
        disk->startupPolicy = VIR_DOMAIN_STARTUP_POLICY_DEFAULT;
10753
    } else {
10754
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
10755 10756
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_DROP_MISSING_ON_START);
10757 10758
        virDomainDiskRemove(vm->def, diskIndex);
        virDomainDiskDefFree(disk);
10759 10760
    }

10761
    virObjectEventStateQueue(driver->domainEventState, event);
10762 10763
}

10764 10765 10766 10767 10768 10769 10770 10771 10772 10773 10774 10775 10776 10777 10778 10779 10780

/**
 * 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
10781 10782
qemuDomainCheckDiskStartupPolicy(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
10783
                                 size_t diskIndex,
10784 10785
                                 bool cold_boot)
{
10786
    int startupPolicy = vm->def->disks[diskIndex]->startupPolicy;
10787
    int device = vm->def->disks[diskIndex]->device;
10788

10789
    switch ((virDomainStartupPolicy) startupPolicy) {
10790
        case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
10791 10792 10793 10794 10795 10796
            /* 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)
10797
                return -1;
10798 10799
            break;

10800
        case VIR_DOMAIN_STARTUP_POLICY_DEFAULT:
10801
        case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
10802
            return -1;
10803 10804

        case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
10805
            if (cold_boot)
10806
                return -1;
10807 10808 10809 10810 10811 10812 10813
            break;

        case VIR_DOMAIN_STARTUP_POLICY_LAST:
            /* this should never happen */
            break;
    }

10814
    qemuDomainCheckRemoveOptionalDisk(driver, vm, diskIndex);
10815
    virResetLastError();
10816 10817 10818
    return 0;
}

10819

10820 10821 10822 10823 10824 10825 10826 10827 10828 10829

/*
 * The vm must be locked when any of the following cleanup functions is
 * called.
 */
int
qemuDomainCleanupAdd(virDomainObjPtr vm,
                     qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
10830
    size_t i;
10831 10832 10833 10834 10835 10836 10837 10838 10839 10840

    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,
10841
                     priv->ncleanupCallbacks, 1) < 0)
10842 10843 10844 10845 10846 10847 10848 10849 10850 10851 10852
        return -1;

    priv->cleanupCallbacks[priv->ncleanupCallbacks++] = cb;
    return 0;
}

void
qemuDomainCleanupRemove(virDomainObjPtr vm,
                        qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
10853
    size_t i;
10854 10855 10856 10857

    VIR_DEBUG("vm=%s, cb=%p", vm->def->name, cb);

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
10858 10859 10860
        if (priv->cleanupCallbacks[i] == cb)
            VIR_DELETE_ELEMENT_INPLACE(priv->cleanupCallbacks,
                                       i, priv->ncleanupCallbacks);
10861 10862 10863 10864 10865 10866 10867 10868
    }

    VIR_SHRINK_N(priv->cleanupCallbacks,
                 priv->ncleanupCallbacks_max,
                 priv->ncleanupCallbacks_max - priv->ncleanupCallbacks);
}

void
10869
qemuDomainCleanupRun(virQEMUDriverPtr driver,
10870 10871 10872
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
10873
    size_t i;
10874 10875 10876 10877

    VIR_DEBUG("driver=%p, vm=%s", driver, vm->def->name);

    /* run cleanup callbacks in reverse order */
10878 10879
    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[priv->ncleanupCallbacks - (i + 1)])
10880 10881 10882 10883 10884 10885 10886
            priv->cleanupCallbacks[i](driver, vm);
    }

    VIR_FREE(priv->cleanupCallbacks);
    priv->ncleanupCallbacks = 0;
    priv->ncleanupCallbacks_max = 0;
}
10887

10888
void
10889 10890
qemuDomainGetImageIds(virQEMUDriverConfigPtr cfg,
                      virDomainObjPtr vm,
10891
                      virStorageSourcePtr src,
10892
                      virStorageSourcePtr parentSrc,
10893 10894 10895 10896 10897 10898 10899 10900 10901 10902 10903 10904 10905 10906 10907 10908 10909 10910
                      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;
    }

10911 10912
    if (vm && (vmlabel = virDomainDefGetSecurityLabelDef(vm->def, "dac")) &&
        vmlabel->label)
10913 10914
        virParseOwnershipIds(vmlabel->label, uid, gid);

10915 10916 10917 10918 10919
    if (parentSrc &&
        (disklabel = virStorageSourceGetSecurityLabelDef(parentSrc, "dac")) &&
        disklabel->label)
        virParseOwnershipIds(disklabel->label, uid, gid);

10920
    if ((disklabel = virStorageSourceGetSecurityLabelDef(src, "dac")) &&
10921
        disklabel->label)
10922 10923 10924 10925
        virParseOwnershipIds(disklabel->label, uid, gid);
}


10926 10927 10928
int
qemuDomainStorageFileInit(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
10929 10930
                          virStorageSourcePtr src,
                          virStorageSourcePtr parent)
10931 10932 10933 10934 10935 10936
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    uid_t uid;
    gid_t gid;
    int ret = -1;

10937
    qemuDomainGetImageIds(cfg, vm, src, parent, &uid, &gid);
10938 10939 10940 10941 10942 10943 10944 10945 10946 10947 10948 10949

    if (virStorageFileInitAs(src, uid, gid) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
}


10950 10951 10952 10953 10954
char *
qemuDomainStorageAlias(const char *device, int depth)
{
    char *alias;

10955
    device = qemuAliasDiskDriveSkipPrefix(device);
10956 10957

    if (!depth)
10958
        alias = g_strdup(device);
10959
    else
10960
        alias = g_strdup_printf("%s.%d", device, depth);
10961 10962 10963 10964
    return alias;
}


10965 10966 10967 10968 10969 10970 10971 10972 10973 10974 10975 10976 10977 10978 10979 10980 10981 10982 10983 10984 10985 10986 10987 10988 10989 10990 10991 10992 10993 10994 10995 10996 10997 10998 10999 11000 11001 11002 11003 11004 11005 11006 11007 11008 11009 11010 11011
/**
 * 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;
}


11012 11013 11014 11015 11016 11017 11018 11019 11020 11021 11022 11023 11024 11025 11026 11027 11028 11029 11030 11031 11032 11033 11034 11035 11036 11037
/**
 * 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;
    }
}


11038 11039 11040 11041 11042 11043 11044 11045 11046 11047 11048 11049 11050
/**
 * 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.
 */
11051
int
11052
qemuDomainDetermineDiskChain(virQEMUDriverPtr driver,
11053
                             virDomainObjPtr vm,
11054
                             virDomainDiskDefPtr disk,
11055
                             virStorageSourcePtr disksrc,
11056
                             bool report_broken)
11057
{
11058
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
11059 11060
    virStorageSourcePtr src; /* iterator for the backing chain declared in XML */
    virStorageSourcePtr n; /* iterator for the backing chain detected from disk */
11061
    qemuDomainObjPrivatePtr priv = vm->privateData;
11062 11063
    uid_t uid;
    gid_t gid;
11064

11065 11066 11067
    if (!disksrc)
        disksrc = disk->src;

11068 11069
    if (virStorageSourceIsEmpty(disksrc))
        return 0;
11070

11071 11072
    /* There is no need to check the backing chain for disks without backing
     * support */
11073 11074 11075
    if (virStorageSourceIsLocalStorage(disksrc) &&
        disksrc->format > VIR_STORAGE_FILE_NONE &&
        disksrc->format < VIR_STORAGE_FILE_BACKING) {
11076

11077
        if (!virFileExists(disksrc->path)) {
11078
            if (report_broken)
11079
                virStorageFileReportBrokenChain(errno, disksrc, disksrc);
11080

11081
            return -1;
11082 11083
        }

11084
        /* terminate the chain for such images as the code below would do */
11085
        if (!disksrc->backingStore &&
11086
            !(disksrc->backingStore = virStorageSourceNew()))
11087
            return -1;
11088

11089 11090 11091
        /* 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 &&
11092 11093 11094 11095
            disksrc->format == VIR_STORAGE_FILE_RAW &&
            virStorageSourceIsBlockLocal(disksrc) &&
            virFileIsCDROM(disksrc->path) == 1)
            disksrc->hostcdrom = true;
11096

11097
        return 0;
11098 11099
    }

11100
    src = disksrc;
11101 11102
    /* skip to the end of the chain if there is any */
    while (virStorageSourceHasBacking(src)) {
11103 11104
        if (report_broken) {
            int rv = virStorageFileSupportsAccess(src);
11105

11106
            if (rv < 0)
11107
                return -1;
11108

11109
            if (rv > 0) {
11110
                if (qemuDomainStorageFileInit(driver, vm, src, disksrc) < 0)
11111
                    return -1;
11112 11113

                if (virStorageFileAccess(src, F_OK) < 0) {
11114
                    virStorageFileReportBrokenChain(errno, src, disksrc);
11115
                    virStorageFileDeinit(src);
11116
                    return -1;
11117 11118
                }

11119
                virStorageFileDeinit(src);
11120 11121
            }
        }
11122
        src = src->backingStore;
11123 11124 11125 11126
    }

    /* We skipped to the end of the chain. Skip detection if there's the
     * terminator. (An allocated but empty backingStore) */
11127 11128 11129 11130
    if (src->backingStore) {
        if (qemuDomainStorageSourceValidateDepth(disksrc, 0, disk->dst) < 0)
            return -1;

11131
        return 0;
11132
    }
11133

11134
    qemuDomainGetImageIds(cfg, vm, src, disksrc, &uid, &gid);
11135

11136
    if (virStorageFileGetMetadata(src, uid, gid, report_broken) < 0)
11137
        return -1;
11138

11139
    for (n = src->backingStore; virStorageSourceIsBacking(n); n = n->backingStore) {
11140
        if (qemuDomainValidateStorageSource(n, priv->qemuCaps) < 0)
11141
            return -1;
11142

11143
        qemuDomainPrepareStorageSourceConfig(n, cfg, priv->qemuCaps);
11144
        qemuDomainPrepareDiskSourceData(disk, n);
11145 11146 11147

        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV) &&
            qemuDomainPrepareStorageSourceBlockdev(disk, n, priv, cfg) < 0)
11148
            return -1;
11149
    }
11150

11151 11152 11153
    if (qemuDomainStorageSourceValidateDepth(disksrc, 0, disk->dst) < 0)
        return -1;

11154
    return 0;
11155
}
11156

11157

11158 11159 11160 11161 11162 11163 11164 11165 11166 11167 11168 11169 11170 11171 11172 11173
/**
 * 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,
11174
                              virQEMUCapsPtr qemuCaps,
11175 11176
                              char **backendAlias)
{
11177 11178
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);
    const char *nodename = NULL;
11179 11180
    *backendAlias = NULL;

11181 11182 11183 11184 11185 11186 11187 11188 11189 11190 11191 11192 11193 11194 11195
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if (!(*backendAlias = qemuAliasDiskDriveFromDisk(disk)))
            return -1;

        return 0;
    }

    if (virStorageSourceIsEmpty(disk->src))
        return 0;

    if (disk->copy_on_read == VIR_TRISTATE_SWITCH_ON)
        nodename = priv->nodeCopyOnRead;
    else
        nodename = disk->src->nodeformat;

11196
    *backendAlias = g_strdup(nodename);
11197 11198 11199 11200 11201

    return 0;
}


11202 11203 11204
typedef enum {
    /* revoke access to the image instead of allowing it */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE = 1 << 0,
11205 11206
    /* operate on full backing chain rather than single image */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN = 1 << 1,
11207 11208 11209 11210
    /* 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,
11211
    /* don't revoke permissions when modification has failed */
11212
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE = 1 << 4,
11213
    /* VM already has access to the source and we are just modifying it */
11214
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS = 1 << 5,
11215 11216 11217
} qemuDomainStorageSourceAccessFlags;


11218
/**
11219
 * qemuDomainStorageSourceAccessModify:
11220 11221 11222
 * @driver: qemu driver struct
 * @vm: domain object
 * @src: Source to prepare
11223
 * @flags: bitwise or of qemuDomainStorageSourceAccessFlags
11224 11225
 *
 * Setup the locks, cgroups and security permissions on a disk source and its
11226
 * backing chain.
11227 11228 11229
 *
 * Returns 0 on success and -1 on error. Reports libvirt error.
 */
11230
static int
11231 11232 11233
qemuDomainStorageSourceAccessModify(virQEMUDriverPtr driver,
                                    virDomainObjPtr vm,
                                    virStorageSourcePtr src,
11234
                                    qemuDomainStorageSourceAccessFlags flags)
11235
{
11236
    g_autoptr(virQEMUDriverConfig) cfg = virQEMUDriverGetConfig(driver);
11237 11238 11239
    const char *srcstr = NULLSTR(src->path);
    int ret = -1;
    virErrorPtr orig_err = NULL;
11240
    bool chain = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
11241 11242 11243
    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;
11244
    int rc;
11245
    bool was_readonly = src->readonly;
11246 11247 11248 11249
    bool revoke_cgroup = false;
    bool revoke_label = false;
    bool revoke_namespace = false;
    bool revoke_lockspace = false;
11250

11251 11252 11253 11254
    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)
11255
        src->readonly = true;
11256

11257
    if (force_rw)
11258 11259
        src->readonly = false;

11260
    /* just tear down the disk access */
11261
    if (revoke) {
11262
        virErrorPreserveLast(&orig_err);
11263 11264 11265 11266
        revoke_cgroup = true;
        revoke_label = true;
        revoke_namespace = true;
        revoke_lockspace = true;
11267
        ret = 0;
11268
        goto revoke;
11269 11270 11271
    }

    if (virDomainLockImageAttach(driver->lockManager, cfg->uri, vm, src) < 0)
11272 11273 11274
        goto revoke;

    revoke_lockspace = true;
11275

11276 11277 11278 11279
    /* When modifying access of existing @src namespace does not need update */
    if (!(flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS)) {
        if (qemuDomainNamespaceSetupDisk(vm, src) < 0)
            goto revoke;
11280

11281 11282
        revoke_namespace = true;
    }
11283

11284
    if (qemuSecuritySetImageLabel(driver, vm, src, chain) < 0)
11285 11286 11287
        goto revoke;

    revoke_label = true;
11288

11289 11290 11291 11292 11293 11294
    if (chain)
        rc = qemuSetupImageChainCgroup(vm, src);
    else
        rc = qemuSetupImageCgroup(vm, src);

    if (rc < 0)
11295 11296 11297
        goto revoke;

    revoke_cgroup = true;
11298 11299 11300 11301

    ret = 0;
    goto cleanup;

11302
 revoke:
11303 11304 11305
    if (flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE)
        goto cleanup;

11306 11307 11308 11309 11310
    if (revoke_cgroup) {
        if (chain)
            rc = qemuTeardownImageChainCgroup(vm, src);
        else
            rc = qemuTeardownImageCgroup(vm, src);
11311

11312 11313 11314 11315 11316 11317 11318 11319 11320 11321 11322 11323 11324 11325 11326 11327 11328 11329
        if (rc < 0)
            VIR_WARN("Unable to tear down cgroup access on %s", srcstr);
    }

    if (revoke_label) {
        if (qemuSecurityRestoreImageLabel(driver, vm, src, chain) < 0)
            VIR_WARN("Unable to restore security label on %s", srcstr);
    }

    if (revoke_namespace) {
        if (qemuDomainNamespaceTeardownDisk(vm, src) < 0)
            VIR_WARN("Unable to remove /dev entry for %s", srcstr);
    }

    if (revoke_lockspace) {
        if (virDomainLockImageDetach(driver->lockManager, vm, src) < 0)
            VIR_WARN("Unable to release lock on %s", srcstr);
    }
11330 11331

 cleanup:
11332
    src->readonly = was_readonly;
11333 11334 11335 11336 11337 11338
    virErrorRestore(&orig_err);

    return ret;
}


11339 11340 11341 11342 11343
int
qemuDomainStorageSourceChainAccessAllow(virQEMUDriverPtr driver,
                                        virDomainObjPtr vm,
                                        virStorageSourcePtr src)
{
11344
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
11345 11346

    return qemuDomainStorageSourceAccessModify(driver, vm, src, flags);
11347 11348 11349 11350 11351 11352 11353 11354
}


int
qemuDomainStorageSourceChainAccessRevoke(virQEMUDriverPtr driver,
                                         virDomainObjPtr vm,
                                         virStorageSourcePtr src)
{
11355 11356
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE |
                                               QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
11357 11358

    return qemuDomainStorageSourceAccessModify(driver, vm, src, flags);
11359 11360 11361
}


11362
/**
11363
 * qemuDomainStorageSourceAccessRevoke:
11364 11365 11366 11367 11368
 *
 * Revoke access to a single backing chain element. This restores the labels,
 * removes cgroup ACLs for devices and removes locks.
 */
void
11369 11370 11371
qemuDomainStorageSourceAccessRevoke(virQEMUDriverPtr driver,
                                    virDomainObjPtr vm,
                                    virStorageSourcePtr elem)
11372
{
11373
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE;
11374

11375
    ignore_value(qemuDomainStorageSourceAccessModify(driver, vm, elem, flags));
11376 11377 11378 11379
}


/**
11380
 * qemuDomainStorageSourceAccessAllow:
11381 11382 11383 11384 11385
 * @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
11386 11387 11388
 *
 * 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
11389 11390 11391 11392 11393
 * 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.
 */
11394
int
11395 11396 11397 11398 11399
qemuDomainStorageSourceAccessAllow(virQEMUDriverPtr driver,
                                   virDomainObjPtr vm,
                                   virStorageSourcePtr elem,
                                   bool readonly,
                                   bool newSource)
11400
{
11401
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE;
11402

