qemu_domain.c 425.7 KB
Newer Older
1
/*
2
 * qemu_domain.c: QEMU domain private state
3
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24
 */

#include <config.h>

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

59 60 61 62 63
#ifdef MAJOR_IN_MKDEV
# include <sys/mkdev.h>
#elif MAJOR_IN_SYSMACROS
# include <sys/sysmacros.h>
#endif
64
#include <sys/time.h>
65
#include <fcntl.h>
66 67 68
#if defined(HAVE_SYS_MOUNT_H)
# include <sys/mount.h>
#endif
69 70 71
#ifdef WITH_SELINUX
# include <selinux/selinux.h>
#endif
72

73
#include <libxml/xpathInternals.h>
74
#include "dosname.h"
75 76 77

#define VIR_FROM_THIS VIR_FROM_QEMU

78 79
VIR_LOG_INIT("qemu.qemu_domain");

80 81
#define QEMU_NAMESPACE_HREF "http://libvirt.org/schemas/domain/qemu/1.0"

82 83 84 85 86 87
VIR_ENUM_IMPL(qemuDomainJob, QEMU_JOB_LAST,
              "none",
              "query",
              "destroy",
              "suspend",
              "modify",
88
              "abort",
89
              "migration operation",
90 91 92 93
              "none",   /* async job is never stored in job.active */
              "async nested",
);

94 95 96 97 98 99
VIR_ENUM_IMPL(qemuDomainAgentJob, QEMU_AGENT_JOB_LAST,
              "none",
              "query",
              "modify",
);

100 101 102 103 104 105
VIR_ENUM_IMPL(qemuDomainAsyncJob, QEMU_ASYNC_JOB_LAST,
              "none",
              "migration out",
              "migration in",
              "save",
              "dump",
106
              "snapshot",
107
              "start",
108 109
);

110 111 112 113 114
VIR_ENUM_IMPL(qemuDomainNamespace, QEMU_DOMAIN_NS_LAST,
              "mount",
);


115 116
#define PROC_MOUNTS "/proc/mounts"
#define DEVPREFIX "/dev/"
117
#define DEV_VFIO "/dev/vfio/vfio"
118
#define DEVICE_MAPPER_CONTROL_PATH "/dev/mapper/control"
119

120

121
struct _qemuDomainLogContext {
122 123
    virObject parent;

124
    int writefd;
125
    int readfd; /* Only used if manager == NULL */
126
    off_t pos;
127
    ino_t inode; /* Only used if manager != NULL */
128
    char *path;
129
    virLogManagerPtr manager;
130 131
};

132
static virClassPtr qemuDomainLogContextClass;
133
static virClassPtr qemuDomainSaveCookieClass;
134 135

static void qemuDomainLogContextDispose(void *obj);
136
static void qemuDomainSaveCookieDispose(void *obj);
137

138 139 140 141 142 143 144 145

static int
qemuDomainPrepareStorageSourceBlockdev(virDomainDiskDefPtr disk,
                                       virStorageSourcePtr src,
                                       qemuDomainObjPrivatePtr priv,
                                       virQEMUDriverConfigPtr cfg);


146
static int
147
qemuDomainOnceInit(void)
148
{
149
    if (!VIR_CLASS_NEW(qemuDomainLogContext, virClassForObject()))
150 151
        return -1;

152
    if (!VIR_CLASS_NEW(qemuDomainSaveCookie, virClassForObject()))
153 154
        return -1;

155 156 157
    return 0;
}

158
VIR_ONCE_GLOBAL_INIT(qemuDomain)
159 160 161 162 163 164 165 166 167 168 169 170 171

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
172
const char *
173
qemuDomainAsyncJobPhaseToString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
174 175 176 177 178
                                int phase ATTRIBUTE_UNUSED)
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
179 180
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
181 182
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
183
    case QEMU_ASYNC_JOB_SNAPSHOT:
184
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
185
    case QEMU_ASYNC_JOB_NONE:
M
Marc Hartmayer 已提交
186
        ATTRIBUTE_FALLTHROUGH;
187 188
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
189 190 191 192 193 194
    }

    return "none";
}

int
195
qemuDomainAsyncJobPhaseFromString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
196 197 198 199 200 201 202 203
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
204 205
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
206 207
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
208
    case QEMU_ASYNC_JOB_SNAPSHOT:
209
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
210
    case QEMU_ASYNC_JOB_NONE:
M
Marc Hartmayer 已提交
211
        ATTRIBUTE_FALLTHROUGH;
212 213
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
214 215 216 217 218 219 220 221
    }

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

222

223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254
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;
}


255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
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;
        }
    }
}


271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
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);
291
    virObjectEventStateQueue(driver->domainEventState, event);
292 293 294
}


295 296 297 298 299 300 301 302
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

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

303
    if (virCondInit(&priv->job.asyncCond) < 0) {
304
        virCondDestroy(&priv->job.cond);
305 306 307
        return -1;
    }

308 309 310 311 312 313
    return 0;
}

static void
qemuDomainObjResetJob(qemuDomainObjPrivatePtr priv)
{
314
    qemuDomainJobObjPtr job = &priv->job;
315 316

    job->active = QEMU_JOB_NONE;
317
    job->owner = 0;
318
    job->ownerAPI = NULL;
J
Jiri Denemark 已提交
319
    job->started = 0;
320 321
}

322 323 324 325 326 327 328 329 330 331 332 333 334

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

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


335 336 337
static void
qemuDomainObjResetAsyncJob(qemuDomainObjPrivatePtr priv)
{
338
    qemuDomainJobObjPtr job = &priv->job;
339 340

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
341
    job->asyncOwner = 0;
342
    job->asyncOwnerAPI = NULL;
J
Jiri Denemark 已提交
343
    job->asyncStarted = 0;
J
Jiri Denemark 已提交
344
    job->phase = 0;
345
    job->mask = QEMU_JOB_DEFAULT_MASK;
346
    job->abortJob = false;
347
    job->spiceMigration = false;
348
    job->spiceMigrated = false;
349 350
    job->dumpCompleted = false;
    VIR_FREE(job->error);
J
Jiri Denemark 已提交
351
    VIR_FREE(job->current);
352 353
    qemuMigrationParamsFree(job->migParams);
    job->migParams = NULL;
354
    job->apiFlags = 0;
355 356
}

357 358
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
359
                        qemuDomainJobObjPtr job)
360 361 362 363 364
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
365
    job->owner = priv->job.owner;
366
    job->asyncJob = priv->job.asyncJob;
367
    job->asyncOwner = priv->job.asyncOwner;
J
Jiri Denemark 已提交
368
    job->phase = priv->job.phase;
369
    VIR_STEAL_PTR(job->migParams, priv->job.migParams);
370
    job->apiFlags = priv->job.apiFlags;
371 372 373 374 375

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

376 377 378
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
J
Jiri Denemark 已提交
379
    VIR_FREE(priv->job.current);
380
    VIR_FREE(priv->job.completed);
381 382
    virCondDestroy(&priv->job.cond);
    virCondDestroy(&priv->job.asyncCond);
383 384
}

385
static bool
386
qemuDomainTrackJob(qemuDomainJob job)
387 388 389 390
{
    return (QEMU_DOMAIN_TRACK_JOBS & JOB_MASK(job)) != 0;
}

391

J
Jiri Denemark 已提交
392 393 394 395 396 397 398 399 400 401 402
int
qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->started)
        return 0;

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

403 404 405 406 407 408
    if (now < jobInfo->started) {
        VIR_WARN("Async job starts in the future");
        jobInfo->started = 0;
        return 0;
    }

J
Jiri Denemark 已提交
409 410 411 412
    jobInfo->timeElapsed = now - jobInfo->started;
    return 0;
}

413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429
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;
    }

430 431
    jobInfo->stats.mig.downtime = now - jobInfo->stopped;
    jobInfo->stats.mig.downtime_set = true;
432 433 434
    return 0;
}

435 436 437 438 439 440 441 442
static virDomainJobType
qemuDomainJobStatusToType(qemuDomainJobStatus status)
{
    switch (status) {
    case QEMU_DOMAIN_JOB_STATUS_NONE:
        break;

    case QEMU_DOMAIN_JOB_STATUS_ACTIVE:
443
    case QEMU_DOMAIN_JOB_STATUS_MIGRATING:
444
    case QEMU_DOMAIN_JOB_STATUS_QEMU_COMPLETED:
445
    case QEMU_DOMAIN_JOB_STATUS_POSTCOPY:
446
    case QEMU_DOMAIN_JOB_STATUS_PAUSED:
447 448 449 450 451 452 453 454 455 456 457 458 459 460 461
        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
462 463 464 465
int
qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                        virDomainJobInfoPtr info)
{
466
    info->type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
467 468
    info->timeElapsed = jobInfo->timeElapsed;

469 470 471 472 473 474 475 476 477 478 479 480 481
    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
482

483 484 485 486 487 488
    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;

489
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
490 491 492 493 494
        info->memTotal = jobInfo->stats.dump.total;
        info->memProcessed = jobInfo->stats.dump.completed;
        info->memRemaining = info->memTotal - info->memProcessed;
        break;

495 496 497
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
        break;
    }
J
Jiri Denemark 已提交
498 499 500 501 502 503 504 505

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

    return 0;
}

506 507 508 509 510 511

static int
qemuDomainMigrationJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                                   int *type,
                                   virTypedParameterPtr *params,
                                   int *nparams)
J
Jiri Denemark 已提交
512
{
513
    qemuMonitorMigrationStats *stats = &jobInfo->stats.mig;
514
    qemuDomainMirrorStatsPtr mirrorStats = &jobInfo->mirrorStats;
J
Jiri Denemark 已提交
515 516 517
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;
518 519
    unsigned long long mirrorRemaining = mirrorStats->total -
                                         mirrorStats->transferred;
J
Jiri Denemark 已提交
520

521 522 523 524 525
    if (virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_OPERATION,
                             jobInfo->operation) < 0)
        goto error;

J
Jiri Denemark 已提交
526 527 528 529 530
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED,
                                jobInfo->timeElapsed) < 0)
        goto error;

531 532 533 534 535 536 537
    if (jobInfo->timeDeltaSet &&
        jobInfo->timeElapsed > jobInfo->timeDelta &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED_NET,
                                jobInfo->timeElapsed - jobInfo->timeDelta) < 0)
        goto error;

538
    if (stats->downtime_set &&
J
Jiri Denemark 已提交
539 540
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME,
541
                                stats->downtime) < 0)
J
Jiri Denemark 已提交
542 543
        goto error;

544
    if (stats->downtime_set &&
545
        jobInfo->timeDeltaSet &&
546
        stats->downtime > jobInfo->timeDelta &&
547 548
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME_NET,
549
                                stats->downtime - jobInfo->timeDelta) < 0)
550 551
        goto error;

552
    if (stats->setup_time_set &&
553 554
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_SETUP_TIME,
555
                                stats->setup_time) < 0)
556 557
        goto error;

J
Jiri Denemark 已提交
558 559
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_TOTAL,
560
                                stats->ram_total +
561 562
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
563 564
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_PROCESSED,
565
                                stats->ram_transferred +
566 567
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
568 569
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_REMAINING,
570
                                stats->ram_remaining +
571 572
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
573 574 575 576
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
577
                                stats->ram_total) < 0 ||
J
Jiri Denemark 已提交
578 579
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
580
                                stats->ram_transferred) < 0 ||
J
Jiri Denemark 已提交
581 582
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
583
                                stats->ram_remaining) < 0)
J
Jiri Denemark 已提交
584 585
        goto error;

586
    if (stats->ram_bps &&
587 588
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_BPS,
589
                                stats->ram_bps) < 0)
590 591
        goto error;

592
    if (stats->ram_duplicate_set) {
J
Jiri Denemark 已提交
593 594
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_CONSTANT,
595
                                    stats->ram_duplicate) < 0 ||
J
Jiri Denemark 已提交
596 597
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL,
598
                                    stats->ram_normal) < 0 ||
J
Jiri Denemark 已提交
599 600
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL_BYTES,
601
                                    stats->ram_normal_bytes) < 0)
J
Jiri Denemark 已提交
602 603 604
            goto error;
    }

605 606 607 608 609
    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,
610 611 612 613
                                stats->ram_iteration) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_POSTCOPY_REQS,
                                stats->ram_postcopy_reqs) < 0)
614 615
        goto error;

616 617 618 619 620 621
    if (stats->ram_page_size > 0 &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PAGE_SIZE,
                                stats->ram_page_size) < 0)
        goto error;

622 623 624 625 626
    /* 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
627 628
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_TOTAL,
629 630
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
631 632
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_PROCESSED,
633 634
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
635 636
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_REMAINING,
637 638
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
639 640
        goto error;

641
    if (stats->disk_bps &&
642 643
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_BPS,
644
                                stats->disk_bps) < 0)
645 646
        goto error;

647
    if (stats->xbzrle_set) {
J
Jiri Denemark 已提交
648 649
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE,
650
                                    stats->xbzrle_cache_size) < 0 ||
J
Jiri Denemark 已提交
651 652
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_BYTES,
653
                                    stats->xbzrle_bytes) < 0 ||
J
Jiri Denemark 已提交
654 655
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_PAGES,
656
                                    stats->xbzrle_pages) < 0 ||
J
Jiri Denemark 已提交
657 658
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE_MISSES,
659
                                    stats->xbzrle_cache_miss) < 0 ||
J
Jiri Denemark 已提交
660 661
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_OVERFLOW,
662
                                    stats->xbzrle_overflow) < 0)
J
Jiri Denemark 已提交
663 664 665
            goto error;
    }

666 667 668 669 670 671
    if (stats->cpu_throttle_percentage &&
        virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_AUTO_CONVERGE_THROTTLE,
                             stats->cpu_throttle_percentage) < 0)
        goto error;

672
 done:
673
    *type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
674 675 676 677 678 679 680 681 682 683
    *params = par;
    *nparams = npar;
    return 0;

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


684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726
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;
}


727 728 729 730 731 732 733 734
int
qemuDomainJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                          int *type,
                          virTypedParameterPtr *params,
                          int *nparams)
{
    switch (jobInfo->statsType) {
    case QEMU_DOMAIN_JOB_STATS_TYPE_MIGRATION:
735
    case QEMU_DOMAIN_JOB_STATS_TYPE_SAVEDUMP:
736 737
        return qemuDomainMigrationJobInfoToParams(jobInfo, type, params, nparams);

738
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
739 740
        return qemuDomainDumpJobInfoToParams(jobInfo, type, params, nparams);

741
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
742 743 744 745 746 747
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid job statistics type"));
        break;

    default:
        virReportEnumRangeError(qemuDomainJobStatsType, jobInfo->statsType);
748 749 750 751 752 753 754
        break;
    }

    return -1;
}


J
John Ferlan 已提交
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776
/* 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:
777 778
 * @driver: qemu driver data
 * @vm: Pointer to the vm object
J
John Ferlan 已提交
779 780 781 782 783
 *
 * 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
 */
784
int
M
Martin Kletzander 已提交
785 786
qemuDomainWriteMasterKeyFile(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
J
John Ferlan 已提交
787 788 789 790
{
    char *path;
    int fd = -1;
    int ret = -1;
M
Martin Kletzander 已提交
791
    qemuDomainObjPrivatePtr priv = vm->privateData;
J
John Ferlan 已提交
792

793 794 795 796
    /* Only gets filled in if we have the capability */
    if (!priv->masterKey)
        return 0;

J
John Ferlan 已提交
797 798 799 800 801 802 803 804 805 806 807 808 809 810 811
    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;
    }

812
    if (qemuSecurityDomainSetPathLabel(driver, vm, path, false) < 0)
M
Martin Kletzander 已提交
813 814
        goto cleanup;

J
John Ferlan 已提交
815 816 817 818 819 820 821 822 823 824
    ret = 0;

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

    return ret;
}


825 826 827 828 829 830 831 832 833
static void
qemuDomainMasterKeyFree(qemuDomainObjPrivatePtr priv)
{
    if (!priv->masterKey)
        return;

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

J
John Ferlan 已提交
834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931
/* 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 */
932
    qemuDomainMasterKeyFree(priv);
J
John Ferlan 已提交
933 934 935 936 937 938 939 940 941 942

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

    VIR_FREE(path);
}


/* qemuDomainMasterKeyCreate:
943
 * @vm: Pointer to the domain object
J
John Ferlan 已提交
944 945 946 947 948 949 950 951
 *
 * 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
952
qemuDomainMasterKeyCreate(virDomainObjPtr vm)
J
John Ferlan 已提交
953
{
M
Martin Kletzander 已提交
954 955
    qemuDomainObjPrivatePtr priv = vm->privateData;

J
John Ferlan 已提交
956 957 958 959
    /* If we don't have the capability, then do nothing. */
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

960
    if (VIR_ALLOC_N(priv->masterKey, QEMU_DOMAIN_MASTER_KEY_LEN) < 0)
961
        return -1;
J
John Ferlan 已提交
962 963
    priv->masterKeyLen = QEMU_DOMAIN_MASTER_KEY_LEN;

964
    if (virRandomBytes(priv->masterKey, priv->masterKeyLen) < 0) {
965 966 967 968
        VIR_DISPOSE_N(priv->masterKey, priv->masterKeyLen);
        return -1;
    }

J
John Ferlan 已提交
969 970 971 972
    return 0;
}


973
static void
974
qemuDomainSecretPlainClear(qemuDomainSecretPlainPtr secret)
975
{
976 977
    VIR_FREE(secret->username);
    VIR_DISPOSE_N(secret->secret, secret->secretlen);
978 979 980
}


J
John Ferlan 已提交
981
static void
982
qemuDomainSecretAESClear(qemuDomainSecretAESPtr secret,
983
                         bool keepAlias)
J
John Ferlan 已提交
984
{
985
    if (!keepAlias)
986
        VIR_FREE(secret->alias);
987

988 989 990
    VIR_FREE(secret->username);
    VIR_FREE(secret->iv);
    VIR_FREE(secret->ciphertext);
J
John Ferlan 已提交
991 992 993
}


994 995 996
static void
qemuDomainSecretInfoClear(qemuDomainSecretInfoPtr secinfo,
                          bool keepAlias)
997
{
998
    if (!secinfo)
999 1000
        return;

1001
    switch ((qemuDomainSecretInfoType) secinfo->type) {
J
John Ferlan 已提交
1002
    case VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN:
1003
        qemuDomainSecretPlainClear(&secinfo->s.plain);
J
John Ferlan 已提交
1004 1005
        break;

1006
    case VIR_DOMAIN_SECRET_INFO_TYPE_AES:
1007
        qemuDomainSecretAESClear(&secinfo->s.aes, keepAlias);
J
John Ferlan 已提交
1008 1009 1010 1011 1012
        break;

    case VIR_DOMAIN_SECRET_INFO_TYPE_LAST:
        break;
    }
1013 1014 1015 1016 1017 1018 1019 1020 1021 1022
}


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

    qemuDomainSecretInfoClear(*secinfo, false);
J
John Ferlan 已提交
1023

1024 1025 1026 1027
    VIR_FREE(*secinfo);
}


1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040
/**
 * qemuDomainSecretInfoDestroy:
 * @secinfo: object to destroy
 *
 * Removes any data unnecessary for further use, but keeps alias allocated.
 */
void
qemuDomainSecretInfoDestroy(qemuDomainSecretInfoPtr secinfo)
{
    qemuDomainSecretInfoClear(secinfo, true);
}


1041
static virClassPtr qemuDomainDiskPrivateClass;
1042
static void qemuDomainDiskPrivateDispose(void *obj);
1043 1044 1045 1046

static int
qemuDomainDiskPrivateOnceInit(void)
{
1047
    if (!VIR_CLASS_NEW(qemuDomainDiskPrivate, virClassForObject()))
1048
        return -1;
1049 1050

    return 0;
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068
}

VIR_ONCE_GLOBAL_INIT(qemuDomainDiskPrivate)

static virObjectPtr
qemuDomainDiskPrivateNew(void)
{
    qemuDomainDiskPrivatePtr priv;

    if (qemuDomainDiskPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}

1069 1070 1071 1072 1073
static void
qemuDomainDiskPrivateDispose(void *obj)
{
    qemuDomainDiskPrivatePtr priv = obj;

1074
    virStorageSourceFree(priv->migrSource);
1075
    VIR_FREE(priv->qomName);
1076
    VIR_FREE(priv->nodeCopyOnRead);
1077
    virObjectUnref(priv->blockjob);
1078
}
1079

1080 1081 1082 1083 1084 1085
static virClassPtr qemuDomainStorageSourcePrivateClass;
static void qemuDomainStorageSourcePrivateDispose(void *obj);

static int
qemuDomainStorageSourcePrivateOnceInit(void)
{
1086
    if (!VIR_CLASS_NEW(qemuDomainStorageSourcePrivate, virClassForObject()))
1087
        return -1;
1088 1089

    return 0;
1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118
}

VIR_ONCE_GLOBAL_INIT(qemuDomainStorageSourcePrivate)

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);
}


1119 1120 1121 1122 1123 1124
static virClassPtr qemuDomainVcpuPrivateClass;
static void qemuDomainVcpuPrivateDispose(void *obj);

static int
qemuDomainVcpuPrivateOnceInit(void)
{
1125
    if (!VIR_CLASS_NEW(qemuDomainVcpuPrivate, virClassForObject()))
1126
        return -1;
1127 1128

    return 0;
1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148
}

VIR_ONCE_GLOBAL_INIT(qemuDomainVcpuPrivate)

static virObjectPtr
qemuDomainVcpuPrivateNew(void)
{
    qemuDomainVcpuPrivatePtr priv;

    if (qemuDomainVcpuPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}


static void
1149
qemuDomainVcpuPrivateDispose(void *obj)
1150
{
1151 1152 1153 1154
    qemuDomainVcpuPrivatePtr priv = obj;

    VIR_FREE(priv->type);
    VIR_FREE(priv->alias);
1155 1156 1157 1158
    return;
}


1159 1160
static virClassPtr qemuDomainChrSourcePrivateClass;
static void qemuDomainChrSourcePrivateDispose(void *obj);
1161 1162

static int
1163
qemuDomainChrSourcePrivateOnceInit(void)
1164
{
1165
    if (!VIR_CLASS_NEW(qemuDomainChrSourcePrivate, virClassForObject()))
1166
        return -1;
1167 1168

    return 0;
1169 1170
}

1171
VIR_ONCE_GLOBAL_INIT(qemuDomainChrSourcePrivate)
1172 1173

static virObjectPtr
1174
qemuDomainChrSourcePrivateNew(void)
1175
{
1176
    qemuDomainChrSourcePrivatePtr priv;
1177

1178
    if (qemuDomainChrSourcePrivateInitialize() < 0)
1179 1180
        return NULL;

1181
    if (!(priv = virObjectNew(qemuDomainChrSourcePrivateClass)))
1182 1183 1184 1185 1186 1187 1188
        return NULL;

    return (virObjectPtr) priv;
}


static void
1189
qemuDomainChrSourcePrivateDispose(void *obj)
1190
{
1191
    qemuDomainChrSourcePrivatePtr priv = obj;
1192 1193 1194 1195 1196

    qemuDomainSecretInfoFree(&priv->secinfo);
}


J
Ján Tomko 已提交
1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221
static virClassPtr qemuDomainVsockPrivateClass;
static void qemuDomainVsockPrivateDispose(void *obj);

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

    return 0;
}

VIR_ONCE_GLOBAL_INIT(qemuDomainVsockPrivate)

static virObjectPtr
qemuDomainVsockPrivateNew(void)
{
    qemuDomainVsockPrivatePtr priv;

    if (qemuDomainVsockPrivateInitialize() < 0)
        return NULL;

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

1222 1223
    priv->vhostfd = -1;

J
Ján Tomko 已提交
1224 1225 1226 1227 1228 1229 1230
    return (virObjectPtr) priv;
}


static void
qemuDomainVsockPrivateDispose(void *obj ATTRIBUTE_UNUSED)
{
1231 1232 1233
    qemuDomainVsockPrivatePtr priv = obj;

    VIR_FORCE_CLOSE(priv->vhostfd);
J
Ján Tomko 已提交
1234 1235 1236
}


1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271
static virClassPtr qemuDomainGraphicsPrivateClass;
static void qemuDomainGraphicsPrivateDispose(void *obj);

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

    return 0;
}

VIR_ONCE_GLOBAL_INIT(qemuDomainGraphicsPrivate)

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);
1272
    qemuDomainSecretInfoFree(&priv->secinfo);
1273 1274 1275
}


1276 1277
/* qemuDomainSecretPlainSetup:
 * @secinfo: Pointer to secret info
J
John Ferlan 已提交
1278
 * @usageType: The virSecretUsageType
1279 1280
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1281 1282 1283 1284 1285 1286
 *
 * Taking a secinfo, fill in the plaintext information
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1287
qemuDomainSecretPlainSetup(qemuDomainSecretInfoPtr secinfo,
J
John Ferlan 已提交
1288
                           virSecretUsageType usageType,
1289 1290
                           const char *username,
                           virSecretLookupTypeDefPtr seclookupdef)
1291
{
1292 1293 1294 1295 1296 1297 1298
    virConnectPtr conn;
    int ret = -1;

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

1299
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN;
1300
    if (VIR_STRDUP(secinfo->s.plain.username, username) < 0)
1301
        goto cleanup;
1302

1303 1304 1305 1306 1307 1308 1309
    ret = virSecretGetSecretString(conn, seclookupdef, usageType,
                                   &secinfo->s.plain.secret,
                                   &secinfo->s.plain.secretlen);

 cleanup:
    virObjectUnref(conn);
    return ret;
1310 1311 1312
}


1313 1314 1315 1316
/* 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
1317
 * @usageType: The virSecretUsageType
1318 1319
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1320
 * @isLuks: True/False for is for luks (alias generation)
1321 1322 1323 1324 1325 1326
 *
 * Taking a secinfo, fill in the AES specific information using the
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1327
qemuDomainSecretAESSetup(qemuDomainObjPrivatePtr priv,
1328 1329
                         qemuDomainSecretInfoPtr secinfo,
                         const char *srcalias,
J
John Ferlan 已提交
1330
                         virSecretUsageType usageType,
1331
                         const char *username,
1332 1333
                         virSecretLookupTypeDefPtr seclookupdef,
                         bool isLuks)
1334
{
1335
    virConnectPtr conn;
1336 1337 1338 1339 1340 1341 1342 1343
    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;

1344 1345 1346 1347
    conn = virGetConnectSecret();
    if (!conn)
        return -1;

1348
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_AES;
1349
    if (VIR_STRDUP(secinfo->s.aes.username, username) < 0)
1350
        goto cleanup;
1351

1352
    if (!(secinfo->s.aes.alias = qemuDomainGetSecretAESAlias(srcalias, isLuks)))
1353
        goto cleanup;
1354

1355 1356 1357
    if (VIR_ALLOC_N(raw_iv, ivlen) < 0)
        goto cleanup;

1358
    /* Create a random initialization vector */
1359
    if (virRandomBytes(raw_iv, ivlen) < 0)
1360
        goto cleanup;
1361 1362 1363 1364 1365 1366

    /* Encode the IV and save that since qemu will need it */
    if (!(secinfo->s.aes.iv = virStringEncodeBase64(raw_iv, ivlen)))
        goto cleanup;

    /* Grab the unencoded secret */
J
John Ferlan 已提交
1367
    if (virSecretGetSecretString(conn, seclookupdef, usageType,
1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390
                                 &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 */
    if (!(secinfo->s.aes.ciphertext = virStringEncodeBase64(ciphertext,
                                                            ciphertextlen)))
        goto cleanup;

    ret = 0;

 cleanup:
    VIR_DISPOSE_N(raw_iv, ivlen);
    VIR_DISPOSE_N(secret, secretlen);
    VIR_DISPOSE_N(ciphertext, ciphertextlen);
1391
    virObjectUnref(conn);
1392 1393 1394 1395
    return ret;
}


1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411
/**
 * 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;
}


1412
/* qemuDomainSecretInfoNewPlain:
1413
 * @usageType: Secret usage type
1414 1415
 * @username: username
 * @lookupDef: lookup def describing secret
1416
 *
1417
 * Helper function to create a secinfo to be used for secinfo consumers. This
1418
 * sets up a 'plain' (unencrypted) secret for legacy consumers.
1419 1420 1421 1422 1423
 *
 * Returns @secinfo on success, NULL on failure. Caller is responsible
 * to eventually free @secinfo.
 */
static qemuDomainSecretInfoPtr
1424
qemuDomainSecretInfoNewPlain(virSecretUsageType usageType,
1425
                             const char *username,
1426
                             virSecretLookupTypeDefPtr lookupDef)
1427 1428 1429 1430 1431 1432
{
    qemuDomainSecretInfoPtr secinfo = NULL;

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

1433 1434 1435
    if (qemuDomainSecretPlainSetup(secinfo, usageType, username, lookupDef) < 0) {
        qemuDomainSecretInfoFree(&secinfo);
        return NULL;
1436 1437 1438 1439 1440 1441
    }

    return secinfo;
}


1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484
/* 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;
}


1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495
/**
 * 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.
 */
1496
qemuDomainSecretInfoPtr
1497
qemuDomainSecretInfoTLSNew(qemuDomainObjPrivatePtr priv,
1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510
                           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;

1511 1512 1513
    return qemuDomainSecretInfoNew(priv, srcAlias,
                                   VIR_SECRET_USAGE_TYPE_TLS, NULL,
                                   &seclookupdef, false);
1514 1515 1516
}


1517 1518 1519
void
qemuDomainSecretDiskDestroy(virDomainDiskDefPtr disk)
{
1520 1521
    qemuDomainStorageSourcePrivatePtr srcPriv;
    virStorageSourcePtr n;
1522

1523 1524 1525 1526 1527 1528
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if ((srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(n))) {
            qemuDomainSecretInfoDestroy(srcPriv->secinfo);
            qemuDomainSecretInfoDestroy(srcPriv->encinfo);
        }
    }
1529 1530 1531
}


1532
bool
1533
qemuDomainStorageSourceHasAuth(virStorageSourcePtr src)
1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545
{
    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;
}


1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557
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;
}


1558 1559 1560 1561 1562 1563
/**
 * 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
1564
 *
1565 1566 1567 1568
 * 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.
1569
 *
1570
 * Returns 0 on success; -1 on error while reporting an libvirt error.
1571
 */
1572
static int
1573
qemuDomainSecretStorageSourcePrepare(qemuDomainObjPrivatePtr priv,
1574 1575 1576
                                     virStorageSourcePtr src,
                                     const char *authalias,
                                     const char *encalias)
1577
{
1578
    qemuDomainStorageSourcePrivatePtr srcPriv;
1579
    bool iscsiHasPS = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET);
1580
    bool hasAuth = qemuDomainStorageSourceHasAuth(src);
1581 1582 1583 1584
    bool hasEnc = qemuDomainDiskHasEncryptionSecret(src);

    if (!hasAuth && !hasEnc)
        return 0;
1585

1586
    if (!(src->privateData = qemuDomainStorageSourcePrivateNew()))
1587 1588
        return -1;

1589
    srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
1590

1591
    if (hasAuth) {
J
John Ferlan 已提交
1592
        virSecretUsageType usageType = VIR_SECRET_USAGE_TYPE_ISCSI;
1593

1594
        if (src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD)
J
John Ferlan 已提交
1595
            usageType = VIR_SECRET_USAGE_TYPE_CEPH;
1596

1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611
        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;
1612 1613
    }

1614
    if (hasEnc) {
1615
        if (!(srcPriv->encinfo =
1616 1617 1618 1619
              qemuDomainSecretInfoNew(priv, encalias,
                                      VIR_SECRET_USAGE_TYPE_VOLUME, NULL,
                                      &src->encryption->secrets[0]->seclookupdef,
                                      true)))
1620
              return -1;
1621 1622
    }

1623 1624 1625 1626
    return 0;
}


1627 1628 1629
void
qemuDomainSecretHostdevDestroy(virDomainHostdevDefPtr hostdev)
{
1630
    qemuDomainStorageSourcePrivatePtr srcPriv;
1631

1632 1633 1634
    if (virHostdevIsSCSIDevice(hostdev)) {
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1635