11403
    if (readonly)
11404 11405 11406
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_ONLY;
    else
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_WRITE;
11407

11408
    if (!newSource)
11409
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS;
11410

11411
    return qemuDomainStorageSourceAccessModify(driver, vm, elem, flags);
11412 11413 11414
}


11415 11416 11417 11418 11419 11420 11421 11422 11423 11424
/*
 * 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)
{
11425 11426 11427 11428 11429 11430 11431 11432 11433 11434
#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; \
        } \
11435 11436
    } while (0)

11437 11438 11439 11440 11441 11442 11443 11444 11445 11446 11447 11448
#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)

11449
    CHECK_EQ(device, "device", false);
11450 11451 11452 11453
    CHECK_EQ(bus, "bus", false);
    if (STRNEQ(disk->dst, orig_disk->dst)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
11454
                       "target");
11455 11456 11457 11458
        return false;
    }
    CHECK_EQ(tray_status, "tray", true);
    CHECK_EQ(removable, "removable", true);
11459 11460 11461 11462 11463 11464 11465 11466 11467 11468 11469 11470 11471 11472 11473 11474 11475 11476 11477 11478 11479 11480 11481 11482 11483 11484 11485 11486 11487 11488 11489 11490 11491 11492 11493 11494 11495 11496 11497 11498 11499 11500 11501 11502 11503 11504 11505 11506 11507 11508 11509 11510 11511 11512

    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);
11513 11514
    CHECK_STREQ_NULLABLE(blkdeviotune.group_name,
                         "blkdeviotune group name");
11515

11516 11517 11518 11519 11520 11521 11522 11523
    CHECK_STREQ_NULLABLE(serial,
                         "serial");
    CHECK_STREQ_NULLABLE(wwn,
                         "wwn");
    CHECK_STREQ_NULLABLE(vendor,
                         "vendor");
    CHECK_STREQ_NULLABLE(product,
                         "product");
11524

11525 11526 11527 11528 11529 11530 11531
    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);
11532
    /* "snapshot" is a libvirt internal field and thus can be changed */
11533
    /* startupPolicy is allowed to be updated. Therefore not checked here. */
11534
    CHECK_EQ(transient, "transient", true);
11535 11536 11537 11538 11539 11540 11541 11542 11543 11544 11545 11546 11547

    /* 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;
    }

11548
    /* device alias is checked already in virDomainDefCompatibleDevice */
11549

11550
    CHECK_EQ(info.bootIndex, "boot order", true);
11551 11552 11553 11554 11555
    CHECK_EQ(rawio, "rawio", true);
    CHECK_EQ(sgio, "sgio", true);
    CHECK_EQ(discard, "discard", true);
    CHECK_EQ(iothread, "iothread", true);

11556 11557
    CHECK_STREQ_NULLABLE(domain_name,
                         "backenddomain");
11558

11559 11560 11561 11562 11563 11564
    /* 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);

11565 11566 11567 11568 11569 11570 11571 11572
    if (!virStoragePRDefIsEqual(disk->src->pr,
                                orig_disk->src->pr)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "reservations");
        return false;
    }

11573
#undef CHECK_EQ
11574
#undef CHECK_STREQ_NULLABLE
11575 11576 11577 11578

    return true;
}

11579 11580 11581 11582 11583 11584 11585 11586 11587 11588 11589 11590 11591 11592 11593

/* Return the format node name for a given disk of an online guest */
const char *
qemuDomainDiskNodeFormatLookup(virDomainObjPtr vm,
                               const char *disk)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++) {
        if (STREQ(vm->def->disks[i]->dst, disk))
            return vm->def->disks[i]->src->nodeformat;
    }
    return NULL;
}

11594 11595 11596
bool
qemuDomainDiskBlockJobIsActive(virDomainDiskDefPtr disk)
{
11597 11598
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

11599 11600 11601 11602 11603 11604 11605 11606
    if (disk->mirror) {
        virReportError(VIR_ERR_BLOCK_COPY_ACTIVE,
                       _("disk '%s' already in active block job"),
                       disk->dst);

        return true;
    }

11607
    if (diskPriv->blockjob &&
11608
        qemuBlockJobIsRunning(diskPriv->blockjob)) {
11609 11610 11611 11612 11613 11614 11615 11616 11617 11618
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("disk '%s' already in active block job"),
                       disk->dst);
        return true;
    }

    return false;
}


11619 11620 11621 11622 11623 11624 11625 11626 11627 11628 11629 11630 11631 11632 11633
/**
 * 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++) {
11634 11635 11636
        virDomainDiskDefPtr disk = vm->def->disks[i];
        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

11637 11638
        if (!copy_only && diskPriv->blockjob &&
            qemuBlockJobIsRunning(diskPriv->blockjob))
11639 11640
            return true;

11641
        if (disk->mirror && disk->mirrorJob == VIR_DOMAIN_BLOCK_JOB_TYPE_COPY)
11642 11643 11644 11645 11646 11647 11648
            return true;
    }

    return false;
}


11649 11650
int
qemuDomainUpdateDeviceList(virQEMUDriverPtr driver,
11651 11652
                           virDomainObjPtr vm,
                           int asyncJob)
11653 11654 11655
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char **aliases;
11656
    int rc;
11657

11658 11659
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
11660 11661 11662 11663
    rc = qemuMonitorGetDeviceAliases(priv->mon, &aliases);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
    if (rc < 0)
11664 11665
        return -1;

11666
    virStringListFree(priv->qemuDevices);
11667 11668 11669
    priv->qemuDevices = aliases;
    return 0;
}
11670

11671 11672 11673 11674 11675 11676 11677 11678 11679 11680 11681 11682 11683 11684 11685 11686 11687 11688 11689

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);

11690 11691
    if (qemuDomainObjExitMonitor(driver, vm) < 0) {
        virHashFree(meminfo);
11692
        return -1;
11693
    }
11694 11695 11696 11697 11698 11699 11700 11701 11702 11703 11704 11705 11706 11707 11708 11709 11710 11711 11712 11713 11714 11715 11716 11717 11718 11719 11720 11721

    /* 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;
}


11722 11723 11724 11725
static bool
qemuDomainABIStabilityCheck(const virDomainDef *src,
                            const virDomainDef *dst)
{
11726 11727
    size_t i;

11728 11729 11730 11731 11732 11733 11734 11735 11736
    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;
    }

11737 11738 11739 11740 11741 11742 11743 11744 11745 11746 11747 11748 11749
    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;
        }
    }

11750 11751 11752 11753 11754 11755 11756 11757 11758
    return true;
}


virDomainABIStability virQEMUDriverDomainABIStability = {
    .domain = qemuDomainABIStabilityCheck,
};


11759 11760 11761 11762 11763 11764 11765 11766 11767 11768 11769 11770 11771 11772 11773 11774 11775 11776 11777 11778
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;
}


11779 11780 11781
#define COPY_FLAGS (VIR_DOMAIN_XML_SECURE | \
                    VIR_DOMAIN_XML_MIGRATABLE)

11782 11783
bool
qemuDomainDefCheckABIStability(virQEMUDriverPtr driver,
11784
                               virQEMUCapsPtr qemuCaps,
11785 11786 11787 11788 11789 11790 11791
                               virDomainDefPtr src,
                               virDomainDefPtr dst)
{
    virDomainDefPtr migratableDefSrc = NULL;
    virDomainDefPtr migratableDefDst = NULL;
    bool ret = false;

11792 11793
    if (!(migratableDefSrc = qemuDomainDefCopy(driver, qemuCaps, src, COPY_FLAGS)) ||
        !(migratableDefDst = qemuDomainDefCopy(driver, qemuCaps, dst, COPY_FLAGS)))
11794 11795
        goto cleanup;

11796 11797 11798
    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   src, migratableDefSrc,
                                                   dst, migratableDefDst);
11799

11800
 cleanup:
11801 11802 11803 11804
    virDomainDefFree(migratableDefSrc);
    virDomainDefFree(migratableDefDst);
    return ret;
}
11805

11806 11807 11808 11809 11810 11811

bool
qemuDomainCheckABIStability(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
                            virDomainDefPtr dst)
{
11812
    qemuDomainObjPrivatePtr priv = vm->privateData;
11813 11814 11815 11816 11817 11818
    virDomainDefPtr migratableSrc = NULL;
    virDomainDefPtr migratableDst = NULL;
    char *xml = NULL;
    bool ret = false;

    if (!(xml = qemuDomainFormatXML(driver, vm, COPY_FLAGS)) ||
11819 11820
        !(migratableSrc = qemuDomainDefFromXML(driver, priv->qemuCaps, xml)) ||
        !(migratableDst = qemuDomainDefCopy(driver, priv->qemuCaps, dst, COPY_FLAGS)))
11821 11822 11823 11824 11825 11826 11827 11828 11829 11830 11831 11832 11833 11834 11835 11836
        goto cleanup;

    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   vm->def, migratableSrc,
                                                   dst, migratableDst);

 cleanup:
    VIR_FREE(xml);
    virDomainDefFree(migratableSrc);
    virDomainDefFree(migratableDst);
    return ret;
}

#undef COPY_FLAGS


11837
bool
11838
qemuDomainAgentAvailable(virDomainObjPtr vm,
11839 11840
                         bool reportError)
{
11841 11842
    qemuDomainObjPrivatePtr priv = vm->privateData;

11843 11844 11845 11846 11847 11848 11849
    if (virDomainObjGetState(vm, NULL) != VIR_DOMAIN_RUNNING) {
        if (reportError) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
        }
        return false;
    }
11850 11851 11852 11853 11854 11855 11856 11857 11858
    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) {
11859 11860 11861 11862 11863 11864 11865 11866 11867 11868 11869 11870
        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;
11871 11872 11873 11874
        }
    }
    return true;
}
11875

11876

11877
static unsigned long long
11878
qemuDomainGetMemorySizeAlignment(virDomainDefPtr def)
11879
{
11880 11881 11882 11883 11884
    /* 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;

11885 11886 11887 11888 11889 11890 11891
    /* Align memory size. QEMU requires rounding to next 4KiB block.
     * We'll take the "traditional" path and round it to 1MiB*/

    return 1024;
}


11892 11893
static unsigned long long
qemuDomainGetMemoryModuleSizeAlignment(const virDomainDef *def,
J
Ján Tomko 已提交
11894
                                       const virDomainMemoryDef *mem G_GNUC_UNUSED)
11895 11896 11897 11898 11899 11900 11901 11902 11903 11904 11905 11906
{
    /* 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;
}


11907 11908 11909
int
qemuDomainAlignMemorySizes(virDomainDefPtr def)
{
11910 11911
    unsigned long long maxmemkb = virMemoryMaxValue(false) >> 10;
    unsigned long long maxmemcapped = virMemoryMaxValue(true) >> 10;
11912
    unsigned long long initialmem = 0;
11913
    unsigned long long hotplugmem = 0;
11914
    unsigned long long mem;
11915
    unsigned long long align = qemuDomainGetMemorySizeAlignment(def);
11916 11917 11918 11919 11920
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    /* align NUMA cell sizes if relevant */
    for (i = 0; i < ncells; i++) {
11921 11922
        mem = VIR_ROUND_UP(virDomainNumaGetNodeMemorySize(def->numa, i), align);
        initialmem += mem;
11923 11924 11925 11926 11927 11928 11929

        if (mem > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("memory size of NUMA node '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
11930
        virDomainNumaSetNodeMemorySize(def->numa, i, mem);
11931 11932
    }

11933 11934 11935 11936 11937
    /* 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);

11938 11939 11940 11941 11942 11943
    if (initialmem > maxmemcapped) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("initial memory size overflowed after alignment"));
        return -1;
    }

11944
    def->mem.max_memory = VIR_ROUND_UP(def->mem.max_memory, align);
11945 11946 11947 11948 11949
    if (def->mem.max_memory > maxmemkb) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("maximum memory size overflowed after alignment"));
        return -1;
    }
11950

11951
    /* Align memory module sizes */
11952 11953
    for (i = 0; i < def->nmems; i++) {
        align = qemuDomainGetMemoryModuleSizeAlignment(def, def->mems[i]);
11954
        def->mems[i]->size = VIR_ROUND_UP(def->mems[i]->size, align);
11955
        hotplugmem += def->mems[i]->size;
11956 11957 11958 11959 11960 11961 11962

        if (def->mems[i]->size > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("size of memory module '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
11963
    }
11964

11965 11966
    virDomainDefSetMemoryTotal(def, initialmem + hotplugmem);

11967 11968
    return 0;
}
11969 11970 11971 11972 11973 11974 11975 11976 11977 11978 11979


/**
 * qemuDomainMemoryDeviceAlignSize:
 * @mem: memory device definition object
 *
 * Aligns the size of the memory module as qemu enforces it. The size is updated
 * inplace. Default rounding is now to 1 MiB (qemu requires rouding to page,
 * size so this should be safe).
 */
void
11980 11981
qemuDomainMemoryDeviceAlignSize(virDomainDefPtr def,
                                virDomainMemoryDefPtr mem)
11982
{
11983
    mem->size = VIR_ROUND_UP(mem->size, qemuDomainGetMemorySizeAlignment(def));
11984
}
11985 11986 11987 11988 11989 11990 11991 11992 11993 11994 11995 11996 11997


/**
 * qemuDomainGetMonitor:
 * @vm: domain object
 *
 * Returns the monitor pointer corresponding to the domain object @vm.
 */
qemuMonitorPtr
qemuDomainGetMonitor(virDomainObjPtr vm)
{
    return ((qemuDomainObjPrivatePtr) vm->privateData)->mon;
}
11998 11999


12000 12001 12002 12003 12004 12005 12006
/**
 * 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.
 */
12007
virDomainChrDefPtr
12008 12009 12010 12011 12012 12013 12014 12015 12016 12017
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;

12018 12019
        if (STREQ_NULLABLE(channel->target.name, "org.qemu.guest_agent.0"))
            return channel;
12020 12021
    }

12022
    return NULL;
12023
}
12024 12025


12026
static bool
12027
qemuDomainMachineIsQ35(const char *machine,
12028
                       const virArch arch)
12029
{
12030 12031 12032
    if (!ARCH_IS_X86(arch))
        return false;

12033 12034 12035 12036 12037 12038
    if (STREQ(machine, "q35") ||
        STRPREFIX(machine, "pc-q35-")) {
        return true;
    }

    return false;
12039
}
12040 12041


12042
static bool
12043
qemuDomainMachineIsI440FX(const char *machine,
12044
                          const virArch arch)
12045
{
12046 12047 12048
    if (!ARCH_IS_X86(arch))
        return false;

12049 12050 12051 12052 12053 12054 12055 12056 12057
    if (STREQ(machine, "pc") ||
        STRPREFIX(machine, "pc-0.") ||
        STRPREFIX(machine, "pc-1.") ||
        STRPREFIX(machine, "pc-i440fx-") ||
        STRPREFIX(machine, "rhel")) {
        return true;
    }

    return false;
12058 12059 12060
}


12061
static bool
12062
qemuDomainMachineIsS390CCW(const char *machine,
12063
                           const virArch arch)
12064
{
12065 12066 12067
    if (!ARCH_IS_S390(arch))
        return false;

12068 12069 12070 12071
    if (STRPREFIX(machine, "s390-ccw"))
        return true;

    return false;
A
Andrea Bolognani 已提交
12072
}
12073

A
Andrea Bolognani 已提交
12074

12075 12076
/* You should normally avoid this function and use
 * qemuDomainIsARMVirt() instead. */
A
Andrea Bolognani 已提交
12077 12078 12079 12080 12081 12082
bool
qemuDomainMachineIsARMVirt(const char *machine,
                           const virArch arch)
{
    if (arch != VIR_ARCH_ARMV6L &&
        arch != VIR_ARCH_ARMV7L &&
12083
        arch != VIR_ARCH_AARCH64) {
12084
        return false;
12085
    }
12086

12087 12088 12089 12090
    if (STREQ(machine, "virt") ||
        STRPREFIX(machine, "virt-")) {
        return true;
    }
12091

12092
    return false;
12093 12094 12095
}


12096
static bool
A
Andrea Bolognani 已提交
12097 12098
qemuDomainMachineIsRISCVVirt(const char *machine,
                             const virArch arch)
12099
{
A
Andrea Bolognani 已提交
12100 12101
    if (!ARCH_IS_RISCV(arch))
        return false;
12102

12103 12104 12105 12106
    if (STREQ(machine, "virt") ||
        STRPREFIX(machine, "virt-")) {
        return true;
    }
12107

12108
    return false;
A
Andrea Bolognani 已提交
12109 12110 12111
}


12112 12113
/* You should normally avoid this function and use
 * qemuDomainIsPSeries() instead. */
A
Andrea Bolognani 已提交
12114 12115 12116 12117 12118 12119 12120
bool
qemuDomainMachineIsPSeries(const char *machine,
                           const virArch arch)
{
    if (!ARCH_IS_PPC64(arch))
        return false;

12121 12122 12123 12124
    if (STREQ(machine, "pseries") ||
        STRPREFIX(machine, "pseries-")) {
        return true;
    }
12125