1636 1637 1638 1639 1640 1641
        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);
        }
    }
1642 1643 1644 1645
}


/* qemuDomainSecretHostdevPrepare:
J
John Ferlan 已提交
1646
 * @priv: pointer to domain private object
1647 1648 1649 1650 1651 1652 1653
 * @hostdev: Pointer to a hostdev definition
 *
 * For the right host device, generate the qemuDomainSecretInfo structure.
 *
 * Returns 0 on success, -1 on failure
 */
int
1654
qemuDomainSecretHostdevPrepare(qemuDomainObjPrivatePtr priv,
1655 1656
                               virDomainHostdevDefPtr hostdev)
{
1657
    if (virHostdevIsSCSIDevice(hostdev)) {
1658 1659
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1660
        virStorageSourcePtr src = iscsisrc->src;
1661 1662

        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI &&
1663
            src->auth) {
1664 1665
            if (qemuDomainSecretStorageSourcePrepare(priv, src,
                                                     hostdev->info->alias, NULL) < 0)
1666 1667 1668 1669 1670 1671 1672 1673
                return -1;
        }
    }

    return 0;
}


1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698
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
1699
qemuDomainSecretChardevPrepare(virQEMUDriverConfigPtr cfg,
1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714
                               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)))
1715
            return -1;
1716

1717
        chrSourcePriv->secinfo =
1718
            qemuDomainSecretInfoTLSNew(priv, charAlias,
1719
                                       cfg->chardevTLSx509secretUUID);
1720
        VIR_FREE(charAlias);
1721 1722 1723

        if (!chrSourcePriv->secinfo)
            return -1;
1724 1725 1726 1727 1728 1729
    }

    return 0;
}


1730 1731 1732 1733 1734 1735 1736 1737 1738
static void
qemuDomainSecretGraphicsDestroy(virDomainGraphicsDefPtr graphics)
{
    qemuDomainGraphicsPrivatePtr gfxPriv = QEMU_DOMAIN_GRAPHICS_PRIVATE(graphics);

    if (!gfxPriv)
        return;

    VIR_FREE(gfxPriv->tlsAlias);
1739
    qemuDomainSecretInfoFree(&gfxPriv->secinfo);
1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762
}


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;

    if (VIR_STRDUP(gfxPriv->tlsAlias, "vnc-tls-creds0") < 0)
        return -1;

1763 1764 1765 1766 1767 1768 1769
    if (cfg->vncTLSx509secretUUID) {
        gfxPriv->secinfo = qemuDomainSecretInfoTLSNew(priv, gfxPriv->tlsAlias,
                                                      cfg->vncTLSx509secretUUID);
        if (!gfxPriv->secinfo)
            return -1;
    }

1770 1771 1772 1773
    return 0;
}


1774 1775 1776
/* qemuDomainSecretDestroy:
 * @vm: Domain object
 *
1777
 * Removes all unnecessary data which was needed to generate 'secret' objects.
1778 1779 1780 1781 1782 1783 1784 1785
 */
void
qemuDomainSecretDestroy(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++)
        qemuDomainSecretDiskDestroy(vm->def->disks[i]);
1786 1787 1788

    for (i = 0; i < vm->def->nhostdevs; i++)
        qemuDomainSecretHostdevDestroy(vm->def->hostdevs[i]);
1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814

    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);
1815 1816 1817

    for (i = 0; i < vm->def->ngraphics; i++)
        qemuDomainSecretGraphicsDestroy(vm->def->graphics[i]);
1818 1819 1820 1821
}


/* qemuDomainSecretPrepare:
1822
 * @driver: Pointer to driver object
1823 1824 1825
 * @vm: Domain object
 *
 * For any objects that may require an auth/secret setup, create a
1826
 * qemuDomainSecretInfo and save it in the appropriate place within
1827 1828 1829 1830 1831 1832 1833
 * 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
1834
qemuDomainSecretPrepare(virQEMUDriverPtr driver,
1835 1836
                        virDomainObjPtr vm)
{
J
John Ferlan 已提交
1837
    qemuDomainObjPrivatePtr priv = vm->privateData;
1838
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1839
    size_t i;
1840
    int ret = -1;
1841

1842
    /* disk secrets are prepared when preparing disks */
1843

1844
    for (i = 0; i < vm->def->nhostdevs; i++) {
1845
        if (qemuDomainSecretHostdevPrepare(priv,
J
John Ferlan 已提交
1846
                                           vm->def->hostdevs[i]) < 0)
1847
            goto cleanup;
1848 1849
    }

1850
    for (i = 0; i < vm->def->nserials; i++) {
1851
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1852 1853 1854 1855 1856 1857
                                           vm->def->serials[i]->info.alias,
                                           vm->def->serials[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nparallels; i++) {
1858
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1859 1860 1861 1862 1863 1864
                                           vm->def->parallels[i]->info.alias,
                                           vm->def->parallels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nchannels; i++) {
1865
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1866 1867 1868 1869 1870 1871
                                           vm->def->channels[i]->info.alias,
                                           vm->def->channels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nconsoles; i++) {
1872
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1873 1874 1875 1876 1877 1878 1879 1880
                                           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 &&
1881
            qemuDomainSecretChardevPrepare(cfg, priv,
1882 1883 1884 1885 1886 1887
                                           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 &&
1888
            qemuDomainSecretChardevPrepare(cfg, priv,
1889 1890 1891 1892 1893 1894
                                           vm->def->rngs[i]->info.alias,
                                           vm->def->rngs[i]->source.chardev) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nredirdevs; i++) {
1895
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1896 1897 1898 1899 1900
                                           vm->def->redirdevs[i]->info.alias,
                                           vm->def->redirdevs[i]->source) < 0)
            goto cleanup;
    }

1901 1902 1903 1904 1905
    for (i = 0; i < vm->def->ngraphics; i++) {
        if (qemuDomainSecretGraphicsPrepare(cfg, priv, vm->def->graphics[i]) < 0)
            goto cleanup;
    }

1906 1907 1908 1909 1910
    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
1911 1912 1913
}


1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940
/* This is the old way of setting up per-domain directories */
static int
qemuDomainSetPrivatePathsOld(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

    if (!priv->libDir &&
        virAsprintf(&priv->libDir, "%s/domain-%s",
                    cfg->libDir, vm->def->name) < 0)
        goto cleanup;

    if (!priv->channelTargetDir &&
        virAsprintf(&priv->channelTargetDir, "%s/domain-%s",
                    cfg->channelTargetDir, vm->def->name) < 0)
        goto cleanup;

    ret = 0;
 cleanup:
    virObjectUnref(cfg);
    return ret;
}


int
1941 1942
qemuDomainSetPrivatePaths(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
1943
{
1944 1945
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainObjPrivatePtr priv = vm->privateData;
1946
    char *domname = virDomainDefGetShortName(vm->def);
1947
    int ret = -1;
1948

M
Martin Kletzander 已提交
1949 1950 1951
    if (!domname)
        goto cleanup;

1952
    if (!priv->libDir &&
M
Martin Kletzander 已提交
1953
        virAsprintf(&priv->libDir, "%s/domain-%s", cfg->libDir, domname) < 0)
1954
        goto cleanup;
1955

1956
    if (!priv->channelTargetDir &&
M
Martin Kletzander 已提交
1957 1958
        virAsprintf(&priv->channelTargetDir, "%s/domain-%s",
                    cfg->channelTargetDir, domname) < 0)
1959
        goto cleanup;
1960

1961 1962 1963
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
M
Martin Kletzander 已提交
1964
    VIR_FREE(domname);
1965
    return ret;
1966 1967 1968
}


1969
static void *
1970
qemuDomainObjPrivateAlloc(void *opaque)
1971 1972 1973 1974 1975 1976
{
    qemuDomainObjPrivatePtr priv;

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

1977 1978 1979
    if (qemuDomainObjInitJob(priv) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to init qemu driver mutexes"));
1980
        goto error;
1981
    }
1982

1983
    if (!(priv->devs = virChrdevAlloc()))
1984 1985
        goto error;

1986
    priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
1987
    priv->driver = opaque;
1988

1989
    return priv;
1990

1991
 error:
1992 1993
    VIR_FREE(priv);
    return NULL;
1994 1995
}

1996 1997 1998 1999 2000 2001 2002 2003 2004
/**
 * 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)
2005
{
2006 2007 2008
    virStringListFree(priv->qemuDevices);
    priv->qemuDevices = NULL;

2009
    virCgroupFree(&priv->cgroup);
2010 2011 2012 2013 2014

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

    VIR_FREE(priv->machineName);
2015

2016
    virObjectUnref(priv->qemuCaps);
2017
    priv->qemuCaps = NULL;
2018

2019
    VIR_FREE(priv->pidfile);
2020

2021 2022 2023
    VIR_FREE(priv->libDir);
    VIR_FREE(priv->channelTargetDir);

2024 2025
    priv->memPrealloc = false;

2026 2027 2028 2029 2030 2031 2032
    /* remove automatic pinning data */
    virBitmapFree(priv->autoNodeset);
    priv->autoNodeset = NULL;
    virBitmapFree(priv->autoCpuset);
    priv->autoCpuset = NULL;

    /* remove address data */
2033
    virDomainPCIAddressSetFree(priv->pciaddrs);
2034
    priv->pciaddrs = NULL;
J
Ján Tomko 已提交
2035
    virDomainUSBAddressSetFree(priv->usbaddrs);
2036 2037 2038 2039 2040 2041 2042 2043
    priv->usbaddrs = NULL;

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

    /* clear previously used namespaces */
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
2044

2045 2046
    priv->rememberOwner = false;

2047
    priv->reconnectBlockjobs = VIR_TRISTATE_BOOL_ABSENT;
2048
    priv->allowReboot = VIR_TRISTATE_BOOL_ABSENT;
2049 2050 2051

    virBitmapFree(priv->migrationCaps);
    priv->migrationCaps = NULL;
2052 2053 2054

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
2055 2056 2057 2058 2059 2060 2061 2062 2063 2064
}


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

    qemuDomainObjPrivateDataClear(priv);

2065
    virDomainChrSourceDefFree(priv->monConfig);
2066
    qemuDomainObjFreeJob(priv);
2067
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
2068
    VIR_FREE(priv->origname);
2069

2070
    virChrdevFree(priv->devs);
2071

2072 2073
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
2074
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
2075 2076
        qemuMonitorClose(priv->mon);
    }
D
Daniel P. Berrange 已提交
2077 2078 2079 2080
    if (priv->agent) {
        VIR_ERROR(_("Unexpected QEMU agent still active during domain deletion"));
        qemuAgentClose(priv->agent);
    }
2081
    VIR_FREE(priv->cleanupCallbacks);
2082 2083

    qemuDomainSecretInfoFree(&priv->migSecinfo);
2084
    qemuDomainMasterKeyFree(priv);
2085

2086 2087 2088 2089
    VIR_FREE(priv);
}


2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110
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)
        VIR_STEAL_PTR((*secinfo)->s.aes.alias, *alias);

    return 0;
}


2111 2112 2113 2114
static int
qemuStorageSourcePrivateDataParse(xmlXPathContextPtr ctxt,
                                  virStorageSourcePtr src)
{
2115 2116 2117 2118 2119
    qemuDomainStorageSourcePrivatePtr priv;
    char *authalias = NULL;
    char *encalias = NULL;
    int ret = -1;

2120 2121
    src->nodestorage = virXPathString("string(./nodenames/nodename[@type='storage']/@name)", ctxt);
    src->nodeformat = virXPathString("string(./nodenames/nodename[@type='format']/@name)", ctxt);
2122
    src->tlsAlias = virXPathString("string(./objects/TLSx509/@alias)", ctxt);
2123

2124 2125 2126
    if (src->pr)
        src->pr->mgralias = virXPathString("string(./reservations/@mgralias)", ctxt);

2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143
    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;
    }

2144
    if (virStorageSourcePrivateDataParseRelPath(ctxt, src) < 0)
2145
        goto cleanup;
2146

2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168
    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);
2169 2170 2171 2172 2173 2174 2175
}


static int
qemuStorageSourcePrivateDataFormat(virStorageSourcePtr src,
                                   virBufferPtr buf)
{
2176 2177 2178 2179
    virBuffer tmp = VIR_BUFFER_INITIALIZER;
    qemuDomainStorageSourcePrivatePtr srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
    int ret = -1;

2180 2181 2182 2183 2184 2185 2186 2187 2188
    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");
    }

2189 2190 2191
    if (src->pr)
        virBufferAsprintf(buf, "<reservations mgralias='%s'/>\n", src->pr->mgralias);

2192
    if (virStorageSourcePrivateDataFormatRelPath(src, buf) < 0)
2193
        goto cleanup;
2194

2195 2196 2197 2198 2199 2200 2201
    virBufferSetChildIndent(&tmp, buf);

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

2202 2203 2204
    if (src->tlsAlias)
        virBufferAsprintf(&tmp, "<TLSx509 alias='%s'/>\n", src->tlsAlias);

2205 2206 2207 2208 2209 2210 2211 2212
    if (virXMLFormatElement(buf, "objects", NULL, &tmp) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virBufferFreeAndReset(&tmp);
    return ret;
2213 2214 2215
}


2216 2217 2218 2219 2220 2221 2222
static int
qemuDomainDiskPrivateParse(xmlXPathContextPtr ctxt,
                           virDomainDiskDefPtr disk)
{
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);

    priv->qomName = virXPathString("string(./qom/@name)", ctxt);
2223
    priv->nodeCopyOnRead = virXPathString("string(./nodenames/nodename[@type='copyOnRead']/@name)", ctxt);
2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236

    return 0;
}


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

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

2237 2238 2239 2240 2241 2242 2243 2244 2245
    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");
    }

2246 2247 2248 2249
    return 0;
}


2250 2251
static void
qemuDomainObjPrivateXMLFormatVcpus(virBufferPtr buf,
2252
                                   virDomainDefPtr def)
2253 2254
{
    size_t i;
2255 2256 2257
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    pid_t tid;
2258 2259 2260 2261

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

2262 2263 2264 2265 2266 2267 2268 2269 2270
    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);
    }
2271 2272 2273 2274 2275 2276

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


2277
static int
2278
qemuDomainObjPrivateXMLFormatAutomaticPlacement(virBufferPtr buf,
2279
                                                qemuDomainObjPrivatePtr priv)
2280 2281
{
    char *nodeset = NULL;
2282
    char *cpuset = NULL;
2283 2284
    int ret = -1;

2285
    if (!priv->autoNodeset && !priv->autoCpuset)
2286 2287
        return 0;

2288 2289
    if (priv->autoNodeset &&
        !((nodeset = virBitmapFormat(priv->autoNodeset))))
2290 2291
        goto cleanup;

2292 2293 2294 2295 2296 2297 2298 2299
    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");
2300 2301 2302 2303 2304

    ret = 0;

 cleanup:
    VIR_FREE(nodeset);
2305
    VIR_FREE(cpuset);
2306 2307 2308 2309
    return ret;
}


2310 2311 2312 2313 2314 2315 2316 2317 2318 2319
static int
qemuDomainObjPrivateXMLFormatBlockjobs(virBufferPtr buf,
                                       virDomainObjPtr vm)
{
    virBuffer attrBuf = VIR_BUFFER_INITIALIZER;
    bool bj = qemuDomainHasBlockjob(vm, false);

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

2320
    return virXMLFormatElement(buf, "blockjobs", &attrBuf, NULL);
2321 2322 2323
}


2324
void
2325 2326 2327 2328 2329 2330 2331 2332 2333
qemuDomainObjPrivateXMLFormatAllowReboot(virBufferPtr buf,
                                         virTristateBool allowReboot)
{
    virBufferAsprintf(buf, "<allowReboot value='%s'/>\n",
                      virTristateBoolTypeToString(allowReboot));

}


2334 2335 2336 2337 2338 2339 2340 2341 2342
static void
qemuDomainObjPrivateXMLFormatPR(virBufferPtr buf,
                                qemuDomainObjPrivatePtr priv)
{
    if (priv->prDaemonRunning)
        virBufferAddLit(buf, "<prDaemon/>\n");
}


2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383
static int
qemuDomainObjPrivateXMLFormatNBDMigrationSource(virBufferPtr buf,
                                                virStorageSourcePtr src)
{
    virBuffer attrBuf = VIR_BUFFER_INITIALIZER;
    virBuffer childBuf = VIR_BUFFER_INITIALIZER;
    virBuffer privateDataBuf = VIR_BUFFER_INITIALIZER;
    int ret = -1;

    virBufferSetChildIndent(&childBuf, buf);
    virBufferSetChildIndent(&privateDataBuf, &childBuf);

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

    if (virDomainStorageSourceFormat(&attrBuf, &childBuf, src,
                                     VIR_DOMAIN_DEF_FORMAT_STATUS, false) < 0)
        goto cleanup;

    if (qemuStorageSourcePrivateDataFormat(src, &privateDataBuf) < 0)
        goto cleanup;

    if (virXMLFormatElement(&childBuf, "privateData", NULL, &privateDataBuf) < 0)
        goto cleanup;

    if (virXMLFormatElement(buf, "migrationSource", &attrBuf, &childBuf) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virBufferFreeAndReset(&attrBuf);
    virBufferFreeAndReset(&childBuf);
    virBufferFreeAndReset(&privateDataBuf);

    return ret;
}


static int
2384 2385 2386
qemuDomainObjPrivateXMLFormatNBDMigration(virBufferPtr buf,
                                          virDomainObjPtr vm)
{
2387 2388
    virBuffer attrBuf = VIR_BUFFER_INITIALIZER;
    virBuffer childBuf = VIR_BUFFER_INITIALIZER;
2389 2390 2391
    size_t i;
    virDomainDiskDefPtr disk;
    qemuDomainDiskPrivatePtr diskPriv;
2392
    int ret = -1;
2393 2394 2395 2396

    for (i = 0; i < vm->def->ndisks; i++) {
        disk = vm->def->disks[i];
        diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
2397 2398 2399 2400

        virBufferSetChildIndent(&childBuf, buf);

        virBufferAsprintf(&attrBuf, " dev='%s' migrating='%s'",
2401
                          disk->dst, diskPriv->migrating ? "yes" : "no");
2402 2403 2404 2405 2406 2407 2408 2409

        if (diskPriv->migrSource &&
            qemuDomainObjPrivateXMLFormatNBDMigrationSource(&childBuf,
                                                            diskPriv->migrSource) < 0)
            goto cleanup;

        if (virXMLFormatElement(buf, "disk", &attrBuf, &childBuf) < 0)
            goto cleanup;
2410
    }
2411 2412 2413 2414 2415 2416 2417 2418

    ret = 0;

 cleanup:
    virBufferFreeAndReset(&attrBuf);
    virBufferFreeAndReset(&childBuf);

    return ret;
2419 2420 2421
}


2422
static int
2423 2424 2425 2426
qemuDomainObjPrivateXMLFormatJob(virBufferPtr buf,
                                 virDomainObjPtr vm,
                                 qemuDomainObjPrivatePtr priv)
{
2427 2428
    virBuffer attrBuf = VIR_BUFFER_INITIALIZER;
    virBuffer childBuf = VIR_BUFFER_INITIALIZER;
2429
    qemuDomainJob job = priv->job.active;
2430
    int ret = -1;
2431 2432

    if (!qemuDomainTrackJob(job))
2433
        job = QEMU_JOB_NONE;
2434

2435 2436
    if (job == QEMU_JOB_NONE &&
        priv->job.asyncJob == QEMU_ASYNC_JOB_NONE)
2437 2438 2439
        return 0;

    virBufferSetChildIndent(&childBuf, buf);
2440

2441
    virBufferAsprintf(&attrBuf, " type='%s' async='%s'",
2442 2443
                      qemuDomainJobTypeToString(job),
                      qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
2444

2445
    if (priv->job.phase) {
2446
        virBufferAsprintf(&attrBuf, " phase='%s'",
2447 2448 2449
                          qemuDomainAsyncJobPhaseToString(priv->job.asyncJob,
                                                          priv->job.phase));
    }
2450

2451 2452 2453
    if (priv->job.asyncJob != QEMU_ASYNC_JOB_NONE)
        virBufferAsprintf(&attrBuf, " flags='0x%lx'", priv->job.apiFlags);

2454 2455 2456
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_OUT &&
        qemuDomainObjPrivateXMLFormatNBDMigration(&childBuf, vm) < 0)
        goto cleanup;
2457

2458 2459 2460
    if (priv->job.migParams)
        qemuMigrationParamsFormat(&childBuf, priv->job.migParams);

2461 2462 2463 2464 2465 2466 2467 2468 2469 2470
    if (virXMLFormatElement(buf, "job", &attrBuf, &childBuf) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virBufferFreeAndReset(&attrBuf);
    virBufferFreeAndReset(&childBuf);

    return ret;
2471 2472 2473
}


2474
static int
2475 2476
qemuDomainObjPrivateXMLFormat(virBufferPtr buf,
                              virDomainObjPtr vm)
2477
{
2478
    qemuDomainObjPrivatePtr priv = vm->privateData;
2479 2480 2481 2482
    const char *monitorpath;

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
2483
        switch (priv->monConfig->type) {
2484
        case VIR_DOMAIN_CHR_TYPE_UNIX:
2485
            monitorpath = priv->monConfig->data.nix.path;
2486 2487 2488
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
2489
            monitorpath = priv->monConfig->data.file.path;
2490 2491 2492
            break;
        }

2493
        virBufferEscapeString(buf, "<monitor path='%s'", monitorpath);
2494 2495
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
2496
        virBufferAsprintf(buf, " type='%s'/>\n",
2497
                          virDomainChrTypeToString(priv->monConfig->type));
2498 2499
    }

2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510
    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");
    }

2511
    qemuDomainObjPrivateXMLFormatVcpus(buf, vm->def);
2512

2513
    if (priv->qemuCaps) {
2514
        size_t i;
2515 2516
        virBufferAddLit(buf, "<qemuCaps>\n");
        virBufferAdjustIndent(buf, 2);
2517
        for (i = 0; i < QEMU_CAPS_LAST; i++) {
2518
            if (virQEMUCapsGet(priv->qemuCaps, i)) {
2519
                virBufferAsprintf(buf, "<flag name='%s'/>\n",
2520
                                  virQEMUCapsTypeToString(i));
2521 2522
            }
        }
2523 2524
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</qemuCaps>\n");
2525 2526
    }

2527
    if (priv->lockState)
2528
        virBufferAsprintf(buf, "<lockstate>%s</lockstate>\n", priv->lockState);
2529

2530 2531
    if (qemuDomainObjPrivateXMLFormatJob(buf, vm, priv) < 0)
        return -1;
2532

2533
    if (priv->fakeReboot)
2534
        virBufferAddLit(buf, "<fakereboot/>\n");
2535

2536 2537
    if (priv->qemuDevices && *priv->qemuDevices) {
        char **tmp = priv->qemuDevices;
2538 2539
        virBufferAddLit(buf, "<devices>\n");
        virBufferAdjustIndent(buf, 2);
2540
        while (*tmp) {
2541
            virBufferAsprintf(buf, "<device alias='%s'/>\n", *tmp);
2542 2543
            tmp++;
        }
2544 2545
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</devices>\n");
2546 2547
    }

2548
    if (qemuDomainObjPrivateXMLFormatAutomaticPlacement(buf, priv) < 0)
2549
        return -1;
2550

2551 2552 2553 2554 2555
    /* Various per-domain paths */
    virBufferEscapeString(buf, "<libDir path='%s'/>\n", priv->libDir);
    virBufferEscapeString(buf, "<channelTargetDir path='%s'/>\n",
                          priv->channelTargetDir);

2556
    virCPUDefFormatBufFull(buf, priv->origCPU, NULL);
2557

2558
    if (priv->chardevStdioLogd)
2559
        virBufferAddLit(buf, "<chardevStdioLogd/>\n");
2560

2561 2562 2563
    if (priv->rememberOwner)
        virBufferAddLit(buf, "<rememberOwner/>\n");

2564 2565
    qemuDomainObjPrivateXMLFormatAllowReboot(buf, priv->allowReboot);

2566 2567
    qemuDomainObjPrivateXMLFormatPR(buf, priv);

2568 2569 2570
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV))
        virBufferAsprintf(buf, "<nodename index='%llu'/>\n", priv->nodenameindex);

2571 2572 2573
    if (priv->memPrealloc)
        virBufferAddLit(buf, "<memPrealloc/>\n");

2574 2575 2576
    if (qemuDomainObjPrivateXMLFormatBlockjobs(buf, vm) < 0)
        return -1;

2577 2578 2579
    return 0;
}

2580 2581 2582 2583

static int
qemuDomainObjPrivateXMLParseVcpu(xmlNodePtr node,
                                 unsigned int idx,
2584
                                 virDomainDefPtr def)
2585
{
2586
    virDomainVcpuDefPtr vcpu;
2587
    char *idstr;
2588
    char *pidstr;
2589
    unsigned int tmp;
2590 2591
    int ret = -1;

2592 2593
    idstr = virXMLPropString(node, "id");

2594 2595
    if (idstr &&
        (virStrToLong_uip(idstr, NULL, 10, &idx) < 0)) {
2596
        virReportError(VIR_ERR_INTERNAL_ERROR,
2597 2598 2599 2600 2601 2602
                       _("cannot parse vcpu index '%s'"), idstr);
        goto cleanup;
    }
    if (!(vcpu = virDomainDefGetVcpu(def, idx))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("invalid vcpu index '%u'"), idx);
2603
        goto cleanup;
2604 2605
    }

2606 2607 2608
    if (!(pidstr = virXMLPropString(node, "pid")))
        goto cleanup;

2609
    if (virStrToLong_uip(pidstr, NULL, 10, &tmp) < 0)
2610 2611
        goto cleanup;

2612 2613
    QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid = tmp;

2614 2615 2616
    ret = 0;

 cleanup:
2617
    VIR_FREE(idstr);
2618 2619 2620 2621 2622
    VIR_FREE(pidstr);
    return ret;
}


2623 2624 2625 2626 2627 2628 2629
static int
qemuDomainObjPrivateXMLParseAutomaticPlacement(xmlXPathContextPtr ctxt,
                                               qemuDomainObjPrivatePtr priv,
                                               virQEMUDriverPtr driver)
{
    virCapsPtr caps = NULL;
    char *nodeset;
2630
    char *cpuset;
2631 2632
    int nodesetSize = 0;
    size_t i;
2633 2634 2635
    int ret = -1;

    nodeset = virXPathString("string(./numad/@nodeset)", ctxt);
2636
    cpuset = virXPathString("string(./numad/@cpuset)", ctxt);
2637

2638
    if (!nodeset && !cpuset)
2639 2640 2641 2642 2643
        return 0;

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

2644 2645 2646 2647 2648 2649
    /* Figure out how big the nodeset bitmap needs to be.
     * This is necessary because NUMA node IDs are not guaranteed to
     * start from 0 or be densely allocated */
    for (i = 0; i < caps->host.nnumaCell; i++)
        nodesetSize = MAX(nodesetSize, caps->host.numaCell[i]->num + 1);

2650
    if (nodeset &&
2651
        virBitmapParse(nodeset, &priv->autoNodeset, nodesetSize) < 0)
2652 2653
        goto cleanup;

2654 2655 2656 2657 2658 2659 2660 2661 2662 2663
    if (cpuset) {
        if (virBitmapParse(cpuset, &priv->autoCpuset, VIR_DOMAIN_CPUMASK_LEN) < 0)
            goto cleanup;
    } else {
        /* autoNodeset is present in this case, since otherwise we wouldn't
         * reach this code */
        if (!(priv->autoCpuset = virCapabilitiesGetCpusForNodemask(caps,
                                                                   priv->autoNodeset)))
            goto cleanup;
    }
2664 2665 2666 2667 2668 2669

    ret = 0;

 cleanup:
    virObjectUnref(caps);
    VIR_FREE(nodeset);
2670
    VIR_FREE(cpuset);
2671 2672 2673 2674 2675

    return ret;
}


2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691
static int
qemuDomainObjPrivateXMLParseBlockjobs(qemuDomainObjPrivatePtr priv,
                                      xmlXPathContextPtr ctxt)
{
    char *active;
    int tmp;

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

    VIR_FREE(active);
    return 0;
}


2692
int
2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716
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;
}


2717 2718 2719 2720 2721 2722 2723 2724
static void
qemuDomainObjPrivateXMLParsePR(xmlXPathContextPtr ctxt,
                               bool *prDaemonRunning)
{
    *prDaemonRunning = virXPathBoolean("boolean(./prDaemon)", ctxt) > 0;
}


2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754
static int
qemuDomainObjPrivateXMLParseJobNBDSource(xmlNodePtr node,
                                         xmlXPathContextPtr ctxt,
                                         virDomainDiskDefPtr disk)
{
    xmlNodePtr savedNode = ctxt->node;
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
    virStorageSourcePtr migrSource = NULL;
    char *format = NULL;
    char *type = NULL;
    int ret = -1;

    ctxt->node = node;

    if (!(ctxt->node = virXPathNode("./migrationSource", ctxt))) {
        ret = 0;
        goto cleanup;
    }

    if (VIR_ALLOC(migrSource) < 0)
        goto cleanup;

    if (!(type = virXMLPropString(ctxt->node, "type"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing storage source type"));
        goto cleanup;
    }

    if (!(format = virXMLPropString(ctxt->node, "format"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
2755
                       _("missing storage source format"));
2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790
        goto cleanup;
    }

    if ((migrSource->type = virStorageTypeFromString(type)) <= 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("unknown storage source type '%s'"), type);
        goto cleanup;
    }

    if ((migrSource->format = virStorageFileFormatTypeFromString(format)) <= 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("unknown storage source format '%s'"), format);
        goto cleanup;
    }

    if (virDomainStorageSourceParse(ctxt->node, ctxt, migrSource,
                                    VIR_DOMAIN_DEF_PARSE_STATUS) < 0)
        goto cleanup;

    if ((ctxt->node = virXPathNode("./privateData", ctxt)) &&
        qemuStorageSourcePrivateDataParse(ctxt, migrSource) < 0)
        goto cleanup;

    VIR_STEAL_PTR(diskPriv->migrSource, migrSource);
    ret = 0;

 cleanup:
    virStorageSourceFree(migrSource);
    VIR_FREE(format);
    VIR_FREE(type);
    ctxt->node = savedNode;
    return ret;
}


2791 2792 2793 2794 2795 2796
static int
qemuDomainObjPrivateXMLParseJobNBD(virDomainObjPtr vm,
                                   qemuDomainObjPrivatePtr priv,
                                   xmlXPathContextPtr ctxt)
{
    xmlNodePtr *nodes = NULL;
2797
    char *dst = NULL;
2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813
    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;

2814 2815
            if ((dst = virXMLPropString(nodes[i], "dev")) &&
                (disk = virDomainDiskByName(vm->def, dst, false))) {
2816
                QEMU_DOMAIN_DISK_PRIVATE(disk)->migrating = true;
2817 2818 2819 2820 2821 2822

                if (qemuDomainObjPrivateXMLParseJobNBDSource(nodes[i], ctxt,
                                                             disk) < 0)
                    goto cleanup;
            }

2823 2824 2825 2826 2827 2828 2829 2830
            VIR_FREE(dst);
        }
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
2831
    VIR_FREE(dst);
2832 2833 2834 2835
    return ret;
}


2836 2837 2838 2839 2840
static int
qemuDomainObjPrivateXMLParseJob(virDomainObjPtr vm,
                                qemuDomainObjPrivatePtr priv,
                                xmlXPathContextPtr ctxt)
{
2841
    xmlNodePtr savedNode = ctxt->node;
2842 2843 2844
    char *tmp = NULL;
    int ret = -1;

2845 2846 2847 2848 2849 2850
    if (!(ctxt->node = virXPathNode("./job[1]", ctxt))) {
        ret = 0;
        goto cleanup;
    }

    if ((tmp = virXPathString("string(@type)", ctxt))) {
2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861
        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;
    }

2862
    if ((tmp = virXPathString("string(@async)", ctxt))) {
2863 2864 2865 2866 2867 2868 2869 2870 2871 2872
        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;

2873
        if ((tmp = virXPathString("string(@phase)", ctxt))) {
2874 2875 2876 2877 2878 2879 2880 2881 2882 2883
            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);
        }
    }