12126
    return false;
12127 12128 12129
}


12130 12131
/* You should normally avoid this function and use
 * qemuDomainHasBuiltinIDE() instead. */
12132
bool
12133 12134
qemuDomainMachineHasBuiltinIDE(const char *machine,
                               const virArch arch)
12135
{
12136
    return qemuDomainMachineIsI440FX(machine, arch) ||
A
Andrea Bolognani 已提交
12137 12138 12139
        STREQ(machine, "malta") ||
        STREQ(machine, "sun4u") ||
        STREQ(machine, "g3beige");
12140 12141 12142
}


12143
static bool
12144
qemuDomainMachineNeedsFDC(const char *machine,
12145
                          const virArch arch)
12146 12147
{
    const char *p = STRSKIP(machine, "pc-q35-");
12148

12149 12150 12151
    if (!ARCH_IS_X86(arch))
        return false;

12152 12153 12154 12155 12156 12157 12158 12159 12160
    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;
12161
    }
12162 12163

    return true;
12164 12165 12166
}


12167
bool
A
Andrea Bolognani 已提交
12168
qemuDomainIsQ35(const virDomainDef *def)
12169
{
12170
    return qemuDomainMachineIsQ35(def->os.machine, def->os.arch);
12171 12172
}

12173

J
Ján Tomko 已提交
12174
bool
A
Andrea Bolognani 已提交
12175
qemuDomainIsI440FX(const virDomainDef *def)
J
Ján Tomko 已提交
12176
{
12177
    return qemuDomainMachineIsI440FX(def->os.machine, def->os.arch);
12178 12179 12180 12181
}


bool
A
Andrea Bolognani 已提交
12182
qemuDomainIsS390CCW(const virDomainDef *def)
12183
{
12184
    return qemuDomainMachineIsS390CCW(def->os.machine, def->os.arch);
12185 12186 12187 12188
}


bool
A
Andrea Bolognani 已提交
12189
qemuDomainIsARMVirt(const virDomainDef *def)
12190
{
A
Andrea Bolognani 已提交
12191
    return qemuDomainMachineIsARMVirt(def->os.machine, def->os.arch);
J
Ján Tomko 已提交
12192 12193 12194
}


12195 12196 12197 12198 12199 12200 12201 12202
bool
qemuDomainIsRISCVVirt(const virDomainDef *def)
{
    return qemuDomainMachineIsRISCVVirt(def->os.machine, def->os.arch);
}


bool
A
Andrea Bolognani 已提交
12203
qemuDomainIsPSeries(const virDomainDef *def)
12204
{
A
Andrea Bolognani 已提交
12205 12206 12207 12208 12209 12210 12211 12212 12213 12214
    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)
12215 12216
        return false;

A
Andrea Bolognani 已提交
12217
    if (def->controllers[root]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)
12218 12219 12220 12221 12222 12223
        return false;

    return true;
}


12224
bool
A
Andrea Bolognani 已提交
12225
qemuDomainHasPCIeRoot(const virDomainDef *def)
12226
{
A
Andrea Bolognani 已提交
12227 12228 12229 12230 12231 12232 12233 12234 12235
    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;
12236 12237 12238 12239
}


bool
A
Andrea Bolognani 已提交
12240
qemuDomainHasBuiltinIDE(const virDomainDef *def)
12241
{
12242
    return qemuDomainMachineHasBuiltinIDE(def->os.machine, def->os.arch);
A
Andrea Bolognani 已提交
12243
}
12244 12245


A
Andrea Bolognani 已提交
12246 12247 12248
bool
qemuDomainNeedsFDC(const virDomainDef *def)
{
12249
    return qemuDomainMachineNeedsFDC(def->os.machine, def->os.arch);
12250 12251 12252
}


12253 12254 12255 12256
bool
qemuDomainSupportsPCI(virDomainDefPtr def,
                      virQEMUCapsPtr qemuCaps)
{
12257 12258 12259 12260
    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)) {
12261
        return true;
12262
    }
12263 12264 12265 12266 12267 12268

    if (STREQ(def->os.machine, "versatilepb"))
        return true;

    if ((qemuDomainIsARMVirt(def) ||
         qemuDomainIsRISCVVirt(def)) &&
12269
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX)) {
12270
        return true;
12271
    }
12272 12273 12274 12275 12276

    return false;
}


12277 12278 12279 12280 12281 12282 12283 12284 12285 12286 12287 12288 12289 12290 12291 12292 12293 12294 12295 12296 12297 12298 12299 12300 12301 12302 12303 12304 12305 12306 12307 12308 12309 12310 12311 12312 12313 12314 12315
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
12316
    case VIR_DOMAIN_MEMORY_MODEL_NVDIMM:
12317 12318 12319 12320 12321 12322 12323 12324
        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;
        }

12325 12326 12327
        if (virDomainNumaGetNodeCount(def->numa) != 0) {
            if (mem->targetNode == -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
Y
Yuri Chornoivan 已提交
12328
                               _("target NUMA node needs to be specified for "
12329 12330 12331
                                 "memory device"));
                return -1;
            }
12332 12333
        }

12334 12335 12336 12337 12338 12339 12340 12341 12342 12343 12344 12345 12346 12347 12348 12349 12350 12351 12352 12353 12354 12355 12356 12357
        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;
}


12358 12359 12360 12361 12362 12363 12364 12365 12366 12367 12368 12369 12370 12371 12372 12373 12374 12375 12376 12377 12378
/**
 * 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
12379 12380
    bool needPCDimmCap = false;
    bool needNvdimmCap = false;
12381 12382 12383 12384 12385 12386 12387
    size_t i;

    hotplugSpace = def->mem.max_memory - virDomainDefGetMemoryInitial(def);

    if (mem) {
        nmems++;
        hotplugMemory = mem->size;
12388 12389 12390

        if (qemuDomainDefValidateMemoryHotplugDevice(mem, def) < 0)
            return -1;
12391 12392 12393 12394 12395 12396 12397 12398 12399 12400 12401 12402 12403
    }

    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;
    }

12404 12405 12406 12407 12408 12409 12410 12411 12412 12413 12414
    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;
        }
12415 12416 12417 12418 12419 12420 12421 12422 12423
    }

    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;
    }

12424
    for (i = 0; i < def->nmems; i++) {
12425 12426
        hotplugMemory += def->mems[i]->size;

M
Michal Privoznik 已提交
12427 12428 12429 12430 12431 12432 12433 12434 12435 12436 12437 12438 12439 12440
        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;
        }

12441 12442 12443 12444 12445 12446
        /* 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
12447 12448 12449 12450 12451 12452 12453 12454 12455 12456 12457 12458 12459 12460
    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;
    }

12461 12462 12463 12464 12465 12466 12467 12468 12469 12470
    if (hotplugMemory > hotplugSpace) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory device total size exceeds hotplug space"));
        return -1;
    }

    return 0;
}


12471 12472 12473
/**
 * qemuDomainUpdateCurrentMemorySize:
 *
12474 12475
 * In case when the balloon is not present for the domain, the function
 * recalculates the maximum size to reflect possible changes.
12476
 */
12477 12478
void
qemuDomainUpdateCurrentMemorySize(virDomainObjPtr vm)
12479 12480 12481
{
    /* inactive domain doesn't need size update */
    if (!virDomainObjIsActive(vm))
12482
        return;
12483 12484 12485

    /* if no balloning is available, the current size equals to the current
     * full memory size */
12486
    if (!virDomainDefHasMemballoon(vm->def))
12487
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
12488
}
12489 12490


12491 12492 12493 12494 12495 12496 12497 12498
/**
 * 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.
 */
12499
static bool
12500 12501 12502 12503 12504 12505
ppc64VFIODeviceIsNV2Bridge(const char *device)
{
    const char *nvlink2Files[] = {"ibm,gpu", "ibm,nvlink",
                                  "ibm,nvlink-speed", "memory-region"};
    size_t i;

12506
    for (i = 0; i < G_N_ELEMENTS(nvlink2Files); i++) {
12507
        g_autofree char *file = NULL;
12508

12509 12510
        file = g_strdup_printf("/sys/bus/pci/devices/%s/of_node/%s",
                               device, nvlink2Files[i]);
12511 12512 12513 12514 12515 12516 12517 12518 12519

        if (!virFileExists(file))
            return false;
    }

    return true;
}


12520 12521 12522 12523 12524 12525 12526 12527 12528 12529 12530 12531 12532 12533 12534 12535
/**
 * getPPC64MemLockLimitBytes:
 * @def: domain definition
 *
 * A PPC64 helper that calculates the memory locking limit in order for
 * the guest to operate properly.
 */
static unsigned long long
getPPC64MemLockLimitBytes(virDomainDefPtr def)
{
    unsigned long long memKB = 0;
    unsigned long long baseLimit = 0;
    unsigned long long memory = 0;
    unsigned long long maxMemory = 0;
    unsigned long long passthroughLimit = 0;
    size_t i, nPCIHostBridges = 0;
12536
    virPCIDeviceAddressPtr pciAddr;
12537
    bool usesVFIO = false;
12538
    bool nvlink2Capable = false;
12539 12540 12541 12542 12543 12544 12545 12546 12547 12548 12549 12550 12551

    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];

12552
        if (virHostdevIsVFIODevice(dev)) {
12553
            usesVFIO = true;
12554 12555 12556

            pciAddr = &dev->source.subsys.u.pci.addr;
            if (virPCIDeviceAddressIsValid(pciAddr, false)) {
12557
                g_autofree char *pciAddrStr = NULL;
12558 12559 12560 12561 12562 12563 12564

                pciAddrStr = virPCIDeviceAddressAsString(pciAddr);
                if (ppc64VFIODeviceIsNV2Bridge(pciAddrStr)) {
                    nvlink2Capable = true;
                    break;
                }
            }
12565 12566 12567 12568 12569 12570 12571 12572 12573 12574 12575 12576 12577 12578 12579 12580 12581 12582 12583 12584 12585 12586 12587 12588 12589 12590
        }
    }

    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;

12591 12592 12593 12594 12595
    /* 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.
12596
     *
12597 12598 12599
     * 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.
12600
     *
12601 12602 12603 12604
     * 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.
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 12633 12634 12635 12636 12637 12638 12639
     * Note that the NVIDIA RAM window must be accounted for the TCE
     * table size, but *not* for the main RAM (maxMemory). This gives
     * us the following passthroughLimit for the NVLink2 case:
     *
     * passthroughLimit = maxMemory +
     *                    128TiB/512KiB * #PHBs + 8 MiB */
    if (nvlink2Capable) {
        passthroughLimit = maxMemory +
                           128 * (1ULL<<30) / 512 * nPCIHostBridges +
                           8192;
    } else if (usesVFIO) {
        /* For regular (non-NVLink2 present) VFIO passthrough, the value
         * of passthroughLimit is:
         *
         * passthroughLimit := max( 2 GiB * #PHBs,                       (c)
         *                          memory                               (d)
         *                          + memory * 1/512 * #PHBs + 8 MiB )   (e)
         *
         * (c) is the pre-DDW VFIO DMA window accounting. We're allowing 2
         * GiB rather than 1 GiB
         *
         * (d) is the with-DDW (and memory pre-registration and related
         * features) DMA window accounting - assuming that we only account
         * RAM once, even if mapped to multiple PHBs
         *
         * (e) is the with-DDW userspace view and overhead for the 64-bit
         * DMA window. This is based a bit on expected guest behaviour, but
         * there really isn't a way to completely avoid that. We assume the
         * guest requests a 64-bit DMA window (per PHB) just big enough to
         * map all its RAM. 4 kiB page size gives the 1/512; it will be
         * less with 64 kiB pages, less still if the guest is mapped with
         * hugepages (unlike the default 32-bit DMA window, DDW windows
         * can use large IOMMU pages). 8 MiB is for second and further level
         * overheads, like (b) */
12640 12641 12642
        passthroughLimit = MAX(2 * 1024 * 1024 * nPCIHostBridges,
                               memory +
                               memory / 512 * nPCIHostBridges + 8192);
12643
    }
12644 12645 12646 12647 12648 12649 12650

    memKB = baseLimit + passthroughLimit;

    return memKB << 10;
}


12651
/**
12652
 * qemuDomainGetMemLockLimitBytes:
12653 12654
 * @def: domain definition
 *
12655 12656 12657 12658 12659 12660
 * Calculate the memory locking limit that needs to be set in order for
 * the guest to operate properly. The limit depends on a number of factors,
 * including certain configuration options and less immediately apparent ones
 * such as the guest architecture or the use of certain devices.
 *
 * Returns: the memory locking limit, or 0 if setting the limit is not needed
12661 12662
 */
unsigned long long
12663
qemuDomainGetMemLockLimitBytes(virDomainDefPtr def)
12664
{
12665 12666
    unsigned long long memKB = 0;
    size_t i;
12667

12668 12669 12670 12671 12672 12673
    /* prefer the hard limit */
    if (virMemoryLimitIsSet(def->mem.hard_limit)) {
        memKB = def->mem.hard_limit;
        goto done;
    }

12674 12675 12676 12677 12678 12679 12680
    /* 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;
12681

12682 12683
    if (ARCH_IS_PPC64(def->os.arch) && def->virtType == VIR_DOMAIN_VIRT_KVM)
        return getPPC64MemLockLimitBytes(def);
12684

12685 12686 12687 12688 12689 12690 12691 12692 12693 12694 12695 12696 12697 12698 12699 12700 12701 12702
    /* 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.
     */
12703
    for (i = 0; i < def->nhostdevs; i++) {
12704 12705
        if (virHostdevIsVFIODevice(def->hostdevs[i]) ||
            virHostdevIsMdevDevice(def->hostdevs[i])) {
12706 12707 12708
            memKB = virDomainDefGetMemoryTotal(def) + 1024 * 1024;
            goto done;
        }
12709 12710
    }

12711 12712
 done:
    return memKB << 10;
12713
}
12714

12715

12716 12717 12718 12719 12720 12721 12722
/**
 * qemuDomainAdjustMaxMemLock:
 * @vm: domain
 *
 * Adjust the memory locking limit for the QEMU process associated to @vm, in
 * order to comply with VFIO or architecture requirements.
 *
12723 12724 12725 12726
 * 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.
12727 12728 12729 12730 12731 12732 12733 12734 12735
 *
 * Returns: 0 on success, <0 on failure
 */
int
qemuDomainAdjustMaxMemLock(virDomainObjPtr vm)
{
    unsigned long long bytes = 0;
    int ret = -1;

12736 12737 12738
    bytes = qemuDomainGetMemLockLimitBytes(vm->def);

    if (bytes) {
12739 12740 12741 12742 12743 12744 12745 12746 12747 12748 12749 12750 12751 12752
        /* 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;
    }
12753 12754 12755 12756 12757 12758 12759 12760 12761 12762

    /* Trying to set the memory locking limit to zero is a no-op */
    if (virProcessSetMaxMemLock(vm->pid, bytes) < 0)
        goto out;

    ret = 0;

 out:
     return ret;
}
12763

12764 12765 12766 12767 12768 12769 12770 12771 12772 12773 12774 12775 12776 12777 12778 12779 12780 12781 12782 12783 12784 12785 12786 12787 12788 12789 12790 12791 12792 12793

/**
 * qemuDomainAdjustMaxMemLockHostdev:
 * @vm: domain
 * @hostdev: device
 *
 * Temporarily add the hostdev to the domain definition. This is needed
 * because qemuDomainAdjustMaxMemLock() requires the hostdev to be already
 * part of the domain definition, but other functions like
 * qemuAssignDeviceHostdevAlias() expect it *not* to be there.
 * A better way to handle this would be nice
 *
 * Returns: 0 on success, <0 on failure
 */
int
qemuDomainAdjustMaxMemLockHostdev(virDomainObjPtr vm,
                                  virDomainHostdevDefPtr hostdev)
{
    int ret = 0;

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;
    if (qemuDomainAdjustMaxMemLock(vm) < 0)
        ret = -1;

    vm->def->hostdevs[--(vm->def->nhostdevs)] = NULL;

    return ret;
}


12794 12795 12796 12797 12798 12799 12800 12801 12802
/**
 * qemuDomainHasVcpuPids:
 * @vm: Domain object
 *
 * Returns true if we were able to successfully detect vCPU pids for the VM.
 */
bool
qemuDomainHasVcpuPids(virDomainObjPtr vm)
{
12803 12804 12805 12806 12807 12808
    size_t i;
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    virDomainVcpuDefPtr vcpu;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
12809

12810 12811 12812 12813 12814
        if (QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid > 0)
            return true;
    }

    return false;
12815
}
12816 12817 12818 12819 12820 12821 12822 12823 12824 12825 12826


/**
 * 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,
12827
                     unsigned int vcpuid)
12828
{
12829 12830
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
    return QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid;
12831
}
12832 12833


12834 12835 12836 12837 12838 12839 12840 12841 12842 12843 12844 12845 12846 12847 12848 12849 12850 12851 12852 12853 12854 12855 12856 12857 12858 12859 12860 12861 12862 12863 12864 12865 12866 12867 12868 12869 12870 12871 12872 12873 12874 12875
/**
 * 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;
}


12876 12877 12878 12879 12880 12881 12882 12883 12884
bool
qemuDomainSupportsNewVcpuHotplug(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
}