2884 2885 2886 2887 2888
    if (virXPathULongHex("string(@flags)", ctxt, &priv->job.apiFlags) == -2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Invalid job flags"));
        goto cleanup;
    }

2889
    if (qemuDomainObjPrivateXMLParseJobNBD(vm, priv, ctxt) < 0)
2890 2891
        goto cleanup;

2892 2893 2894
    if (qemuMigrationParamsParse(ctxt, &priv->job.migParams) < 0)
        goto cleanup;

2895 2896 2897
    ret = 0;

 cleanup:
2898
    ctxt->node = savedNode;
2899 2900 2901 2902 2903
    VIR_FREE(tmp);
    return ret;
}


2904
static int
2905
qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt,
2906
                             virDomainObjPtr vm,
2907
                             virDomainDefParserConfigPtr config)
2908
{
2909
    qemuDomainObjPrivatePtr priv = vm->privateData;
2910
    virQEMUDriverPtr driver = config->priv;
2911
    char *monitorpath;
2912
    char *tmp = NULL;
2913 2914
    int n;
    size_t i;
2915
    xmlNodePtr *nodes = NULL;
2916
    xmlNodePtr node = NULL;
2917
    virQEMUCapsPtr qemuCaps = NULL;
2918

2919
    if (!(priv->monConfig = virDomainChrSourceDefNew(NULL)))
2920 2921 2922 2923
        goto error;

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
2924 2925
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("no monitor path"));
2926 2927 2928 2929 2930
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
2931
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
2932
    else
2933
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
2934 2935
    VIR_FREE(tmp);

E
Eric Blake 已提交
2936 2937
    priv->monJSON = virXPathBoolean("count(./monitor[@json = '1']) > 0",
                                    ctxt) > 0;
2938

2939
    switch (priv->monConfig->type) {
2940
    case VIR_DOMAIN_CHR_TYPE_PTY:
2941
        priv->monConfig->data.file.path = monitorpath;
2942 2943
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
2944
        priv->monConfig->data.nix.path = monitorpath;
2945 2946 2947
        break;
    default:
        VIR_FREE(monitorpath);
2948 2949 2950
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported monitor type '%s'"),
                       virDomainChrTypeToString(priv->monConfig->type));
2951 2952 2953
        goto error;
    }

2954 2955 2956 2957
    if ((node = virXPathNode("./namespaces", ctxt))) {
        xmlNodePtr next;

        for (next = node->children; next; next = next->next) {
2958
            int ns = qemuDomainNamespaceTypeFromString((const char *)next->name);
2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977

            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;
    }

2978 2979
    priv->rememberOwner = virXPathBoolean("count(./rememberOwner) > 0", ctxt);

2980
    if ((n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes)) < 0)
2981 2982
        goto error;

2983
    for (i = 0; i < n; i++) {
2984
        if (qemuDomainObjPrivateXMLParseVcpu(nodes[i], i, vm->def) < 0)
2985
            goto error;
2986
    }
2987
    VIR_FREE(nodes);
2988

2989
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
2990 2991
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse qemu capabilities flags"));
2992 2993 2994
        goto error;
    }
    if (n > 0) {
2995
        if (!(qemuCaps = virQEMUCapsNew()))
2996 2997
            goto error;

2998
        for (i = 0; i < n; i++) {
2999 3000
            char *str = virXMLPropString(nodes[i], "name");
            if (str) {
3001
                int flag = virQEMUCapsTypeFromString(str);
3002
                if (flag < 0) {
3003 3004
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Unknown qemu capabilities flag %s"), str);
3005
                    VIR_FREE(str);
3006 3007
                    goto error;
                }
3008
                VIR_FREE(str);
3009
                virQEMUCapsSet(qemuCaps, flag);
3010 3011 3012
            }
        }

M
Marc Hartmayer 已提交
3013
        VIR_STEAL_PTR(priv->qemuCaps, qemuCaps);
3014 3015 3016
    }
    VIR_FREE(nodes);

3017
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
3018

3019
    if (qemuDomainObjPrivateXMLParseJob(vm, priv, ctxt) < 0)
3020 3021
        goto error;

3022 3023
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044
    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);

3045
    if (qemuDomainObjPrivateXMLParseAutomaticPlacement(ctxt, priv, driver) < 0)
3046 3047
        goto error;

3048 3049 3050 3051 3052 3053 3054 3055 3056
    if ((tmp = virXPathString("string(./libDir/@path)", ctxt)))
        priv->libDir = tmp;
    if ((tmp = virXPathString("string(./channelTargetDir/@path)", ctxt)))
        priv->channelTargetDir = tmp;
    tmp = NULL;

    if (qemuDomainSetPrivatePathsOld(driver, vm) < 0)
        goto error;

3057 3058 3059
    if (virCPUDefParseXML(ctxt, "./cpu", VIR_CPU_TYPE_GUEST, &priv->origCPU) < 0)
        goto error;

3060 3061 3062
    priv->chardevStdioLogd = virXPathBoolean("boolean(./chardevStdioLogd)",
                                             ctxt) == 1;

3063 3064
    qemuDomainObjPrivateXMLParseAllowReboot(ctxt, &priv->allowReboot);

3065 3066
    qemuDomainObjPrivateXMLParsePR(ctxt, &priv->prDaemonRunning);

3067 3068 3069
    if (qemuDomainObjPrivateXMLParseBlockjobs(priv, ctxt) < 0)
        goto error;

3070 3071 3072 3073 3074 3075 3076 3077
    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;
    }

3078 3079
    priv->memPrealloc = virXPathBoolean("boolean(./memPrealloc)", ctxt) == 1;

3080 3081
    return 0;

3082
 error:
3083
    VIR_FREE(nodes);
3084
    VIR_FREE(tmp);
3085 3086 3087 3088
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
    virDomainChrSourceDefFree(priv->monConfig);
    priv->monConfig = NULL;
3089
    virStringListFree(priv->qemuDevices);
3090
    priv->qemuDevices = NULL;
3091
    virObjectUnref(qemuCaps);
3092 3093 3094 3095
    return -1;
}


3096 3097 3098 3099 3100 3101 3102 3103 3104
static void *
qemuDomainObjPrivateXMLGetParseOpaque(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return priv->qemuCaps;
}


3105 3106 3107
virDomainXMLPrivateDataCallbacks virQEMUDriverPrivateDataCallbacks = {
    .alloc = qemuDomainObjPrivateAlloc,
    .free = qemuDomainObjPrivateFree,
3108
    .diskNew = qemuDomainDiskPrivateNew,
3109 3110
    .diskParse = qemuDomainDiskPrivateParse,
    .diskFormat = qemuDomainDiskPrivateFormat,
3111
    .vcpuNew = qemuDomainVcpuPrivateNew,
3112
    .chrSourceNew = qemuDomainChrSourcePrivateNew,
J
Ján Tomko 已提交
3113
    .vsockNew = qemuDomainVsockPrivateNew,
3114
    .graphicsNew = qemuDomainGraphicsPrivateNew,
3115 3116
    .parse = qemuDomainObjPrivateXMLParse,
    .format = qemuDomainObjPrivateXMLFormat,
3117
    .getParseOpaque = qemuDomainObjPrivateXMLGetParseOpaque,
3118 3119
    .storageParse = qemuStorageSourcePrivateDataParse,
    .storageFormat = qemuStorageSourcePrivateDataFormat,
3120 3121 3122
};


3123 3124 3125 3126 3127
static void
qemuDomainDefNamespaceFree(void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;

3128
    qemuDomainCmdlineDefFree(cmd);
3129 3130 3131
}

static int
P
Philipp Hahn 已提交
3132 3133
qemuDomainDefNamespaceParse(xmlDocPtr xml ATTRIBUTE_UNUSED,
                            xmlNodePtr root ATTRIBUTE_UNUSED,
3134 3135 3136 3137
                            xmlXPathContextPtr ctxt,
                            void **data)
{
    qemuDomainCmdlineDefPtr cmd = NULL;
P
Philipp Hahn 已提交
3138
    bool uses_qemu_ns = false;
3139
    xmlNodePtr *nodes = NULL;
3140 3141
    int n;
    size_t i;
3142

P
Philipp Hahn 已提交
3143
    if (xmlXPathRegisterNs(ctxt, BAD_CAST "qemu", BAD_CAST QEMU_NAMESPACE_HREF) < 0) {
3144 3145 3146
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to register xml namespace '%s'"),
                       QEMU_NAMESPACE_HREF);
3147 3148 3149
        return -1;
    }

3150
    if (VIR_ALLOC(cmd) < 0)
3151 3152 3153 3154 3155 3156
        return -1;

    /* first handle the extra command-line arguments */
    n = virXPathNodeSet("./qemu:commandline/qemu:arg", ctxt, &nodes);
    if (n < 0)
        goto error;
P
Philipp Hahn 已提交
3157
    uses_qemu_ns |= n > 0;
3158 3159

    if (n && VIR_ALLOC_N(cmd->args, n) < 0)
3160
        goto error;
3161 3162 3163 3164

    for (i = 0; i < n; i++) {
        cmd->args[cmd->num_args] = virXMLPropString(nodes[i], "value");
        if (cmd->args[cmd->num_args] == NULL) {
3165 3166
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu command-line argument specified"));
3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177
            goto error;
        }
        cmd->num_args++;
    }

    VIR_FREE(nodes);

    /* now handle the extra environment variables */
    n = virXPathNodeSet("./qemu:commandline/qemu:env", ctxt, &nodes);
    if (n < 0)
        goto error;
P
Philipp Hahn 已提交
3178
    uses_qemu_ns |= n > 0;
3179 3180

    if (n && VIR_ALLOC_N(cmd->env_name, n) < 0)
3181
        goto error;
3182 3183

    if (n && VIR_ALLOC_N(cmd->env_value, n) < 0)
3184
        goto error;
3185 3186 3187 3188 3189 3190

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

        tmp = virXMLPropString(nodes[i], "name");
        if (tmp == NULL) {
3191 3192
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu environment name specified"));
3193 3194 3195
            goto error;
        }
        if (tmp[0] == '\0') {
3196 3197
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Empty qemu environment name specified"));
3198 3199 3200
            goto error;
        }
        if (!c_isalpha(tmp[0]) && tmp[0] != '_') {
3201 3202
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must begin with a letter or underscore"));
3203 3204 3205
            goto error;
        }
        if (strspn(tmp, "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789_") != strlen(tmp)) {
3206 3207
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must contain only alphanumerics and underscore"));
3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219
            goto error;
        }

        cmd->env_name[cmd->num_env] = tmp;

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

    VIR_FREE(nodes);

P
Philipp Hahn 已提交
3220 3221 3222 3223
    if (uses_qemu_ns)
        *data = cmd;
    else
        VIR_FREE(cmd);
3224 3225 3226

    return 0;

3227
 error:
3228 3229 3230 3231 3232 3233 3234 3235 3236 3237
    VIR_FREE(nodes);
    qemuDomainDefNamespaceFree(cmd);
    return -1;
}

static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;
3238
    size_t i;
3239 3240 3241 3242

    if (!cmd->num_args && !cmd->num_env)
        return 0;

3243 3244 3245
    virBufferAddLit(buf, "<qemu:commandline>\n");
    virBufferAdjustIndent(buf, 2);

3246
    for (i = 0; i < cmd->num_args; i++)
3247
        virBufferEscapeString(buf, "<qemu:arg value='%s'/>\n",
3248 3249
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
3250
        virBufferAsprintf(buf, "<qemu:env name='%s'", cmd->env_name[i]);
3251 3252 3253 3254 3255
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }

3256 3257
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:commandline>\n");
3258 3259 3260 3261 3262 3263 3264 3265 3266 3267
    return 0;
}

static const char *
qemuDomainDefNamespaceHref(void)
{
    return "xmlns:qemu='" QEMU_NAMESPACE_HREF "'";
}


3268 3269 3270 3271 3272 3273
virDomainXMLNamespace virQEMUDriverDomainXMLNamespace = {
    .parse = qemuDomainDefNamespaceParse,
    .free = qemuDomainDefNamespaceFree,
    .format = qemuDomainDefNamespaceFormatXML,
    .href = qemuDomainDefNamespaceHref,
};
3274

3275

P
Pavel Hrdina 已提交
3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294
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;
}


3295
static int
3296 3297
qemuDomainDefAddDefaultDevices(virDomainDefPtr def,
                               virQEMUCapsPtr qemuCaps)
3298
{
3299
    bool addDefaultUSB = true;
3300
    int usbModel = -1; /* "default for machinetype" */
3301
    int pciRoot;       /* index within def->controllers */
3302
    bool addImplicitSATA = false;
3303
    bool addPCIRoot = false;
L
Laine Stump 已提交
3304
    bool addPCIeRoot = false;
3305
    bool addDefaultMemballoon = true;
3306 3307
    bool addDefaultUSBKBD = false;
    bool addDefaultUSBMouse = false;
3308
    bool addPanicDevice = false;
3309
    int ret = -1;
3310

P
Pavel Hrdina 已提交
3311 3312 3313 3314
    /* add implicit input devices */
    if (qemuDomainDefAddImplicitInputDevice(def) < 0)
        goto cleanup;

3315 3316 3317 3318
    /* 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
3319
        if (STREQ(def->os.machine, "isapc")) {
3320
            addDefaultUSB = false;
3321
            break;
3322
        }
3323
        if (qemuDomainIsQ35(def)) {
3324 3325
            addPCIeRoot = true;
            addImplicitSATA = true;
3326

3327 3328 3329
            /* 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.
3330
             */
3331 3332 3333
            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))
3334 3335
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
            else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_USB_EHCI1))
3336 3337 3338
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_ICH9_EHCI1;
            else
                addDefaultUSB = false;
3339
            break;
L
Laine Stump 已提交
3340
        }
3341
        if (qemuDomainIsI440FX(def))
3342
            addPCIRoot = true;
3343 3344
        break;

3345 3346 3347 3348 3349 3350 3351
    case VIR_ARCH_ARMV6L:
        addDefaultUSB = false;
        addDefaultMemballoon = false;
        if (STREQ(def->os.machine, "versatilepb"))
            addPCIRoot = true;
        break;

3352
    case VIR_ARCH_ARMV7L:
3353
    case VIR_ARCH_AARCH64:
3354 3355
        addDefaultUSB = false;
        addDefaultMemballoon = false;
3356
        if (qemuDomainIsARMVirt(def))
3357 3358
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
        break;
3359

3360
    case VIR_ARCH_PPC64:
3361
    case VIR_ARCH_PPC64LE:
3362 3363 3364
        addPCIRoot = true;
        addDefaultUSBKBD = true;
        addDefaultUSBMouse = true;
3365 3366 3367
        /* For pSeries guests, the firmware provides the same
         * functionality as the pvpanic device, so automatically
         * add the definition if not already present */
3368
        if (qemuDomainIsPSeries(def))
3369
            addPanicDevice = true;
3370 3371
        break;

3372 3373 3374 3375 3376 3377 3378
    case VIR_ARCH_ALPHA:
    case VIR_ARCH_PPC:
    case VIR_ARCH_PPCEMB:
    case VIR_ARCH_SH4:
    case VIR_ARCH_SH4EB:
        addPCIRoot = true;
        break;
3379

3380 3381 3382 3383 3384
    case VIR_ARCH_RISCV32:
    case VIR_ARCH_RISCV64:
        addDefaultUSB = false;
        break;

3385 3386 3387
    case VIR_ARCH_S390:
    case VIR_ARCH_S390X:
        addDefaultUSB = false;
3388
        addPanicDevice = true;
3389
        addPCIRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_ZPCI);
3390
        break;
3391 3392 3393 3394 3395 3396

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

3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416
    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:
3417 3418 3419 3420
    default:
        break;
    }

3421
    if (addDefaultUSB &&
3422 3423
        virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_USB, 0) < 0 &&
        virDomainDefAddUSBController(def, 0, usbModel) < 0)
3424
        goto cleanup;
3425

3426 3427 3428
    if (addImplicitSATA &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_SATA, 0, -1) < 0)
3429
        goto cleanup;
3430

3431 3432
    pciRoot = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

3433 3434 3435
    /* NB: any machine that sets addPCIRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
     */
3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450
    if (addPCIRoot) {
        if (pciRoot >= 0) {
            if (def->controllers[pciRoot]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT) {
                virReportError(VIR_ERR_XML_ERROR,
                               _("The PCI controller with index='0' must be "
                                 "model='pci-root' for this machine type, "
                                 "but model='%s' was found instead"),
                               virDomainControllerModelPCITypeToString(def->controllers[pciRoot]->model));
                goto cleanup;
            }
        } else if (!virDomainDefAddController(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
                                              VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)) {
            goto cleanup;
        }
    }
3451

3452 3453 3454
    /* 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
3455 3456 3457
     *
     * NB: any machine that sets addPCIeRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
3458 3459
     */
    if (addPCIeRoot) {
3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472
        if (pciRoot >= 0) {
            if (def->controllers[pciRoot]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT) {
                virReportError(VIR_ERR_XML_ERROR,
                               _("The PCI controller with index='0' must be "
                                 "model='pcie-root' for this machine type, "
                                 "but model='%s' was found instead"),
                               virDomainControllerModelPCITypeToString(def->controllers[pciRoot]->model));
                goto cleanup;
            }
        } else if (!virDomainDefAddController(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
                                             VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT)) {
            goto cleanup;
        }
3473
    }
3474

3475
    if (addDefaultMemballoon && !def->memballoon) {
3476 3477
        virDomainMemballoonDefPtr memballoon;
        if (VIR_ALLOC(memballoon) < 0)
3478
            goto cleanup;
3479 3480 3481 3482 3483

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

3484 3485 3486 3487
    if (STRPREFIX(def->os.machine, "s390-virtio") &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_S390) && def->memballoon)
        def->memballoon->model = VIR_DOMAIN_MEMBALLOON_MODEL_NONE;

3488 3489 3490 3491 3492
    if (addDefaultUSBKBD &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_KBD,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
3493
        goto cleanup;
3494 3495 3496 3497 3498 3499

    if (addDefaultUSBMouse &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
3500
        goto cleanup;
3501

D
Dmitry Andreev 已提交
3502 3503 3504 3505
    if (addPanicDevice) {
        size_t j;
        for (j = 0; j < def->npanics; j++) {
            if (def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT ||
3506 3507 3508 3509
                (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
3510 3511
                break;
        }
3512

D
Dmitry Andreev 已提交
3513 3514 3515 3516 3517 3518 3519 3520 3521
        if (j == def->npanics) {
            virDomainPanicDefPtr panic;
            if (VIR_ALLOC(panic) < 0 ||
                VIR_APPEND_ELEMENT_COPY(def->panics,
                                        def->npanics, panic) < 0) {
                VIR_FREE(panic);
                goto cleanup;
            }
        }
3522 3523
    }

3524 3525 3526 3527 3528 3529
    ret = 0;
 cleanup:
    return ret;
}


A
Andrea Bolognani 已提交
3530 3531 3532
/**
 * qemuDomainDefEnableDefaultFeatures:
 * @def: domain definition
3533
 * @qemuCaps: QEMU capabilities
A
Andrea Bolognani 已提交
3534 3535 3536 3537 3538
 *
 * Make sure that features that should be enabled by default are actually
 * enabled and configure default values related to those features.
 */
static void
3539 3540
qemuDomainDefEnableDefaultFeatures(virDomainDefPtr def,
                                   virQEMUCapsPtr qemuCaps)
A
Andrea Bolognani 已提交
3541
{
3542
    /* The virt machine type always uses GIC: if the relevant information
3543 3544
     * was not included in the domain XML, we need to choose a suitable
     * GIC version ourselves */
3545
    if ((def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ABSENT &&
3546
         qemuDomainIsARMVirt(def)) ||
3547 3548 3549
        (def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ON &&
         def->gic_version == VIR_GIC_VERSION_NONE)) {
        virGICVersion version;
3550 3551 3552 3553 3554

        VIR_DEBUG("Looking for usable GIC version in domain capabilities");
        for (version = VIR_GIC_VERSION_LAST - 1;
             version > VIR_GIC_VERSION_NONE;
             version--) {
3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569

            /* 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;
            }

3570 3571 3572 3573 3574 3575 3576 3577
            if (virQEMUCapsSupportsGICVersion(qemuCaps,
                                              def->virtType,
                                              version)) {
                VIR_DEBUG("Using GIC version %s",
                          virGICVersionTypeToString(version));
                def->gic_version = version;
                break;
            }
3578 3579
        }

3580 3581 3582 3583 3584 3585 3586
        /* 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;
        }

3587 3588 3589
        /* 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;
3590
    }
A
Andrea Bolognani 已提交
3591 3592 3593
}


3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613
static int
qemuCanonicalizeMachine(virDomainDefPtr def, virQEMUCapsPtr qemuCaps)
{
    const char *canon;

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

    if (STRNEQ(canon, def->os.machine)) {
        char *tmp;
        if (VIR_STRDUP(tmp, canon) < 0)
            return -1;
        VIR_FREE(def->os.machine);
        def->os.machine = tmp;
    }

    return 0;
}


3614
static int
3615 3616 3617 3618 3619
qemuDomainRecheckInternalPaths(virDomainDefPtr def,
                               virQEMUDriverConfigPtr cfg,
                               unsigned int flags)
{
    size_t i = 0;
3620
    size_t j = 0;
3621 3622 3623 3624

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

3625 3626 3627 3628 3629 3630
        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
3631
             * the auto-generated socket based on config option from qemu.conf
3632 3633 3634 3635
             * 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 &&
3636
                !glisten->autoGenerated &&
3637 3638 3639 3640 3641 3642 3643
                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;
                }
3644
            }
3645 3646
        }
    }
3647 3648

    return 0;
3649 3650 3651
}


3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721
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;
}


3722 3723 3724 3725 3726 3727
static int
qemuDomainDefCPUPostParse(virDomainDefPtr def)
{
    if (!def->cpu)
        return 0;

3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781
    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;
        }
    }

3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814
    /* 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;
}


3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846
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;
}


3847 3848 3849 3850 3851 3852 3853 3854
static int
qemuDomainDefPostParseBasic(virDomainDefPtr def,
                            virCapsPtr caps,
                            void *opaque ATTRIBUTE_UNUSED)
{
    /* check for emulator and create a default one if needed */
    if (!def->emulator &&
        !(def->emulator = virDomainDefGetDefaultEmulator(def, caps)))
3855
        return 1;
3856 3857 3858 3859 3860

    return 0;
}


3861 3862
static int
qemuDomainDefPostParse(virDomainDefPtr def,
3863
                       virCapsPtr caps ATTRIBUTE_UNUSED,
3864
                       unsigned int parseFlags,
3865
                       void *opaque,
3866
                       void *parseOpaque)
3867 3868
{
    virQEMUDriverPtr driver = opaque;
3869
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3870 3871 3872
    /* 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. */
3873
    virQEMUCapsPtr qemuCaps = parseOpaque;
3874 3875 3876 3877 3878
    int ret = -1;

    if (def->os.bootloader || def->os.bootloaderArgs) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bootloader is not supported by QEMU"));
3879
        goto cleanup;
3880 3881
    }

3882 3883 3884
    if (!def->os.machine) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing machine type"));
3885
        goto cleanup;
3886 3887
    }

3888 3889 3890 3891 3892 3893 3894 3895 3896
    if (def->os.loader &&
        def->os.loader->type == VIR_DOMAIN_LOADER_TYPE_PFLASH &&
        def->os.loader->readonly == VIR_TRISTATE_SWITCH_ON &&
        !def->os.loader->nvram) {
        if (virAsprintf(&def->os.loader->nvram, "%s/%s_VARS.fd",
                        cfg->nvramDir, def->name) < 0)
            goto cleanup;
    }

3897 3898 3899
    if (qemuDomainDefAddDefaultDevices(def, qemuCaps) < 0)
        goto cleanup;

3900 3901 3902
    if (qemuCanonicalizeMachine(def, qemuCaps) < 0)
        goto cleanup;

3903
    qemuDomainDefEnableDefaultFeatures(def, qemuCaps);
A
Andrea Bolognani 已提交
3904

3905 3906
    if (qemuDomainRecheckInternalPaths(def, cfg, parseFlags) < 0)
        goto cleanup;
3907

3908
    if (qemuSecurityVerify(driver->securityManager, def) < 0)
3909 3910
        goto cleanup;

3911 3912 3913
    if (qemuDomainDefVcpusPostParse(def) < 0)
        goto cleanup;

3914 3915 3916
    if (qemuDomainDefCPUPostParse(def) < 0)
        goto cleanup;

3917 3918 3919
    if (qemuDomainDefTsegPostParse(def, qemuCaps) < 0)
        goto cleanup;

3920 3921
    ret = 0;
 cleanup:
3922
    virObjectUnref(cfg);
3923
    return ret;
3924 3925
}

3926

3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956
/**
 * 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;
}


3957 3958 3959
#define QEMU_MAX_VCPUS_WITHOUT_EIM 255


3960
static int
3961 3962
qemuDomainDefValidateFeatures(const virDomainDef *def,
                              virQEMUCapsPtr qemuCaps)
3963
{
3964
    size_t i;
3965

3966 3967 3968 3969 3970
    for (i = 0; i < VIR_DOMAIN_FEATURE_LAST; i++) {
        const char *featureName = virDomainFeatureTypeToString(i);

        switch ((virDomainFeature) i) {
        case VIR_DOMAIN_FEATURE_IOAPIC:
3971
            if (def->features[i] != VIR_DOMAIN_IOAPIC_NONE &&
3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983
                !ARCH_IS_X86(def->os.arch)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("The '%s' feature is not supported for "
                                 "architecture '%s' or machine type '%s'"),
                               featureName,
                               virArchToString(def->os.arch),
                               def->os.machine);
                return -1;
            }
            break;

        case VIR_DOMAIN_FEATURE_HPT:
3984
        case VIR_DOMAIN_FEATURE_HTM:
3985
        case VIR_DOMAIN_FEATURE_NESTED_HV:
3986
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997
                !qemuDomainIsPSeries(def)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("The '%s' feature is not supported for "
                                 "architecture '%s' or machine type '%s'"),
                               featureName,
                               virArchToString(def->os.arch),
                               def->os.machine);
                return -1;
            }
            break;

3998 3999
        case VIR_DOMAIN_FEATURE_GIC:
            if (def->features[i] == VIR_TRISTATE_SWITCH_ON &&
4000
                !qemuDomainIsARMVirt(def)) {
4001 4002 4003 4004 4005 4006 4007 4008 4009 4010
                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;

4011 4012
        case VIR_DOMAIN_FEATURE_SMM:
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
4013
                !virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT)) {
4014 4015 4016 4017 4018 4019
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("smm is not available with this QEMU binary"));
                return -1;
            }
            break;

4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035
        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_KVM:
        case VIR_DOMAIN_FEATURE_PVSPINLOCK:
        case VIR_DOMAIN_FEATURE_CAPABILITIES:
        case VIR_DOMAIN_FEATURE_PMU:
        case VIR_DOMAIN_FEATURE_VMPORT:
        case VIR_DOMAIN_FEATURE_VMCOREINFO:
        case VIR_DOMAIN_FEATURE_LAST:
            break;
        }
4036 4037 4038 4039 4040 4041
    }

    return 0;
}


4042
static int
M
Marc-André Lureau 已提交
4043 4044
qemuDomainDefValidateMemory(const virDomainDef *def,
                            virQEMUCapsPtr qemuCaps)
4045 4046
{
    const long system_page_size = virGetSystemPageSizeKB();
4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064
    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;
    }
4065

M
Marc-André Lureau 已提交
4066 4067 4068
    if (mem->source == VIR_DOMAIN_MEMORY_SOURCE_MEMFD &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4069
                       _("hugepages is not supported with memfd memory source"));
M
Marc-André Lureau 已提交
4070 4071 4072
        return -1;
    }

4073 4074
    /* We can't guarantee any other mem.access
     * if no guest NUMA nodes are defined. */
4075
    if (mem->hugepages[0].size != system_page_size &&
4076
        virDomainNumaGetNodeCount(def->numa) == 0 &&
4077 4078
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_DEFAULT &&
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_PRIVATE) {
4079 4080 4081
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("memory access mode '%s' not supported "
                         "without guest numa node"),
4082
                       virDomainMemoryAccessTypeToString(mem->access));
4083 4084 4085 4086 4087 4088 4089
        return -1;
    }

    return 0;
}


4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112
static int
qemuDomainValidateCpuCount(const virDomainDef *def,
                            virQEMUCapsPtr qemuCaps)
{
    unsigned int maxCpus = virQEMUCapsGetMachineMaxCpus(qemuCaps, def->os.machine);

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

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

    return 0;
}


4113 4114
static int
qemuDomainDefValidate(const virDomainDef *def,
4115
                      virCapsPtr caps ATTRIBUTE_UNUSED,
4116
                      void *opaque)
4117
{
4118
    virQEMUDriverPtr driver = opaque;
4119
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4120 4121
    virQEMUCapsPtr qemuCaps = NULL;
    int ret = -1;
4122
    size_t i;
4123

4124
    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
4125 4126 4127
                                            def->emulator)))
        goto cleanup;

4128 4129 4130
    if (def->mem.min_guarantee) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Parameter 'min_guarantee' not supported by QEMU."));
4131
        goto cleanup;
4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151
    }

    /* On x86, UEFI requires ACPI */
    if (def->os.loader &&
        def->os.loader->type == VIR_DOMAIN_LOADER_TYPE_PFLASH &&
        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 &&
        (!def->os.loader ||
         def->os.loader->type != VIR_DOMAIN_LOADER_TYPE_PFLASH)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("ACPI requires UEFI on this architecture"));
        goto cleanup;
4152 4153
    }

M
Michal Privoznik 已提交
4154 4155 4156 4157 4158
    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. */

4159
        if (!qemuDomainIsQ35(def)) {
M
Michal Privoznik 已提交
4160 4161
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot is supported with q35 machine types only"));
4162
            goto cleanup;
M
Michal Privoznik 已提交
4163 4164 4165 4166 4167 4168 4169 4170
        }

        /* 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"));
4171
            goto cleanup;
M
Michal Privoznik 已提交
4172 4173 4174 4175 4176
        }

        if (def->features[VIR_DOMAIN_FEATURE_SMM] != VIR_TRISTATE_SWITCH_ON) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot requires SMM feature enabled"));
4177
            goto cleanup;
M
Michal Privoznik 已提交
4178 4179 4180
        }
    }

4181 4182 4183 4184 4185
    /* 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;
4186
        unsigned int granularity;
4187 4188 4189 4190 4191 4192

        /* 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 */
        if (virDomainDefGetVcpusTopology(def, &topologycpus) == 0 &&
            topologycpus != virDomainDefGetVcpusMax(def)) {
4193 4194 4195
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CPU topology doesn't match maximum vcpu count"));
            goto cleanup;
4196
        }
4197 4198 4199 4200 4201 4202 4203 4204 4205 4206

        /* 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;
        }
4207 4208
    }

4209
    if (qemuDomainValidateCpuCount(def, qemuCaps) < 0)
4210 4211
        goto cleanup;

4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229
    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
4230
    if (def->nresctrls &&
4231 4232 4233 4234 4235 4236
        def->virtType != VIR_DOMAIN_VIRT_KVM) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("cachetune is only supported for KVM domains"));
        goto cleanup;
    }

4237
    if (qemuDomainDefValidateFeatures(def, qemuCaps) < 0)
4238 4239
        goto cleanup;

M
Marc-André Lureau 已提交
4240
    if (qemuDomainDefValidateMemory(def, qemuCaps) < 0)
4241 4242
        goto cleanup;

4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254
    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;
            }
        }
    }

4255 4256 4257 4258
    ret = 0;

 cleanup:
    virObjectUnref(qemuCaps);
4259
    virObjectUnref(cfg);
4260
    return ret;
4261 4262 4263
}


4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287
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;
}


4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334
static int
qemuDomainChrSourceReconnectDefValidate(const virDomainChrSourceReconnectDef *def)
{
    if (def->enabled == VIR_TRISTATE_BOOL_YES &&
        def->timeout == 0) {
        virReportError(VIR_ERR_INVALID_ARG, "%s",
                       _("chardev reconnect source timeout cannot be '0'"));
        return -1;
    }

    return 0;
}


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

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

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

    return 0;
}


4335 4336 4337
static int
qemuDomainChrSerialTargetTypeToAddressType(int targetType)
{
4338
    switch ((virDomainChrSerialTargetType)targetType) {
4339 4340 4341 4342 4343 4344
    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;
4345 4346
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
        return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO;
4347
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
4348
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
4349 4350 4351 4352 4353 4354 4355 4356 4357
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        break;
    }

    return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE;
}


4358 4359 4360 4361 4362 4363 4364 4365 4366 4367
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;
4368 4369
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
4370
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
4371
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
4372
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
4373 4374 4375
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
4376 4377 4378 4379 4380 4381 4382 4383 4384
    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;
}


4385
static int
4386
qemuDomainChrTargetDefValidate(const virDomainChrDef *chr)
4387
{
4388 4389
    int expected;

4390
    switch ((virDomainChrDeviceType)chr->deviceType) {
4391 4392 4393
    case VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL:

        /* Validate target type */
4394
        switch ((virDomainChrSerialTargetType)chr->targetType) {
4395
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA:
4396 4397
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI:
4398
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
4399

4400
            expected = qemuDomainChrSerialTargetTypeToAddressType(chr->targetType);
4401 4402

            if (chr->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
4403 4404 4405 4406 4407
                chr->info.type != expected) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Target type '%s' requires address type '%s'"),
                               virDomainChrSerialTargetTypeToString(chr->targetType),
                               virDomainDeviceAddressTypeToString(expected));
4408 4409 4410 4411
                return -1;
            }
            break;

4412
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
4413
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
4414 4415 4416 4417 4418 4419 4420 4421 4422
            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;

4423 4424 4425 4426
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            break;
        }
4427 4428 4429 4430 4431 4432

        /* 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:
4433
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
4434
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
4435 4436
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
4437
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453

            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;
        }
4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467
        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;
}


4468
static int
4469
qemuDomainChrDefValidate(const virDomainChrDef *dev,
4470
                         const virDomainDef *def)
4471
{
4472
    if (qemuDomainChrSourceDefValidate(dev->source) < 0)
4473 4474
        return -1;

4475
    if (qemuDomainChrTargetDefValidate(dev) < 0)
4476 4477
        return -1;

4478
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL &&
4479
        (ARCH_IS_S390(def->os.arch) || qemuDomainIsPSeries(def))) {
4480 4481 4482 4483 4484
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("parallel ports are not supported"));
            return -1;
    }

4485 4486 4487
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL) {
        bool isCompatible = true;

4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498
        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;
            }
        }

4499 4500 4501 4502 4503 4504
        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;
        }

4505 4506 4507 4508 4509 4510 4511
        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;
        }

4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522
        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;
        }
    }

4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538
    return 0;
}


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

    return 0;
}


static int
4539 4540
qemuDomainRNGDefValidate(const virDomainRNGDef *def,
                         virQEMUCapsPtr qemuCaps)
4541
{
4542 4543
    bool modelIsSupported = false;

4544 4545 4546 4547
    if (def->backend == VIR_DOMAIN_RNG_BACKEND_EGD &&
        qemuDomainChrSourceDefValidate(def->source.chardev) < 0)
        return -1;

4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563
    switch ((virDomainRNGModel) def->model) {
    case VIR_DOMAIN_RNG_MODEL_VIRTIO:
        modelIsSupported = virQEMUCapsGet(qemuCaps,
                                          QEMU_CAPS_DEVICE_VIRTIO_RNG);
        break;
    case VIR_DOMAIN_RNG_MODEL_LAST:
        break;
    }

    if (!modelIsSupported) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("this qemu doesn't support RNG device type '%s'"),
                       virDomainRNGModelTypeToString(def->model));
        return -1;
    }

4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577
    return 0;
}


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

    return 0;
}


4578 4579 4580 4581 4582 4583 4584 4585 4586
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
4587
                           _("%s model of watchdog can go only on PCI bus"),
4588 4589 4590 4591 4592 4593 4594 4595 4596
                           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
4597
                           _("%s model of watchdog can go only on ISA bus"),
4598 4599 4600 4601 4602 4603 4604 4605
                           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
4606
                           _("%s model of watchdog is virtual and cannot go on any bus."),
4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625
                           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;
}


4626
static int
4627
qemuDomainDeviceDefValidateNetwork(const virDomainNetDef *net)
4628
{
4629 4630
    bool hasIPv4 = false;
    bool hasIPv6 = false;
4631
    size_t i;
4632

4633 4634 4635 4636 4637 4638 4639 4640 4641 4642
    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];
4643

4644
            if (VIR_SOCKET_ADDR_VALID(&net->guestIP.ips[i]->peer)) {
4645
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4646 4647
                               _("Invalid attempt to set peer IP for guest"));
                return -1;
4648
            }
4649

4650 4651
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ip->address, AF_INET)) {
                if (hasIPv4) {
4652
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4653 4654 4655
                                   _("Only one IPv4 address per "
                                     "interface is allowed"));
                    return -1;
4656
                }
4657
                hasIPv4 = true;
4658

4659 4660
                if (ip->prefix > 0 &&
                    (ip->prefix < 4 || ip->prefix > 27)) {
4661
                    virReportError(VIR_ERR_XML_ERROR, "%s",
4662
                                   _("invalid prefix, must be in range of 4-27"));
4663
                    return -1;
4664
                }
4665
            }
4666

4667 4668 4669 4670 4671 4672
            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;
4673
                }
4674
                hasIPv6 = true;
4675

4676 4677 4678 4679 4680
                if (ip->prefix > 120) {
                    virReportError(VIR_ERR_XML_ERROR, "%s",
                                   _("prefix too long"));
                    return -1;
                }
4681
            }
4682
        }
4683 4684 4685 4686 4687 4688 4689
    } 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;
    }
4690

4691
    if (virDomainNetIsVirtioModel(net)) {
4692 4693 4694 4695 4696 4697 4698 4699 4700
        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;
4701
        }
4702
    }
4703

4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722
    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;
}


4723
static int
4724 4725 4726
qemuDomainMdevDefVFIOPCIValidate(const virDomainHostdevSubsysMediatedDev *dev,
                                 const virDomainDef *def,
                                 virQEMUCapsPtr qemuCaps)
4727
{
4728
    if (dev->display == VIR_TRISTATE_SWITCH_ABSENT)
4729 4730 4731 4732 4733 4734 4735 4736 4737
        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;
    }

4738
    if (dev->model != VIR_MDEV_MODEL_TYPE_VFIO_PCI) {
4739 4740 4741 4742 4743 4744 4745
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("<hostdev> attribute 'display' is only supported"
                         " with model='vfio-pci'"));

        return -1;
    }

4746
    if (dev->display == VIR_TRISTATE_SWITCH_ON) {
4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758
        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;
}


4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784
static int
qemuDomainMdevDefVFIOAPValidate(const virDomainDef *def)
{
    size_t i;
    bool vfioap_found = false;

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

        if (virHostdevIsMdevDevice(hostdev) &&
            hostdev->source.subsys.u.mdev.model == VIR_MDEV_MODEL_TYPE_VFIO_AP) {
            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;
}


4785 4786 4787 4788 4789 4790 4791 4792 4793
static int
qemuDomainMdevDefValidate(const virDomainHostdevSubsysMediatedDev *mdevsrc,
                          const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
{
    switch ((virMediatedDeviceModelType) mdevsrc->model) {
    case VIR_MDEV_MODEL_TYPE_VFIO_PCI:
        return qemuDomainMdevDefVFIOPCIValidate(mdevsrc, def, qemuCaps);
    case VIR_MDEV_MODEL_TYPE_VFIO_AP:
4794
        return qemuDomainMdevDefVFIOAPValidate(def);
4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807
    case VIR_MDEV_MODEL_TYPE_VFIO_CCW:
        break;
    case VIR_MDEV_MODEL_TYPE_LAST:
    default:
        virReportEnumRangeError(virMediatedDeviceModelType,
                                mdevsrc->model);
        return -1;
    }

    return 0;
}


4808 4809
static int
qemuDomainDeviceDefValidateHostdev(const virDomainHostdevDef *hostdev,
4810 4811
                                   const virDomainDef *def,
                                   virQEMUCapsPtr qemuCaps)
4812
{
4813 4814
    const virDomainHostdevSubsysMediatedDev *mdevsrc;

4815 4816 4817 4818 4819 4820 4821 4822 4823
    /* 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;
    }

4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841
    if (hostdev->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
        switch ((virDomainHostdevSubsysType) hostdev->source.subsys.type) {
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI:
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI_HOST:
            break;
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
            mdevsrc = &hostdev->source.subsys.u.mdev;
            return qemuDomainMdevDefValidate(mdevsrc, def, qemuCaps);
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
        default:
            virReportEnumRangeError(virDomainHostdevSubsysType,
                                    hostdev->source.subsys.type);
            return -1;
        }
    }

4842 4843 4844 4845
    return 0;
}


4846 4847 4848
static int
qemuDomainDeviceDefValidateVideo(const virDomainVideoDef *video)
{
4849
    switch ((virDomainVideoType) video->type) {
4850 4851
    case VIR_DOMAIN_VIDEO_TYPE_NONE:
        return 0;
4852 4853 4854
    case VIR_DOMAIN_VIDEO_TYPE_XEN:
    case VIR_DOMAIN_VIDEO_TYPE_VBOX:
    case VIR_DOMAIN_VIDEO_TYPE_PARALLELS:
4855
    case VIR_DOMAIN_VIDEO_TYPE_GOP:
4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896
    case VIR_DOMAIN_VIDEO_TYPE_DEFAULT:
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("video type '%s' is not supported with QEMU"),
                       virDomainVideoTypeToString(video->type));
        return -1;
    case VIR_DOMAIN_VIDEO_TYPE_VGA:
    case VIR_DOMAIN_VIDEO_TYPE_CIRRUS:
    case VIR_DOMAIN_VIDEO_TYPE_VMVGA:
    case VIR_DOMAIN_VIDEO_TYPE_QXL:
    case VIR_DOMAIN_VIDEO_TYPE_VIRTIO:
    case VIR_DOMAIN_VIDEO_TYPE_LAST:
        break;
    }

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

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

    if (video->type == VIR_DOMAIN_VIDEO_TYPE_QXL) {
        if (video->vram > (UINT_MAX / 1024)) {
            virReportError(VIR_ERR_OVERFLOW,
                           _("value for 'vram' must be less than '%u'"),
                           UINT_MAX / 1024);
            return -1;
        }
        if (video->ram > (UINT_MAX / 1024)) {
            virReportError(VIR_ERR_OVERFLOW,
                           _("value for 'ram' must be less than '%u'"),
                           UINT_MAX / 1024);
4897 4898
            return -1;
        }
4899 4900 4901 4902 4903 4904 4905
        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;
            }
4906

4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920
            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;
            }
        }
    }

    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)"));
4921 4922 4923 4924 4925 4926 4927 4928
            return -1;
        }
    }

    return 0;
}


4929
int
4930 4931
qemuDomainValidateStorageSource(virStorageSourcePtr src,
                                virQEMUCapsPtr qemuCaps)
4932
{
4933 4934
    int actualType = virStorageSourceGetActualType(src);

4935 4936 4937 4938 4939 4940
    if (src->format == VIR_STORAGE_FILE_COW) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                      _("'cow' storage format is not supported"));
        return -1;
    }

4941 4942 4943 4944 4945 4946 4947
    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;
    }

4948 4949 4950 4951 4952 4953 4954
    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;
    }

4955 4956 4957 4958 4959 4960 4961 4962 4963 4964
    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;
    }

4965 4966 4967 4968 4969
    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",
4970
                       _("LUKS encrypted QCOW2 images are not supported by this QEMU"));
4971 4972 4973
        return -1;
    }

4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996
    if (src->format == VIR_STORAGE_FILE_FAT &&
        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;
        }
    }

4997 4998 4999 5000 5001
    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;
5002 5003
    }

5004 5005 5006 5007 5008 5009 5010 5011 5012
    /* 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;
    }

5013 5014 5015 5016
    return 0;
}


5017
int
5018 5019
qemuDomainDeviceDefValidateDisk(const virDomainDiskDef *disk,
                                virQEMUCapsPtr qemuCaps)
5020
{
5021
    const char *driverName = virDomainDiskGetDriver(disk);
5022 5023
    virStorageSourcePtr n;

5024 5025 5026 5027 5028 5029
    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;
5030 5031
    }

5032 5033 5034 5035 5036 5037 5038
    if (disk->src->readonly && disk->copy_on_read == VIR_TRISTATE_SWITCH_ON) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("copy_on_read is not compatible with read-only disk '%s'"),
                       disk->dst);
        return -1;
    }

5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065
    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;
    }

5066 5067 5068 5069 5070 5071 5072
    if (driverName && STRNEQ(driverName, "qemu")) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("unsupported driver name '%s' for disk '%s'"),
                       driverName, disk->dst);
        return -1;
    }

5073
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
5074
        if (qemuDomainValidateStorageSource(n, qemuCaps) < 0)
5075 5076 5077
            return -1;
    }

5078 5079 5080 5081
    return 0;
}


5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106
static int
qemuDomainDeviceDefValidateControllerAttributes(const virDomainControllerDef *controller)
{
    if (!(controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
          controller->model == VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI)) {
        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;
        }
5107 5108 5109 5110 5111
        if (controller->iothread) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'iothread' is only supported for virtio-scsi controller"));
            return -1;
        }
5112 5113 5114 5115 5116 5117
    }

    return 0;
}


5118 5119 5120 5121 5122 5123
/**
 * @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
5124
 * Returns true if acceptable, false otherwise with error message set.
5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172
 */
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:
        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;
5173 5174 5175 5176 5177 5178
    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;
5179 5180 5181 5182 5183 5184
    }

    return true;
}


5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210
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;
}


5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227
/* 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;

5228 5229
    if (controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI &&
5230 5231
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
       virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
5232 5233
                       _("virtio-scsi IOThreads only available for virtio "
                         "pci and virtio ccw controllers"));
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
       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:
            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:
5266
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT:
5267 5268 5269 5270 5271 5272 5273 5274
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LAST:
            break;
    }

    return 0;
}


5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306
/**
 * 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;
5307 5308
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCIE_PCI_BRIDGE:
        return QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE;
5309 5310 5311 5312 5313
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE:
        return 0;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_LAST:
    default:
        return -1;
5314 5315
    }

5316
    return -1;
5317 5318 5319
}


5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336
#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));


5337 5338 5339 5340 5341 5342 5343 5344 5345
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);
5346
    int cap = virDomainControllerPCIModelNameToQEMUCaps(pciopts->modelName);
5347 5348 5349 5350 5351 5352 5353 5354 5355 5356

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

5357 5358 5359 5360 5361 5362 5363 5364 5365
    /* 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:
5366
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5367 5368 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 5466 5467 5468 5469 5470
        /* 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;

5471 5472 5473 5474 5475 5476 5477
    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;

5478 5479 5480 5481 5482 5483 5484
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

5485 5486 5487 5488 5489 5490 5491 5492 5493
    /* 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:
5494
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528
        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;
    }

5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555
    /* 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:
5556
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569
        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;
    }

5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589
    /* 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:
5590
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604
        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;
    }

5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621
    /* 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:
5622
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635
        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;
    }

5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666
    /* 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:
5667
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680
        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;
    }

5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697
    /* 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:
5698
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711
        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;
    }

5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732
    /* 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:
5733
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749
        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);
    }

5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775
    /* 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;
    }

5776
    return 0;
5777 5778 5779
}


5780 5781 5782 5783 5784
#undef virReportControllerInvalidValue
#undef virReportControllerInvalidOption
#undef virReportControllerMissingOption


5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802
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;
}


5803
static int
5804 5805 5806
qemuDomainDeviceDefValidateController(const virDomainControllerDef *controller,
                                      const virDomainDef *def,
                                      virQEMUCapsPtr qemuCaps)
5807
{
5808 5809
    int ret = 0;

5810 5811 5812 5813
    if (!qemuDomainCheckCCWS390AddressSupport(def, controller->info, qemuCaps,
                                              "controller"))
        return -1;

5814 5815 5816 5817
    if (controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
        !qemuDomainCheckSCSIControllerModel(qemuCaps, controller->model))
        return -1;

5818 5819 5820
    if (qemuDomainDeviceDefValidateControllerAttributes(controller) < 0)
        return -1;

5821
    switch ((virDomainControllerType)controller->type) {
5822
    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
5823 5824 5825
        ret = qemuDomainDeviceDefValidateControllerIDE(controller, def);
        break;

5826
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
5827 5828 5829
        ret = qemuDomainDeviceDefValidateControllerSCSI(controller, def);
        break;

5830
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
5831 5832
        ret = qemuDomainDeviceDefValidateControllerPCI(controller, def,
                                                       qemuCaps);
5833 5834
        break;

5835
    case VIR_DOMAIN_CONTROLLER_TYPE_SATA:
5836 5837 5838 5839 5840
        ret = qemuDomainDeviceDefValidateControllerSATA(controller, def,
                                                        qemuCaps);
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
5841 5842 5843 5844 5845 5846 5847
    case VIR_DOMAIN_CONTROLLER_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_CONTROLLER_TYPE_CCID:
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
    }

5848
    return ret;
5849 5850 5851
}


5852
static int
5853 5854
qemuDomainDeviceDefValidateVsock(const virDomainVsockDef *vsock,
                                 const virDomainDef *def,
5855 5856 5857 5858 5859 5860 5861 5862
                                 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;
    }
5863 5864 5865 5866 5867

    if (!qemuDomainCheckCCWS390AddressSupport(def, vsock->info, qemuCaps,
                                              "vsock"))
        return -1;

5868 5869
    return 0;
}
5870

5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895

static int
qemuDomainDeviceDefValidateTPM(virDomainTPMDef *tpm,
                               const virDomainDef *def ATTRIBUTE_UNUSED)
{
    /* 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;
    }
5896 5897 5898 5899
    return 0;
}


5900 5901 5902 5903 5904 5905 5906 5907 5908
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++) {
5909
        if (def->graphics[i]->type == VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS) {
5910 5911 5912 5913 5914 5915 5916 5917 5918 5919 5920 5921 5922 5923 5924 5925 5926 5927 5928 5929 5930 5931 5932 5933 5934 5935 5936 5937 5938 5939 5940 5941 5942 5943 5944 5945 5946 5947 5948 5949 5950 5951
            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;
}


5952 5953 5954 5955 5956 5957 5958 5959 5960 5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001 6002 6003 6004 6005
static int
qemuDomainDeviceDefValidateInput(const virDomainInputDef *input,
                                 const virDomainDef *def ATTRIBUTE_UNUSED,
                                 virQEMUCapsPtr qemuCaps)
{
    if (input->bus != VIR_DOMAIN_INPUT_BUS_VIRTIO)
        return 0;

    switch ((virDomainInputType)input->type) {
    case VIR_DOMAIN_INPUT_TYPE_MOUSE:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_MOUSE) ||
            (input->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW &&
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_MOUSE_CCW))) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtio-mouse is not supported by this QEMU binary"));
            return -1;
        }
        break;
    case VIR_DOMAIN_INPUT_TYPE_TABLET:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_TABLET) ||
            (input->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW &&
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_TABLET_CCW))) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtio-tablet is not supported by this QEMU binary"));
            return -1;
        }
        break;
    case VIR_DOMAIN_INPUT_TYPE_KBD:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_KEYBOARD) ||
            (input->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW &&
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_KEYBOARD_CCW))) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtio-keyboard is not supported by this QEMU binary"));
            return -1;
        }
        break;
    case VIR_DOMAIN_INPUT_TYPE_PASSTHROUGH:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_INPUT_HOST)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtio-input-host is not supported by this QEMU binary"));
            return -1;
        }
        break;
    case VIR_DOMAIN_INPUT_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainInputType,
                                input->type);
        return -1;
    }

    return 0;
}


6006 6007 6008 6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027 6028 6029 6030 6031 6032
static int
qemuDomainDeviceDefValidateMemballoon(const virDomainMemballoonDef *memballoon,
                                      virQEMUCapsPtr qemuCaps)
{
    if (!memballoon ||
        memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_NONE) {
        return 0;
    }

    if (memballoon->model != VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO) {
        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;
}


6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064
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;

    if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)
        return qemuDomainDeviceDefValidateZPCIAddress(info, qemuCaps);

    return 0;
}


6065 6066 6067
static int
qemuDomainDeviceDefValidate(const virDomainDeviceDef *dev,
                            const virDomainDef *def,
6068
                            void *opaque)
6069
{
6070
    int ret = 0;
6071 6072 6073 6074 6075 6076
    virQEMUDriverPtr driver = opaque;
    virQEMUCapsPtr qemuCaps = NULL;

    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
                                            def->emulator)))
        return -1;
6077

6078 6079 6080
    if ((ret = qemuDomainDeviceDefValidateAddress(dev, qemuCaps)) < 0)
        goto cleanup;

6081
    switch ((virDomainDeviceType)dev->type) {
6082 6083 6084 6085 6086 6087 6088 6089 6090 6091 6092 6093 6094
    case VIR_DOMAIN_DEVICE_NET:
        ret = qemuDomainDeviceDefValidateNetwork(dev->data.net);
        break;

    case VIR_DOMAIN_DEVICE_CHR:
        ret = qemuDomainChrDefValidate(dev->data.chr, def);
        break;

    case VIR_DOMAIN_DEVICE_SMARTCARD:
        ret = qemuDomainSmartcardDefValidate(dev->data.smartcard);
        break;

    case VIR_DOMAIN_DEVICE_RNG:
6095
        ret = qemuDomainRNGDefValidate(dev->data.rng, qemuCaps);
6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106
        break;

    case VIR_DOMAIN_DEVICE_REDIRDEV:
        ret = qemuDomainRedirdevDefValidate(dev->data.redirdev);
        break;

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

    case VIR_DOMAIN_DEVICE_HOSTDEV:
6107 6108
        ret = qemuDomainDeviceDefValidateHostdev(dev->data.hostdev, def,
                                                 qemuCaps);
6109 6110 6111 6112 6113 6114 6115
        break;

    case VIR_DOMAIN_DEVICE_VIDEO:
        ret = qemuDomainDeviceDefValidateVideo(dev->data.video);
        break;

    case VIR_DOMAIN_DEVICE_DISK:
6116
        ret = qemuDomainDeviceDefValidateDisk(dev->data.disk, qemuCaps);
6117 6118
        break;

6119
    case VIR_DOMAIN_DEVICE_CONTROLLER:
6120 6121
        ret = qemuDomainDeviceDefValidateController(dev->data.controller, def,
                                                    qemuCaps);
6122 6123
        break;

6124
    case VIR_DOMAIN_DEVICE_VSOCK:
6125
        ret = qemuDomainDeviceDefValidateVsock(dev->data.vsock, def, qemuCaps);
6126 6127
        break;

6128 6129 6130 6131
    case VIR_DOMAIN_DEVICE_TPM:
        ret = qemuDomainDeviceDefValidateTPM(dev->data.tpm, def);
        break;

6132 6133 6134 6135 6136
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainDeviceDefValidateGraphics(dev->data.graphics, def,
                                                  qemuCaps);
        break;

6137 6138 6139 6140
    case VIR_DOMAIN_DEVICE_INPUT:
        ret = qemuDomainDeviceDefValidateInput(dev->data.input, def, qemuCaps);
        break;

6141 6142 6143 6144
    case VIR_DOMAIN_DEVICE_MEMBALLOON:
        ret = qemuDomainDeviceDefValidateMemballoon(dev->data.memballoon, qemuCaps);
        break;

6145 6146 6147 6148 6149 6150
    case VIR_DOMAIN_DEVICE_LEASE:
    case VIR_DOMAIN_DEVICE_FS:
    case VIR_DOMAIN_DEVICE_SOUND:
    case VIR_DOMAIN_DEVICE_HUB:
    case VIR_DOMAIN_DEVICE_NVRAM:
    case VIR_DOMAIN_DEVICE_SHMEM:
6151
    case VIR_DOMAIN_DEVICE_MEMORY:
6152 6153 6154 6155 6156
    case VIR_DOMAIN_DEVICE_PANIC:
    case VIR_DOMAIN_DEVICE_IOMMU:
    case VIR_DOMAIN_DEVICE_NONE:
    case VIR_DOMAIN_DEVICE_LAST:
        break;
6157 6158
    }

6159
 cleanup:
6160
    virObjectUnref(qemuCaps);
6161 6162 6163 6164
    return ret;
}


6165 6166 6167 6168 6169 6170 6171 6172 6173
/**
 * 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.
 */
6174
static const char *
6175 6176
qemuDomainDefaultNetModel(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
6177
{
6178
    if (ARCH_IS_S390(def->os.arch))
6179 6180
        return "virtio";

S
Stefan Schallenberg 已提交
6181 6182
    if (def->os.arch == VIR_ARCH_ARMV6L ||
        def->os.arch == VIR_ARCH_ARMV7L ||
6183
        def->os.arch == VIR_ARCH_AARCH64) {
6184 6185 6186
        if (STREQ(def->os.machine, "versatilepb"))
            return "smc91c111";

6187
        if (qemuDomainIsARMVirt(def))
6188 6189
            return "virtio";

6190 6191 6192 6193 6194
        /* Incomplete. vexpress (and a few others) use this, but not all
         * arm boards */
        return "lan9118";
    }

6195 6196 6197 6198
    /* virtio is a sensible default for RISC-V virt guests */
    if (qemuDomainIsRISCVVirt(def))
        return "virtio";

6199 6200 6201 6202 6203
    /* In all other cases the model depends on the capabilities. If they were
     * not provided don't report any default. */
    if (!qemuCaps)
        return NULL;

6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215
    /* 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))
        return "rtl8139";
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_E1000))
        return "e1000";
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_NET))
        return "virtio";

    /* We've had no luck detecting support for any network device,
     * but we have to return something: might as well be rtl8139 */
6216 6217
    return "rtl8139";
}
6218

6219 6220

/*
6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236
 * 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.
6237
 */
6238
static int
6239 6240 6241
qemuDomainChrDefDropDefaultPath(virDomainChrDefPtr chr,
                                virQEMUDriverPtr driver)
{
6242 6243 6244 6245
    virQEMUDriverConfigPtr cfg;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *regexp = NULL;
    int ret = -1;
6246

6247 6248 6249 6250 6251
    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;
6252 6253
    }

6254 6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267 6268 6269 6270
    cfg = virQEMUDriverGetConfig(driver);

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

    if (virBufferCheckError(&buf) < 0)
        goto cleanup;

    regexp = virBufferContentAndReset(&buf);

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

    ret = 0;
 cleanup:
    VIR_FREE(regexp);
6271
    virObjectUnref(cfg);
6272
    return ret;
6273 6274 6275
}


6276 6277 6278 6279 6280 6281 6282 6283 6284 6285 6286 6287 6288 6289 6290 6291 6292 6293 6294 6295 6296 6297 6298 6299 6300 6301 6302 6303 6304 6305 6306 6307 6308 6309 6310 6311 6312 6313 6314 6315 6316 6317 6318 6319 6320 6321 6322 6323 6324 6325
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;
}


6326
#define QEMU_USB_XHCI_MAXPORTS 15
6327 6328


6329 6330 6331
static int
qemuDomainControllerDefPostParse(virDomainControllerDefPtr cont,
                                 const virDomainDef *def,
6332 6333
                                 virQEMUCapsPtr qemuCaps,
                                 unsigned int parseFlags)
6334
{
6335 6336
    switch ((virDomainControllerType)cont->type) {
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
6337 6338 6339
        /* Set the default SCSI controller model if not already set */
        if (qemuDomainSetSCSIControllerModel(def, cont, qemuCaps) < 0)
            return -1;
6340
        break;
6341

6342
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
6343
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT && qemuCaps) {
6344
            /* Pick a suitable default model for the USB controller if none
6345 6346
             * has been selected by the user and we have the qemuCaps for
             * figuring out which contollers are supported.
6347 6348 6349 6350 6351 6352 6353
             *
             * 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() */
6354 6355 6356 6357 6358 6359 6360 6361 6362 6363 6364

            /* 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;
                }
6365
            } else if (ARCH_IS_PPC64(def->os.arch)) {
6366 6367
                /* To not break migration we need to set default USB controller
                 * for ppc64 to pci-ohci if we cannot change ABI of the VM.
6368 6369
                 * The nec-usb-xhci or qemu-xhci controller is used as default
                 * only for newly defined domains or devices. */
6370
                if ((parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE) &&
6371 6372 6373
                    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) &&
6374 6375 6376
                    virQEMUCapsGet(qemuCaps, QEMU_CAPS_NEC_USB_XHCI)) {
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
                } else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_OHCI)) {
6377
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_PCI_OHCI;
6378 6379 6380
                } else {
                    /* Explicitly fallback to legacy USB controller for PPC64. */
                    cont->model = -1;
6381
                }
6382
            } else if (def->os.arch == VIR_ARCH_AARCH64) {
6383 6384 6385
                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))
6386
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397
            }
        }
        /* 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;
        }
6398 6399 6400
        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) {
6401
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
6402 6403 6404
                           _("'%s' controller only supports up to '%u' ports"),
                           virDomainControllerModelUSBTypeToString(cont->model),
                           QEMU_USB_XHCI_MAXPORTS);
6405 6406 6407
            return -1;
        }
        break;
6408

6409
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
6410 6411 6412 6413 6414 6415 6416

        /* 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) {
6417 6418 6419 6420 6421 6422
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("pci-root and pcie-root controllers "
                             "should have index 0"));
            return -1;
        }

6423
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS &&
6424
            !qemuDomainIsI440FX(def)) {
6425 6426 6427 6428 6429 6430
            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 &&
6431
            !qemuDomainIsQ35(def)) {
6432 6433 6434 6435 6436 6437
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("pcie-expander-bus controllers are only supported "
                             "on q35-based machinetypes"));
            return -1;
        }

6438 6439 6440
        /* 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
6441 6442
         * from 0 .. size-1.
         */
6443 6444
        if (cont->opts.pciopts.numaNode >= 0 &&
            cont->opts.pciopts.numaNode >=
6445
            (int)virDomainNumaGetNodeCount(def->numa)) {
6446 6447 6448 6449 6450 6451 6452 6453 6454 6455
            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;
        }
6456 6457 6458 6459 6460 6461 6462 6463 6464
        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:
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
6465 6466 6467 6468 6469
    }

    return 0;
}

6470 6471 6472 6473 6474 6475
static int
qemuDomainChrDefPostParse(virDomainChrDefPtr chr,
                          const virDomainDef *def,
                          virQEMUDriverPtr driver,
                          unsigned int parseFlags)
{
6476 6477 6478 6479 6480 6481 6482 6483 6484 6485 6486 6487 6488 6489
    /* 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)) {
6490
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
6491
        } else if (qemuDomainIsARMVirt(def) || qemuDomainIsRISCVVirt(def)) {
6492
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
6493 6494
        } else if (ARCH_IS_S390(def->os.arch)) {
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
6495 6496 6497
        }
    }

6498 6499 6500
    /* Set the default target model */
    if (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL &&
        chr->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_NONE) {
6501
        switch ((virDomainChrSerialTargetType)chr->targetType) {
6502 6503 6504 6505 6506 6507 6508 6509 6510
        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;
6511 6512 6513
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY;
            break;
6514
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
6515 6516 6517 6518 6519
            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;
            }
6520
            break;
6521 6522 6523
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE;
            break;
6524 6525 6526 6527 6528 6529 6530
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            /* Nothing to do */
            break;
        }
    }

6531 6532 6533 6534 6535 6536 6537 6538 6539 6540 6541 6542 6543 6544 6545 6546
    /* 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;
}
6547

6548

6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562 6563 6564 6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608 6609 6610 6611 6612 6613 6614 6615 6616 6617 6618 6619 6620 6621 6622 6623 6624 6625 6626 6627 6628 6629
/**
 * 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;
}