12885
/**
12886
 * qemuDomainRefreshVcpuInfo:
12887 12888 12889
 * @driver: qemu driver data
 * @vm: domain object
 * @asyncJob: current asynchronous job type
12890
 * @state: refresh vcpu state
12891
 *
12892 12893
 * Updates vCPU information private data of @vm. Due to historical reasons this
 * function returns success even if some data were not reported by qemu.
12894
 *
12895 12896
 * If @state is true, the vcpu state is refreshed as reported by the monitor.
 *
12897
 * Returns 0 on success and -1 on fatal error.
12898 12899
 */
int
12900 12901
qemuDomainRefreshVcpuInfo(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
12902 12903
                          int asyncJob,
                          bool state)
12904
{
12905
    virDomainVcpuDefPtr vcpu;
12906 12907
    qemuDomainVcpuPrivatePtr vcpupriv;
    qemuMonitorCPUInfoPtr info = NULL;
12908
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
12909
    size_t i, j;
12910
    bool hotplug;
12911
    bool fast;
12912
    bool validTIDs = true;
12913
    int rc;
12914
    int ret = -1;
12915

12916
    hotplug = qemuDomainSupportsNewVcpuHotplug(vm);
12917 12918
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
12919

12920 12921
    VIR_DEBUG("Maxvcpus %zu hotplug %d fast query %d", maxvcpus, hotplug, fast);

12922 12923
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
12924

12925 12926
    rc = qemuMonitorGetCPUInfo(qemuDomainGetMonitor(vm), &info, maxvcpus,
                               hotplug, fast);
12927

12928
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
12929 12930
        goto cleanup;

12931
    if (rc < 0)
12932 12933
        goto cleanup;

12934 12935 12936 12937 12938 12939 12940 12941 12942 12943 12944 12945 12946 12947 12948 12949 12950 12951 12952 12953 12954 12955 12956 12957 12958 12959 12960 12961 12962 12963 12964 12965 12966
    /*
     * 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++) {
12967
            if (info[i].tid != 0 && info[i].tid == info[j].tid) {
12968 12969 12970 12971 12972 12973 12974 12975 12976 12977 12978 12979
                VIR_DEBUG("vCPU[%zu] PID %llu duplicates vCPU[%zu]",
                          i, (unsigned long long)info[i].tid, j);
                validTIDs = false;
            }
        }

        if (validTIDs)
            VIR_DEBUG("vCPU[%zu] PID %llu is valid",
                      i, (unsigned long long)info[i].tid);
    }

    VIR_DEBUG("Extracting vCPU information validTIDs=%d", validTIDs);
12980 12981
    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
12982
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
12983

12984
        if (validTIDs)
12985 12986 12987 12988 12989
            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;
12990
        vcpupriv->node_id = info[i].node_id;
12991 12992
        vcpupriv->vcpus = info[i].vcpus;
        VIR_FREE(vcpupriv->type);
12993
        vcpupriv->type = g_steal_pointer(&info[i].type);
12994
        VIR_FREE(vcpupriv->alias);
12995
        vcpupriv->alias = g_steal_pointer(&info[i].alias);
12996
        virJSONValueFree(vcpupriv->props);
12997
        vcpupriv->props = g_steal_pointer(&info[i].props);
12998
        vcpupriv->enable_id = info[i].id;
12999
        vcpupriv->qemu_id = info[i].qemu_id;
13000

13001
        if (hotplug && state) {
13002 13003 13004 13005 13006
            vcpu->online = info[i].online;
            if (info[i].hotpluggable)
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_YES;
            else
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_NO;
13007
        }
13008 13009
    }

13010
    ret = 0;
13011 13012

 cleanup:
13013
    qemuMonitorCPUInfoFree(info, maxvcpus);
13014
    return ret;
13015
}
13016

13017 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
/**
 * 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;
13048
    qemuDomainVcpuPrivatePtr vcpupriv;
13049
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
13050
    virBitmapPtr haltedmap = NULL;
13051 13052
    size_t i;
    int ret = -1;
13053
    bool fast;
13054 13055 13056 13057 13058

    /* Not supported currently for TCG, see qemuDomainRefreshVcpuInfo */
    if (vm->def->virtType == VIR_DOMAIN_VIRT_QEMU)
        return 0;

13059
    /* The halted state is interresting only on s390(x). On other platforms
13060 13061 13062 13063 13064 13065 13066
     * 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))
13067 13068
        return 0;

13069 13070 13071
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;

13072 13073 13074 13075
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
    haltedmap = qemuMonitorGetCpuHalted(qemuDomainGetMonitor(vm), maxvcpus,
                                        fast);
13076
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || !haltedmap)
13077 13078 13079 13080
        goto cleanup;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
13081
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
13082 13083
        vcpupriv->halted = virTristateBoolFromBool(virBitmapIsBitSet(haltedmap,
                                                                     vcpupriv->qemu_id));
13084 13085 13086 13087 13088
    }

    ret = 0;

 cleanup:
13089
    virBitmapFree(haltedmap);
13090 13091
    return ret;
}
13092 13093 13094 13095 13096 13097

bool
qemuDomainSupportsNicdev(virDomainDefPtr def,
                         virDomainNetDefPtr net)
{
    /* non-virtio ARM nics require legacy -net nic */
S
Stefan Schallenberg 已提交
13098 13099
    if (((def->os.arch == VIR_ARCH_ARMV6L) ||
        (def->os.arch == VIR_ARCH_ARMV7L) ||
13100 13101 13102 13103 13104 13105 13106 13107
        (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;
}

13108 13109 13110 13111
bool
qemuDomainNetSupportsMTU(virDomainNetType type)
{
    switch (type) {
13112 13113
    case VIR_DOMAIN_NET_TYPE_NETWORK:
    case VIR_DOMAIN_NET_TYPE_BRIDGE:
13114 13115
    case VIR_DOMAIN_NET_TYPE_ETHERNET:
    case VIR_DOMAIN_NET_TYPE_VHOSTUSER:
13116 13117
        return true;
    case VIR_DOMAIN_NET_TYPE_USER:
13118 13119 13120 13121 13122 13123 13124 13125 13126 13127 13128 13129
    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
13130

P
Peter Krempa 已提交
13131 13132 13133 13134 13135 13136 13137 13138

virDomainDiskDefPtr
qemuDomainDiskByName(virDomainDefPtr def,
                     const char *name)
{
    virDomainDiskDefPtr ret;

    if (!(ret = virDomainDiskByName(def, name, true))) {
13139 13140
        virReportError(VIR_ERR_INVALID_ARG,
                       _("disk '%s' not found in domain"), name);
P
Peter Krempa 已提交
13141 13142 13143 13144 13145
        return NULL;
    }

    return ret;
}
13146 13147 13148 13149 13150 13151 13152 13153 13154 13155 13156 13157 13158 13159 13160 13161 13162 13163 13164 13165 13166 13167 13168 13169 13170 13171 13172 13173 13174 13175 13176


/**
 * 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;
}
13177 13178 13179 13180 13181 13182


int
qemuDomainPrepareChannel(virDomainChrDefPtr channel,
                         const char *domainChannelTargetDir)
{
S
Scott Garfinkle 已提交
13183 13184 13185 13186 13187 13188
    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) {
13189 13190 13191
        channel->source->data.nix.path = g_strdup_printf("%s/%s",
                                                         domainChannelTargetDir,
                                                         channel->target.name);
13192 13193
    } else {
        /* Generate a unique name */
13194 13195 13196 13197 13198
        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);
13199 13200 13201 13202
    }

    return 0;
}
13203 13204


13205
/* qemuDomainPrepareChardevSourceTLS:
13206 13207 13208 13209 13210 13211 13212 13213 13214 13215 13216 13217 13218 13219 13220 13221 13222
 * @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;
13223
            source->data.tcp.tlsFromConfig = true;
13224 13225 13226 13227 13228
        }
    }
}


13229
/* qemuDomainPrepareChardevSource:
13230
 * @def: live domain definition
13231
 * @cfg: driver configuration
13232 13233 13234 13235 13236 13237
 *
 * Iterate through all devices that use virDomainChrSourceDefPtr as host
 * interface part.
 */
void
qemuDomainPrepareChardevSource(virDomainDefPtr def,
13238
                               virQEMUDriverConfigPtr cfg)
13239 13240 13241 13242 13243 13244 13245 13246 13247 13248 13249 13250 13251 13252 13253 13254 13255 13256 13257 13258 13259 13260 13261 13262 13263 13264 13265 13266 13267
{
    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);
}


13268 13269 13270 13271 13272 13273 13274 13275 13276 13277 13278 13279 13280 13281 13282
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;
    }

13283 13284
    if (src->haveTLS == VIR_TRISTATE_BOOL_YES)
        src->tlsCertdir = g_strdup(cfg->vxhsTLSx509certdir);
13285 13286 13287 13288 13289

    return 0;
}


13290 13291 13292 13293 13294 13295 13296 13297 13298 13299 13300 13301 13302 13303 13304 13305 13306 13307 13308 13309
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;
        }

13310
        src->tlsCertdir = g_strdup(cfg->nbdTLSx509certdir);
13311 13312 13313 13314 13315 13316
    }

    return 0;
}


13317
/* qemuProcessPrepareStorageSourceTLS:
13318
 * @source: source for a disk
13319
 * @cfg: driver configuration
13320
 * @parentAlias: alias of the parent device
13321 13322 13323 13324 13325 13326 13327
 *
 * 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
 */
13328
static int
13329
qemuDomainPrepareStorageSourceTLS(virStorageSourcePtr src,
13330
                                  virQEMUDriverConfigPtr cfg,
13331 13332
                                  const char *parentAlias,
                                  virQEMUCapsPtr qemuCaps)
13333
{
13334 13335 13336 13337 13338 13339 13340 13341
    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;
13342

13343
    case VIR_STORAGE_NET_PROTOCOL_NBD:
13344 13345 13346 13347
        if (qemuProcessPrepareStorageSourceTLSNBD(src, cfg, qemuCaps) < 0)
            return -1;
        break;

13348 13349 13350 13351 13352 13353 13354 13355 13356 13357
    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:
13358 13359 13360 13361 13362 13363
        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;
        }
13364 13365 13366 13367 13368 13369
        break;

    case VIR_STORAGE_NET_PROTOCOL_NONE:
    case VIR_STORAGE_NET_PROTOCOL_LAST:
    default:
        virReportEnumRangeError(virStorageNetProtocol, src->protocol);
13370
        return -1;
13371
    }
13372

13373 13374 13375 13376
    if (src->haveTLS == VIR_TRISTATE_BOOL_YES &&
        !(src->tlsAlias = qemuAliasTLSObjFromSrcAlias(parentAlias)))
        return -1;

13377 13378 13379 13380
    return 0;
}


13381
void
13382 13383 13384 13385
qemuDomainPrepareShmemChardev(virDomainShmemDefPtr shmem)
{
    if (!shmem->server.enabled ||
        shmem->server.chr.data.nix.path)
13386
        return;
13387

13388 13389
    shmem->server.chr.data.nix.path = g_strdup_printf("/var/lib/libvirt/shmem-%s-sock",
                                                      shmem->name);
13390 13391 13392
}


13393 13394 13395 13396 13397 13398 13399 13400 13401 13402 13403 13404 13405 13406 13407 13408 13409 13410 13411 13412 13413 13414 13415 13416 13417 13418 13419 13420 13421 13422 13423 13424 13425
/**
 * 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);
}
13426 13427 13428 13429 13430 13431 13432 13433 13434 13435 13436 13437 13438 13439 13440 13441 13442 13443 13444 13445 13446 13447 13448 13449 13450 13451 13452 13453 13454 13455 13456 13457 13458 13459


/**
 * 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;
        }
    }
}
13460 13461 13462 13463 13464 13465 13466 13467 13468 13469 13470 13471 13472 13473 13474 13475 13476 13477 13478 13479


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;
}
13480 13481 13482 13483 13484 13485


bool
qemuDomainSupportsVideoVga(virDomainVideoDefPtr video,
                           virQEMUCapsPtr qemuCaps)
{
13486 13487 13488 13489 13490 13491 13492 13493
    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;
        }
    }
13494 13495 13496

    return true;
}
13497 13498


13499 13500 13501 13502 13503 13504 13505 13506
bool
qemuDomainNeedsVFIO(const virDomainDef *def)
{
    return virDomainDefHasVFIOHostdev(def) ||
        virDomainDefHasMdevHostdev(def);
}


13507 13508
/**
 * qemuDomainGetHostdevPath:
13509
 * @def: domain definition
13510
 * @dev: host device definition
13511
 * @teardown: true if device will be removed
13512
 * @npaths: number of items in @path and @perms arrays
13513 13514 13515
 * @path: resulting path to @dev
 * @perms: Optional pointer to VIR_CGROUP_DEVICE_* perms
 *
13516 13517 13518 13519 13520 13521
 * For given device @dev fetch its host path and store it at
 * @path. If a device requires other paths to be present/allowed
 * they are stored in the @path array after the actual path.
 * Optionally, caller can get @perms on the path (e.g. rw/ro).
 *
 * The caller is responsible for freeing the memory.
13522 13523 13524 13525
 *
 * Returns 0 on success, -1 otherwise.
 */
int
13526 13527 13528
qemuDomainGetHostdevPath(virDomainDefPtr def,
                         virDomainHostdevDefPtr dev,
                         bool teardown,
13529 13530 13531
                         size_t *npaths,
                         char ***path,
                         int **perms)
13532 13533 13534 13535 13536 13537
{
    int ret = -1;
    virDomainHostdevSubsysUSBPtr usbsrc = &dev->source.subsys.u.usb;
    virDomainHostdevSubsysPCIPtr pcisrc = &dev->source.subsys.u.pci;
    virDomainHostdevSubsysSCSIPtr scsisrc = &dev->source.subsys.u.scsi;
    virDomainHostdevSubsysSCSIVHostPtr hostsrc = &dev->source.subsys.u.scsi_host;
13538
    virDomainHostdevSubsysMediatedDevPtr mdevsrc = &dev->source.subsys.u.mdev;
13539 13540 13541 13542 13543
    virPCIDevicePtr pci = NULL;
    virUSBDevicePtr usb = NULL;
    virSCSIDevicePtr scsi = NULL;
    virSCSIVHostDevicePtr host = NULL;
    char *tmpPath = NULL;
13544 13545 13546
    bool includeVFIO = false;
    char **tmpPaths = NULL;
    int *tmpPerms = NULL;
13547
    size_t tmpNpaths = 0;
13548
    int perm = 0;
13549

13550
    *npaths = 0;
13551 13552 13553

    switch ((virDomainHostdevMode) dev->mode) {
    case VIR_DOMAIN_HOSTDEV_MODE_SUBSYS:
13554
        switch ((virDomainHostdevSubsysType)dev->source.subsys.type) {
13555 13556 13557 13558 13559 13560 13561 13562 13563 13564 13565
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
            if (pcisrc->backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO) {
                pci = virPCIDeviceNew(pcisrc->addr.domain,
                                      pcisrc->addr.bus,
                                      pcisrc->addr.slot,
                                      pcisrc->addr.function);
                if (!pci)
                    goto cleanup;

                if (!(tmpPath = virPCIDeviceGetIOMMUGroupDev(pci)))
                    goto cleanup;
13566 13567

                perm = VIR_CGROUP_DEVICE_RW;
13568
                if (teardown) {
13569
                    if (!virDomainDefHasVFIOHostdev(def))
13570 13571 13572 13573
                        includeVFIO = true;
                } else {
                    includeVFIO = true;
                }
13574 13575 13576 13577 13578 13579 13580 13581 13582 13583 13584 13585
            }
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
            if (dev->missing)
                break;
            usb = virUSBDeviceNew(usbsrc->bus,
                                  usbsrc->device,
                                  NULL);
            if (!usb)
                goto cleanup;

J
Ján Tomko 已提交
13586
            tmpPath = g_strdup(virUSBDeviceGetPath(usb));
13587
            perm = VIR_CGROUP_DEVICE_RW;
13588 13589 13590 13591 13592
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI:
            if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI) {
                virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
13593
                VIR_DEBUG("Not updating /dev for hostdev iSCSI path '%s'", iscsisrc->src->path);
13594 13595 13596 13597 13598 13599 13600 13601 13602 13603 13604 13605 13606
            } else {
                virDomainHostdevSubsysSCSIHostPtr scsihostsrc = &scsisrc->u.host;
                scsi = virSCSIDeviceNew(NULL,
                                        scsihostsrc->adapter,
                                        scsihostsrc->bus,
                                        scsihostsrc->target,
                                        scsihostsrc->unit,
                                        dev->readonly,
                                        dev->shareable);

                if (!scsi)
                    goto cleanup;

J
Ján Tomko 已提交
13607
                tmpPath = g_strdup(virSCSIDeviceGetPath(scsi));
13608 13609
                perm = virSCSIDeviceGetReadonly(scsi) ?
                    VIR_CGROUP_DEVICE_READ : VIR_CGROUP_DEVICE_RW;
13610 13611 13612 13613 13614 13615 13616 13617 13618
            }
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI_HOST: {
            if (hostsrc->protocol ==
                VIR_DOMAIN_HOSTDEV_SUBSYS_SCSI_HOST_PROTOCOL_TYPE_VHOST) {
                if (!(host = virSCSIVHostDeviceNew(hostsrc->wwpn)))
                    goto cleanup;

J
Ján Tomko 已提交
13619
                tmpPath = g_strdup(virSCSIVHostDeviceGetPath(host));
13620
                perm = VIR_CGROUP_DEVICE_RW;
13621 13622 13623 13624
            }
            break;
        }