6630 6631
static int
qemuDomainDeviceDiskDefPostParse(virDomainDiskDefPtr disk,
6632
                                 virQEMUCapsPtr qemuCaps,
6633
                                 unsigned int parseFlags)
6634 6635
{
    /* set default disk types and drivers */
6636 6637 6638
    if (!virDomainDiskGetDriver(disk) &&
        virDomainDiskSetDriver(disk, "qemu") < 0)
        return -1;
6639

6640 6641 6642 6643 6644
    /* default disk format for drives */
    if (virDomainDiskGetFormat(disk) == VIR_STORAGE_FILE_NONE &&
        (virDomainDiskGetType(disk) == VIR_STORAGE_TYPE_FILE ||
         virDomainDiskGetType(disk) == VIR_STORAGE_TYPE_BLOCK))
        virDomainDiskSetFormat(disk, VIR_STORAGE_FILE_RAW);
6645

6646 6647 6648 6649
    /* default disk format for mirrored drive */
    if (disk->mirror &&
        disk->mirror->format == VIR_STORAGE_FILE_NONE)
        disk->mirror->format = VIR_STORAGE_FILE_RAW;
6650

6651 6652 6653 6654
    if (qemuDomainDeviceDiskDefPostParseRestoreSecAlias(disk, qemuCaps,
                                                        parseFlags) < 0)
        return -1;

6655 6656 6657 6658 6659 6660 6661
    /* 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;

6662 6663 6664 6665
    return 0;
}


6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676 6677 6678 6679 6680 6681
static int
qemuDomainDeviceNetDefPostParse(virDomainNetDefPtr net,
                                const virDomainDef *def,
                                virQEMUCapsPtr qemuCaps)
{
    if (net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
        !net->model) {
        if (VIR_STRDUP(net->model,
                       qemuDomainDefaultNetModel(def, qemuCaps)) < 0)
            return -1;
    }

    return 0;
}


6682 6683 6684 6685 6686 6687 6688
static int
qemuDomainDeviceVideoDefPostParse(virDomainVideoDefPtr video,
                                  const virDomainDef *def)
{
    if (video->type == VIR_DOMAIN_VIDEO_TYPE_DEFAULT) {
        if (ARCH_IS_PPC64(def->os.arch))
            video->type = VIR_DOMAIN_VIDEO_TYPE_VGA;
6689 6690 6691
        else if (qemuDomainIsARMVirt(def) ||
                 qemuDomainIsRISCVVirt(def) ||
                 ARCH_IS_S390(def->os.arch))
6692 6693 6694 6695 6696 6697 6698 6699 6700 6701 6702 6703 6704 6705
            video->type = VIR_DOMAIN_VIDEO_TYPE_VIRTIO;
        else
            video->type = VIR_DOMAIN_VIDEO_TYPE_CIRRUS;
    }

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

    return 0;
}


6706 6707 6708 6709 6710 6711 6712 6713 6714 6715 6716 6717 6718 6719 6720 6721 6722
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
6723 6724 6725 6726 6727 6728 6729 6730 6731 6732
static int
qemuDomainVsockDefPostParse(virDomainVsockDefPtr vsock)
{
    if (vsock->model == VIR_DOMAIN_VSOCK_MODEL_DEFAULT)
        vsock->model = VIR_DOMAIN_VSOCK_MODEL_VIRTIO;

    return 0;
}


6733 6734 6735 6736 6737 6738 6739
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) &&
6740
        mdevsrc->model == VIR_MDEV_MODEL_TYPE_VFIO_PCI &&
6741 6742 6743 6744 6745 6746 6747 6748 6749 6750 6751 6752 6753 6754 6755 6756 6757 6758 6759 6760 6761 6762
        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;
}


6763 6764
static int
qemuDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
6765
                             const virDomainDef *def,
6766
                             virCapsPtr caps ATTRIBUTE_UNUSED,
6767
                             unsigned int parseFlags,
6768
                             void *opaque,
6769
                             void *parseOpaque)
6770
{
6771
    virQEMUDriverPtr driver = opaque;
6772 6773 6774
    /* 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. */
6775
    virQEMUCapsPtr qemuCaps = parseOpaque;
6776
    int ret = -1;
6777

6778 6779 6780 6781
    switch ((virDomainDeviceType) dev->type) {
    case VIR_DOMAIN_DEVICE_NET:
        ret = qemuDomainDeviceNetDefPostParse(dev->data.net, def, qemuCaps);
        break;
6782

6783
    case VIR_DOMAIN_DEVICE_DISK:
6784
        ret = qemuDomainDeviceDiskDefPostParse(dev->data.disk, qemuCaps,
6785
                                               parseFlags);
6786
        break;
6787

6788 6789 6790
    case VIR_DOMAIN_DEVICE_VIDEO:
        ret = qemuDomainDeviceVideoDefPostParse(dev->data.video, def);
        break;
6791

6792 6793 6794
    case VIR_DOMAIN_DEVICE_PANIC:
        ret = qemuDomainDevicePanicDefPostParse(dev->data.panic, def);
        break;
6795

6796 6797 6798 6799
    case VIR_DOMAIN_DEVICE_CONTROLLER:
        ret = qemuDomainControllerDefPostParse(dev->data.controller, def,
                                               qemuCaps, parseFlags);
        break;
6800

6801 6802 6803
    case VIR_DOMAIN_DEVICE_SHMEM:
        ret = qemuDomainShmemDefPostParse(dev->data.shmem);
        break;
6804

6805 6806 6807
    case VIR_DOMAIN_DEVICE_CHR:
        ret = qemuDomainChrDefPostParse(dev->data.chr, def, driver, parseFlags);
        break;
6808

J
Ján Tomko 已提交
6809 6810 6811 6812
    case VIR_DOMAIN_DEVICE_VSOCK:
        ret = qemuDomainVsockDefPostParse(dev->data.vsock);
        break;

6813 6814 6815 6816
    case VIR_DOMAIN_DEVICE_HOSTDEV:
        ret = qemuDomainHostdevDefPostParse(dev->data.hostdev, qemuCaps);
        break;

6817 6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836 6837 6838 6839 6840 6841 6842 6843 6844
    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;
    }
6845 6846

    return ret;
6847 6848 6849
}


6850 6851
static int
qemuDomainDefAssignAddresses(virDomainDef *def,
6852
                             virCapsPtr caps ATTRIBUTE_UNUSED,
6853
                             unsigned int parseFlags ATTRIBUTE_UNUSED,
6854
                             void *opaque,
6855
                             void *parseOpaque)
6856 6857
{
    virQEMUDriverPtr driver = opaque;
6858 6859 6860
    /* 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. */
6861
    virQEMUCapsPtr qemuCaps = parseOpaque;
6862
    bool newDomain = parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE;
6863

6864 6865 6866 6867 6868 6869 6870
    /* 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;

6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882 6883 6884
    return qemuDomainAssignAddresses(def, qemuCaps, driver, NULL, newDomain);
}


static int
qemuDomainPostParseDataAlloc(const virDomainDef *def,
                             virCapsPtr caps ATTRIBUTE_UNUSED,
                             unsigned int parseFlags ATTRIBUTE_UNUSED,
                             void *opaque,
                             void **parseOpaque)
{
    virQEMUDriverPtr driver = opaque;

    if (!(*parseOpaque = virQEMUCapsCacheLookup(driver->qemuCapsCache,
6885
                                                def->emulator)))
6886
        return 1;
6887

6888 6889 6890 6891 6892 6893 6894 6895
    return 0;
}


static void
qemuDomainPostParseDataFree(void *parseOpaque)
{
    virQEMUCapsPtr qemuCaps = parseOpaque;
6896 6897 6898 6899 6900

    virObjectUnref(qemuCaps);
}


6901
virDomainDefParserConfig virQEMUDriverDomainDefParserConfig = {
6902
    .domainPostParseBasicCallback = qemuDomainDefPostParseBasic,
6903 6904
    .domainPostParseDataAlloc = qemuDomainPostParseDataAlloc,
    .domainPostParseDataFree = qemuDomainPostParseDataFree,
6905
    .devicesPostParseCallback = qemuDomainDeviceDefPostParse,
6906
    .domainPostParseCallback = qemuDomainDefPostParse,
6907
    .assignAddressesCallback = qemuDomainDefAssignAddresses,
6908
    .domainValidateCallback = qemuDomainDefValidate,
6909 6910
    .deviceValidateCallback = qemuDomainDeviceDefValidate,

6911
    .features = VIR_DOMAIN_DEF_FEATURE_MEMORY_HOTPLUG |
6912
                VIR_DOMAIN_DEF_FEATURE_OFFLINE_VCPUPIN |
6913 6914
                VIR_DOMAIN_DEF_FEATURE_INDIVIDUAL_VCPUS |
                VIR_DOMAIN_DEF_FEATURE_USER_ALIAS,
6915 6916 6917
};


6918
static void
6919
qemuDomainObjSaveJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
6920
{
6921 6922 6923
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    if (virDomainObjIsActive(obj)) {
6924
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj, driver->caps) < 0)
6925
            VIR_WARN("Failed to save status on vm %s", obj->def->name);
6926
    }
6927

6928
    virObjectUnref(cfg);
6929 6930
}

J
Jiri Denemark 已提交
6931
void
6932
qemuDomainObjSetJobPhase(virQEMUDriverPtr driver,
J
Jiri Denemark 已提交
6933 6934 6935 6936
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
6937
    unsigned long long me = virThreadSelfID();
J
Jiri Denemark 已提交
6938 6939 6940 6941

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

6942 6943 6944 6945 6946
    VIR_DEBUG("Setting '%s' phase to '%s'",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob, phase));

    if (priv->job.asyncOwner && me != priv->job.asyncOwner) {
6947
        VIR_WARN("'%s' async job is owned by thread %llu",
6948 6949 6950 6951
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }

J
Jiri Denemark 已提交
6952
    priv->job.phase = phase;
6953
    priv->job.asyncOwner = me;
J
Jiri Denemark 已提交
6954 6955 6956
    qemuDomainObjSaveJob(driver, obj);
}

6957
void
6958 6959
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
6960 6961 6962
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

6963 6964 6965 6966 6967 6968 6969
    if (!priv->job.asyncJob)
        return;

    priv->job.mask = allowedJobs | JOB_MASK(QEMU_JOB_DESTROY);
}

void
6970
qemuDomainObjDiscardAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
6971 6972 6973 6974 6975 6976
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
6977
    qemuDomainObjSaveJob(driver, obj);
6978 6979
}

6980 6981 6982 6983 6984 6985 6986 6987 6988
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()) {
6989
        VIR_WARN("'%s' async job is owned by thread %llu",
6990 6991 6992 6993 6994 6995
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }
    priv->job.asyncOwner = 0;
}

6996
static bool
6997
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
6998
{
6999 7000 7001
    return !priv->job.asyncJob ||
           job == QEMU_JOB_NONE ||
           (priv->job.mask & JOB_MASK(job)) != 0;
7002 7003
}

7004
bool
7005
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
7006 7007 7008 7009
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020
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));
}

7021 7022 7023
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

7024 7025 7026 7027 7028 7029
/**
 * qemuDomainObjBeginJobInternal:
 * @driver: qemu driver
 * @obj: domain object
 * @job: qemuDomainJob to start
 * @asyncJob: qemuDomainAsyncJob to start
7030
 * @nowait: don't wait trying to acquire @job
7031 7032 7033 7034
 *
 * 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
7035 7036 7037 7038 7039
 * 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.
7040 7041 7042 7043 7044
 *
 * Returns: 0 on success,
 *         -2 if unable to start job because of timeout or
 *            maxQueuedJobs limit,
 *         -1 otherwise.
7045
 */
7046
static int ATTRIBUTE_NONNULL(1)
7047
qemuDomainObjBeginJobInternal(virQEMUDriverPtr driver,
7048
                              virDomainObjPtr obj,
7049
                              qemuDomainJob job,
7050
                              qemuDomainAgentJob agentJob,
7051 7052
                              qemuDomainAsyncJob asyncJob,
                              bool nowait)
7053 7054
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
7055
    unsigned long long now;
7056
    unsigned long long then;
7057
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
J
Jiri Denemark 已提交
7058
    bool async = job == QEMU_JOB_ASYNC;
7059
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
7060
    const char *blocker = NULL;
7061
    const char *agentBlocker = NULL;
J
Jiri Denemark 已提交
7062
    int ret = -1;
J
Jiri Denemark 已提交
7063
    unsigned long long duration = 0;
7064
    unsigned long long agentDuration = 0;
J
Jiri Denemark 已提交
7065
    unsigned long long asyncDuration = 0;
7066

7067 7068 7069 7070 7071 7072
    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
7073
              qemuDomainJobTypeToString(priv->job.active),
7074
              qemuDomainAgentJobTypeToString(priv->job.agentActive),
J
Jiri Denemark 已提交
7075
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
7076

7077 7078
    if (virTimeMillisNow(&now) < 0) {
        virObjectUnref(cfg);
7079
        return -1;
7080 7081
    }

7082
    priv->jobs_queued++;
J
Jiri Denemark 已提交
7083
    then = now + QEMU_JOB_WAIT_TIME;
7084

7085
 retry:
7086 7087
    if ((!async && job != QEMU_JOB_DESTROY) &&
        cfg->maxQueuedJobs &&
7088
        priv->jobs_queued > cfg->maxQueuedJobs) {
7089 7090 7091
        goto error;
    }

7092
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
7093 7094 7095
        if (nowait)
            goto cleanup;

7096
        VIR_DEBUG("Waiting for async job (vm=%p name=%s)", obj, obj->def->name);
7097
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->parent.lock, then) < 0)
7098 7099 7100
            goto error;
    }

7101
    while (!qemuDomainObjCanSetJob(priv, job, agentJob)) {
7102 7103 7104
        if (nowait)
            goto cleanup;

7105
        VIR_DEBUG("Waiting for job (vm=%p name=%s)", obj, obj->def->name);
7106
        if (virCondWaitUntil(&priv->job.cond, &obj->parent.lock, then) < 0)
7107
            goto error;
7108
    }
7109 7110 7111

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
7112
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
7113 7114
        goto retry;

J
Jiri Denemark 已提交
7115 7116
    ignore_value(virTimeMillisNow(&now));

7117 7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131 7132 7133 7134 7135 7136 7137 7138 7139 7140 7141 7142 7143 7144 7145 7146 7147 7148 7149 7150 7151 7152 7153 7154 7155 7156
    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;
7157
    }
7158

7159 7160
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
7161

7162
    virObjectUnref(cfg);
7163
    return 0;
7164

7165
 error:
J
Jiri Denemark 已提交
7166 7167 7168
    ignore_value(virTimeMillisNow(&now));
    if (priv->job.active && priv->job.started)
        duration = now - priv->job.started;
7169 7170
    if (priv->job.agentActive && priv->job.agentStarted)
        agentDuration = now - priv->job.agentStarted;
J
Jiri Denemark 已提交
7171 7172 7173
    if (priv->job.asyncJob && priv->job.asyncStarted)
        asyncDuration = now - priv->job.asyncStarted;

7174 7175 7176 7177
    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)",
7178
             qemuDomainJobTypeToString(job),
7179
             qemuDomainAgentJobTypeToString(agentJob),
7180 7181 7182
             qemuDomainAsyncJobTypeToString(asyncJob),
             obj->def->name,
             qemuDomainJobTypeToString(priv->job.active),
7183
             qemuDomainAgentJobTypeToString(priv->job.agentActive),
7184
             qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
7185
             priv->job.owner, NULLSTR(priv->job.ownerAPI),
7186
             priv->job.agentOwner, NULLSTR(priv->job.agentOwnerAPI),
J
Jiri Denemark 已提交
7187
             priv->job.asyncOwner, NULLSTR(priv->job.asyncOwnerAPI),
7188
             priv->job.apiFlags,
7189
             duration / 1000, agentDuration / 1000, asyncDuration / 1000);
7190

7191 7192 7193 7194 7195 7196 7197 7198 7199
    if (job) {
        if (nested || qemuDomainNestedJobAllowed(priv, job))
            blocker = priv->job.ownerAPI;
        else
            blocker = priv->job.asyncOwnerAPI;
    }

    if (agentJob)
        agentBlocker = priv->job.agentOwnerAPI;
7200

7201
    if (errno == ETIMEDOUT) {
7202 7203 7204 7205 7206 7207
        if (blocker && agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by monitor=%s agent=%s)"),
                           blocker, agentBlocker);
        } else if (blocker) {
7208
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
7209 7210
                           _("cannot acquire state change "
                             "lock (held by monitor=%s)"),
7211
                           blocker);
7212 7213 7214 7215 7216
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by agent=%s)"),
                           agentBlocker);
7217 7218 7219 7220
        } else {
            virReportError(VIR_ERR_OPERATION_TIMEOUT, "%s",
                           _("cannot acquire state change lock"));
        }
7221 7222 7223
        ret = -2;
    } else if (cfg->maxQueuedJobs &&
               priv->jobs_queued > cfg->maxQueuedJobs) {
7224 7225 7226 7227 7228 7229 7230
        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) {
7231
            virReportError(VIR_ERR_OPERATION_FAILED,
7232 7233
                           _("cannot acquire state change "
                             "lock (held by monitor=%s) "
7234 7235
                             "due to max_queued limit"),
                           blocker);
7236 7237 7238 7239 7240 7241
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_FAILED,
                           _("cannot acquire state change "
                             "lock (held by agent=%s) "
                             "due to max_queued limit"),
                           agentBlocker);
7242 7243 7244 7245 7246
        } else {
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("cannot acquire state change lock "
                             "due to max_queued limit"));
        }
7247 7248
        ret = -2;
    } else {
7249
        virReportSystemError(errno, "%s", _("cannot acquire job mutex"));
7250
    }
J
Jiri Denemark 已提交
7251 7252

 cleanup:
7253
    priv->jobs_queued--;
7254
    virObjectUnref(cfg);
7255
    return ret;
7256 7257 7258
}

/*
7259
 * obj must be locked before calling
7260 7261 7262 7263
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
7264
 * Successful calls must be followed by EndJob eventually
7265
 */
7266
int qemuDomainObjBeginJob(virQEMUDriverPtr driver,
7267
                          virDomainObjPtr obj,
7268
                          qemuDomainJob job)
7269
{
7270
    if (qemuDomainObjBeginJobInternal(driver, obj, job,
7271
                                      QEMU_AGENT_JOB_NONE,
7272
                                      QEMU_ASYNC_JOB_NONE, false) < 0)
7273 7274 7275
        return -1;
    else
        return 0;
7276 7277
}

7278 7279 7280 7281 7282 7283 7284 7285 7286 7287 7288 7289 7290 7291 7292 7293 7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311 7312 7313 7314
/**
 * 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);
}

7315
int qemuDomainObjBeginAsyncJob(virQEMUDriverPtr driver,
7316
                               virDomainObjPtr obj,
7317
                               qemuDomainAsyncJob asyncJob,
7318 7319
                               virDomainJobOperation operation,
                               unsigned long apiFlags)
7320
{
7321 7322
    qemuDomainObjPrivatePtr priv;

7323
    if (qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_ASYNC,
7324
                                      QEMU_AGENT_JOB_NONE,
7325
                                      asyncJob, false) < 0)
7326
        return -1;
7327 7328 7329

    priv = obj->privateData;
    priv->job.current->operation = operation;
7330
    priv->job.apiFlags = apiFlags;
7331
    return 0;
7332 7333
}

7334
int
7335 7336
qemuDomainObjBeginNestedJob(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
7337
                            qemuDomainAsyncJob asyncJob)
7338 7339 7340 7341 7342
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (asyncJob != priv->job.asyncJob) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
7343 7344
                       _("unexpected async job %d type expected %d"),
                       asyncJob, priv->job.asyncJob);
7345 7346 7347 7348
        return -1;
    }

    if (priv->job.asyncOwner != virThreadSelfID()) {
7349
        VIR_WARN("This thread doesn't seem to be the async job owner: %llu",
7350 7351 7352 7353 7354
                 priv->job.asyncOwner);
    }

    return qemuDomainObjBeginJobInternal(driver, obj,
                                         QEMU_JOB_ASYNC_NESTED,
7355
                                         QEMU_AGENT_JOB_NONE,
7356 7357
                                         QEMU_ASYNC_JOB_NONE,
                                         false);
7358 7359
}

7360 7361 7362 7363 7364 7365 7366 7367 7368 7369 7370 7371 7372 7373 7374 7375 7376 7377 7378
/**
 * 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,
7379
                                         QEMU_AGENT_JOB_NONE,
7380 7381
                                         QEMU_ASYNC_JOB_NONE, true);
}
7382

7383
/*
7384
 * obj must be locked and have a reference before calling
7385 7386 7387 7388
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 */
7389 7390
void
qemuDomainObjEndJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
7391 7392
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
7393
    qemuDomainJob job = priv->job.active;
7394

7395 7396
    priv->jobs_queued--;

7397
    VIR_DEBUG("Stopping job: %s (async=%s vm=%p name=%s)",
7398
              qemuDomainJobTypeToString(job),
7399 7400
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
7401

7402
    qemuDomainObjResetJob(priv);
7403 7404
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
7405 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 7432 7433 7434 7435 7436 7437 7438 7439 7440 7441 7442 7443 7444 7445 7446 7447 7448 7449 7450 7451
    /* 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))
        qemuDomainObjSaveJob(driver, obj);
    /* We indeed need to wake up ALL threads waiting because
     * grabbing a job requires checking more variables. */
    virCondBroadcast(&priv->job.cond);
7452 7453
}

7454
void
7455
qemuDomainObjEndAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
7456 7457
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
7458

7459 7460
    priv->jobs_queued--;

7461 7462 7463
    VIR_DEBUG("Stopping async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
7464

7465
    qemuDomainObjResetAsyncJob(priv);
7466
    qemuDomainObjSaveJob(driver, obj);
7467 7468 7469
    virCondBroadcast(&priv->job.asyncCond);
}

7470 7471 7472 7473 7474
void
qemuDomainObjAbortAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

7475 7476 7477
    VIR_DEBUG("Requesting abort of async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
7478

7479 7480
    priv->job.abortJob = true;
    virDomainObjBroadcast(obj);
7481 7482
}

7483 7484 7485 7486
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU monitor API call
7487 7488 7489
 * Must have already either called qemuDomainObjBeginJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active; may not be used for nested async jobs.
7490 7491 7492
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
7493
static int
7494
qemuDomainObjEnterMonitorInternal(virQEMUDriverPtr driver,
7495
                                  virDomainObjPtr obj,
7496
                                  qemuDomainAsyncJob asyncJob)
7497 7498 7499
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

7500
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
7501 7502 7503
        int ret;
        if ((ret = qemuDomainObjBeginNestedJob(driver, obj, asyncJob)) < 0)
            return ret;
7504
        if (!virDomainObjIsActive(obj)) {
7505 7506
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
7507
            qemuDomainObjEndJob(driver, obj);
7508 7509
            return -1;
        }
7510 7511 7512
    } 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");
7513 7514
    }

7515 7516
    VIR_DEBUG("Entering monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
7517
    virObjectLock(priv->mon);
7518
    virObjectRef(priv->mon);
7519
    ignore_value(virTimeMillisNow(&priv->monStart));
7520
    virObjectUnlock(obj);
7521 7522

    return 0;
7523 7524
}

7525
static void ATTRIBUTE_NONNULL(1)
7526
qemuDomainObjExitMonitorInternal(virQEMUDriverPtr driver,
7527
                                 virDomainObjPtr obj)
7528 7529
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
7530
    bool hasRefs;
7531

7532
    hasRefs = virObjectUnref(priv->mon);
7533

7534
    if (hasRefs)
7535
        virObjectUnlock(priv->mon);
7536

7537
    virObjectLock(obj);
7538 7539
    VIR_DEBUG("Exited monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
7540

7541
    priv->monStart = 0;
7542
    if (!hasRefs)
7543
        priv->mon = NULL;
7544

J
Jiri Denemark 已提交
7545 7546
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjEndJob(driver, obj);
7547 7548
}

7549
void qemuDomainObjEnterMonitor(virQEMUDriverPtr driver,
7550
                               virDomainObjPtr obj)
7551
{
7552
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, obj,
7553
                                                   QEMU_ASYNC_JOB_NONE));
7554 7555
}

7556
/* obj must NOT be locked before calling
7557 7558
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
7559 7560 7561 7562 7563 7564
 *
 * 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.
7565
 */
7566 7567
int qemuDomainObjExitMonitor(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
7568
{
7569
    qemuDomainObjExitMonitorInternal(driver, obj);
7570
    if (!virDomainObjIsActive(obj)) {
7571
        if (virGetLastErrorCode() == VIR_ERR_OK)
7572 7573
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
7574 7575 7576
        return -1;
    }
    return 0;
7577
}
7578 7579

/*
7580
 * obj must be locked before calling
7581 7582
 *
 * To be called immediately before any QEMU monitor API call.
7583
 * Must have already either called qemuDomainObjBeginJob()
7584 7585 7586 7587 7588
 * 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
7589 7590 7591
 * 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).
7592 7593
 */
int
7594
qemuDomainObjEnterMonitorAsync(virQEMUDriverPtr driver,
7595
                               virDomainObjPtr obj,
7596
                               qemuDomainAsyncJob asyncJob)
7597
{
7598
    return qemuDomainObjEnterMonitorInternal(driver, obj, asyncJob);
7599 7600
}

D
Daniel P. Berrange 已提交
7601

7602 7603 7604 7605
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU agent API call.
7606 7607 7608
 * Must have already called qemuDomainObjBeginAgentJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active.
7609 7610 7611
 *
 * To be followed with qemuDomainObjExitAgent() once complete
 */
7612
qemuAgentPtr
7613
qemuDomainObjEnterAgent(virDomainObjPtr obj)
D
Daniel P. Berrange 已提交
7614 7615
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
7616
    qemuAgentPtr agent = priv->agent;
D
Daniel P. Berrange 已提交
7617

7618 7619
    VIR_DEBUG("Entering agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
7620 7621 7622

    virObjectLock(agent);
    virObjectRef(agent);
7623
    virObjectUnlock(obj);
7624 7625

    return agent;
D
Daniel P. Berrange 已提交
7626 7627
}

7628 7629 7630 7631 7632 7633

/* obj must NOT be locked before calling
 *
 * Should be paired with an earlier qemuDomainObjEnterAgent() call
 */
void
7634
qemuDomainObjExitAgent(virDomainObjPtr obj, qemuAgentPtr agent)
D
Daniel P. Berrange 已提交
7635
{
7636 7637
    virObjectUnlock(agent);
    virObjectUnref(agent);
7638
    virObjectLock(obj);
D
Daniel P. Berrange 已提交
7639

7640 7641
    VIR_DEBUG("Exited agent (agent=%p vm=%p name=%s)",
              agent, obj, obj->def->name);
D
Daniel P. Berrange 已提交
7642 7643
}

7644
void qemuDomainObjEnterRemote(virDomainObjPtr obj)
7645
{
7646 7647
    VIR_DEBUG("Entering remote (vm=%p name=%s)",
              obj, obj->def->name);
7648
    virObjectUnlock(obj);
7649 7650
}

7651 7652 7653 7654

int
qemuDomainObjExitRemote(virDomainObjPtr obj,
                        bool checkActive)
7655
{
7656
    virObjectLock(obj);
7657 7658
    VIR_DEBUG("Exited remote (vm=%p name=%s)",
              obj, obj->def->name);
7659 7660 7661 7662 7663 7664 7665 7666 7667

    if (checkActive && !virDomainObjIsActive(obj)) {
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("domain '%s' is not running"),
                       obj->def->name);
        return -1;
    }

    return 0;
7668
}
7669 7670


7671 7672 7673 7674 7675 7676 7677 7678 7679 7680 7681 7682 7683 7684 7685 7686 7687 7688 7689
static virDomainDefPtr
qemuDomainDefFromXML(virQEMUDriverPtr driver,
                     const char *xml)
{
    virCapsPtr caps;
    virDomainDefPtr def;

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        return NULL;

    def = virDomainDefParseString(xml, caps, driver->xmlopt, NULL,
                                  VIR_DOMAIN_DEF_PARSE_INACTIVE |
                                  VIR_DOMAIN_DEF_PARSE_SKIP_VALIDATE);

    virObjectUnref(caps);
    return def;
}


7690 7691 7692 7693 7694 7695
virDomainDefPtr
qemuDomainDefCopy(virQEMUDriverPtr driver,
                  virDomainDefPtr src,
                  unsigned int flags)
{
    virDomainDefPtr ret = NULL;
7696
    char *xml;
7697

7698
    if (!(xml = qemuDomainDefFormatXML(driver, src, flags)))
7699
        return NULL;
7700

7701
    ret = qemuDomainDefFromXML(driver, xml);
7702 7703 7704 7705 7706

    VIR_FREE(xml);
    return ret;
}

7707 7708 7709 7710 7711 7712 7713

static int
qemuDomainDefFormatBufInternal(virQEMUDriverPtr driver,
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags,
                               virBuffer *buf)
7714
{
7715
    int ret = -1;
7716
    virDomainDefPtr copy = NULL;
7717
    virCapsPtr caps = NULL;
7718
    virQEMUCapsPtr qemuCaps = NULL;
7719 7720 7721

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;
7722

7723 7724 7725
    if (!(flags & (VIR_DOMAIN_XML_UPDATE_CPU | VIR_DOMAIN_XML_MIGRATABLE)))
        goto format;

7726
    if (!(copy = virDomainDefCopy(def, caps, driver->xmlopt, NULL,
7727 7728 7729 7730 7731
                                  flags & VIR_DOMAIN_XML_MIGRATABLE)))
        goto cleanup;

    def = copy;

7732
    /* Update guest CPU requirements according to host CPU */
7733
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) &&
7734
        def->cpu &&
7735 7736
        (def->cpu->mode != VIR_CPU_MODE_CUSTOM ||
         def->cpu->model)) {
7737 7738 7739 7740 7741 7742 7743 7744
        if (!(qemuCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
                                                    def->emulator,
                                                    def->os.machine)))
            goto cleanup;

        if (virCPUUpdate(def->os.arch, def->cpu,
                         virQEMUCapsGetHostModel(qemuCaps, def->virtType,
                                                 VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE)) < 0)
7745 7746 7747
            goto cleanup;
    }

7748
    if ((flags & VIR_DOMAIN_XML_MIGRATABLE)) {
7749
        size_t i;
7750
        int toremove = 0;
7751
        virDomainControllerDefPtr usb = NULL, pci = NULL;
7752 7753 7754 7755 7756 7757 7758 7759 7760 7761 7762 7763 7764 7765 7766

        /* 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];
            }
        }
7767 7768 7769 7770 7771 7772 7773 7774 7775

        /* 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'/>.
7776
         */
7777
        if (ARCH_IS_X86(def->os.arch) && qemuDomainIsI440FX(def) &&
7778
            usb && usb->idx == 0 &&
7779
            (usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT ||
7780 7781
             usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_PIIX3_UHCI) &&
            !virDomainDeviceAliasIsUserAlias(usb->info.alias)) {
7782 7783
            VIR_DEBUG("Removing default USB controller from domain '%s'"
                      " for migration compatibility", def->name);
7784
            toremove++;
7785 7786 7787 7788 7789 7790 7791 7792 7793 7794 7795 7796 7797 7798 7799 7800 7801
        } 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 &&
7802 7803 7804
            pci->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT &&
            !virDomainDeviceAliasIsUserAlias(pci->info.alias) &&
            !pci->opts.pciopts.pcihole64) {
L
Laine Stump 已提交
7805
            VIR_DEBUG("Removing default pci-root from domain '%s'"
7806
                      " for migration compatibility", def->name);
7807
            toremove++;
7808 7809 7810 7811
        } else {
            pci = NULL;
        }

7812
        if (toremove) {
7813 7814 7815
            virDomainControllerDefPtr *controllers = def->controllers;
            int ncontrollers = def->ncontrollers;

7816
            if (VIR_ALLOC_N(def->controllers, ncontrollers - toremove) < 0) {
7817
                def->controllers = controllers;
7818 7819 7820 7821 7822
                goto cleanup;
            }

            def->ncontrollers = 0;
            for (i = 0; i < ncontrollers; i++) {
7823
                if (controllers[i] != usb && controllers[i] != pci)
7824 7825
                    def->controllers[def->ncontrollers++] = controllers[i];
            }
7826 7827 7828 7829