13625
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
13626
            if (!(tmpPath = virMediatedDeviceGetIOMMUGroupDev(mdevsrc->uuidstr)))
13627 13628 13629 13630 13631
                goto cleanup;

            includeVFIO = true;
            perm = VIR_CGROUP_DEVICE_RW;
            break;
13632 13633 13634 13635 13636 13637 13638 13639 13640 13641 13642
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
            break;
        }
        break;

    case VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES:
    case VIR_DOMAIN_HOSTDEV_MODE_LAST:
        /* nada */
        break;
    }

13643 13644 13645 13646 13647
    if (tmpPath) {
        size_t toAlloc = 1;

        if (includeVFIO)
            toAlloc = 2;
13648

13649
        if (VIR_ALLOC_N(tmpPaths, toAlloc) < 0 ||
13650
            VIR_ALLOC_N(tmpPerms, toAlloc) < 0)
13651
            goto cleanup;
13652
        tmpPaths[0] = g_strdup(tmpPath);
13653 13654 13655 13656
        tmpNpaths = toAlloc;
        tmpPerms[0] = perm;

        if (includeVFIO) {
13657
            tmpPaths[1] = g_strdup(QEMU_DEV_VFIO);
13658 13659 13660 13661 13662 13663 13664 13665 13666 13667 13668 13669
            tmpPerms[1] = VIR_CGROUP_DEVICE_RW;
        }
    }

    *npaths = tmpNpaths;
    tmpNpaths = 0;
    *path = tmpPaths;
    tmpPaths = NULL;
    if (perms) {
        *perms = tmpPerms;
        tmpPerms = NULL;
    }
13670 13671
    ret = 0;
 cleanup:
13672
    virStringListFreeCount(tmpPaths, tmpNpaths);
13673
    VIR_FREE(tmpPerms);
13674 13675 13676 13677
    virPCIDeviceFree(pci);
    virUSBDeviceFree(usb);
    virSCSIDeviceFree(scsi);
    virSCSIVHostDeviceFree(host);
13678
    VIR_FREE(tmpPath);
13679 13680 13681 13682
    return ret;
}


13683 13684 13685 13686
/**
 * qemuDomainGetPreservedMountPath:
 * @cfg: driver configuration data
 * @vm: domain object
13687
 * @mountpoint: mount point path to convert
13688
 *
13689
 * For given @mountpoint return new path where the mount point
13690 13691 13692 13693 13694 13695 13696 13697
 * 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,
13698
                                const char *mountpoint)
13699 13700 13701
{
    char *path = NULL;
    char *tmp;
13702
    const char *suffix = mountpoint + strlen(QEMU_DEVPREFIX);
13703
    char *domname = virDomainDefGetShortName(vm->def);
13704 13705
    size_t off;

13706 13707 13708
    if (!domname)
        return NULL;

13709
    if (STREQ(mountpoint, "/dev"))
13710 13711
        suffix = "dev";

13712
    path = g_strdup_printf("%s/%s.%s", cfg->stateDir, domname, suffix);
13713

13714
    /* Now consider that @mountpoint is "/dev/blah/blah2".
13715 13716 13717 13718 13719 13720 13721 13722 13723 13724 13725 13726 13727
     * @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++;
    }

13728
    VIR_FREE(domname);
13729 13730 13731 13732
    return path;
}


13733 13734 13735 13736 13737 13738 13739 13740 13741 13742 13743 13744
/**
 * 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
13745
qemuDomainGetPreservedMounts(virQEMUDriverConfigPtr cfg,
13746 13747 13748 13749 13750 13751
                             virDomainObjPtr vm,
                             char ***devPath,
                             char ***devSavePath,
                             size_t *ndevPath)
{
    char **paths = NULL, **mounts = NULL;
13752
    size_t i, j, nmounts;
13753

13754
    if (virFileGetMountSubtree(QEMU_PROC_MOUNTS, "/dev",
13755 13756 13757 13758 13759 13760 13761 13762 13763
                               &mounts, &nmounts) < 0)
        goto error;

    if (!nmounts) {
        if (ndevPath)
            *ndevPath = 0;
        return 0;
    }

13764 13765 13766 13767 13768 13769 13770 13771 13772 13773 13774 13775
    /* 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) {
13776 13777
            char *c = STRSKIP(mounts[j], mounts[i]);

13778
            if (c && (*c == '/' || *c == '\0')) {
13779 13780 13781 13782 13783 13784 13785 13786
                VIR_DEBUG("Dropping path %s because of %s", mounts[j], mounts[i]);
                VIR_DELETE_ELEMENT(mounts, j, nmounts);
            } else {
                j++;
            }
        }
    }

13787 13788 13789 13790
    if (VIR_ALLOC_N(paths, nmounts) < 0)
        goto error;

    for (i = 0; i < nmounts; i++) {
13791
        if (!(paths[i] = qemuDomainGetPreservedMountPath(cfg, vm, mounts[i])))
13792 13793 13794 13795 13796 13797 13798 13799 13800 13801 13802 13803 13804 13805 13806 13807 13808 13809 13810 13811 13812 13813 13814 13815 13816
            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;
}


13817 13818
struct qemuDomainCreateDeviceData {
    const char *path;     /* Path to temp new /dev location */
13819 13820
    char * const *devMountsPath;
    size_t ndevMountsPath;
13821 13822 13823
};


13824
static int
13825
qemuDomainCreateDeviceRecursive(const char *device,
13826
                                const struct qemuDomainCreateDeviceData *data,
13827 13828
                                bool allow_noent,
                                unsigned int ttl)
13829 13830
{
    char *devicePath = NULL;
13831
    char *target = NULL;
13832 13833
    struct stat sb;
    int ret = -1;
13834
    bool isLink = false;
13835
    bool isDev = false;
13836
    bool isReg = false;
13837
    bool isDir = false;
13838
    bool create = false;
13839 13840 13841
#ifdef WITH_SELINUX
    char *tcon = NULL;
#endif
13842

13843 13844 13845 13846 13847 13848 13849
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             device);
        return ret;
    }

13850
    if (lstat(device, &sb) < 0) {
13851 13852
        if (errno == ENOENT && allow_noent) {
            /* Ignore non-existent device. */
13853
            return 0;
13854
        }
13855 13856
        virReportSystemError(errno, _("Unable to stat %s"), device);
        return ret;
13857 13858
    }

13859
    isLink = S_ISLNK(sb.st_mode);
13860
    isDev = S_ISCHR(sb.st_mode) || S_ISBLK(sb.st_mode);
13861
    isReg = S_ISREG(sb.st_mode) || S_ISFIFO(sb.st_mode) || S_ISSOCK(sb.st_mode);
13862
    isDir = S_ISDIR(sb.st_mode);
13863 13864 13865 13866 13867 13868 13869 13870 13871 13872 13873 13874 13875 13876 13877 13878

    /* 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.
     */
13879
    if (STRPREFIX(device, QEMU_DEVPREFIX)) {
13880
        size_t i;
13881

13882 13883 13884 13885 13886 13887 13888 13889 13890 13891
        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. */
13892 13893
            devicePath = g_strdup_printf("%s/%s", data->path,
                                         device + strlen(QEMU_DEVPREFIX));
13894 13895 13896 13897 13898 13899 13900 13901 13902 13903 13904 13905

            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]);
13906
        }
13907 13908
    }

13909
    if (isLink) {
13910 13911
        g_autoptr(GError) gerr = NULL;

13912 13913
        /* We are dealing with a symlink. Create a dangling symlink and descend
         * down one level which hopefully creates the symlink's target. */
13914 13915 13916
        if (!(target = g_file_read_link(device, &gerr))) {
            virReportError(VIR_ERR_SYSTEM_ERROR,
                           _("failed to resolve symlink %s: %s"), device, gerr->message);
13917 13918
            goto cleanup;
        }
13919

13920 13921 13922 13923 13924 13925 13926 13927 13928
        if (create &&
            symlink(target, devicePath) < 0) {
            if (errno == EEXIST) {
                ret = 0;
            } else {
                virReportSystemError(errno,
                                     _("unable to create symlink %s"),
                                     devicePath);
            }
13929 13930 13931
            goto cleanup;
        }

13932 13933 13934 13935 13936 13937 13938 13939 13940 13941
        /* Tricky part. If the target starts with a slash then we need to take
         * it as it is. Otherwise we need to replace the last component in the
         * original path with the link target:
         * /dev/rtc -> rtc0 (want /dev/rtc0)
         * /dev/disk/by-id/ata-SanDisk_SDSSDXPS480G_161101402485 -> ../../sda
         *   (want /dev/disk/by-id/../../sda)
         * /dev/stdout -> /proc/self/fd/1 (no change needed)
         */
        if (IS_RELATIVE_FILE_NAME(target)) {
            char *c = NULL, *tmp = NULL, *devTmp = NULL;
13942

13943
            devTmp = g_strdup(device);
13944

13945 13946 13947
            if ((c = strrchr(devTmp, '/')))
                *(c + 1) = '\0';

13948
            tmp = g_strdup_printf("%s%s", devTmp, target);
13949 13950
            VIR_FREE(devTmp);
            VIR_FREE(target);
13951
            target = g_steal_pointer(&tmp);
13952 13953
        }

13954
        if (qemuDomainCreateDeviceRecursive(target, data,
13955
                                            allow_noent, ttl - 1) < 0)
13956
            goto cleanup;
13957
    } else if (isDev) {
13958 13959 13960
        if (create) {
            unlink(devicePath);
            if (mknod(devicePath, sb.st_mode, sb.st_rdev) < 0) {
13961 13962 13963
                virReportSystemError(errno,
                                     _("Failed to make device %s"),
                                     devicePath);
13964
                goto cleanup;
13965
            }
13966
        }
13967
    } else if (isReg) {
13968
        if (create &&
13969
            virFileTouch(devicePath, sb.st_mode) < 0)
13970
            goto cleanup;
13971 13972
        /* Just create the file here so that code below sets
         * proper owner and mode. Bind mount only after that. */
13973 13974 13975 13976
    } else if (isDir) {
        if (create &&
            virFileMakePathWithMode(devicePath, sb.st_mode) < 0)
            goto cleanup;
13977 13978 13979 13980 13981
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       device, sb.st_mode);
        goto cleanup;
13982 13983 13984 13985
    }

    if (!create) {
        ret = 0;
13986 13987 13988
        goto cleanup;
    }

13989
    if (lchown(devicePath, sb.st_uid, sb.st_gid) < 0) {
13990 13991 13992 13993 13994 13995
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             devicePath);
        goto cleanup;
    }

13996 13997 13998 13999 14000 14001 14002 14003 14004
    /* 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;
    }

14005 14006 14007
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileCopyACLs(device, devicePath) < 0 &&
14008 14009 14010 14011 14012 14013 14014
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Failed to copy ACLs on device %s"),
                             devicePath);
        goto cleanup;
    }

14015
#ifdef WITH_SELINUX
14016
    if (lgetfilecon_raw(device, &tcon) < 0 &&
14017 14018 14019
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"),
14020
                             device);
14021 14022 14023 14024
        goto cleanup;
    }

    if (tcon &&
14025
        lsetfilecon_raw(devicePath, (const char *)tcon) < 0) {
14026 14027 14028 14029 14030 14031 14032 14033 14034 14035 14036
        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

14037
    /* Finish mount process started earlier. */
14038
    if ((isReg || isDir) &&
14039 14040 14041
        virFileBindMountDevice(device, devicePath) < 0)
        goto cleanup;

14042 14043
    ret = 0;
 cleanup:
14044
    VIR_FREE(target);
14045
    VIR_FREE(devicePath);
14046 14047 14048
#ifdef WITH_SELINUX
    freecon(tcon);
#endif
14049 14050 14051 14052
    return ret;
}


14053 14054
static int
qemuDomainCreateDevice(const char *device,
14055
                       const struct qemuDomainCreateDeviceData *data,
14056 14057 14058 14059
                       bool allow_noent)
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

14060
    return qemuDomainCreateDeviceRecursive(device, data,
14061 14062 14063
                                           allow_noent, symloop_max);
}

14064 14065

static int
14066
qemuDomainPopulateDevices(virQEMUDriverConfigPtr cfg,
J
Ján Tomko 已提交
14067
                          virDomainObjPtr vm G_GNUC_UNUSED,
14068
                          const struct qemuDomainCreateDeviceData *data)
14069 14070 14071 14072 14073 14074 14075 14076
{
    const char *const *devices = (const char *const *) cfg->cgroupDeviceACL;
    size_t i;

    if (!devices)
        devices = defaultDeviceACL;

    for (i = 0; devices[i]; i++) {
14077
        if (qemuDomainCreateDevice(devices[i], data, true) < 0)
14078
            return -1;
14079 14080
    }

14081
    return 0;
14082 14083 14084 14085
}


static int
14086 14087
qemuDomainSetupDev(virQEMUDriverConfigPtr cfg,
                   virSecurityManagerPtr mgr,
14088
                   virDomainObjPtr vm,
14089
                   const struct qemuDomainCreateDeviceData *data)
14090 14091 14092 14093 14094 14095 14096
{
    char *mount_options = NULL;
    char *opts = NULL;
    int ret = -1;

    VIR_DEBUG("Setting up /dev/ for domain %s", vm->def->name);

14097
    mount_options = qemuSecurityGetMountOptions(mgr, vm->def);
14098

14099 14100
    if (!mount_options)
        mount_options = g_strdup("");
14101 14102 14103 14104 14105

    /*
     * tmpfs is limited to 64kb, since we only have device nodes in there
     * and don't want to DOS the entire OS RAM usage
     */
14106
    opts = g_strdup_printf("mode=755,size=65536%s", mount_options);
14107

14108
    if (virFileSetupDev(data->path, opts) < 0)
14109 14110
        goto cleanup;

14111
    if (qemuDomainPopulateDevices(cfg, vm, data) < 0)
14112 14113 14114 14115 14116 14117 14118 14119 14120 14121
        goto cleanup;

    ret = 0;
 cleanup:
    VIR_FREE(opts);
    VIR_FREE(mount_options);
    return ret;
}


14122
static int
J
Ján Tomko 已提交
14123
qemuDomainSetupDisk(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14124
                    virDomainDiskDefPtr disk,
14125
                    const struct qemuDomainCreateDeviceData *data)
14126 14127 14128 14129 14130
{
    virStorageSourcePtr next;
    char *dst = NULL;
    int ret = -1;

14131
    for (next = disk->src; virStorageSourceIsBacking(next); next = next->backingStore) {
14132
        if (!next->path || !virStorageSourceIsLocalStorage(next)) {
14133 14134 14135 14136
            /* Not creating device. Just continue. */
            continue;
        }

14137
        if (qemuDomainCreateDevice(next->path, data, false) < 0)
14138 14139 14140
            goto cleanup;
    }

14141
    /* qemu-pr-helper might require access to /dev/mapper/control. */
14142
    if (disk->src->pr &&
14143
        qemuDomainCreateDevice(QEMU_DEVICE_MAPPER_CONTROL_PATH, data, true) < 0)
14144 14145
        goto cleanup;

14146 14147 14148 14149 14150 14151 14152 14153
    ret = 0;
 cleanup:
    VIR_FREE(dst);
    return ret;
}


static int
14154
qemuDomainSetupAllDisks(virQEMUDriverConfigPtr cfg,
14155
                        virDomainObjPtr vm,
14156
                        const struct qemuDomainCreateDeviceData *data)
14157 14158 14159 14160 14161
{
    size_t i;
    VIR_DEBUG("Setting up disks");

    for (i = 0; i < vm->def->ndisks; i++) {
14162
        if (qemuDomainSetupDisk(cfg,
14163
                                vm->def->disks[i],
14164
                                data) < 0)
14165 14166 14167 14168 14169 14170 14171 14172
            return -1;
    }

    VIR_DEBUG("Setup all disks");
    return 0;
}


14173
static int
J
Ján Tomko 已提交
14174
qemuDomainSetupHostdev(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14175
                       virDomainHostdevDefPtr dev,
14176
                       const struct qemuDomainCreateDeviceData *data)
14177 14178
{
    int ret = -1;
14179 14180
    char **path = NULL;
    size_t i, npaths = 0;
14181

14182
    if (qemuDomainGetHostdevPath(NULL, dev, false, &npaths, &path, NULL) < 0)
14183 14184
        goto cleanup;

14185
    for (i = 0; i < npaths; i++) {
14186
        if (qemuDomainCreateDevice(path[i], data, false) < 0)
14187
            goto cleanup;
14188 14189 14190 14191
    }

    ret = 0;
 cleanup:
14192 14193
    for (i = 0; i < npaths; i++)
        VIR_FREE(path[i]);
14194 14195 14196 14197 14198 14199
    VIR_FREE(path);
    return ret;
}


static int
14200
qemuDomainSetupAllHostdevs(virQEMUDriverConfigPtr cfg,
14201
                           virDomainObjPtr vm,
14202
                           const struct qemuDomainCreateDeviceData *data)
14203 14204 14205 14206 14207
{
    size_t i;

    VIR_DEBUG("Setting up hostdevs");
    for (i = 0; i < vm->def->nhostdevs; i++) {
14208
        if (qemuDomainSetupHostdev(cfg,
14209
                                   vm->def->hostdevs[i],
14210
                                   data) < 0)
14211 14212 14213 14214 14215 14216 14217
            return -1;
    }
    VIR_DEBUG("Setup all hostdevs");
    return 0;
}


M
Michal Privoznik 已提交
14218
static int
J
Ján Tomko 已提交
14219
qemuDomainSetupMemory(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
M
Michal Privoznik 已提交
14220
                      virDomainMemoryDefPtr mem,
14221
                      const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
14222 14223 14224 14225
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