            VIR_FREE(controllers);
            virDomainControllerDefFree(pci);
            virDomainControllerDefFree(usb);
7830
        }
7831

7832 7833 7834 7835 7836 7837 7838 7839 7840
        /* 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;
            }
        }

7841 7842 7843 7844
        for (i = 0; i < def->nchannels; i++) {
            if (qemuDomainChrDefDropDefaultPath(def->channels[i], driver) < 0)
                goto cleanup;
        }
7845

7846 7847 7848 7849 7850 7851 7852 7853 7854
        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) {
7855
                switch ((virDomainChrSerialTargetType)serial->targetType) {
7856
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
7857
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
7858 7859 7860 7861 7862 7863
                    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:
7864
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
7865 7866 7867 7868 7869 7870 7871 7872
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
                    /* Nothing to do */
                    break;
                }
            }
        }

7873 7874 7875 7876 7877 7878 7879 7880 7881
        /* 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;
        }
7882 7883
    }

7884 7885
 format:
    ret = virDomainDefFormatInternal(def, caps,
7886
                                     virDomainDefFormatConvertXMLFlags(flags),
7887
                                     buf, driver->xmlopt);
7888

7889
 cleanup:
7890
    virDomainDefFree(copy);
7891
    virObjectUnref(caps);
7892
    virObjectUnref(qemuCaps);
7893 7894
    return ret;
}
7895

7896 7897 7898 7899 7900 7901 7902 7903 7904 7905 7906 7907 7908 7909 7910 7911

int
qemuDomainDefFormatBuf(virQEMUDriverPtr driver,
                       virDomainDefPtr def,
                       unsigned int flags,
                       virBufferPtr buf)
{
    return qemuDomainDefFormatBufInternal(driver, def, NULL, flags, buf);
}


static char *
qemuDomainDefFormatXMLInternal(virQEMUDriverPtr driver,
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags)
7912 7913 7914
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

7915
    if (qemuDomainDefFormatBufInternal(driver, def, origCPU, flags, &buf) < 0)
7916 7917 7918 7919 7920
        return NULL;

    return virBufferContentAndReset(&buf);
}

7921 7922 7923 7924 7925 7926 7927 7928 7929 7930

char *
qemuDomainDefFormatXML(virQEMUDriverPtr driver,
                       virDomainDefPtr def,
                       unsigned int flags)
{
    return qemuDomainDefFormatXMLInternal(driver, def, NULL, flags);
}


7931
char *qemuDomainFormatXML(virQEMUDriverPtr driver,
7932
                          virDomainObjPtr vm,
7933
                          unsigned int flags)
7934 7935
{
    virDomainDefPtr def;
7936 7937
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCPUDefPtr origCPU = NULL;
7938

7939
    if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef) {
7940
        def = vm->newDef;
7941
    } else {
7942
        def = vm->def;
7943
        origCPU = priv->origCPU;
7944
    }
7945

7946
    return qemuDomainDefFormatXMLInternal(driver, def, origCPU, flags);
7947 7948
}

7949
char *
7950
qemuDomainDefFormatLive(virQEMUDriverPtr driver,
7951
                        virDomainDefPtr def,
7952
                        virCPUDefPtr origCPU,
7953 7954
                        bool inactive,
                        bool compatible)
7955 7956 7957 7958 7959
{
    unsigned int flags = QEMU_DOMAIN_FORMAT_LIVE_FLAGS;

    if (inactive)
        flags |= VIR_DOMAIN_XML_INACTIVE;
7960 7961
    if (compatible)
        flags |= VIR_DOMAIN_XML_MIGRATABLE;
7962

7963
    return qemuDomainDefFormatXMLInternal(driver, def, origCPU, flags);
7964 7965
}

7966

7967
void qemuDomainObjTaint(virQEMUDriverPtr driver,
7968
                        virDomainObjPtr obj,
7969
                        virDomainTaintFlags taint,
7970
                        qemuDomainLogContextPtr logCtxt)
7971
{
7972
    virErrorPtr orig_err = NULL;
7973
    bool closeLog = false;
7974 7975
    char *timestamp = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
7976

7977 7978 7979 7980 7981 7982 7983 7984 7985 7986 7987 7988 7989 7990 7991
    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 */
    orig_err = virSaveLastError();
7992

7993 7994 7995 7996 7997 7998 7999 8000 8001
    if (!(timestamp = virTimeStringNow()))
        goto cleanup;

    if (logCtxt == NULL) {
        logCtxt = qemuDomainLogContextNew(driver, obj,
                                          QEMU_DOMAIN_LOG_CONTEXT_MODE_ATTACH);
        if (!logCtxt) {
            VIR_WARN("Unable to open domainlog");
            goto cleanup;
8002
        }
8003 8004 8005 8006 8007 8008 8009 8010 8011 8012 8013 8014 8015
        closeLog = true;
    }

    if (qemuDomainLogContextWrite(logCtxt,
                                  "%s: Domain id=%d is tainted: %s\n",
                                  timestamp,
                                  obj->def->id,
                                  virDomainTaintTypeToString(taint)) < 0)
        virResetLastError();

 cleanup:
    VIR_FREE(timestamp);
    if (closeLog)
8016
        virObjectUnref(logCtxt);
8017 8018 8019
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
8020 8021 8022 8023
    }
}


8024
void qemuDomainObjCheckTaint(virQEMUDriverPtr driver,
8025
                             virDomainObjPtr obj,
8026
                             qemuDomainLogContextPtr logCtxt)
8027
{
8028
    size_t i;
8029
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8030
    qemuDomainObjPrivatePtr priv = obj->privateData;
8031

8032
    if (virQEMUDriverIsPrivileged(driver) &&
8033 8034 8035
        (!cfg->clearEmulatorCapabilities ||
         cfg->user == 0 ||
         cfg->group == 0))
8036
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
8037

8038
    if (priv->hookRun)
8039
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOOK, logCtxt);
8040

8041 8042 8043
    if (obj->def->namespaceData) {
        qemuDomainCmdlineDefPtr qemucmd = obj->def->namespaceData;
        if (qemucmd->num_args || qemucmd->num_env)
8044
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logCtxt);
8045 8046
    }

8047
    if (obj->def->cpu && obj->def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
8048
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOST_CPU, logCtxt);
8049

8050
    for (i = 0; i < obj->def->ndisks; i++)
8051
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logCtxt);
8052

8053 8054
    for (i = 0; i < obj->def->nhostdevs; i++)
        qemuDomainObjCheckHostdevTaint(driver, obj, obj->def->hostdevs[i],
8055
                                       logCtxt);
8056

8057
    for (i = 0; i < obj->def->nnets; i++)
8058
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logCtxt);
8059

8060
    if (obj->def->os.dtb)
8061
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_DTB, logCtxt);
8062

8063
    virObjectUnref(cfg);
8064 8065 8066
}


8067
void qemuDomainObjCheckDiskTaint(virQEMUDriverPtr driver,
8068
                                 virDomainObjPtr obj,
8069
                                 virDomainDiskDefPtr disk,
8070
                                 qemuDomainLogContextPtr logCtxt)
8071
{
8072
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8073

8074 8075
    if (disk->rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
8076
                           logCtxt);
8077

8078 8079
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
        virStorageSourceGetActualType(disk->src) == VIR_STORAGE_TYPE_BLOCK &&
8080
        disk->src->path && virFileIsCDROM(disk->src->path) == 1)
8081
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CDROM_PASSTHROUGH,
8082
                           logCtxt);
8083

8084
    virObjectUnref(cfg);
8085 8086 8087
}


8088 8089 8090
void qemuDomainObjCheckHostdevTaint(virQEMUDriverPtr driver,
                                    virDomainObjPtr obj,
                                    virDomainHostdevDefPtr hostdev,
8091
                                    qemuDomainLogContextPtr logCtxt)
8092
{
8093
    if (!virHostdevIsSCSIDevice(hostdev))
8094
        return;
8095

8096 8097
    if (hostdev->source.subsys.u.scsi.rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
8098 8099 8100
}


8101
void qemuDomainObjCheckNetTaint(virQEMUDriverPtr driver,
8102
                                virDomainObjPtr obj,
8103
                                virDomainNetDefPtr net,
8104
                                qemuDomainLogContextPtr logCtxt)
8105
{
8106 8107 8108 8109 8110 8111
    /* 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)
8112
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logCtxt);
8113
}
8114 8115


8116 8117 8118 8119 8120 8121 8122
qemuDomainLogContextPtr qemuDomainLogContextNew(virQEMUDriverPtr driver,
                                                virDomainObjPtr vm,
                                                qemuDomainLogContextMode mode)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainLogContextPtr ctxt = NULL;

8123
    if (qemuDomainInitialize() < 0)
8124 8125 8126 8127
        goto cleanup;

    if (!(ctxt = virObjectNew(qemuDomainLogContextClass)))
        goto cleanup;
8128

8129
    VIR_DEBUG("Context new %p stdioLogD=%d", ctxt, cfg->stdioLogD);
8130 8131 8132
    ctxt->writefd = -1;
    ctxt->readfd = -1;

8133 8134 8135
    if (virAsprintf(&ctxt->path, "%s/%s.log", cfg->logDir, vm->def->name) < 0)
        goto error;

8136 8137 8138 8139
    if (cfg->stdioLogD) {
        ctxt->manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver));
        if (!ctxt->manager)
            goto error;
8140

8141 8142 8143 8144
        ctxt->writefd = virLogManagerDomainOpenLogFile(ctxt->manager,
                                                       "qemu",
                                                       vm->def->uuid,
                                                       vm->def->name,
8145
                                                       ctxt->path,
8146 8147 8148 8149 8150 8151
                                                       0,
                                                       &ctxt->inode,
                                                       &ctxt->pos);
        if (ctxt->writefd < 0)
            goto error;
    } else {
8152
        if ((ctxt->writefd = open(ctxt->path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
8153
            virReportSystemError(errno, _("failed to create logfile %s"),
8154
                                 ctxt->path);
8155 8156
            goto error;
        }
8157
        if (virSetCloseExec(ctxt->writefd) < 0) {
8158
            virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
8159
                                 ctxt->path);
8160 8161 8162
            goto error;
        }

8163 8164 8165 8166 8167 8168 8169
        /* 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"),
8170
                                 ctxt->path);
8171 8172 8173 8174
            goto error;
        }

        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START) {
8175
            if ((ctxt->readfd = open(ctxt->path, O_RDONLY, S_IRUSR | S_IWUSR)) < 0) {
8176
                virReportSystemError(errno, _("failed to open logfile %s"),
8177
                                     ctxt->path);
8178 8179 8180 8181
                goto error;
            }
            if (virSetCloseExec(ctxt->readfd) < 0) {
                virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
8182
                                     ctxt->path);
8183 8184 8185 8186
                goto error;
            }
        }

8187 8188
        if ((ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END)) < 0) {
            virReportSystemError(errno, _("failed to seek in log file %s"),
8189
                                 ctxt->path);
8190 8191 8192 8193
            goto error;
        }
    }

8194
 cleanup:
8195 8196 8197 8198
    virObjectUnref(cfg);
    return ctxt;

 error:
8199
    virObjectUnref(ctxt);
8200 8201
    ctxt = NULL;
    goto cleanup;
8202 8203 8204 8205 8206 8207 8208 8209 8210 8211 8212 8213 8214 8215
}


int qemuDomainLogContextWrite(qemuDomainLogContextPtr ctxt,
                              const char *fmt, ...)
{
    va_list argptr;
    char *message = NULL;
    int ret = -1;

    va_start(argptr, fmt);

    if (virVasprintf(&message, fmt, argptr) < 0)
        goto cleanup;
8216 8217
    if (!ctxt->manager &&
        lseek(ctxt->writefd, 0, SEEK_END) < 0) {
8218
        virReportSystemError(errno, "%s",
8219
                             _("Unable to seek to end of domain logfile"));
8220 8221 8222 8223 8224 8225 8226 8227 8228 8229 8230 8231 8232 8233 8234 8235 8236 8237 8238 8239
        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)
{
8240 8241 8242 8243
    VIR_DEBUG("Context read %p manager=%p inode=%llu pos=%llu",
              ctxt, ctxt->manager,
              (unsigned long long)ctxt->inode,
              (unsigned long long)ctxt->pos);
8244
    char *buf;
8245 8246 8247
    size_t buflen;
    if (ctxt->manager) {
        buf = virLogManagerDomainReadLogFile(ctxt->manager,
8248
                                             ctxt->path,
8249 8250 8251 8252 8253 8254 8255 8256 8257
                                             ctxt->inode,
                                             ctxt->pos,
                                             1024 * 128,
                                             0);
        if (!buf)
            return -1;
        buflen = strlen(buf);
    } else {
        ssize_t got;
8258

8259
        buflen = 1024 * 128;
8260

8261 8262
        /* Best effort jump to start of messages */
        ignore_value(lseek(ctxt->readfd, ctxt->pos, SEEK_SET));
8263

8264 8265
        if (VIR_ALLOC_N(buf, buflen) < 0)
            return -1;
8266

8267 8268 8269 8270 8271 8272 8273 8274 8275 8276 8277 8278 8279
        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;
    }
8280 8281 8282

    *msg = buf;

8283
    return buflen;
8284 8285 8286
}


8287 8288 8289 8290 8291 8292 8293 8294 8295 8296 8297 8298 8299 8300 8301 8302 8303 8304 8305 8306 8307 8308 8309 8310 8311 8312 8313 8314 8315 8316 8317 8318 8319 8320 8321 8322 8323 8324 8325 8326 8327 8328 8329 8330 8331 8332 8333 8334 8335 8336 8337 8338 8339 8340 8341 8342 8343 8344 8345 8346 8347 8348
/**
 * qemuDomainLogAppendMessage:
 *
 * This is a best-effort attempt to add a log message to the qemu log file
 * either by using virtlogd or the legacy approach */
int
qemuDomainLogAppendMessage(virQEMUDriverPtr driver,
                           virDomainObjPtr vm,
                           const char *fmt,
                           ...)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    virLogManagerPtr manager = NULL;
    va_list ap;
    char *path = NULL;
    int writefd = -1;
    char *message = NULL;
    int ret = -1;

    va_start(ap, fmt);

    if (virVasprintf(&message, fmt, ap) < 0)
        goto cleanup;

    VIR_DEBUG("Append log message (vm='%s' message='%s) stdioLogD=%d",
              vm->def->name, message, cfg->stdioLogD);

    if (virAsprintf(&path, "%s/%s.log", cfg->logDir, vm->def->name) < 0)
        goto cleanup;

    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;
}


8349 8350 8351 8352 8353 8354 8355 8356
int qemuDomainLogContextGetWriteFD(qemuDomainLogContextPtr ctxt)
{
    return ctxt->writefd;
}


void qemuDomainLogContextMarkPosition(qemuDomainLogContextPtr ctxt)
{
8357 8358
    if (ctxt->manager)
        virLogManagerDomainGetLogFilePosition(ctxt->manager,
8359
                                              ctxt->path,
8360 8361 8362 8363 8364
                                              0,
                                              &ctxt->inode,
                                              &ctxt->pos);
    else
        ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END);
8365 8366 8367
}


8368 8369 8370 8371 8372 8373
virLogManagerPtr qemuDomainLogContextGetManager(qemuDomainLogContextPtr ctxt)
{
    return ctxt->manager;
}


8374 8375
/* Locate an appropriate 'qemu-img' binary.  */
const char *
8376
qemuFindQemuImgBinary(virQEMUDriverPtr driver)
8377
{
8378 8379
    if (!driver->qemuImgBinary)
        virReportError(VIR_ERR_INTERNAL_ERROR,
8380
                       "%s", _("unable to find qemu-img"));
8381 8382 8383 8384 8385 8386 8387

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
                                virDomainSnapshotObjPtr snapshot,
8388
                                virCapsPtr caps,
8389
                                virDomainXMLOptionPtr xmlopt,
8390 8391 8392 8393 8394 8395 8396 8397 8398
                                char *snapshotDir)
{
    char *newxml = NULL;
    int ret = -1;
    char *snapDir = NULL;
    char *snapFile = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

    virUUIDFormat(vm->def->uuid, uuidstr);
8399
    newxml = virDomainSnapshotDefFormat(
8400
        uuidstr, snapshot->def, caps, xmlopt,
8401 8402
        virDomainDefFormatConvertXMLFlags(QEMU_DOMAIN_FORMAT_LIVE_FLAGS),
        1);
8403
    if (newxml == NULL)
8404 8405
        return -1;

8406
    if (virAsprintf(&snapDir, "%s/%s", snapshotDir, vm->def->name) < 0)
8407 8408 8409 8410 8411 8412 8413
        goto cleanup;
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
        goto cleanup;
    }

8414
    if (virAsprintf(&snapFile, "%s/%s.xml", snapDir, snapshot->def->name) < 0)
8415 8416
        goto cleanup;

J
Ján Tomko 已提交
8417
    ret = virXMLSaveFile(snapFile, NULL, "snapshot-edit", newxml);
8418

8419
 cleanup:
8420 8421 8422 8423 8424 8425 8426 8427
    VIR_FREE(snapFile);
    VIR_FREE(snapDir);
    VIR_FREE(newxml);
    return ret;
}

/* The domain is expected to be locked and inactive. Return -1 on normal
 * failure, 1 if we skipped a disk due to try_all.  */
8428
static int
8429
qemuDomainSnapshotForEachQcow2Raw(virQEMUDriverPtr driver,
8430 8431 8432 8433 8434
                                  virDomainDefPtr def,
                                  const char *name,
                                  const char *op,
                                  bool try_all,
                                  int ndisks)
8435 8436
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
8437
    size_t i;
8438 8439 8440 8441 8442 8443 8444 8445 8446
    bool skipped = false;

    qemuimgarg[0] = qemuFindQemuImgBinary(driver);
    if (qemuimgarg[0] == NULL) {
        /* qemuFindQemuImgBinary set the error */
        return -1;
    }

    qemuimgarg[2] = op;
8447
    qemuimgarg[3] = name;
8448

8449
    for (i = 0; i < ndisks; i++) {
8450
        /* FIXME: we also need to handle LVM here */
8451
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
8452 8453 8454
            int format = virDomainDiskGetFormat(def->disks[i]);

            if (format > 0 && format != VIR_STORAGE_FILE_QCOW2) {
8455 8456 8457 8458 8459
                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",
8460
                             def->disks[i]->dst);
8461 8462
                    skipped = true;
                    continue;
8463 8464 8465 8466 8467
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
8468
                }
8469 8470 8471 8472
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("Disk device '%s' does not support"
                                 " snapshotting"),
                               def->disks[i]->dst);
8473 8474 8475
                return -1;
            }

8476
            qemuimgarg[4] = virDomainDiskGetSource(def->disks[i]);
8477 8478 8479 8480

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
8481
                             def->disks[i]->dst);
8482 8483
                    skipped = true;
                    continue;
8484 8485 8486 8487 8488
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
8489 8490 8491 8492 8493 8494 8495 8496 8497
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

8498 8499 8500
/* 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
8501
qemuDomainSnapshotForEachQcow2(virQEMUDriverPtr driver,
8502 8503 8504 8505 8506 8507 8508 8509 8510 8511 8512 8513 8514 8515 8516 8517
                               virDomainObjPtr vm,
                               virDomainSnapshotObjPtr snap,
                               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.  */
    virDomainDefPtr def = snap->def->dom;

    if (!def)
        def = vm->def;
    return qemuDomainSnapshotForEachQcow2Raw(driver, def, snap->def->name,
                                             op, try_all, def->ndisks);
}

8518 8519
/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
8520
qemuDomainSnapshotDiscard(virQEMUDriverPtr driver,
8521 8522
                          virDomainObjPtr vm,
                          virDomainSnapshotObjPtr snap,
8523
                          bool update_parent,
8524 8525 8526 8527 8528 8529
                          bool metadata_only)
{
    char *snapFile = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;
    virDomainSnapshotObjPtr parentsnap = NULL;
8530
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8531 8532 8533 8534 8535 8536 8537 8538 8539

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
                goto cleanup;
        } else {
            priv = vm->privateData;
8540
            qemuDomainObjEnterMonitor(driver, vm);
8541 8542
            /* we continue on even in the face of error */
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
8543
            ignore_value(qemuDomainObjExitMonitor(driver, vm));
8544 8545 8546
        }
    }

8547
    if (virAsprintf(&snapFile, "%s/%s/%s.xml", cfg->snapshotDir,
8548
                    vm->def->name, snap->def->name) < 0)
8549 8550 8551
        goto cleanup;

    if (snap == vm->current_snapshot) {
8552
        if (update_parent && snap->def->parent) {
8553
            parentsnap = virDomainSnapshotFindByName(vm->snapshots,
8554 8555 8556 8557 8558 8559
                                                     snap->def->parent);
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
                         snap->def->parent);
            } else {
                parentsnap->def->current = true;
8560
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap, driver->caps,
8561
                                                    driver->xmlopt,
8562
                                                    cfg->snapshotDir) < 0) {
8563 8564 8565 8566 8567 8568 8569 8570 8571 8572 8573 8574
                    VIR_WARN("failed to set parent snapshot '%s' as current",
                             snap->def->parent);
                    parentsnap->def->current = false;
                    parentsnap = NULL;
                }
            }
        }
        vm->current_snapshot = parentsnap;
    }

    if (unlink(snapFile) < 0)
        VIR_WARN("Failed to unlink %s", snapFile);
8575 8576
    if (update_parent)
        virDomainSnapshotDropParent(snap);
8577
    virDomainSnapshotObjListRemove(vm->snapshots, snap);
8578 8579 8580

    ret = 0;

8581
 cleanup:
8582
    VIR_FREE(snapFile);
8583
    virObjectUnref(cfg);
8584 8585 8586 8587
    return ret;
}

/* Hash iterator callback to discard multiple snapshots.  */
8588 8589 8590
int qemuDomainSnapshotDiscardAll(void *payload,
                                 const void *name ATTRIBUTE_UNUSED,
                                 void *data)
8591 8592
{
    virDomainSnapshotObjPtr snap = payload;
8593
    virQEMUSnapRemovePtr curr = data;
8594 8595 8596 8597 8598 8599 8600 8601
    int err;

    if (snap->def->current)
        curr->current = true;
    err = qemuDomainSnapshotDiscard(curr->driver, curr->vm, snap, false,
                                    curr->metadata_only);
    if (err && !curr->err)
        curr->err = err;
8602
    return 0;
8603 8604 8605
}

int
8606
qemuDomainSnapshotDiscardAllMetadata(virQEMUDriverPtr driver,
8607 8608
                                     virDomainObjPtr vm)
{
8609
    virQEMUSnapRemove rem;
8610 8611 8612 8613 8614

    rem.driver = driver;
    rem.vm = vm;
    rem.metadata_only = true;
    rem.err = 0;
8615 8616
    virDomainSnapshotForEach(vm->snapshots, qemuDomainSnapshotDiscardAll,
                             &rem);
8617 8618 8619 8620

    return rem.err;
}

8621

8622 8623 8624
static void
qemuDomainRemoveInactiveCommon(virQEMUDriverPtr driver,
                               virDomainObjPtr vm)
8625
{
8626
    char *snapDir;
8627 8628 8629
    virQEMUDriverConfigPtr cfg;

    cfg = virQEMUDriverGetConfig(driver);
8630

8631 8632 8633 8634 8635
    /* 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);
    }
8636
    else if (virAsprintf(&snapDir, "%s/%s", cfg->snapshotDir,
8637 8638
                         vm->def->name) < 0) {
        VIR_WARN("unable to remove snapshot directory %s/%s",
8639
                 cfg->snapshotDir, vm->def->name);
8640 8641 8642 8643 8644
    } else {
        if (rmdir(snapDir) < 0 && errno != ENOENT)
            VIR_WARN("unable to remove snapshot directory %s", snapDir);
        VIR_FREE(snapDir);
    }
8645
    qemuExtDevicesCleanupHost(driver, vm->def);
8646

8647
    virObjectUnref(cfg);
8648 8649 8650
}


8651 8652 8653 8654 8655 8656 8657 8658 8659 8660 8661 8662 8663 8664 8665 8666 8667 8668 8669 8670
/**
 * 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);
}


8671 8672 8673 8674 8675 8676 8677 8678 8679 8680 8681 8682 8683 8684 8685 8686 8687 8688 8689 8690 8691 8692
/**
 * 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);
}


8693 8694 8695 8696 8697 8698 8699 8700 8701 8702 8703 8704 8705 8706 8707 8708 8709
/**
 * 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);
8710 8711

    if (haveJob)
8712
        qemuDomainObjEndJob(driver, vm);
8713
}
8714

8715

8716 8717 8718 8719 8720 8721 8722 8723 8724 8725 8726 8727 8728 8729 8730 8731 8732 8733 8734 8735 8736
/**
 * 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);
}


8737
void
8738
qemuDomainSetFakeReboot(virQEMUDriverPtr driver,
8739 8740 8741 8742
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
8743
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8744 8745

    if (priv->fakeReboot == value)
8746
        goto cleanup;
8747 8748 8749

    priv->fakeReboot = value;

8750
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
8751
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
8752

8753
 cleanup:
8754
    virObjectUnref(cfg);
8755
}
M
Michal Privoznik 已提交
8756

8757
static void
8758 8759
qemuDomainCheckRemoveOptionalDisk(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
8760
                                  size_t diskIndex)
8761 8762
{
    char uuid[VIR_UUID_STRING_BUFLEN];
8763
    virObjectEventPtr event = NULL;
8764
    virDomainDiskDefPtr disk = vm->def->disks[diskIndex];
8765
    const char *src = virDomainDiskGetSource(disk);
8766 8767 8768 8769 8770

    virUUIDFormat(vm->def->uuid, uuid);

    VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
              "due to inaccessible source '%s'",
8771
              disk->dst, vm->def->name, uuid, src);
8772 8773 8774 8775

    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM ||
        disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY) {

8776
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
8777 8778
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
8779
        virDomainDiskEmptySource(disk);
8780 8781
        /* keeping the old startup policy would be invalid for new images */
        disk->startupPolicy = VIR_DOMAIN_STARTUP_POLICY_DEFAULT;
8782
    } else {
8783
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
8784 8785
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_DROP_MISSING_ON_START);
8786 8787
        virDomainDiskRemove(vm->def, diskIndex);
        virDomainDiskDefFree(disk);
8788 8789
    }

8790
    virObjectEventStateQueue(driver->domainEventState, event);
8791 8792
}

8793 8794 8795 8796 8797 8798 8799 8800 8801 8802 8803 8804 8805 8806 8807 8808 8809

/**
 * 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
8810 8811
qemuDomainCheckDiskStartupPolicy(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
8812
                                 size_t diskIndex,
8813 8814
                                 bool cold_boot)
{
8815
    int startupPolicy = vm->def->disks[diskIndex]->startupPolicy;
8816
    int device = vm->def->disks[diskIndex]->device;
8817

8818
    switch ((virDomainStartupPolicy) startupPolicy) {
8819
        case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
8820 8821 8822 8823 8824 8825
            /* 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)
8826
                return -1;
8827 8828
            break;

8829
        case VIR_DOMAIN_STARTUP_POLICY_DEFAULT:
8830
        case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
8831
            return -1;
8832 8833

        case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
8834
            if (cold_boot)
8835
                return -1;
8836 8837 8838 8839 8840 8841 8842
            break;

        case VIR_DOMAIN_STARTUP_POLICY_LAST:
            /* this should never happen */
            break;
    }

8843
    qemuDomainCheckRemoveOptionalDisk(driver, vm, diskIndex);
8844
    virResetLastError();
8845 8846 8847
    return 0;
}

8848

8849 8850 8851 8852 8853 8854 8855 8856 8857 8858

/*
 * The vm must be locked when any of the following cleanup functions is
 * called.
 */
int
qemuDomainCleanupAdd(virDomainObjPtr vm,
                     qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
8859
    size_t i;
8860 8861 8862 8863 8864 8865 8866 8867 8868 8869

    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,
8870
                     priv->ncleanupCallbacks, 1) < 0)
8871 8872 8873 8874 8875 8876 8877 8878 8879 8880 8881
        return -1;

    priv->cleanupCallbacks[priv->ncleanupCallbacks++] = cb;
    return 0;
}

void
qemuDomainCleanupRemove(virDomainObjPtr vm,
                        qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
8882
    size_t i;
8883 8884 8885 8886

    VIR_DEBUG("vm=%s, cb=%p", vm->def->name, cb);

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
8887 8888 8889
        if (priv->cleanupCallbacks[i] == cb)
            VIR_DELETE_ELEMENT_INPLACE(priv->cleanupCallbacks,
                                       i, priv->ncleanupCallbacks);
8890 8891 8892 8893 8894 8895 8896 8897
    }

    VIR_SHRINK_N(priv->cleanupCallbacks,
                 priv->ncleanupCallbacks_max,
                 priv->ncleanupCallbacks_max - priv->ncleanupCallbacks);
}

void
8898
qemuDomainCleanupRun(virQEMUDriverPtr driver,
8899 8900 8901
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
8902
    size_t i;
8903 8904 8905 8906

    VIR_DEBUG("driver=%p, vm=%s", driver, vm->def->name);

    /* run cleanup callbacks in reverse order */
8907 8908
    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[priv->ncleanupCallbacks - (i + 1)])
8909 8910 8911 8912 8913 8914 8915
            priv->cleanupCallbacks[i](driver, vm);
    }

    VIR_FREE(priv->cleanupCallbacks);
    priv->ncleanupCallbacks = 0;
    priv->ncleanupCallbacks_max = 0;
}
8916

8917 8918 8919
static void
qemuDomainGetImageIds(virQEMUDriverConfigPtr cfg,
                      virDomainObjPtr vm,
8920
                      virStorageSourcePtr src,
8921
                      virStorageSourcePtr parentSrc,
8922 8923 8924 8925 8926 8927 8928 8929 8930 8931 8932 8933 8934 8935 8936 8937 8938 8939
                      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;
    }

8940 8941
    if (vm && (vmlabel = virDomainDefGetSecurityLabelDef(vm->def, "dac")) &&
        vmlabel->label)
8942 8943
        virParseOwnershipIds(vmlabel->label, uid, gid);

8944 8945 8946 8947 8948
    if (parentSrc &&
        (disklabel = virStorageSourceGetSecurityLabelDef(parentSrc, "dac")) &&
        disklabel->label)
        virParseOwnershipIds(disklabel->label, uid, gid);

8949
    if ((disklabel = virStorageSourceGetSecurityLabelDef(src, "dac")) &&
8950
        disklabel->label)
8951 8952 8953 8954
        virParseOwnershipIds(disklabel->label, uid, gid);
}


8955 8956 8957
int
qemuDomainStorageFileInit(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
8958 8959
                          virStorageSourcePtr src,
                          virStorageSourcePtr parent)
8960 8961 8962 8963 8964 8965
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    uid_t uid;
    gid_t gid;
    int ret = -1;

8966
    qemuDomainGetImageIds(cfg, vm, src, parent, &uid, &gid);
8967 8968 8969 8970 8971 8972 8973 8974 8975 8976 8977 8978

    if (virStorageFileInitAs(src, uid, gid) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
}


8979 8980 8981 8982 8983
char *
qemuDomainStorageAlias(const char *device, int depth)
{
    char *alias;

8984
    device = qemuAliasDiskDriveSkipPrefix(device);
8985 8986 8987 8988 8989 8990 8991 8992 8993

    if (!depth)
        ignore_value(VIR_STRDUP(alias, device));
    else
        ignore_value(virAsprintf(&alias, "%s.%d", device, depth));
    return alias;
}


8994 8995 8996 8997 8998 8999 9000 9001 9002 9003 9004 9005 9006
/**
 * 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.
 */
9007
int
9008
qemuDomainDetermineDiskChain(virQEMUDriverPtr driver,
9009
                             virDomainObjPtr vm,
9010
                             virDomainDiskDefPtr disk,
9011
                             virStorageSourcePtr disksrc,
9012
                             bool report_broken)
9013
{
9014
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
9015 9016
    virStorageSourcePtr src; /* iterator for the backing chain declared in XML */
    virStorageSourcePtr n; /* iterator for the backing chain detected from disk */
9017
    qemuDomainObjPrivatePtr priv = vm->privateData;
9018
    int ret = -1;
9019 9020
    uid_t uid;
    gid_t gid;
9021