14226
    return qemuDomainCreateDevice(mem->nvdimmPath, data, false);
M
Michal Privoznik 已提交
14227 14228 14229 14230 14231 14232
}


static int
qemuDomainSetupAllMemories(virQEMUDriverConfigPtr cfg,
                           virDomainObjPtr vm,
14233
                           const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
14234 14235 14236 14237 14238 14239 14240
{
    size_t i;

    VIR_DEBUG("Setting up memories");
    for (i = 0; i < vm->def->nmems; i++) {
        if (qemuDomainSetupMemory(cfg,
                                  vm->def->mems[i],
14241
                                  data) < 0)
M
Michal Privoznik 已提交
14242 14243 14244 14245 14246 14247 14248
            return -1;
    }
    VIR_DEBUG("Setup all memories");
    return 0;
}


14249
static int
J
Ján Tomko 已提交
14250
qemuDomainSetupChardev(virDomainDefPtr def G_GNUC_UNUSED,
14251 14252 14253
                       virDomainChrDefPtr dev,
                       void *opaque)
{
14254
    const struct qemuDomainCreateDeviceData *data = opaque;
14255
    const char *path = NULL;
14256

14257
    if (!(path = virDomainChrSourceDefGetPath(dev->source)))
14258 14259
        return 0;

14260 14261 14262 14263 14264 14265
    /* 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);
14266 14267 14268 14269
}


static int
J
Ján Tomko 已提交
14270
qemuDomainSetupAllChardevs(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14271
                           virDomainObjPtr vm,
14272
                           const struct qemuDomainCreateDeviceData *data)
14273 14274 14275 14276 14277 14278
{
    VIR_DEBUG("Setting up chardevs");

    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuDomainSetupChardev,
14279
                               (void *)data) < 0)
14280 14281 14282 14283 14284 14285 14286
        return -1;

    VIR_DEBUG("Setup all chardevs");
    return 0;
}


14287
static int
J
Ján Tomko 已提交
14288
qemuDomainSetupTPM(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14289
                   virDomainObjPtr vm,
14290
                   const struct qemuDomainCreateDeviceData *data)
14291 14292 14293 14294 14295 14296 14297 14298 14299 14300 14301
{
    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,
14302
                                   data, false) < 0)
14303 14304 14305
            return -1;
        break;

14306
    case VIR_DOMAIN_TPM_TYPE_EMULATOR:
14307 14308 14309 14310 14311 14312 14313 14314 14315 14316
    case VIR_DOMAIN_TPM_TYPE_LAST:
        /* nada */
        break;
    }

    VIR_DEBUG("Setup TPM");
    return 0;
}


14317
static int
J
Ján Tomko 已提交
14318
qemuDomainSetupGraphics(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14319
                        virDomainGraphicsDefPtr gfx,
14320
                        const struct qemuDomainCreateDeviceData *data)
14321
{
14322
    const char *rendernode = virDomainGraphicsGetRenderNode(gfx);
14323

14324
    if (!rendernode)
14325 14326
        return 0;

14327
    return qemuDomainCreateDevice(rendernode, data, false);
14328 14329 14330 14331
}


static int
14332
qemuDomainSetupAllGraphics(virQEMUDriverConfigPtr cfg,
14333
                           virDomainObjPtr vm,
14334
                           const struct qemuDomainCreateDeviceData *data)
14335 14336 14337 14338 14339
{
    size_t i;

    VIR_DEBUG("Setting up graphics");
    for (i = 0; i < vm->def->ngraphics; i++) {
14340
        if (qemuDomainSetupGraphics(cfg,
14341
                                    vm->def->graphics[i],
14342
                                    data) < 0)
14343 14344 14345 14346 14347 14348 14349 14350
            return -1;
    }

    VIR_DEBUG("Setup all graphics");
    return 0;
}


14351
static int
J
Ján Tomko 已提交
14352
qemuDomainSetupInput(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14353
                     virDomainInputDefPtr input,
14354
                     const struct qemuDomainCreateDeviceData *data)
14355
{
J
Ján Tomko 已提交
14356
    const char *path = virDomainInputDefGetPath(input);
14357

J
Ján Tomko 已提交
14358 14359
    if (path && qemuDomainCreateDevice(path, data, false) < 0)
        return -1;
14360

J
Ján Tomko 已提交
14361
    return 0;
14362 14363 14364 14365
}


static int
14366
qemuDomainSetupAllInputs(virQEMUDriverConfigPtr cfg,
14367
                         virDomainObjPtr vm,
14368
                         const struct qemuDomainCreateDeviceData *data)
14369 14370 14371
{
    size_t i;

14372
    VIR_DEBUG("Setting up inputs");
14373
    for (i = 0; i < vm->def->ninputs; i++) {
14374
        if (qemuDomainSetupInput(cfg,
14375
                                 vm->def->inputs[i],
14376
                                 data) < 0)
14377 14378
            return -1;
    }
14379
    VIR_DEBUG("Setup all inputs");
14380 14381 14382 14383
    return 0;
}


14384
static int
J
Ján Tomko 已提交
14385
qemuDomainSetupRNG(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14386
                   virDomainRNGDefPtr rng,
14387
                   const struct qemuDomainCreateDeviceData *data)
14388 14389 14390
{
    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
14391
        if (qemuDomainCreateDevice(rng->source.file, data, false) < 0)
14392 14393 14394 14395 14396 14397 14398 14399 14400 14401 14402 14403 14404
            return -1;

    case VIR_DOMAIN_RNG_BACKEND_EGD:
    case VIR_DOMAIN_RNG_BACKEND_LAST:
        /* nada */
        break;
    }

    return 0;
}


static int
14405
qemuDomainSetupAllRNGs(virQEMUDriverConfigPtr cfg,
14406
                       virDomainObjPtr vm,
14407
                       const struct qemuDomainCreateDeviceData *data)
14408 14409 14410 14411 14412
{
    size_t i;

    VIR_DEBUG("Setting up RNGs");
    for (i = 0; i < vm->def->nrngs; i++) {
14413
        if (qemuDomainSetupRNG(cfg,
14414
                               vm->def->rngs[i],
14415
                               data) < 0)
14416 14417 14418 14419 14420 14421 14422 14423
            return -1;
    }

    VIR_DEBUG("Setup all RNGs");
    return 0;
}


14424
static int
J
Ján Tomko 已提交
14425
qemuDomainSetupLoader(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14426 14427 14428 14429 14430 14431 14432 14433 14434 14435 14436
                      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)
14437
                return -1;
14438 14439 14440 14441
            break;

        case VIR_DOMAIN_LOADER_TYPE_PFLASH:
            if (qemuDomainCreateDevice(loader->path, data, false) < 0)
14442
                return -1;
14443 14444 14445

            if (loader->nvram &&
                qemuDomainCreateDevice(loader->nvram, data, false) < 0)
14446
                return -1;
14447 14448
            break;

14449
        case VIR_DOMAIN_LOADER_TYPE_NONE:
14450 14451 14452 14453 14454 14455
        case VIR_DOMAIN_LOADER_TYPE_LAST:
            break;
        }
    }

    VIR_DEBUG("Setup loader");
14456
    return 0;
14457 14458 14459
}


14460
static int
J
Ján Tomko 已提交
14461
qemuDomainSetupLaunchSecurity(virQEMUDriverConfigPtr cfg G_GNUC_UNUSED,
14462 14463 14464 14465 14466 14467 14468 14469 14470 14471
                              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");

14472
    if (qemuDomainCreateDevice(QEMU_DEV_SEV, data, false) < 0)
14473 14474 14475 14476 14477 14478 14479
        return -1;

    VIR_DEBUG("Set up launch security");
    return 0;
}


14480
int
14481 14482
qemuDomainBuildNamespace(virQEMUDriverConfigPtr cfg,
                         virSecurityManagerPtr mgr,
14483 14484
                         virDomainObjPtr vm)
{
14485
    struct qemuDomainCreateDeviceData data;
14486
    char *devPath = NULL;
14487
    char **devMountsPath = NULL, **devMountsSavePath = NULL;
14488 14489 14490 14491 14492 14493 14494 14495
    size_t ndevMountsPath = 0, i;
    int ret = -1;

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT)) {
        ret = 0;
        goto cleanup;
    }

14496
    if (qemuDomainGetPreservedMounts(cfg, vm,
14497 14498
                                     &devMountsPath, &devMountsSavePath,
                                     &ndevMountsPath) < 0)
14499 14500
        goto cleanup;

14501 14502 14503 14504 14505 14506 14507 14508 14509 14510 14511 14512 14513
    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;
    }

14514
    data.path = devPath;
14515 14516
    data.devMountsPath = devMountsPath;
    data.ndevMountsPath = ndevMountsPath;
14517

14518 14519 14520
    if (virProcessSetupPrivateMountNS() < 0)
        goto cleanup;

14521
    if (qemuDomainSetupDev(cfg, mgr, vm, &data) < 0)
14522 14523
        goto cleanup;

14524
    if (qemuDomainSetupAllDisks(cfg, vm, &data) < 0)
14525 14526
        goto cleanup;

14527
    if (qemuDomainSetupAllHostdevs(cfg, vm, &data) < 0)
14528 14529
        goto cleanup;

14530
    if (qemuDomainSetupAllMemories(cfg, vm, &data) < 0)
14531 14532
        goto cleanup;

14533
    if (qemuDomainSetupAllChardevs(cfg, vm, &data) < 0)
14534 14535
        goto cleanup;

14536
    if (qemuDomainSetupTPM(cfg, vm, &data) < 0)
14537 14538
        goto cleanup;

14539
    if (qemuDomainSetupAllGraphics(cfg, vm, &data) < 0)
14540 14541
        goto cleanup;

14542
    if (qemuDomainSetupAllInputs(cfg, vm, &data) < 0)
14543 14544
        goto cleanup;

14545
    if (qemuDomainSetupAllRNGs(cfg, vm, &data) < 0)
14546 14547
        goto cleanup;

14548 14549 14550
    if (qemuDomainSetupLoader(cfg, vm, &data) < 0)
        goto cleanup;

14551 14552 14553
    if (qemuDomainSetupLaunchSecurity(cfg, vm, &data) < 0)
        goto cleanup;

14554 14555
    /* Save some mount points because we want to share them with the host */
    for (i = 0; i < ndevMountsPath; i++) {
14556 14557
        struct stat sb;

14558 14559 14560
        if (devMountsSavePath[i] == devPath)
            continue;

14561 14562 14563 14564 14565 14566 14567
        if (stat(devMountsPath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsPath[i]);
            goto cleanup;
        }

14568 14569 14570
        /* At this point, devMountsPath is either:
         * a file (regular or special), or
         * a directory. */
14571
        if ((S_ISDIR(sb.st_mode) && virFileMakePath(devMountsSavePath[i]) < 0) ||
14572
            (!S_ISDIR(sb.st_mode) && virFileTouch(devMountsSavePath[i], sb.st_mode) < 0)) {
14573 14574 14575 14576 14577 14578
            virReportSystemError(errno,
                                 _("Failed to create %s"),
                                 devMountsSavePath[i]);
            goto cleanup;
        }

14579
        if (virFileMoveMount(devMountsPath[i], devMountsSavePath[i]) < 0)
14580 14581 14582
            goto cleanup;
    }

14583
    if (virFileMoveMount(devPath, "/dev") < 0)
14584 14585 14586
        goto cleanup;

    for (i = 0; i < ndevMountsPath; i++) {
14587 14588
        struct stat sb;

14589 14590 14591
        if (devMountsSavePath[i] == devPath)
            continue;

14592 14593 14594 14595
        if (stat(devMountsSavePath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsSavePath[i]);
14596 14597 14598
            goto cleanup;
        }

14599 14600 14601 14602 14603 14604 14605 14606 14607 14608 14609 14610 14611 14612 14613
        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;
            }
        }

14614
        if (virFileMoveMount(devMountsSavePath[i], devMountsPath[i]) < 0)
14615 14616 14617 14618 14619
            goto cleanup;
    }

    ret = 0;
 cleanup:
14620 14621 14622 14623 14624 14625 14626
    for (i = 0; i < ndevMountsPath; i++) {
        /* The path can be either a regular file or a dir. */
        if (virFileIsDir(devMountsSavePath[i]))
            rmdir(devMountsSavePath[i]);
        else
            unlink(devMountsSavePath[i]);
    }
14627
    virStringListFreeCount(devMountsPath, ndevMountsPath);
14628
    virStringListFreeCount(devMountsSavePath, ndevMountsPath);
14629 14630 14631 14632 14633 14634 14635 14636 14637 14638 14639
    return ret;
}


int
qemuDomainCreateNamespace(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

14640 14641
    if (virBitmapIsBitSet(cfg->namespaces, QEMU_DOMAIN_NS_MOUNT) &&
        qemuDomainEnableNamespace(vm, QEMU_DOMAIN_NS_MOUNT) < 0)
M
Michal Privoznik 已提交
14642
        goto cleanup;
14643 14644 14645 14646 14647 14648 14649 14650

    ret = 0;
 cleanup:
    virObjectUnref(cfg);
    return ret;
}


14651
void
J
Ján Tomko 已提交
14652
qemuDomainDestroyNamespace(virQEMUDriverPtr driver G_GNUC_UNUSED,
14653 14654 14655 14656 14657 14658 14659
                           virDomainObjPtr vm)
{
    if (qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        qemuDomainDisableNamespace(vm, QEMU_DOMAIN_NS_MOUNT);
}


14660
bool
J
Ján Tomko 已提交
14661
qemuDomainNamespaceAvailable(qemuDomainNamespace ns G_GNUC_UNUSED)
14662 14663 14664 14665 14666 14667 14668 14669 14670 14671 14672 14673 14674 14675 14676 14677 14678 14679 14680 14681 14682 14683 14684 14685 14686 14687 14688
{
#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__) */
}


14689 14690 14691 14692
struct qemuDomainAttachDeviceMknodData {
    virQEMUDriverPtr driver;
    virDomainObjPtr vm;
    const char *file;
14693
    const char *target;
14694 14695
    struct stat sb;
    void *acl;
14696 14697 14698
#ifdef WITH_SELINUX
    char *tcon;
#endif
14699 14700 14701
};


14702 14703
/* Our way of creating devices is highly linux specific */
#if defined(__linux__)
14704
static int
J
Ján Tomko 已提交
14705
qemuDomainAttachDeviceMknodHelper(pid_t pid G_GNUC_UNUSED,
14706 14707 14708 14709
                                  void *opaque)
{
    struct qemuDomainAttachDeviceMknodData *data = opaque;
    int ret = -1;
14710
    bool delDevice = false;
14711
    bool isLink = S_ISLNK(data->sb.st_mode);
14712
    bool isDev = S_ISCHR(data->sb.st_mode) || S_ISBLK(data->sb.st_mode);
14713
    bool isReg = S_ISREG(data->sb.st_mode) || S_ISFIFO(data->sb.st_mode) || S_ISSOCK(data->sb.st_mode);
14714
    bool isDir = S_ISDIR(data->sb.st_mode);
14715

14716
    qemuSecurityPostFork(data->driver->securityManager);
14717 14718 14719 14720 14721 14722 14723

    if (virFileMakeParentPath(data->file) < 0) {
        virReportSystemError(errno,
                             _("Unable to create %s"), data->file);
        goto cleanup;
    }

14724 14725
    if (isLink) {
        VIR_DEBUG("Creating symlink %s -> %s", data->file, data->target);
14726 14727 14728 14729 14730 14731 14732 14733 14734 14735 14736 14737

        /* 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;
        }

14738
        if (symlink(data->target, data->file) < 0) {
14739 14740 14741 14742
            virReportSystemError(errno,
                                 _("Unable to create symlink %s (pointing to %s)"),
                                 data->file, data->target);
            goto cleanup;
14743 14744
        } else {
            delDevice = true;
14745
        }
14746
    } else if (isDev) {
14747 14748
        VIR_DEBUG("Creating dev %s (%d,%d)",
                  data->file, major(data->sb.st_rdev), minor(data->sb.st_rdev));
14749
        unlink(data->file);
14750
        if (mknod(data->file, data->sb.st_mode, data->sb.st_rdev) < 0) {
14751 14752 14753 14754
            virReportSystemError(errno,
                                 _("Unable to create device %s"),
                                 data->file);
            goto cleanup;
14755 14756 14757
        } else {
            delDevice = true;
        }
14758
    } else if (isReg || isDir) {
14759 14760 14761 14762 14763
        /* 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 &&
14764
            errno != ENOENT && errno != EINVAL) {
14765 14766 14767 14768 14769
            virReportSystemError(errno,
                                 _("Unable to umount %s"),
                                 data->file);
            goto cleanup;
        }
14770 14771
        if ((isReg && virFileTouch(data->file, data->sb.st_mode) < 0) ||
            (isDir && virFileMakePathWithMode(data->file, data->sb.st_mode) < 0))
14772 14773 14774 14775
            goto cleanup;
        delDevice = true;
        /* Just create the file here so that code below sets
         * proper owner and mode. Move the mount only after that. */
14776 14777 14778 14779 14780
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       data->file, data->sb.st_mode);
        goto cleanup;
14781 14782
    }

14783 14784 14785 14786 14787 14788 14789
    if (lchown(data->file, data->sb.st_uid, data->sb.st_gid) < 0) {
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             data->file);
        goto cleanup;
    }