9022 9023 9024
    if (!disksrc)
        disksrc = disk->src;

9025
    if (virStorageSourceIsEmpty(disksrc)) {
9026
        ret = 0;
9027
        goto cleanup;
9028
    }
9029

9030 9031
    /* There is no need to check the backing chain for disks without backing
     * support */
9032 9033 9034
    if (virStorageSourceIsLocalStorage(disksrc) &&
        disksrc->format > VIR_STORAGE_FILE_NONE &&
        disksrc->format < VIR_STORAGE_FILE_BACKING) {
9035

9036
        if (!virFileExists(disksrc->path)) {
9037
            if (report_broken)
9038
                virStorageFileReportBrokenChain(errno, disksrc, disksrc);
9039 9040 9041 9042

            goto cleanup;
        }

9043
        /* terminate the chain for such images as the code below would do */
9044 9045
        if (!disksrc->backingStore &&
            VIR_ALLOC(disksrc->backingStore) < 0)
9046 9047
            goto cleanup;

9048 9049 9050
        /* 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 &&
9051 9052 9053 9054
            disksrc->format == VIR_STORAGE_FILE_RAW &&
            virStorageSourceIsBlockLocal(disksrc) &&
            virFileIsCDROM(disksrc->path) == 1)
            disksrc->hostcdrom = true;
9055

9056 9057 9058 9059
        ret = 0;
        goto cleanup;
    }

9060
    src = disksrc;
9061 9062
    /* skip to the end of the chain if there is any */
    while (virStorageSourceHasBacking(src)) {
9063 9064
        if (report_broken) {
            int rv = virStorageFileSupportsAccess(src);
9065

9066
            if (rv < 0)
9067
                goto cleanup;
9068

9069
            if (rv > 0) {
9070
                if (qemuDomainStorageFileInit(driver, vm, src, disksrc) < 0)
9071 9072 9073
                    goto cleanup;

                if (virStorageFileAccess(src, F_OK) < 0) {
9074
                    virStorageFileReportBrokenChain(errno, src, disksrc);
9075 9076 9077 9078
                    virStorageFileDeinit(src);
                    goto cleanup;
                }

9079
                virStorageFileDeinit(src);
9080 9081
            }
        }
9082
        src = src->backingStore;
9083 9084 9085 9086 9087 9088 9089
    }

    /* We skipped to the end of the chain. Skip detection if there's the
     * terminator. (An allocated but empty backingStore) */
    if (src->backingStore) {
        ret = 0;
        goto cleanup;
9090
    }
9091

9092
    qemuDomainGetImageIds(cfg, vm, src, disksrc, &uid, &gid);
9093

9094
    if (virStorageFileGetMetadata(src, uid, gid, report_broken) < 0)
9095 9096
        goto cleanup;

9097
    for (n = src->backingStore; virStorageSourceIsBacking(n); n = n->backingStore) {
9098 9099 9100
        if (qemuDomainValidateStorageSource(n, priv->qemuCaps) < 0)
            goto cleanup;

9101 9102
        if (qemuDomainPrepareDiskSourceData(disk, n, cfg, priv->qemuCaps) < 0)
            goto cleanup;
9103 9104 9105 9106

        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV) &&
            qemuDomainPrepareStorageSourceBlockdev(disk, n, priv, cfg) < 0)
            goto cleanup;
9107
    }
9108

9109
    ret = 0;
9110

9111
 cleanup:
9112 9113
    virObjectUnref(cfg);
    return ret;
9114
}
9115

9116

9117 9118 9119 9120 9121 9122 9123 9124 9125 9126 9127 9128 9129 9130 9131 9132
/**
 * 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,
9133
                              virQEMUCapsPtr qemuCaps,
9134 9135
                              char **backendAlias)
{
9136 9137
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);
    const char *nodename = NULL;
9138 9139
    *backendAlias = NULL;

9140 9141 9142 9143 9144 9145 9146 9147 9148 9149 9150 9151 9152 9153 9154 9155
    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;

    if (VIR_STRDUP(*backendAlias, nodename) < 0)
9156 9157 9158 9159 9160 9161
        return -1;

    return 0;
}


9162 9163 9164 9165 9166 9167 9168 9169 9170 9171 9172
/**
 * qemuDomainDiskChainElementRevoke:
 *
 * Revoke access to a single backing chain element. This restores the labels,
 * removes cgroup ACLs for devices and removes locks.
 */
void
qemuDomainDiskChainElementRevoke(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
                                 virStorageSourcePtr elem)
{
9173 9174 9175 9176
    if (qemuTeardownImageCgroup(vm, elem) < 0)
        VIR_WARN("Failed to teardown cgroup for disk path %s",
                 NULLSTR(elem->path));

9177
    if (qemuSecurityRestoreImageLabel(driver, vm, elem, false) < 0)
9178 9179
        VIR_WARN("Unable to restore security label on %s", NULLSTR(elem->path));

9180
    if (qemuDomainNamespaceTeardownDisk(vm, elem) < 0)
9181
        VIR_WARN("Unable to remove /dev entry for %s", NULLSTR(elem->path));
9182 9183 9184 9185 9186 9187 9188 9189

    if (virDomainLockImageDetach(driver->lockManager, vm, elem) < 0)
        VIR_WARN("Unable to release lock on %s", NULLSTR(elem->path));
}


/**
 * qemuDomainDiskChainElementPrepare:
9190 9191 9192 9193 9194
 * @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
9195 9196 9197
 *
 * 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
9198 9199 9200 9201 9202
 * 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.
 */
9203 9204 9205 9206
int
qemuDomainDiskChainElementPrepare(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
                                  virStorageSourcePtr elem,
9207 9208
                                  bool readonly,
                                  bool newSource)
9209 9210 9211 9212 9213 9214 9215 9216 9217 9218 9219 9220
{
    bool was_readonly = elem->readonly;
    virQEMUDriverConfigPtr cfg = NULL;
    int ret = -1;

    cfg = virQEMUDriverGetConfig(driver);

    elem->readonly = readonly;

    if (virDomainLockImageAttach(driver->lockManager, cfg->uri, vm, elem) < 0)
        goto cleanup;

9221
    if (newSource &&
9222
        qemuDomainNamespaceSetupDisk(vm, elem) < 0)
9223 9224
        goto cleanup;

9225 9226 9227
    if (qemuSetupImageCgroup(vm, elem) < 0)
        goto cleanup;

9228
    if (qemuSecuritySetImageLabel(driver, vm, elem, false) < 0)
9229 9230 9231 9232 9233 9234 9235 9236 9237 9238 9239
        goto cleanup;

    ret = 0;

 cleanup:
    elem->readonly = was_readonly;
    virObjectUnref(cfg);
    return ret;
}


9240 9241 9242 9243 9244 9245 9246 9247 9248 9249
/*
 * 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)
{
9250 9251 9252 9253 9254 9255 9256 9257 9258 9259
#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; \
        } \
9260 9261 9262
    } while (0)

    CHECK_EQ(device, "device", false);
9263 9264 9265 9266
    CHECK_EQ(bus, "bus", false);
    if (STRNEQ(disk->dst, orig_disk->dst)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
9267
                       "target");
9268 9269 9270 9271
        return false;
    }
    CHECK_EQ(tray_status, "tray", true);
    CHECK_EQ(removable, "removable", true);
9272 9273 9274 9275 9276 9277 9278 9279 9280 9281 9282 9283 9284 9285 9286 9287 9288 9289 9290 9291 9292 9293 9294 9295 9296 9297 9298 9299 9300 9301 9302 9303 9304 9305 9306 9307 9308 9309 9310 9311 9312 9313 9314 9315 9316 9317 9318 9319 9320 9321 9322 9323 9324 9325

    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);
9326 9327 9328
    CHECK_EQ(blkdeviotune.group_name,
             "blkdeviotune group_name",
             true);
9329 9330 9331 9332 9333 9334 9335 9336 9337 9338 9339 9340 9341 9342 9343 9344 9345 9346 9347 9348 9349 9350 9351 9352 9353 9354 9355 9356 9357

    if (disk->serial && STRNEQ_NULLABLE(disk->serial, orig_disk->serial)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "serial");
        return false;
    }

    if (disk->wwn && STRNEQ_NULLABLE(disk->wwn, orig_disk->wwn)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "wwn");
        return false;
    }

    if (disk->vendor && STRNEQ_NULLABLE(disk->vendor, orig_disk->vendor)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "vendor");
        return false;
    }

    if (disk->product && STRNEQ_NULLABLE(disk->product, orig_disk->product)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "product");
        return false;
    }

9358 9359 9360 9361 9362 9363 9364
    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);
9365
    /* "snapshot" is a libvirt internal field and thus can be changed */
9366
    /* startupPolicy is allowed to be updated. Therefore not checked here. */
9367
    CHECK_EQ(transient, "transient", true);
9368 9369 9370 9371 9372 9373 9374 9375 9376 9377 9378 9379 9380

    /* 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;
    }

9381
    /* device alias is checked already in virDomainDefCompatibleDevice */
9382

9383
    CHECK_EQ(info.bootIndex, "boot order", true);
9384 9385 9386 9387 9388 9389 9390 9391 9392 9393 9394 9395
    CHECK_EQ(rawio, "rawio", true);
    CHECK_EQ(sgio, "sgio", true);
    CHECK_EQ(discard, "discard", true);
    CHECK_EQ(iothread, "iothread", true);

    if (disk->domain_name &&
        STRNEQ_NULLABLE(disk->domain_name, orig_disk->domain_name)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "backenddomain");
        return false;
    }
9396

9397 9398 9399 9400 9401 9402
    /* 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);

9403 9404 9405 9406 9407 9408 9409 9410
    if (!virStoragePRDefIsEqual(disk->src->pr,
                                orig_disk->src->pr)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "reservations");
        return false;
    }

9411 9412 9413 9414 9415
#undef CHECK_EQ

    return true;
}

9416 9417 9418
bool
qemuDomainDiskBlockJobIsActive(virDomainDiskDefPtr disk)
{
9419 9420
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

9421 9422 9423 9424 9425 9426 9427 9428
    if (disk->mirror) {
        virReportError(VIR_ERR_BLOCK_COPY_ACTIVE,
                       _("disk '%s' already in active block job"),
                       disk->dst);

        return true;
    }

9429
    if (diskPriv->blockjob &&
9430
        qemuBlockJobIsRunning(diskPriv->blockjob)) {
9431 9432 9433 9434 9435 9436 9437 9438 9439 9440
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("disk '%s' already in active block job"),
                       disk->dst);
        return true;
    }

    return false;
}


9441 9442 9443 9444 9445 9446 9447 9448 9449 9450 9451 9452 9453 9454 9455
/**
 * 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++) {
9456 9457 9458
        virDomainDiskDefPtr disk = vm->def->disks[i];
        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

9459 9460
        if (!copy_only && diskPriv->blockjob &&
            qemuBlockJobIsRunning(diskPriv->blockjob))
9461 9462
            return true;

9463
        if (disk->mirror && disk->mirrorJob == VIR_DOMAIN_BLOCK_JOB_TYPE_COPY)
9464 9465 9466 9467 9468 9469 9470
            return true;
    }

    return false;
}


9471 9472
int
qemuDomainUpdateDeviceList(virQEMUDriverPtr driver,
9473 9474
                           virDomainObjPtr vm,
                           int asyncJob)
9475 9476 9477
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char **aliases;
9478
    int rc;
9479 9480 9481 9482

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

9483 9484
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
9485 9486 9487 9488
    rc = qemuMonitorGetDeviceAliases(priv->mon, &aliases);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
    if (rc < 0)
9489 9490
        return -1;

9491
    virStringListFree(priv->qemuDevices);
9492 9493 9494
    priv->qemuDevices = aliases;
    return 0;
}
9495

9496 9497 9498 9499 9500 9501 9502 9503 9504 9505 9506 9507 9508 9509 9510 9511 9512 9513 9514

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);

9515 9516
    if (qemuDomainObjExitMonitor(driver, vm) < 0) {
        virHashFree(meminfo);
9517
        return -1;
9518
    }
9519 9520 9521 9522 9523 9524 9525 9526 9527 9528 9529 9530 9531 9532 9533 9534 9535 9536 9537 9538 9539 9540 9541 9542 9543 9544 9545 9546

    /* 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;
}


9547 9548 9549 9550
static bool
qemuDomainABIStabilityCheck(const virDomainDef *src,
                            const virDomainDef *dst)
{
9551 9552
    size_t i;

9553 9554 9555 9556 9557 9558 9559 9560 9561
    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;
    }

9562 9563 9564 9565 9566 9567 9568 9569 9570 9571 9572 9573 9574
    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;
        }
    }

9575 9576 9577 9578 9579 9580 9581 9582 9583
    return true;
}


virDomainABIStability virQEMUDriverDomainABIStability = {
    .domain = qemuDomainABIStabilityCheck,
};


9584 9585 9586 9587 9588 9589 9590 9591 9592 9593 9594 9595 9596 9597 9598 9599 9600 9601 9602 9603
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;
}


9604 9605 9606
#define COPY_FLAGS (VIR_DOMAIN_XML_SECURE | \
                    VIR_DOMAIN_XML_MIGRATABLE)

9607 9608 9609 9610 9611 9612 9613 9614 9615
bool
qemuDomainDefCheckABIStability(virQEMUDriverPtr driver,
                               virDomainDefPtr src,
                               virDomainDefPtr dst)
{
    virDomainDefPtr migratableDefSrc = NULL;
    virDomainDefPtr migratableDefDst = NULL;
    bool ret = false;

9616 9617
    if (!(migratableDefSrc = qemuDomainDefCopy(driver, src, COPY_FLAGS)) ||
        !(migratableDefDst = qemuDomainDefCopy(driver, dst, COPY_FLAGS)))
9618 9619
        goto cleanup;

9620 9621 9622
    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   src, migratableDefSrc,
                                                   dst, migratableDefDst);
9623

9624
 cleanup:
9625 9626 9627 9628
    virDomainDefFree(migratableDefSrc);
    virDomainDefFree(migratableDefDst);
    return ret;
}
9629

9630 9631 9632 9633 9634 9635 9636 9637 9638 9639 9640 9641 9642 9643 9644 9645 9646 9647 9648 9649 9650 9651 9652 9653 9654 9655 9656 9657 9658 9659

bool
qemuDomainCheckABIStability(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
                            virDomainDefPtr dst)
{
    virDomainDefPtr migratableSrc = NULL;
    virDomainDefPtr migratableDst = NULL;
    char *xml = NULL;
    bool ret = false;

    if (!(xml = qemuDomainFormatXML(driver, vm, COPY_FLAGS)) ||
        !(migratableSrc = qemuDomainDefFromXML(driver, xml)) ||
        !(migratableDst = qemuDomainDefCopy(driver, dst, COPY_FLAGS)))
        goto cleanup;

    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   vm->def, migratableSrc,
                                                   dst, migratableDst);

 cleanup:
    VIR_FREE(xml);
    virDomainDefFree(migratableSrc);
    virDomainDefFree(migratableDst);
    return ret;
}

#undef COPY_FLAGS


9660
bool
9661
qemuDomainAgentAvailable(virDomainObjPtr vm,
9662 9663
                         bool reportError)
{
9664 9665
    qemuDomainObjPrivatePtr priv = vm->privateData;

9666 9667 9668 9669 9670 9671 9672
    if (virDomainObjGetState(vm, NULL) != VIR_DOMAIN_RUNNING) {
        if (reportError) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
        }
        return false;
    }
9673 9674 9675 9676 9677 9678 9679 9680 9681
    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) {
9682 9683 9684 9685 9686 9687 9688 9689 9690 9691 9692 9693
        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;
9694 9695 9696 9697
        }
    }
    return true;
}
9698

9699

9700
static unsigned long long
9701
qemuDomainGetMemorySizeAlignment(virDomainDefPtr def)
9702
{
9703 9704 9705 9706 9707
    /* 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;

9708 9709 9710 9711 9712 9713 9714
    /* Align memory size. QEMU requires rounding to next 4KiB block.
     * We'll take the "traditional" path and round it to 1MiB*/

    return 1024;
}


9715 9716 9717 9718 9719 9720 9721 9722 9723 9724 9725 9726 9727 9728 9729
static unsigned long long
qemuDomainGetMemoryModuleSizeAlignment(const virDomainDef *def,
                                       const virDomainMemoryDef *mem ATTRIBUTE_UNUSED)
{
    /* 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;
}


9730 9731 9732
int
qemuDomainAlignMemorySizes(virDomainDefPtr def)
{
9733 9734
    unsigned long long maxmemkb = virMemoryMaxValue(false) >> 10;
    unsigned long long maxmemcapped = virMemoryMaxValue(true) >> 10;
9735
    unsigned long long initialmem = 0;
9736
    unsigned long long hotplugmem = 0;
9737
    unsigned long long mem;
9738
    unsigned long long align = qemuDomainGetMemorySizeAlignment(def);
9739 9740 9741 9742 9743
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    /* align NUMA cell sizes if relevant */
    for (i = 0; i < ncells; i++) {
9744 9745
        mem = VIR_ROUND_UP(virDomainNumaGetNodeMemorySize(def->numa, i), align);
        initialmem += mem;
9746 9747 9748 9749 9750 9751 9752

        if (mem > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("memory size of NUMA node '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
9753
        virDomainNumaSetNodeMemorySize(def->numa, i, mem);
9754 9755
    }

9756 9757 9758 9759 9760
    /* 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);

9761 9762 9763 9764 9765 9766
    if (initialmem > maxmemcapped) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("initial memory size overflowed after alignment"));
        return -1;
    }

9767
    def->mem.max_memory = VIR_ROUND_UP(def->mem.max_memory, align);
9768 9769 9770 9771 9772
    if (def->mem.max_memory > maxmemkb) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("maximum memory size overflowed after alignment"));
        return -1;
    }
9773

9774
    /* Align memory module sizes */
9775 9776
    for (i = 0; i < def->nmems; i++) {
        align = qemuDomainGetMemoryModuleSizeAlignment(def, def->mems[i]);
9777
        def->mems[i]->size = VIR_ROUND_UP(def->mems[i]->size, align);
9778
        hotplugmem += def->mems[i]->size;
9779 9780 9781 9782 9783 9784 9785

        if (def->mems[i]->size > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("size of memory module '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
9786
    }
9787

9788 9789
    virDomainDefSetMemoryTotal(def, initialmem + hotplugmem);

9790 9791
    return 0;
}
9792 9793 9794 9795 9796 9797 9798 9799 9800 9801 9802


/**
 * 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
9803 9804
qemuDomainMemoryDeviceAlignSize(virDomainDefPtr def,
                                virDomainMemoryDefPtr mem)
9805
{
9806
    mem->size = VIR_ROUND_UP(mem->size, qemuDomainGetMemorySizeAlignment(def));
9807
}
9808 9809 9810 9811 9812 9813 9814 9815 9816 9817 9818 9819 9820


/**
 * qemuDomainGetMonitor:
 * @vm: domain object
 *
 * Returns the monitor pointer corresponding to the domain object @vm.
 */
qemuMonitorPtr
qemuDomainGetMonitor(virDomainObjPtr vm)
{
    return ((qemuDomainObjPrivatePtr) vm->privateData)->mon;
}
9821 9822 9823 9824 9825 9826 9827


/**
 * qemuDomainSupportsBlockJobs:
 * @vm: domain object
 *
 * Returns -1 in case when qemu does not support block jobs at all. Otherwise
9828
 * returns 0.
9829 9830
 */
int
9831
qemuDomainSupportsBlockJobs(virDomainObjPtr vm)
9832 9833
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
9834
    bool asynchronous = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKJOB_ASYNC);
9835

9836
    if (!asynchronous) {
9837 9838 9839 9840 9841 9842 9843
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("block jobs not supported with this QEMU binary"));
        return -1;
    }

    return 0;
}
9844 9845 9846 9847 9848 9849 9850 9851 9852


/**
 * 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.
 */
9853
virDomainChrDefPtr
9854 9855 9856 9857 9858 9859 9860 9861 9862 9863
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;

9864 9865
        if (STREQ_NULLABLE(channel->target.name, "org.qemu.guest_agent.0"))
            return channel;
9866 9867
    }

9868
    return NULL;
9869
}
9870 9871 9872


bool
9873
qemuDomainIsQ35(const virDomainDef *def)
9874
{
9875
    return qemuDomainMachineIsQ35(def->os.machine);
9876 9877 9878 9879
}


bool
9880
qemuDomainMachineIsQ35(const char *machine)
9881
{
9882 9883
    return (STRPREFIX(machine, "pc-q35") ||
            STREQ(machine, "q35"));
9884
}
9885 9886


9887
bool
9888 9889 9890 9891 9892 9893 9894 9895 9896 9897 9898 9899 9900 9901 9902 9903 9904 9905 9906
qemuDomainIsI440FX(const virDomainDef *def)
{
    return qemuDomainMachineIsI440FX(def->os.machine);
}


bool
qemuDomainMachineIsI440FX(const char *machine)
{
    return (STREQ(machine, "pc") ||
            STRPREFIX(machine, "pc-0.") ||
            STRPREFIX(machine, "pc-1.") ||
            STRPREFIX(machine, "pc-i440") ||
            STRPREFIX(machine, "rhel"));
}


bool
qemuDomainHasPCIRoot(const virDomainDef *def)
9907 9908 9909 9910 9911 9912 9913 9914 9915 9916 9917 9918 9919 9920
{
    int root = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

    if (root < 0)
        return false;

    if (def->controllers[root]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)
        return false;

    return true;
}


bool
9921
qemuDomainHasPCIeRoot(const virDomainDef *def)
9922 9923 9924 9925 9926 9927 9928 9929 9930 9931 9932 9933 9934
{
    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;
}


9935
bool
9936
qemuDomainNeedsFDC(const virDomainDef *def)
9937
{
9938 9939 9940 9941 9942 9943 9944 9945
    return qemuDomainMachineNeedsFDC(def->os.machine);
}


bool
qemuDomainMachineNeedsFDC(const char *machine)
{
    const char *p = STRSKIP(machine, "pc-q35-");
9946 9947 9948

    if (p) {
        if (STRPREFIX(p, "1.") ||
9949 9950 9951 9952
            STREQ(p, "2.0") ||
            STREQ(p, "2.1") ||
            STREQ(p, "2.2") ||
            STREQ(p, "2.3"))
9953 9954 9955 9956 9957 9958 9959
            return false;
        return true;
    }
    return false;
}


9960
bool
9961
qemuDomainIsS390CCW(const virDomainDef *def)
9962
{
9963
    return qemuDomainMachineIsS390CCW(def->os.machine);
9964 9965
}

9966

J
Ján Tomko 已提交
9967
bool
9968
qemuDomainMachineIsS390CCW(const char *machine)
J
Ján Tomko 已提交
9969
{
9970 9971 9972 9973 9974
    return STRPREFIX(machine, "s390-ccw");
}


bool
9975
qemuDomainIsARMVirt(const virDomainDef *def)
9976
{
9977
    return qemuDomainMachineIsARMVirt(def->os.machine, def->os.arch);
9978 9979 9980 9981
}


bool
9982 9983
qemuDomainMachineIsARMVirt(const char *machine,
                           const virArch arch)
9984
{
S
Stefan Schallenberg 已提交
9985 9986
    if (arch != VIR_ARCH_ARMV6L &&
        arch != VIR_ARCH_ARMV7L &&
9987
        arch != VIR_ARCH_AARCH64)
9988 9989
        return false;

9990 9991
    if (STRNEQ(machine, "virt") &&
        !STRPREFIX(machine, "virt-"))
9992 9993 9994
        return false;

    return true;
J
Ján Tomko 已提交
9995 9996 9997
}


9998 9999 10000 10001 10002 10003 10004 10005 10006 10007 10008 10009 10010 10011 10012 10013 10014 10015 10016 10017 10018 10019
bool
qemuDomainIsRISCVVirt(const virDomainDef *def)
{
    return qemuDomainMachineIsRISCVVirt(def->os.machine, def->os.arch);
}


bool
qemuDomainMachineIsRISCVVirt(const char *machine,
                             const virArch arch)
{
    if (!ARCH_IS_RISCV(arch))
        return false;

    if (STRNEQ(machine, "virt") &&
        !STRPREFIX(machine, "virt-"))
        return false;

    return true;
}


10020
bool
10021
qemuDomainIsPSeries(const virDomainDef *def)
10022
{
10023 10024 10025 10026 10027 10028 10029 10030 10031
    return qemuDomainMachineIsPSeries(def->os.machine, def->os.arch);
}


bool
qemuDomainMachineIsPSeries(const char *machine,
                           const virArch arch)
{
    if (!ARCH_IS_PPC64(arch))
10032 10033
        return false;

10034 10035
    if (STRNEQ(machine, "pseries") &&
        !STRPREFIX(machine, "pseries-"))
10036 10037 10038 10039 10040 10041
        return false;

    return true;
}


10042 10043 10044 10045 10046 10047 10048 10049 10050 10051 10052 10053 10054 10055 10056 10057 10058 10059 10060 10061 10062 10063 10064 10065 10066 10067 10068 10069 10070 10071 10072 10073 10074 10075 10076 10077 10078 10079 10080
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
10081
    case VIR_DOMAIN_MEMORY_MODEL_NVDIMM:
10082 10083 10084 10085 10086 10087 10088 10089
        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;
        }

10090 10091 10092
        if (virDomainNumaGetNodeCount(def->numa) != 0) {
            if (mem->targetNode == -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
Y
Yuri Chornoivan 已提交
10093
                               _("target NUMA node needs to be specified for "
10094 10095 10096
                                 "memory device"));
                return -1;
            }
10097 10098
        }

10099 10100 10101 10102 10103 10104 10105 10106 10107 10108 10109 10110 10111 10112 10113 10114 10115 10116 10117 10118 10119 10120 10121 10122
        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;
}


10123 10124 10125 10126 10127 10128 10129 10130 10131 10132 10133 10134 10135 10136 10137 10138 10139 10140 10141 10142 10143
/**
 * 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
10144 10145
    bool needPCDimmCap = false;
    bool needNvdimmCap = false;
10146 10147 10148 10149 10150 10151 10152
    size_t i;

    hotplugSpace = def->mem.max_memory - virDomainDefGetMemoryInitial(def);

    if (mem) {
        nmems++;
        hotplugMemory = mem->size;
10153 10154 10155

        if (qemuDomainDefValidateMemoryHotplugDevice(mem, def) < 0)
            return -1;
10156 10157 10158 10159 10160 10161 10162 10163 10164 10165 10166 10167 10168
    }

    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;
    }

10169 10170 10171 10172 10173 10174 10175 10176 10177 10178 10179
    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;
        }
10180 10181 10182 10183 10184 10185 10186 10187 10188
    }

    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;
    }

10189
    for (i = 0; i < def->nmems; i++) {
10190 10191
        hotplugMemory += def->mems[i]->size;

M
Michal Privoznik 已提交
10192 10193 10194 10195 10196 10197 10198 10199 10200 10201 10202 10203 10204 10205
        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;
        }

10206 10207 10208 10209 10210 10211
        /* 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
10212 10213 10214 10215 10216 10217 10218 10219 10220 10221 10222 10223 10224 10225
    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;
    }

10226 10227 10228 10229 10230 10231 10232 10233 10234 10235
    if (hotplugMemory > hotplugSpace) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory device total size exceeds hotplug space"));
        return -1;
    }

    return 0;
}


10236
bool
10237 10238 10239 10240 10241 10242 10243 10244
qemuDomainHasBuiltinIDE(const virDomainDef *def)
{
    return qemuDomainMachineHasBuiltinIDE(def->os.machine);
}


bool
qemuDomainMachineHasBuiltinIDE(const char *machine)
10245
{
10246 10247 10248 10249
    return qemuDomainMachineIsI440FX(machine) ||
        STREQ(machine, "malta") ||
        STREQ(machine, "sun4u") ||
        STREQ(machine, "g3beige");
10250 10251 10252
}


10253 10254 10255 10256 10257 10258 10259 10260 10261 10262 10263 10264 10265 10266 10267 10268 10269 10270 10271 10272 10273 10274 10275 10276
/**
 * qemuDomainUpdateCurrentMemorySize:
 *
 * Updates the current balloon size from the monitor if necessary. In case when
 * the balloon is not present for the domain, the function recalculates the
 * maximum size to reflect possible changes.
 *
 * Returns 0 on success and updates vm->def->mem.cur_balloon if necessary, -1 on
 * error and reports libvirt error.
 */
int
qemuDomainUpdateCurrentMemorySize(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    unsigned long long balloon;
    int ret = -1;

    /* inactive domain doesn't need size update */
    if (!virDomainObjIsActive(vm))
        return 0;

    /* if no balloning is available, the current size equals to the current
     * full memory size */
10277
    if (!virDomainDefHasMemballoon(vm->def)) {
10278
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
10279 10280 10281 10282 10283 10284 10285 10286 10287 10288 10289 10290 10291 10292 10293
        return 0;
    }

    /* current size is always automagically updated via the event */
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BALLOON_EVENT))
        return 0;

    /* here we need to ask the monitor */

    /* Don't delay if someone's using the monitor, just use existing most
     * recent data instead */
    if (qemuDomainJobAllowed(priv, QEMU_JOB_QUERY)) {
        if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_QUERY) < 0)
            return -1;

10294
        if (virDomainObjCheckActive(vm) < 0)
10295 10296 10297 10298 10299 10300 10301 10302 10303 10304 10305 10306 10307
            goto endjob;

        qemuDomainObjEnterMonitor(driver, vm);
        ret = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
        if (qemuDomainObjExitMonitor(driver, vm) < 0)
            ret = -1;

 endjob:
        qemuDomainObjEndJob(driver, vm);

        if (ret < 0)
            return -1;

10308 10309
        vm->def->mem.cur_balloon = balloon;
    }
10310 10311 10312

    return 0;
}
10313 10314 10315


/**
10316
 * qemuDomainGetMemLockLimitBytes:
10317 10318
 * @def: domain definition
 *
10319 10320 10321 10322 10323 10324
 * 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
10325 10326
 */
unsigned long long
10327
qemuDomainGetMemLockLimitBytes(virDomainDefPtr def)
10328
{
10329 10330
    unsigned long long memKB = 0;
    size_t i;
10331

10332 10333 10334 10335 10336 10337
    /* prefer the hard limit */
    if (virMemoryLimitIsSet(def->mem.hard_limit)) {
        memKB = def->mem.hard_limit;
        goto done;
    }

10338 10339 10340 10341 10342 10343 10344
    /* 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;
10345 10346

    if (ARCH_IS_PPC64(def->os.arch) && def->virtType == VIR_DOMAIN_VIRT_KVM) {
10347 10348 10349 10350
        unsigned long long maxMemory;
        unsigned long long memory;
        unsigned long long baseLimit;
        unsigned long long passthroughLimit;
10351
        size_t nPCIHostBridges = 0;
10352 10353
        bool usesVFIO = false;

10354 10355 10356
        for (i = 0; i < def->ncontrollers; i++) {
            virDomainControllerDefPtr cont = def->controllers[i];

10357
            if (!virDomainControllerIsPSeriesPHB(cont))
10358 10359 10360 10361
                continue;

            nPCIHostBridges++;
        }
10362 10363 10364 10365 10366 10367 10368 10369 10370 10371 10372 10373

        for (i = 0; i < def->nhostdevs; i++) {
            virDomainHostdevDefPtr dev = def->hostdevs[i];

            if (dev->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS &&
                dev->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI &&
                dev->source.subsys.u.pci.backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO) {
                usesVFIO = true;
                break;
            }
        }

10374
        memory = virDomainDefGetMemoryTotal(def);
10375 10376 10377 10378 10379 10380 10381 10382 10383 10384 10385 10386 10387 10388 10389 10390 10391 10392 10393 10394 10395 10396 10397 10398 10399 10400 10401 10402 10403 10404 10405 10406 10407 10408 10409 10410 10411 10412 10413 10414 10415 10416 10417 10418 10419 10420 10421 10422 10423 10424 10425 10426 10427