14790 14791 14792 14793 14794 14795 14796 14797 14798
    /* 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;
    }

14799 14800 14801
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileSetACLs(data->file, data->acl) < 0 &&
14802 14803 14804 14805 14806 14807
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to set ACLs on %s"), data->file);
        goto cleanup;
    }

14808
# ifdef WITH_SELINUX
14809
    if (data->tcon &&
14810
        lsetfilecon_raw(data->file, (const char *)data->tcon) < 0) {
14811 14812 14813 14814 14815 14816 14817 14818 14819
        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;
        }
    }
14820
# endif
14821

14822
    /* Finish mount process started earlier. */
14823
    if ((isReg || isDir) &&
14824 14825 14826
        virFileMoveMount(data->target, data->file) < 0)
        goto cleanup;

14827 14828
    ret = 0;
 cleanup:
14829 14830 14831 14832 14833 14834
    if (ret < 0 && delDevice) {
        if (isDir)
            virFileDeleteTree(data->file);
        else
            unlink(data->file);
    }
14835
# ifdef WITH_SELINUX
14836
    freecon(data->tcon);
14837
# endif
14838 14839 14840 14841 14842 14843
    virFileFreeACLs(&data->acl);
    return ret;
}


static int
14844 14845 14846
qemuDomainAttachDeviceMknodRecursive(virQEMUDriverPtr driver,
                                     virDomainObjPtr vm,
                                     const char *file,
14847 14848
                                     char * const *devMountsPath,
                                     size_t ndevMountsPath,
14849
                                     unsigned int ttl)
14850
{
14851
    virQEMUDriverConfigPtr cfg = NULL;
14852 14853
    struct qemuDomainAttachDeviceMknodData data;
    int ret = -1;
14854 14855
    char *target = NULL;
    bool isLink;
14856
    bool isReg;
14857
    bool isDir;
14858

14859 14860 14861 14862 14863 14864 14865
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             file);
        return ret;
    }

14866 14867 14868 14869 14870 14871
    memset(&data, 0, sizeof(data));

    data.driver = driver;
    data.vm = vm;
    data.file = file;

14872
    if (lstat(file, &data.sb) < 0) {
14873 14874 14875 14876 14877
        virReportSystemError(errno,
                             _("Unable to access %s"), file);
        return ret;
    }

14878
    isLink = S_ISLNK(data.sb.st_mode);
14879
    isReg = S_ISREG(data.sb.st_mode) || S_ISFIFO(data.sb.st_mode) || S_ISSOCK(data.sb.st_mode);
14880
    isDir = S_ISDIR(data.sb.st_mode);
14881

14882
    if ((isReg || isDir) && STRPREFIX(file, QEMU_DEVPREFIX)) {
14883 14884 14885 14886 14887 14888 14889 14890 14891
        cfg = virQEMUDriverGetConfig(driver);
        if (!(target = qemuDomainGetPreservedMountPath(cfg, vm, file)))
            goto cleanup;

        if (virFileBindMountDevice(file, target) < 0)
            goto cleanup;

        data.target = target;
    } else if (isLink) {
14892 14893 14894 14895 14896
        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);
14897 14898 14899 14900 14901 14902
            return ret;
        }

        if (IS_RELATIVE_FILE_NAME(target)) {
            char *c = NULL, *tmp = NULL, *fileTmp = NULL;

14903
            fileTmp = g_strdup(file);
14904 14905 14906 14907

            if ((c = strrchr(fileTmp, '/')))
                *(c + 1) = '\0';

14908
            tmp = g_strdup_printf("%s%s", fileTmp, target);
14909 14910
            VIR_FREE(fileTmp);
            VIR_FREE(target);
14911
            target = g_steal_pointer(&tmp);
14912 14913 14914 14915 14916 14917 14918 14919
        }

        data.target = target;
    }

    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileGetACLs(file, &data.acl) < 0 &&
14920 14921 14922
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to get ACLs on %s"), file);
14923
        goto cleanup;
14924 14925
    }

14926
# ifdef WITH_SELINUX
14927
    if (lgetfilecon_raw(file, &data.tcon) < 0 &&
14928 14929 14930 14931 14932
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"), file);
        goto cleanup;
    }
14933
# endif
14934

14935
    if (STRPREFIX(file, QEMU_DEVPREFIX)) {
14936 14937 14938 14939 14940 14941 14942 14943 14944 14945 14946 14947
        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;
14948

14949 14950 14951 14952 14953 14954
            if (virProcessRunInMountNamespace(vm->pid,
                                              qemuDomainAttachDeviceMknodHelper,
                                              &data) < 0) {
                qemuSecurityPostFork(driver->securityManager);
                goto cleanup;
            }
14955
            qemuSecurityPostFork(driver->securityManager);
14956 14957 14958
        } else {
            VIR_DEBUG("Skipping dev %s because of %s mount point",
                      file, devMountsPath[i]);
14959
        }
14960 14961
    }

14962
    if (isLink &&
14963 14964 14965
        qemuDomainAttachDeviceMknodRecursive(driver, vm, target,
                                             devMountsPath, ndevMountsPath,
                                             ttl -1) < 0)
14966
        goto cleanup;
14967 14968 14969

    ret = 0;
 cleanup:
14970
# ifdef WITH_SELINUX
14971
    freecon(data.tcon);
14972
# endif
14973
    virFileFreeACLs(&data.acl);
14974 14975
    if (isReg && target)
        umount(target);
14976
    VIR_FREE(target);
14977
    virObjectUnref(cfg);
14978
    return ret;
14979 14980 14981
}


14982 14983 14984 14985
#else /* !defined(__linux__) */


static int
J
Ján Tomko 已提交
14986 14987 14988 14989 14990 14991
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)
14992 14993 14994 14995 14996 14997 14998 14999 15000 15001
{
    virReportSystemError(ENOSYS, "%s",
                         _("Namespaces are not supported on this platform."));
    return -1;
}


#endif /* !defined(__linux__) */


15002 15003 15004
static int
qemuDomainAttachDeviceMknod(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
15005 15006 15007
                            const char *file,
                            char * const *devMountsPath,
                            size_t ndevMountsPath)
15008 15009 15010
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

15011 15012 15013
    return qemuDomainAttachDeviceMknodRecursive(driver, vm, file,
                                                devMountsPath, ndevMountsPath,
                                                symloop_max);
15014 15015 15016
}


15017
static int
J
Ján Tomko 已提交
15018
qemuDomainDetachDeviceUnlinkHelper(pid_t pid G_GNUC_UNUSED,
15019 15020 15021 15022 15023 15024 15025 15026 15027 15028 15029 15030 15031 15032 15033 15034
                                   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
15035
qemuDomainDetachDeviceUnlink(virQEMUDriverPtr driver G_GNUC_UNUSED,
15036
                             virDomainObjPtr vm,
15037 15038 15039
                             const char *file,
                             char * const *devMountsPath,
                             size_t ndevMountsPath)
15040
{
15041
    size_t i;
15042

15043
    if (STRPREFIX(file, QEMU_DEVPREFIX)) {
15044 15045 15046 15047 15048 15049 15050 15051 15052 15053 15054
        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)
15055
                return -1;
15056 15057 15058
        }
    }

15059
    return 0;
15060 15061 15062
}


15063 15064 15065 15066
static int
qemuDomainNamespaceMknodPaths(virDomainObjPtr vm,
                              const char **paths,
                              size_t npaths)
15067
{
15068 15069 15070
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    virQEMUDriverConfigPtr cfg;
15071 15072
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0;
15073
    int ret = -1;
15074
    size_t i;
15075

15076 15077
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
15078 15079
        return 0;

15080 15081 15082 15083 15084 15085
    cfg = virQEMUDriverGetConfig(driver);
    if (qemuDomainGetPreservedMounts(cfg, vm,
                                     &devMountsPath, NULL,
                                     &ndevMountsPath) < 0)
        goto cleanup;

15086 15087 15088 15089 15090 15091 15092 15093 15094 15095 15096 15097 15098 15099 15100 15101
    for (i = 0; i < npaths; i++) {
        if (qemuDomainAttachDeviceMknod(driver,
                                        vm,
                                        paths[i],
                                        devMountsPath, ndevMountsPath) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    virObjectUnref(cfg);
    return ret;
}


15102 15103 15104 15105 15106 15107 15108 15109 15110 15111
static int
qemuDomainNamespaceMknodPath(virDomainObjPtr vm,
                             const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceMknodPaths(vm, paths, 1);
}


15112 15113 15114 15115 15116 15117 15118 15119 15120 15121 15122 15123 15124
static int
qemuDomainNamespaceUnlinkPaths(virDomainObjPtr vm,
                               const char **paths,
                               size_t npaths)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    virQEMUDriverConfigPtr cfg;
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0;
    size_t i;
    int ret = -1;

15125 15126
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
15127 15128 15129 15130 15131 15132 15133 15134 15135 15136 15137 15138 15139 15140 15141
        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;
    }

15142
    ret = 0;
15143 15144 15145 15146 15147 15148 15149
 cleanup:
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    virObjectUnref(cfg);
    return ret;
}


15150 15151 15152 15153 15154 15155 15156 15157 15158 15159
static int
qemuDomainNamespaceUnlinkPath(virDomainObjPtr vm,
                              const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceUnlinkPaths(vm, paths, 1);
}


15160
int
15161
qemuDomainNamespaceSetupDisk(virDomainObjPtr vm,
15162 15163 15164
                             virStorageSourcePtr src)
{
    virStorageSourcePtr next;
15165
    const char **paths = NULL;
15166
    size_t npaths = 0;
15167
    char *dmPath = NULL;
15168 15169
    int ret = -1;

15170
    for (next = src; virStorageSourceIsBacking(next); next = next->backingStore) {
15171 15172
        if (virStorageSourceIsEmpty(next) ||
            !virStorageSourceIsLocalStorage(next)) {
15173 15174 15175 15176
            /* Not creating device. Just continue. */
            continue;
        }

15177
        if (VIR_APPEND_ELEMENT_COPY(paths, npaths, next->path) < 0)
15178 15179 15180
            goto cleanup;
    }

15181
    /* qemu-pr-helper might require access to /dev/mapper/control. */
15182 15183 15184 15185 15186
    if (src->pr) {
        dmPath = g_strdup(QEMU_DEVICE_MAPPER_CONTROL_PATH);
        if (VIR_APPEND_ELEMENT_COPY(paths, npaths, dmPath) < 0)
            goto cleanup;
    }
15187

15188
    if (qemuDomainNamespaceMknodPaths(vm, paths, npaths) < 0)
15189
        goto cleanup;
15190

15191 15192
    ret = 0;
 cleanup:
15193
    VIR_FREE(dmPath);
15194
    VIR_FREE(paths);
15195 15196 15197 15198 15199
    return ret;
}


int
J
Ján Tomko 已提交
15200 15201
qemuDomainNamespaceTeardownDisk(virDomainObjPtr vm G_GNUC_UNUSED,
                                virStorageSourcePtr src G_GNUC_UNUSED)
15202 15203 15204 15205 15206 15207 15208 15209 15210
{
    /* 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;
}
15211 15212 15213


int
15214
qemuDomainNamespaceSetupHostdev(virDomainObjPtr vm,
15215 15216 15217
                                virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
15218
    char **paths = NULL;
15219
    size_t i, npaths = 0;
15220

15221
    if (qemuDomainGetHostdevPath(NULL, hostdev, false, &npaths, &paths, NULL) < 0)
15222 15223
        goto cleanup;

15224
    if (qemuDomainNamespaceMknodPaths(vm, (const char **)paths, npaths) < 0)
15225 15226
        goto cleanup;

15227 15228
    ret = 0;
 cleanup:
15229
    for (i = 0; i < npaths; i++)
15230 15231
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
15232 15233 15234 15235 15236
    return ret;
}


int
15237
qemuDomainNamespaceTeardownHostdev(virDomainObjPtr vm,
15238 15239 15240
                                   virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
15241
    char **paths = NULL;
15242
    size_t i, npaths = 0;
15243

15244
    if (qemuDomainGetHostdevPath(vm->def, hostdev, true,
15245
                                 &npaths, &paths, NULL) < 0)
15246 15247
        goto cleanup;

15248
    if (qemuDomainNamespaceUnlinkPaths(vm, (const char **)paths, npaths) < 0)
15249 15250
        goto cleanup;

15251 15252
    ret = 0;
 cleanup:
15253
    for (i = 0; i < npaths; i++)
15254 15255
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
15256 15257
    return ret;
}
15258 15259


M
Michal Privoznik 已提交
15260
int
15261
qemuDomainNamespaceSetupMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
15262 15263 15264 15265 15266
                               virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

15267 15268
    if (qemuDomainNamespaceMknodPath(vm, mem->nvdimmPath) < 0)
        return -1;
15269

15270
    return 0;
M
Michal Privoznik 已提交
15271 15272 15273 15274
}


int
15275
qemuDomainNamespaceTeardownMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
15276 15277 15278 15279 15280
                                  virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

15281 15282
    if (qemuDomainNamespaceUnlinkPath(vm, mem->nvdimmPath) < 0)
        return -1;
15283

15284
    return 0;
M
Michal Privoznik 已提交
15285 15286 15287
}


15288
int
15289
qemuDomainNamespaceSetupChardev(virDomainObjPtr vm,
15290 15291 15292 15293
                                virDomainChrDefPtr chr)
{
    const char *path;

15294
    if (!(path = virDomainChrSourceDefGetPath(chr->source)))
15295 15296
        return 0;

15297 15298 15299 15300
    /* 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;
15301

15302 15303
    if (qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
15304

15305
    return 0;
15306 15307 15308 15309
}


int
15310
qemuDomainNamespaceTeardownChardev(virDomainObjPtr vm,
15311 15312 15313 15314 15315 15316 15317 15318 15319
                                   virDomainChrDefPtr chr)
{
    const char *path = NULL;

    if (chr->source->type != VIR_DOMAIN_CHR_TYPE_DEV)
        return 0;

    path = chr->source->data.file.path;

15320 15321
    if (qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
15322

15323
    return 0;
15324
}
15325 15326 15327


int
15328
qemuDomainNamespaceSetupRNG(virDomainObjPtr vm,
15329 15330 15331 15332 15333 15334 15335 15336 15337 15338 15339
                            virDomainRNGDefPtr rng)
{
    const char *path = NULL;

    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
        path = rng->source.file;
        break;

    case VIR_DOMAIN_RNG_BACKEND_EGD:
    case VIR_DOMAIN_RNG_BACKEND_LAST:
15340
        break;
15341 15342
    }

15343 15344
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
15345

15346
    return 0;
15347 15348 15349 15350
}


int
15351
qemuDomainNamespaceTeardownRNG(virDomainObjPtr vm,
15352 15353 15354 15355 15356 15357 15358 15359 15360 15361 15362
                               virDomainRNGDefPtr rng)
{
    const char *path = NULL;

    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
        path = rng->source.file;
        break;

    case VIR_DOMAIN_RNG_BACKEND_EGD:
    case VIR_DOMAIN_RNG_BACKEND_LAST:
15363
        break;
15364 15365
    }

15366 15367
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
15368

15369
    return 0;
15370 15371 15372 15373 15374 15375 15376 15377 15378 15379 15380 15381
}


int
qemuDomainNamespaceSetupInput(virDomainObjPtr vm,
                              virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

15382 15383 15384
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
    return 0;
15385 15386 15387 15388 15389 15390 15391 15392 15393 15394 15395 15396
}


int
qemuDomainNamespaceTeardownInput(virDomainObjPtr vm,
                                 virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

15397 15398
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
15399

15400
    return 0;
15401
}
15402 15403 15404 15405 15406 15407 15408 15409 15410 15411 15412 15413 15414 15415 15416 15417 15418 15419 15420 15421 15422 15423 15424 15425 15426 15427 15428 15429 15430 15431 15432 15433 15434 15435 15436 15437 15438 15439 15440 15441 15442 15443 15444


/**
 * 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;
}
15445 15446 15447 15448 15449 15450 15451 15452 15453 15454


/**
 * 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
15455
                                  virStorageSourcePtr src G_GNUC_UNUSED,
15456 15457 15458 15459 15460
                                  unsigned int idx)
{
    char *ret = NULL;

    if (idx)
15461
        ret = g_strdup_printf("%s[%d]", disk->dst, idx);
15462
    else
15463
        ret = g_strdup(disk->dst);
15464 15465 15466

    return ret;
}
15467 15468 15469 15470 15471 15472 15473 15474 15475 15476 15477 15478 15479 15480 15481 15482 15483 15484 15485 15486 15487 15488 15489 15490 15491 15492 15493


virStorageSourcePtr
qemuDomainGetStorageSourceByDevstr(const char *devstr,
                                   virDomainDefPtr def)
{
    virDomainDiskDefPtr disk = NULL;
    virStorageSourcePtr src = NULL;
    char *target = NULL;
    unsigned int idx;
    size_t i;

    if (virStorageFileParseBackingStoreStr(devstr, &target, &idx) < 0) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("failed to parse block device '%s'"), devstr);
        return NULL;
    }

    for (i = 0; i < def->ndisks; i++) {
        if (STREQ(target, def->disks[i]->dst)) {
            disk = def->disks[i];
            break;
        }
    }

    if (!disk) {
        virReportError(VIR_ERR_INVALID_ARG,
Y
Yuri Chornoivan 已提交
15494
                       _("failed to find disk '%s'"), target);
15495 15496 15497
        goto cleanup;
    }

15498 15499 15500 15501
    if (idx == 0)
        src = disk->src;
    else
        src = virStorageFileChainLookup(disk->src, NULL, NULL, idx, NULL);
15502 15503 15504 15505 15506

 cleanup:
    VIR_FREE(target);
    return src;
}
15507 15508 15509 15510 15511 15512 15513 15514


static void
qemuDomainSaveCookieDispose(void *obj)
{
    qemuDomainSaveCookiePtr cookie = obj;

    VIR_DEBUG("cookie=%p", cookie);
15515 15516

    virCPUDefFree(cookie->cpu);
15517 15518 15519 15520
}


qemuDomainSaveCookiePtr
15521
qemuDomainSaveCookieNew(virDomainObjPtr vm)
15522
{
15523
    qemuDomainObjPrivatePtr priv = vm->privateData;
15524 15525 15526 15527 15528 15529 15530 15531
    qemuDomainSaveCookiePtr cookie = NULL;

    if (qemuDomainInitialize() < 0)
        goto error;

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
        goto error;

15532 15533 15534
    if (priv->origCPU && !(cookie->cpu = virCPUDefCopy(vm->def->cpu)))
        goto error;

15535 15536 15537 15538
    cookie->slirpHelper = qemuDomainGetSlirpHelperOk(vm);

    VIR_DEBUG("Save cookie %p, cpu=%p, slirpHelper=%d",
              cookie, cookie->cpu, cookie->slirpHelper);
15539 15540 15541 15542 15543 15544 15545 15546 15547 15548

    return cookie;

 error:
    virObjectUnref(cookie);
    return NULL;
}


static int
J
Ján Tomko 已提交
15549
qemuDomainSaveCookieParse(xmlXPathContextPtr ctxt G_GNUC_UNUSED,
15550 15551 15552 15553 15554 15555 15556 15557 15558 15559
                          virObjectPtr *obj)
{
    qemuDomainSaveCookiePtr cookie = NULL;

    if (qemuDomainInitialize() < 0)
        goto error;

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
        goto error;

15560 15561 15562 15563
    if (virCPUDefParseXML(ctxt, "./cpu[1]", VIR_CPU_TYPE_GUEST,
                          &cookie->cpu) < 0)
        goto error;

15564 15565
    cookie->slirpHelper = virXPathBoolean("boolean(./slirpHelper)", ctxt) > 0;

15566 15567 15568 15569 15570 15571 15572 15573 15574 15575
    *obj = (virObjectPtr) cookie;
    return 0;

 error:
    virObjectUnref(cookie);
    return -1;
}


static int
15576 15577
qemuDomainSaveCookieFormat(virBufferPtr buf,
                           virObjectPtr obj)
15578
{
15579 15580 15581
    qemuDomainSaveCookiePtr cookie = (qemuDomainSaveCookiePtr) obj;

    if (cookie->cpu &&
15582
        virCPUDefFormatBufFull(buf, cookie->cpu, NULL) < 0)
15583 15584
        return -1;

15585 15586 15587
    if (cookie->slirpHelper)
        virBufferAddLit(buf, "<slirpHelper/>\n");

15588 15589 15590 15591 15592 15593 15594 15595
    return 0;
}


virSaveCookieCallbacks virQEMUDriverDomainSaveCookie = {
    .parse = qemuDomainSaveCookieParse,
    .format = qemuDomainSaveCookieFormat,
};
15596 15597 15598 15599 15600 15601 15602 15603 15604 15605 15606 15607 15608 15609 15610 15611 15612 15613 15614 15615 15616 15617 15618 15619 15620 15621 15622 15623 15624 15625 15626 15627 15628 15629 15630 15631 15632 15633 15634 15635


/**
 * 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;
}
15636

15637 15638 15639 15640 15641 15642 15643 15644 15645 15646 15647 15648 15649 15650 15651 15652 15653 15654 15655 15656 15657 15658 15659 15660 15661 15662 15663 15664

/**
 * 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;
15665
    int ret = -1;
15666 15667 15668 15669 15670 15671 15672 15673 15674 15675 15676 15677 15678 15679 15680 15681 15682 15683 15684 15685 15686 15687 15688 15689 15690 15691 15692 15693 15694 15695

    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);
15696
        vm->def->cpu = g_steal_pointer(&fixedCPU);
15697 15698 15699 15700
    }

    if (fixedOrig) {
        virCPUDefFree(*origCPU);
15701
        *origCPU = g_steal_pointer(&fixedOrig);
15702 15703 15704 15705 15706 15707 15708 15709 15710 15711 15712
    }

    ret = 0;

 cleanup:
    virCPUDefFree(fixedCPU);
    virCPUDefFree(fixedOrig);
    return ret;
}


15713 15714 15715 15716 15717 15718 15719
char *
qemuDomainGetMachineName(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    char *ret = NULL;

15720
    if (vm->pid > 0) {
15721 15722 15723 15724 15725 15726 15727 15728 15729 15730 15731
        ret = virSystemdGetMachineNameByPID(vm->pid);
        if (!ret)
            virResetLastError();
    }

    if (!ret)
        ret = virDomainGenerateMachineName("qemu", vm->def->id, vm->def->name,
                                           virQEMUDriverIsPrivileged(driver));

    return ret;
}
15732 15733 15734 15735 15736 15737 15738 15739 15740 15741 15742 15743 15744


/* 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,
15745
                                     const virDomainDeviceInfo *info,
15746 15747 15748
                                     virQEMUCapsPtr qemuCaps,
                                     const char *devicename)
{
15749
    if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
15750 15751 15752 15753 15754 15755
        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;
15756
        } else if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW)) {
15757 15758 15759 15760 15761
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CCW address type is not supported by "
                             "this QEMU"));
            return false;
        }
15762
    } else if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390) {
15763 15764 15765 15766 15767 15768 15769 15770 15771
        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;
}
15772 15773


15774
/**
15775
 * qemuDomainPrepareDiskSourceData:
15776 15777 15778 15779
 *
 * @disk: Disk config object
 * @src: source to start from
 *
15780
 * Prepares various aspects of a storage source belonging to a disk backing
15781 15782
 * chain based on the disk configuration. This function should be also called
 * for detected backing chain members.
15783
 */
15784
void
15785
qemuDomainPrepareDiskSourceData(virDomainDiskDefPtr disk,
15786
                                virStorageSourcePtr src)
15787
{
15788 15789 15790
    if (!disk)
        return;

15791
    /* transfer properties valid only for the top level image */
15792 15793
    if (src == disk->src)
        src->detect_zeroes = disk->detect_zeroes;
15794

15795 15796 15797 15798
    /* transfer properties valid for the full chain */
    src->iomode = disk->iomode;
    src->cachemode = disk->cachemode;
    src->discard = disk->discard;
15799

15800 15801
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY)
        src->floppyimg = true;
15802 15803 15804
}


15805 15806 15807 15808 15809 15810 15811 15812 15813
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;
}


15814 15815
static int
qemuDomainPrepareStorageSourcePR(virStorageSourcePtr src,
15816 15817
                                 qemuDomainObjPrivatePtr priv,
                                 const char *parentalias)
15818 15819 15820 15821 15822
{
    if (!src->pr)
        return 0;

    if (virStoragePRDefIsManaged(src->pr)) {
15823
        VIR_FREE(src->pr->path);
15824 15825
        if (!(src->pr->path = qemuDomainGetManagedPRSocketPath(priv)))
            return -1;
15826
        src->pr->mgralias = g_strdup(qemuDomainGetManagedPRAlias());
15827 15828 15829
    } else {
        if (!(src->pr->mgralias = qemuDomainGetUnmanagedPRAlias(parentalias)))
            return -1;
15830 15831 15832 15833 15834 15835
    }

    return 0;
}


15836 15837 15838 15839 15840 15841 15842 15843 15844 15845 15846 15847
/**
 * 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)
15848
{
15849 15850 15851
    if (qemuDomainValidateStorageSource(disk->src, priv->qemuCaps) < 0)
        return -1;

15852
    qemuDomainPrepareStorageSourceConfig(disk->src, cfg, priv->qemuCaps);
15853
    qemuDomainPrepareDiskSourceData(disk, disk->src);
15854

15855 15856 15857
    if (qemuDomainSecretStorageSourcePrepare(priv, disk->src,
                                             disk->info.alias,
                                             disk->info.alias) < 0)
15858 15859
        return -1;

15860
    if (qemuDomainPrepareStorageSourcePR(disk->src, priv, disk->info.alias) < 0)
15861
        return -1;
15862

15863 15864
    if (qemuDomainPrepareStorageSourceTLS(disk->src, cfg, disk->info.alias,
                                          priv->qemuCaps) < 0)
15865 15866
        return -1;

15867 15868 15869 15870
    return 0;
}


15871
int
15872 15873 15874 15875 15876 15877 15878
qemuDomainPrepareStorageSourceBlockdev(virDomainDiskDefPtr disk,
                                       virStorageSourcePtr src,
                                       qemuDomainObjPrivatePtr priv,
                                       virQEMUDriverConfigPtr cfg)
{
    src->id = qemuDomainStorageIdNew(priv);

15879 15880
    src->nodestorage = g_strdup_printf("libvirt-%u-storage", src->id);
    src->nodeformat = g_strdup_printf("libvirt-%u-format", src->id);
15881 15882 15883 15884

    if (qemuDomainValidateStorageSource(src, priv->qemuCaps) < 0)
        return -1;

15885
    qemuDomainPrepareStorageSourceConfig(src, cfg, priv->qemuCaps);
15886
    qemuDomainPrepareDiskSourceData(disk, src);
15887 15888 15889 15890 15891 15892

    if (qemuDomainSecretStorageSourcePrepare(priv, src,
                                             src->nodestorage,
                                             src->nodeformat) < 0)
        return -1;

15893
    if (qemuDomainPrepareStorageSourcePR(src, priv, src->nodestorage) < 0)
15894 15895
        return -1;

15896
    if (qemuDomainPrepareStorageSourceTLS(src, cfg, src->nodestorage,
15897 15898 15899 15900 15901 15902 15903 15904 15905 15906 15907 15908 15909 15910 15911 15912
                                          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 &&
15913 15914
        !diskPriv->nodeCopyOnRead)
        diskPriv->nodeCopyOnRead = g_strdup_printf("libvirt-CoR-%s", disk->dst);
15915 15916 15917 15918 15919 15920 15921 15922 15923 15924

    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if (qemuDomainPrepareStorageSourceBlockdev(disk, n, priv, cfg) < 0)
            return -1;
    }

    return 0;
}


15925 15926 15927 15928 15929 15930 15931
int
qemuDomainPrepareDiskSource(virDomainDiskDefPtr disk,
                            qemuDomainObjPrivatePtr priv,
                            virQEMUDriverConfigPtr cfg)
{
    qemuDomainPrepareDiskCachemode(disk);

15932
    /* set default format for storage pool based disks */
15933
    if (disk->src->type == VIR_STORAGE_TYPE_VOLUME &&
15934 15935 15936 15937 15938 15939 15940 15941
        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;
    }
15942

15943 15944 15945 15946 15947 15948 15949
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if (qemuDomainPrepareDiskSourceBlockdev(disk, priv, cfg) < 0)
            return -1;
    } else {
        if (qemuDomainPrepareDiskSourceLegacy(disk, priv, cfg) < 0)
            return -1;
    }
15950

15951 15952
    return 0;
}
15953 15954


15955 15956 15957 15958 15959 15960 15961 15962 15963 15964 15965 15966 15967 15968 15969 15970 15971 15972 15973 15974 15975 15976 15977 15978 15979 15980 15981 15982 15983 15984 15985 15986 15987 15988 15989 15990 15991 15992 15993 15994 15995 15996 15997 15998 15999 16000 16001 16002 16003 16004 16005 16006 16007 16008 16009 16010 16011 16012 16013 16014 16015 16016 16017 16018 16019 16020 16021 16022 16023 16024 16025 16026 16027 16028 16029
/**
 * 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;
}


16030 16031 16032 16033 16034 16035 16036 16037 16038 16039
void
qemuProcessEventFree(struct qemuProcessEvent *event)
{
    if (!event)
        return;

    switch (event->eventType) {
    case QEMU_PROCESS_EVENT_GUESTPANIC:
        qemuMonitorEventPanicInfoFree(event->data);
        break;
16040 16041 16042
    case QEMU_PROCESS_EVENT_RDMA_GID_STATUS_CHANGED:
        qemuMonitorEventRdmaGidStatusFree(event->data);
        break;
16043 16044 16045 16046 16047 16048 16049 16050
    case QEMU_PROCESS_EVENT_WATCHDOG:
    case QEMU_PROCESS_EVENT_DEVICE_DELETED:
    case QEMU_PROCESS_EVENT_NIC_RX_FILTER_CHANGED:
    case QEMU_PROCESS_EVENT_SERIAL_CHANGED:
    case QEMU_PROCESS_EVENT_BLOCK_JOB:
    case QEMU_PROCESS_EVENT_MONITOR_EOF:
        VIR_FREE(event->data);
        break;
16051 16052 16053
    case QEMU_PROCESS_EVENT_JOB_STATUS_CHANGE:
        virObjectUnref(event->data);
        break;
16054
    case QEMU_PROCESS_EVENT_PR_DISCONNECT:
16055 16056 16057 16058 16059
    case QEMU_PROCESS_EVENT_LAST:
        break;
    }
    VIR_FREE(event);
}
16060 16061 16062


char *
16063
qemuDomainGetManagedPRSocketPath(qemuDomainObjPrivatePtr priv)
16064 16065 16066
{
    char *ret = NULL;

16067
    ret = g_strdup_printf("%s/%s.sock", priv->libDir, qemuDomainGetManagedPRAlias());
16068 16069 16070

    return ret;
}
16071 16072 16073 16074 16075 16076 16077 16078 16079 16080 16081 16082 16083 16084 16085 16086 16087 16088 16089 16090 16091 16092 16093 16094 16095 16096 16097


/**
 * 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;
}
16098 16099 16100 16101 16102 16103 16104 16105 16106 16107 16108 16109 16110 16111 16112 16113 16114 16115 16116 16117 16118 16119 16120 16121 16122 16123 16124 16125 16126


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
16127 16128 16129 16130 16131


/* qemuDomainIsUsingNoShutdown:
 * @priv: Domain private data
 *
J
Ján Tomko 已提交
16132 16133
 * 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
16134 16135 16136 16137 16138 16139 16140
 *
 * 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
16141
    return priv->allowReboot == VIR_TRISTATE_BOOL_YES;
J
John Ferlan 已提交
16142
}
16143 16144 16145 16146 16147 16148 16149 16150 16151


bool
qemuDomainDiskIsMissingLocalOptional(virDomainDiskDefPtr disk)
{
    return disk->startupPolicy == VIR_DOMAIN_STARTUP_POLICY_OPTIONAL &&
           virStorageSourceIsLocalStorage(disk->src) && disk->src->path &&
           !virFileExists(disk->src->path);
}
16152 16153


16154
void
16155 16156 16157 16158
qemuDomainNVRAMPathFormat(virQEMUDriverConfigPtr cfg,
                            virDomainDefPtr def,
                            char **path)
{
16159
    *path = g_strdup_printf("%s/%s_VARS.fd", cfg->nvramDir, def->name);
16160 16161 16162
}


16163
void
16164 16165 16166 16167 16168
qemuDomainNVRAMPathGenerate(virQEMUDriverConfigPtr cfg,
                            virDomainDefPtr def)
{
    if (def->os.loader &&
        def->os.loader->type == VIR_DOMAIN_LOADER_TYPE_PFLASH &&
16169
        def->os.loader->readonly == VIR_TRISTATE_BOOL_YES &&
16170 16171
        !def->os.loader->nvram)
        qemuDomainNVRAMPathFormat(cfg, def, &def->os.loader->nvram);
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 16203 16204 16205 16206 16207


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;
}
16208 16209 16210 16211


static int
qemuDomainDefHasManagedPRBlockjobIterator(void *payload,
J
Ján Tomko 已提交
16212
                                          const void *name G_GNUC_UNUSED,
16213 16214 16215 16216 16217 16218 16219 16220 16221 16222 16223 16224 16225 16226 16227 16228 16229 16230 16231 16232 16233 16234 16235 16236 16237 16238 16239 16240 16241 16242 16243 16244 16245 16246 16247 16248
                                          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;
}
16249 16250 16251 16252 16253 16254 16255 16256 16257 16258 16259 16260 16261


/**
 * 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)
{
16262 16263 16264 16265
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_INCREMENTAL_BACKUP) &&
        virDomainListCheckpoints(vm->checkpoints, NULL, NULL, NULL, 0) > 0) {
16266 16267 16268 16269 16270 16271 16272
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s",
                       _("cannot perform block operations while checkpoint exists"));
        return -1;
    }

    return 0;
}
16273 16274 16275 16276 16277 16278 16279 16280 16281 16282 16283 16284 16285 16286 16287 16288 16289 16290 16291 16292 16293 16294 16295 16296 16297 16298 16299 16300 16301 16302 16303 16304 16305 16306 16307 16308 16309 16310 16311 16312 16313 16314 16315 16316 16317 16318 16319 16320 16321 16322 16323 16324 16325 16326 16327 16328 16329 16330

/**
 * 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;

    if (!def->os.loader ||
        def->os.loader->type != VIR_DOMAIN_LOADER_TYPE_PFLASH)
        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;
}