        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;

        /* 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) */
        passthroughLimit = MAX(2 * 1024 * 1024 * nPCIHostBridges,
                               memory +
                               memory / 512 * nPCIHostBridges + 8192);

        if (usesVFIO)
            memKB = baseLimit + passthroughLimit;
        else
            memKB = baseLimit;

        goto done;
    }

10428 10429 10430 10431 10432 10433 10434 10435 10436 10437 10438 10439 10440 10441 10442 10443 10444 10445
    /* 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.
     */
10446
    for (i = 0; i < def->nhostdevs; i++) {
10447
        virDomainHostdevSubsysPtr subsys = &def->hostdevs[i]->source.subsys;
10448

10449 10450 10451
        if (def->hostdevs[i]->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS &&
            (subsys->type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV ||
             (subsys->type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI &&
10452 10453 10454 10455
              subsys->u.pci.backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO))) {
            memKB = virDomainDefGetMemoryTotal(def) + 1024 * 1024;
            goto done;
        }
10456 10457
    }

10458 10459
 done:
    return memKB << 10;
10460
}
10461

10462

10463 10464 10465 10466 10467 10468 10469
/**
 * qemuDomainAdjustMaxMemLock:
 * @vm: domain
 *
 * Adjust the memory locking limit for the QEMU process associated to @vm, in
 * order to comply with VFIO or architecture requirements.
 *
10470 10471 10472 10473
 * 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.
10474 10475 10476 10477 10478 10479 10480 10481 10482
 *
 * Returns: 0 on success, <0 on failure
 */
int
qemuDomainAdjustMaxMemLock(virDomainObjPtr vm)
{
    unsigned long long bytes = 0;
    int ret = -1;

10483 10484 10485
    bytes = qemuDomainGetMemLockLimitBytes(vm->def);

    if (bytes) {
10486 10487 10488 10489 10490 10491 10492 10493 10494 10495 10496 10497 10498 10499
        /* 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;
    }
10500 10501 10502 10503 10504 10505 10506 10507 10508 10509

    /* 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;
}
10510 10511 10512 10513 10514 10515 10516 10517 10518 10519

/**
 * qemuDomainHasVcpuPids:
 * @vm: Domain object
 *
 * Returns true if we were able to successfully detect vCPU pids for the VM.
 */
bool
qemuDomainHasVcpuPids(virDomainObjPtr vm)
{
10520 10521 10522 10523 10524 10525
    size_t i;
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    virDomainVcpuDefPtr vcpu;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
10526

10527 10528 10529 10530 10531
        if (QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid > 0)
            return true;
    }

    return false;
10532
}
10533 10534 10535 10536 10537 10538 10539 10540 10541 10542 10543


/**
 * 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,
10544
                     unsigned int vcpuid)
10545
{
10546 10547
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
    return QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid;
10548
}
10549 10550


10551 10552 10553 10554 10555 10556 10557 10558 10559 10560 10561 10562 10563 10564 10565 10566 10567 10568 10569 10570 10571 10572 10573 10574 10575 10576 10577 10578 10579 10580 10581 10582 10583 10584 10585 10586 10587 10588 10589 10590 10591 10592
/**
 * 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;
}


10593 10594 10595 10596 10597 10598 10599 10600 10601
bool
qemuDomainSupportsNewVcpuHotplug(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
}


10602
/**
10603
 * qemuDomainRefreshVcpuInfo:
10604 10605 10606
 * @driver: qemu driver data
 * @vm: domain object
 * @asyncJob: current asynchronous job type
10607
 * @state: refresh vcpu state
10608
 *
10609 10610
 * Updates vCPU information private data of @vm. Due to historical reasons this
 * function returns success even if some data were not reported by qemu.
10611
 *
10612 10613
 * If @state is true, the vcpu state is refreshed as reported by the monitor.
 *
10614
 * Returns 0 on success and -1 on fatal error.
10615 10616
 */
int
10617 10618
qemuDomainRefreshVcpuInfo(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
10619 10620
                          int asyncJob,
                          bool state)
10621
{
10622
    virDomainVcpuDefPtr vcpu;
10623 10624
    qemuDomainVcpuPrivatePtr vcpupriv;
    qemuMonitorCPUInfoPtr info = NULL;
10625
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
10626
    size_t i, j;
10627
    bool hotplug;
10628
    bool fast;
10629
    bool validTIDs = true;
10630
    int rc;
10631
    int ret = -1;
10632

10633
    hotplug = qemuDomainSupportsNewVcpuHotplug(vm);
10634 10635
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
10636

10637 10638
    VIR_DEBUG("Maxvcpus %zu hotplug %d fast query %d", maxvcpus, hotplug, fast);

10639 10640
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
10641

10642 10643
    rc = qemuMonitorGetCPUInfo(qemuDomainGetMonitor(vm), &info, maxvcpus,
                               hotplug, fast);
10644

10645
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
10646 10647
        goto cleanup;

10648
    if (rc < 0)
10649 10650
        goto cleanup;

10651 10652 10653 10654 10655 10656 10657 10658 10659 10660 10661 10662 10663 10664 10665 10666 10667 10668 10669 10670 10671 10672 10673 10674 10675 10676 10677 10678 10679 10680 10681 10682 10683 10684 10685 10686 10687 10688 10689 10690 10691 10692 10693 10694 10695 10696
    /*
     * 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++) {
            if (info[i].tid == info[j].tid) {
                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);
10697 10698
    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
10699
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
10700

10701
        if (validTIDs)
10702 10703 10704 10705 10706
            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;
10707
        vcpupriv->node_id = info[i].node_id;
10708 10709 10710 10711 10712 10713
        vcpupriv->vcpus = info[i].vcpus;
        VIR_FREE(vcpupriv->type);
        VIR_STEAL_PTR(vcpupriv->type, info[i].type);
        VIR_FREE(vcpupriv->alias);
        VIR_STEAL_PTR(vcpupriv->alias, info[i].alias);
        vcpupriv->enable_id = info[i].id;
10714
        vcpupriv->qemu_id = info[i].qemu_id;
10715

10716
        if (hotplug && state) {
10717 10718 10719 10720 10721
            vcpu->online = info[i].online;
            if (info[i].hotpluggable)
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_YES;
            else
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_NO;
10722
        }
10723 10724
    }

10725
    ret = 0;
10726 10727

 cleanup:
10728
    qemuMonitorCPUInfoFree(info, maxvcpus);
10729
    return ret;
10730
}
10731

10732 10733 10734 10735 10736 10737 10738 10739 10740 10741 10742 10743 10744 10745 10746 10747 10748 10749 10750 10751 10752 10753 10754 10755 10756 10757 10758 10759 10760 10761 10762
/**
 * 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;
10763
    qemuDomainVcpuPrivatePtr vcpupriv;
10764
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
10765
    virBitmapPtr haltedmap = NULL;
10766 10767
    size_t i;
    int ret = -1;
10768
    bool fast;
10769 10770 10771 10772 10773

    /* Not supported currently for TCG, see qemuDomainRefreshVcpuInfo */
    if (vm->def->virtType == VIR_DOMAIN_VIRT_QEMU)
        return 0;

10774
    /* The halted state is interresting only on s390(x). On other platforms
10775 10776 10777 10778 10779 10780 10781
     * 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))
10782 10783
        return 0;

10784 10785 10786
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;

10787 10788 10789 10790
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
    haltedmap = qemuMonitorGetCpuHalted(qemuDomainGetMonitor(vm), maxvcpus,
                                        fast);
10791
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || !haltedmap)
10792 10793 10794 10795
        goto cleanup;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
10796
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
10797 10798
        vcpupriv->halted = virTristateBoolFromBool(virBitmapIsBitSet(haltedmap,
                                                                     vcpupriv->qemu_id));
10799 10800 10801 10802 10803
    }

    ret = 0;

 cleanup:
10804
    virBitmapFree(haltedmap);
10805 10806
    return ret;
}
10807 10808 10809 10810 10811 10812

bool
qemuDomainSupportsNicdev(virDomainDefPtr def,
                         virDomainNetDefPtr net)
{
    /* non-virtio ARM nics require legacy -net nic */
S
Stefan Schallenberg 已提交
10813 10814
    if (((def->os.arch == VIR_ARCH_ARMV6L) ||
        (def->os.arch == VIR_ARCH_ARMV7L) ||
10815 10816 10817 10818 10819 10820 10821 10822
        (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;
}

10823 10824 10825 10826
bool
qemuDomainNetSupportsMTU(virDomainNetType type)
{
    switch (type) {
10827 10828
    case VIR_DOMAIN_NET_TYPE_NETWORK:
    case VIR_DOMAIN_NET_TYPE_BRIDGE:
10829 10830
    case VIR_DOMAIN_NET_TYPE_ETHERNET:
    case VIR_DOMAIN_NET_TYPE_VHOSTUSER:
10831 10832
        return true;
    case VIR_DOMAIN_NET_TYPE_USER:
10833 10834 10835 10836 10837 10838 10839 10840 10841 10842 10843 10844
    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
10845

P
Peter Krempa 已提交
10846 10847 10848 10849 10850 10851 10852 10853 10854 10855 10856 10857 10858 10859 10860

virDomainDiskDefPtr
qemuDomainDiskByName(virDomainDefPtr def,
                     const char *name)
{
    virDomainDiskDefPtr ret;

    if (!(ret = virDomainDiskByName(def, name, true))) {
        virReportError(VIR_ERR_INVALID_ARG, "%s",
                       _("No device found for specified path"));
        return NULL;
    }

    return ret;
}
10861 10862 10863 10864 10865 10866 10867 10868 10869 10870 10871 10872 10873 10874 10875 10876 10877 10878 10879 10880 10881 10882 10883 10884 10885 10886 10887 10888 10889 10890 10891


/**
 * 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;
}
10892 10893 10894 10895 10896 10897


int
qemuDomainPrepareChannel(virDomainChrDefPtr channel,
                         const char *domainChannelTargetDir)
{
S
Scott Garfinkle 已提交
10898 10899 10900 10901 10902 10903
    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) {
10904
        if (virAsprintf(&channel->source->data.nix.path,
10905
                        "%s/%s", domainChannelTargetDir,
S
Scott Garfinkle 已提交
10906 10907
                        channel->target.name) < 0)
            return -1;
10908 10909
    } else {
        /* Generate a unique name */
S
Scott Garfinkle 已提交
10910 10911 10912 10913 10914 10915
        if (virAsprintf(&channel->source->data.nix.path,
                        "%s/vioser-%02d-%02d-%02d.sock",
                        domainChannelTargetDir,
                        channel->info.addr.vioserial.controller,
                        channel->info.addr.vioserial.bus,
                        channel->info.addr.vioserial.port) < 0)
10916 10917 10918 10919 10920
            return -1;
    }

    return 0;
}
10921 10922


10923
/* qemuDomainPrepareChardevSourceTLS:
10924 10925 10926 10927 10928 10929 10930 10931 10932 10933 10934 10935 10936 10937 10938 10939 10940
 * @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;
10941
            source->data.tcp.tlsFromConfig = true;
10942 10943 10944 10945 10946
        }
    }
}


10947
/* qemuDomainPrepareChardevSource:
10948
 * @def: live domain definition
10949
 * @cfg: driver configuration
10950 10951 10952 10953 10954 10955
 *
 * Iterate through all devices that use virDomainChrSourceDefPtr as host
 * interface part.
 */
void
qemuDomainPrepareChardevSource(virDomainDefPtr def,
10956
                               virQEMUDriverConfigPtr cfg)
10957 10958 10959 10960 10961 10962 10963 10964 10965 10966 10967 10968 10969 10970 10971 10972 10973 10974 10975 10976 10977 10978 10979 10980 10981 10982 10983 10984 10985
{
    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);
}


10986 10987 10988 10989 10990 10991 10992 10993 10994 10995 10996 10997 10998 10999 11000 11001 11002 11003 11004 11005 11006 11007 11008 11009
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;
    }

    if (src->haveTLS == VIR_TRISTATE_BOOL_YES) {
        if (VIR_STRDUP(src->tlsCertdir, cfg->vxhsTLSx509certdir) < 0)
            return -1;
    }

    return 0;
}


11010 11011 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
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;
        }

        if (VIR_STRDUP(src->tlsCertdir, cfg->nbdTLSx509certdir) < 0)
            return -1;
    }

    return 0;
}


11038
/* qemuProcessPrepareStorageSourceTLS:
11039
 * @source: source for a disk
11040
 * @cfg: driver configuration
11041
 * @parentAlias: alias of the parent device
11042 11043 11044 11045 11046 11047 11048
 *
 * 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
 */
11049
static int
11050
qemuDomainPrepareStorageSourceTLS(virStorageSourcePtr src,
11051
                                  virQEMUDriverConfigPtr cfg,
11052 11053
                                  const char *parentAlias,
                                  virQEMUCapsPtr qemuCaps)
11054
{
11055 11056 11057 11058 11059 11060 11061 11062
    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;
11063

11064
    case VIR_STORAGE_NET_PROTOCOL_NBD:
11065 11066 11067 11068
        if (qemuProcessPrepareStorageSourceTLSNBD(src, cfg, qemuCaps) < 0)
            return -1;
        break;

11069 11070 11071 11072 11073 11074 11075 11076 11077 11078
    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:
11079 11080 11081 11082 11083 11084
        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;
        }
11085 11086 11087 11088 11089 11090
        break;

    case VIR_STORAGE_NET_PROTOCOL_NONE:
    case VIR_STORAGE_NET_PROTOCOL_LAST:
    default:
        virReportEnumRangeError(virStorageNetProtocol, src->protocol);
11091
        return -1;
11092
    }
11093

11094 11095 11096 11097
    if (src->haveTLS == VIR_TRISTATE_BOOL_YES &&
        !(src->tlsAlias = qemuAliasTLSObjFromSrcAlias(parentAlias)))
        return -1;

11098 11099 11100 11101
    return 0;
}


11102 11103 11104 11105 11106 11107 11108 11109 11110 11111 11112 11113 11114
int
qemuDomainPrepareShmemChardev(virDomainShmemDefPtr shmem)
{
    if (!shmem->server.enabled ||
        shmem->server.chr.data.nix.path)
        return 0;

    return virAsprintf(&shmem->server.chr.data.nix.path,
                       "/var/lib/libvirt/shmem-%s-sock",
                       shmem->name);
}


11115 11116 11117 11118 11119 11120 11121 11122 11123 11124 11125 11126 11127 11128 11129 11130 11131 11132 11133 11134 11135 11136 11137 11138 11139 11140 11141 11142 11143 11144 11145 11146 11147
/**
 * 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);
}
11148 11149 11150 11151 11152 11153 11154 11155 11156 11157 11158 11159 11160 11161 11162 11163 11164 11165 11166 11167 11168 11169 11170 11171 11172 11173 11174 11175 11176 11177 11178 11179 11180 11181


/**
 * 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;
        }
    }
}
11182 11183 11184 11185 11186 11187 11188 11189 11190 11191 11192 11193 11194 11195 11196 11197 11198 11199 11200 11201


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;
}
11202 11203 11204 11205 11206 11207 11208 11209 11210 11211 11212 11213


bool
qemuDomainSupportsVideoVga(virDomainVideoDefPtr video,
                           virQEMUCapsPtr qemuCaps)
{
    if (video->type == VIR_DOMAIN_VIDEO_TYPE_VIRTIO &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_VGA))
        return false;

    return true;
}
11214 11215


11216 11217
/**
 * qemuDomainGetHostdevPath:
11218
 * @def: domain definition
11219
 * @dev: host device definition
11220
 * @teardown: true if device will be removed
11221
 * @npaths: number of items in @path and @perms arrays
11222 11223 11224
 * @path: resulting path to @dev
 * @perms: Optional pointer to VIR_CGROUP_DEVICE_* perms
 *
11225 11226 11227 11228 11229 11230
 * 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.
11231 11232 11233 11234
 *
 * Returns 0 on success, -1 otherwise.
 */
int
11235 11236 11237
qemuDomainGetHostdevPath(virDomainDefPtr def,
                         virDomainHostdevDefPtr dev,
                         bool teardown,
11238 11239 11240
                         size_t *npaths,
                         char ***path,
                         int **perms)
11241 11242 11243 11244 11245 11246
{
    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;
11247
    virDomainHostdevSubsysMediatedDevPtr mdevsrc = &dev->source.subsys.u.mdev;
11248 11249 11250 11251 11252 11253
    virPCIDevicePtr pci = NULL;
    virUSBDevicePtr usb = NULL;
    virSCSIDevicePtr scsi = NULL;
    virSCSIVHostDevicePtr host = NULL;
    char *tmpPath = NULL;
    bool freeTmpPath = false;
11254 11255 11256 11257 11258
    bool includeVFIO = false;
    char **tmpPaths = NULL;
    int *tmpPerms = NULL;
    size_t i, tmpNpaths = 0;
    int perm = 0;
11259

11260
    *npaths = 0;
11261 11262 11263

    switch ((virDomainHostdevMode) dev->mode) {
    case VIR_DOMAIN_HOSTDEV_MODE_SUBSYS:
11264
        switch ((virDomainHostdevSubsysType)dev->source.subsys.type) {
11265 11266 11267 11268 11269 11270 11271 11272 11273 11274 11275 11276
        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;
                freeTmpPath = true;
11277 11278

                perm = VIR_CGROUP_DEVICE_RW;
11279 11280 11281 11282 11283 11284 11285 11286 11287 11288 11289 11290 11291 11292 11293
                if (teardown) {
                    size_t nvfios = 0;
                    for (i = 0; i < def->nhostdevs; i++) {
                        virDomainHostdevDefPtr tmp = def->hostdevs[i];
                        if (tmp->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS &&
                            tmp->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI &&
                            tmp->source.subsys.u.pci.backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO)
                            nvfios++;
                    }

                    if (nvfios == 0)
                        includeVFIO = true;
                } else {
                    includeVFIO = true;
                }
11294 11295 11296 11297 11298 11299 11300 11301 11302 11303 11304 11305
            }
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
            if (dev->missing)
                break;
            usb = virUSBDeviceNew(usbsrc->bus,
                                  usbsrc->device,
                                  NULL);
            if (!usb)
                goto cleanup;

11306
            if (!(tmpPath = (char *)virUSBDeviceGetPath(usb)))
11307
                goto cleanup;
11308
            perm = VIR_CGROUP_DEVICE_RW;
11309 11310 11311 11312 11313
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI:
            if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI) {
                virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
11314
                VIR_DEBUG("Not updating /dev for hostdev iSCSI path '%s'", iscsisrc->src->path);
11315 11316 11317 11318 11319 11320 11321 11322 11323 11324 11325 11326 11327
            } 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;

11328
                if (!(tmpPath = (char *)virSCSIDeviceGetPath(scsi)))
11329
                    goto cleanup;
11330 11331
                perm = virSCSIDeviceGetReadonly(scsi) ?
                    VIR_CGROUP_DEVICE_READ : VIR_CGROUP_DEVICE_RW;
11332 11333 11334 11335 11336 11337 11338 11339 11340
            }
            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;

11341
                if (!(tmpPath = (char *)virSCSIVHostDeviceGetPath(host)))
11342
                    goto cleanup;
11343
                perm = VIR_CGROUP_DEVICE_RW;
11344 11345 11346 11347
            }
            break;
        }

11348
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
11349
            if (!(tmpPath = virMediatedDeviceGetIOMMUGroupDev(mdevsrc->uuidstr)))
11350 11351 11352 11353 11354 11355
                goto cleanup;

            freeTmpPath = true;
            includeVFIO = true;
            perm = VIR_CGROUP_DEVICE_RW;
            break;
11356 11357 11358 11359 11360 11361 11362 11363 11364 11365 11366
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
            break;
        }
        break;

    case VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES:
    case VIR_DOMAIN_HOSTDEV_MODE_LAST:
        /* nada */
        break;
    }

11367 11368 11369 11370 11371
    if (tmpPath) {
        size_t toAlloc = 1;

        if (includeVFIO)
            toAlloc = 2;
11372

11373 11374 11375 11376 11377 11378 11379 11380 11381 11382 11383 11384 11385 11386 11387 11388 11389 11390 11391 11392 11393 11394
        if (VIR_ALLOC_N(tmpPaths, toAlloc) < 0 ||
            VIR_ALLOC_N(tmpPerms, toAlloc) < 0 ||
            VIR_STRDUP(tmpPaths[0], tmpPath) < 0)
            goto cleanup;
        tmpNpaths = toAlloc;
        tmpPerms[0] = perm;

        if (includeVFIO) {
            if (VIR_STRDUP(tmpPaths[1], DEV_VFIO) < 0)
                goto cleanup;
            tmpPerms[1] = VIR_CGROUP_DEVICE_RW;
        }
    }

    *npaths = tmpNpaths;
    tmpNpaths = 0;
    *path = tmpPaths;
    tmpPaths = NULL;
    if (perms) {
        *perms = tmpPerms;
        tmpPerms = NULL;
    }
11395 11396
    ret = 0;
 cleanup:
11397
    virStringListFreeCount(tmpPaths, tmpNpaths);
11398
    VIR_FREE(tmpPerms);
11399 11400 11401 11402 11403 11404 11405 11406 11407 11408
    virPCIDeviceFree(pci);
    virUSBDeviceFree(usb);
    virSCSIDeviceFree(scsi);
    virSCSIVHostDeviceFree(host);
    if (freeTmpPath)
        VIR_FREE(tmpPath);
    return ret;
}


11409 11410 11411 11412
/**
 * qemuDomainGetPreservedMountPath:
 * @cfg: driver configuration data
 * @vm: domain object
11413
 * @mountpoint: mount point path to convert
11414
 *
11415
 * For given @mountpoint return new path where the mount point
11416 11417 11418 11419 11420 11421 11422 11423
 * 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,
11424
                                const char *mountpoint)
11425 11426 11427
{
    char *path = NULL;
    char *tmp;
11428
    const char *suffix = mountpoint + strlen(DEVPREFIX);
11429
    char *domname = virDomainDefGetShortName(vm->def);
11430 11431
    size_t off;

11432 11433 11434
    if (!domname)
        return NULL;

11435
    if (STREQ(mountpoint, "/dev"))
11436 11437 11438
        suffix = "dev";

    if (virAsprintf(&path, "%s/%s.%s",
11439 11440
                    cfg->stateDir, domname, suffix) < 0)
        goto cleanup;
11441

11442
    /* Now consider that @mountpoint is "/dev/blah/blah2".
11443 11444 11445 11446 11447 11448 11449 11450 11451 11452 11453 11454 11455
     * @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++;
    }

11456 11457
 cleanup:
    VIR_FREE(domname);
11458 11459 11460 11461
    return path;
}


11462 11463 11464 11465 11466 11467 11468 11469 11470 11471 11472 11473
/**
 * 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
11474
qemuDomainGetPreservedMounts(virQEMUDriverConfigPtr cfg,
11475 11476 11477 11478 11479 11480
                             virDomainObjPtr vm,
                             char ***devPath,
                             char ***devSavePath,
                             size_t *ndevPath)
{
    char **paths = NULL, **mounts = NULL;
11481
    size_t i, j, nmounts;
11482 11483 11484 11485 11486 11487 11488 11489 11490 11491 11492

    if (virFileGetMountSubtree(PROC_MOUNTS, "/dev",
                               &mounts, &nmounts) < 0)
        goto error;

    if (!nmounts) {
        if (ndevPath)
            *ndevPath = 0;
        return 0;
    }

11493 11494 11495 11496 11497 11498 11499 11500 11501 11502 11503 11504
    /* 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) {
11505 11506
            char *c = STRSKIP(mounts[j], mounts[i]);

11507
            if (c && (*c == '/' || *c == '\0')) {
11508 11509 11510 11511 11512 11513 11514 11515
                VIR_DEBUG("Dropping path %s because of %s", mounts[j], mounts[i]);
                VIR_DELETE_ELEMENT(mounts, j, nmounts);
            } else {
                j++;
            }
        }
    }

11516 11517 11518 11519
    if (VIR_ALLOC_N(paths, nmounts) < 0)
        goto error;

    for (i = 0; i < nmounts; i++) {
11520
        if (!(paths[i] = qemuDomainGetPreservedMountPath(cfg, vm, mounts[i])))
11521 11522 11523 11524 11525 11526 11527 11528 11529 11530 11531 11532 11533 11534 11535 11536 11537 11538 11539 11540 11541 11542 11543 11544 11545
            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;
}


11546 11547
struct qemuDomainCreateDeviceData {
    const char *path;     /* Path to temp new /dev location */
11548 11549
    char * const *devMountsPath;
    size_t ndevMountsPath;
11550 11551 11552
};


11553
static int
11554
qemuDomainCreateDeviceRecursive(const char *device,
11555
                                const struct qemuDomainCreateDeviceData *data,
11556 11557
                                bool allow_noent,
                                unsigned int ttl)
11558 11559
{
    char *devicePath = NULL;
11560
    char *target = NULL;
11561 11562
    struct stat sb;
    int ret = -1;
11563
    bool isLink = false;
11564
    bool isDev = false;
11565
    bool isReg = false;
11566
    bool isDir = false;
11567
    bool create = false;
11568 11569 11570
#ifdef WITH_SELINUX
    char *tcon = NULL;
#endif
11571

11572 11573 11574 11575 11576 11577 11578
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             device);
        return ret;
    }

11579
    if (lstat(device, &sb) < 0) {
11580 11581
        if (errno == ENOENT && allow_noent) {
            /* Ignore non-existent device. */
11582
            return 0;
11583
        }
11584 11585
        virReportSystemError(errno, _("Unable to stat %s"), device);
        return ret;
11586 11587
    }

11588
    isLink = S_ISLNK(sb.st_mode);
11589
    isDev = S_ISCHR(sb.st_mode) || S_ISBLK(sb.st_mode);
11590
    isReg = S_ISREG(sb.st_mode) || S_ISFIFO(sb.st_mode) || S_ISSOCK(sb.st_mode);
11591
    isDir = S_ISDIR(sb.st_mode);
11592 11593 11594 11595 11596 11597 11598 11599 11600 11601 11602 11603 11604 11605 11606 11607 11608

    /* 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.
     */
    if (STRPREFIX(device, DEVPREFIX)) {
11609
        size_t i;
11610

11611 11612 11613 11614 11615 11616 11617 11618 11619 11620 11621 11622 11623 11624 11625 11626 11627 11628 11629 11630 11631 11632 11633 11634 11635
        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. */
            if (virAsprintf(&devicePath, "%s/%s",
                            data->path, device + strlen(DEVPREFIX)) < 0)
                goto cleanup;

            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]);
11636
        }
11637 11638
    }

11639 11640 11641 11642 11643 11644 11645 11646 11647
    if (isLink) {
        /* We are dealing with a symlink. Create a dangling symlink and descend
         * down one level which hopefully creates the symlink's target. */
        if (virFileReadLink(device, &target) < 0) {
            virReportSystemError(errno,
                                 _("unable to resolve symlink %s"),
                                 device);
            goto cleanup;
        }
11648

11649 11650 11651 11652 11653 11654 11655 11656 11657
        if (create &&
            symlink(target, devicePath) < 0) {
            if (errno == EEXIST) {
                ret = 0;
            } else {
                virReportSystemError(errno,
                                     _("unable to create symlink %s"),
                                     devicePath);
            }
11658 11659 11660
            goto cleanup;
        }

11661 11662 11663 11664 11665 11666 11667 11668 11669 11670
        /* 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;
11671

11672 11673
            if (VIR_STRDUP(devTmp, device) < 0)
                goto cleanup;
11674

11675 11676 11677 11678 11679 11680 11681 11682 11683
            if ((c = strrchr(devTmp, '/')))
                *(c + 1) = '\0';

            if (virAsprintf(&tmp, "%s%s", devTmp, target) < 0) {
                VIR_FREE(devTmp);
                goto cleanup;
            }
            VIR_FREE(devTmp);
            VIR_FREE(target);
11684
            VIR_STEAL_PTR(target, tmp);
11685 11686
        }

11687
        if (qemuDomainCreateDeviceRecursive(target, data,
11688
                                            allow_noent, ttl - 1) < 0)
11689
            goto cleanup;
11690
    } else if (isDev) {
11691 11692 11693 11694 11695 11696 11697 11698 11699 11700
        if (create &&
            mknod(devicePath, sb.st_mode, sb.st_rdev) < 0) {
            if (errno == EEXIST) {
                ret = 0;
            } else {
                virReportSystemError(errno,
                                     _("Failed to make device %s"),
                                     devicePath);
            }
            goto cleanup;
11701
        }
11702
    } else if (isReg) {
11703
        if (create &&
11704
            virFileTouch(devicePath, sb.st_mode) < 0)
11705
            goto cleanup;
11706 11707
        /* Just create the file here so that code below sets
         * proper owner and mode. Bind mount only after that. */
11708 11709 11710 11711
    } else if (isDir) {
        if (create &&
            virFileMakePathWithMode(devicePath, sb.st_mode) < 0)
            goto cleanup;
11712 11713 11714 11715 11716
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       device, sb.st_mode);
        goto cleanup;
11717 11718 11719 11720
    }

    if (!create) {
        ret = 0;
11721 11722 11723
        goto cleanup;
    }

11724
    if (lchown(devicePath, sb.st_uid, sb.st_gid) < 0) {
11725 11726 11727 11728 11729 11730
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             devicePath);
        goto cleanup;
    }

11731 11732 11733 11734 11735 11736 11737 11738 11739
    /* 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;
    }

11740 11741 11742
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileCopyACLs(device, devicePath) < 0 &&
11743 11744 11745 11746 11747 11748 11749
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Failed to copy ACLs on device %s"),
                             devicePath);
        goto cleanup;
    }

11750
#ifdef WITH_SELINUX
11751
    if (lgetfilecon_raw(device, &tcon) < 0 &&
11752 11753 11754
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"),
11755
                             device);
11756 11757 11758 11759
        goto cleanup;
    }

    if (tcon &&
11760
        lsetfilecon_raw(devicePath, (VIR_SELINUX_CTX_CONST char *)tcon) < 0) {
11761 11762 11763 11764 11765 11766 11767 11768 11769 11770 11771
        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

11772
    /* Finish mount process started earlier. */
11773
    if ((isReg || isDir) &&
11774 11775 11776
        virFileBindMountDevice(device, devicePath) < 0)
        goto cleanup;

11777 11778
    ret = 0;
 cleanup:
11779
    VIR_FREE(target);
11780
    VIR_FREE(devicePath);
11781 11782 11783
#ifdef WITH_SELINUX
    freecon(tcon);
#endif
11784 11785 11786 11787
    return ret;
}


11788 11789
static int
qemuDomainCreateDevice(const char *device,
11790
                       const struct qemuDomainCreateDeviceData *data,
11791 11792 11793 11794
                       bool allow_noent)
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

11795
    return qemuDomainCreateDeviceRecursive(device, data,
11796 11797 11798
                                           allow_noent, symloop_max);
}

11799 11800

static int
11801
qemuDomainPopulateDevices(virQEMUDriverConfigPtr cfg,
11802
                          virDomainObjPtr vm ATTRIBUTE_UNUSED,
11803
                          const struct qemuDomainCreateDeviceData *data)
11804 11805 11806 11807 11808 11809 11810 11811 11812
{
    const char *const *devices = (const char *const *) cfg->cgroupDeviceACL;
    size_t i;
    int ret = -1;

    if (!devices)
        devices = defaultDeviceACL;

    for (i = 0; devices[i]; i++) {
11813
        if (qemuDomainCreateDevice(devices[i], data, true) < 0)
11814 11815 11816 11817 11818 11819 11820 11821 11822 11823
            goto cleanup;
    }

    ret = 0;
 cleanup:
    return ret;
}


static int
11824 11825
qemuDomainSetupDev(virQEMUDriverConfigPtr cfg,
                   virSecurityManagerPtr mgr,
11826
                   virDomainObjPtr vm,
11827
                   const struct qemuDomainCreateDeviceData *data)
11828 11829 11830 11831 11832 11833 11834
{
    char *mount_options = NULL;
    char *opts = NULL;
    int ret = -1;

    VIR_DEBUG("Setting up /dev/ for domain %s", vm->def->name);

11835
    mount_options = qemuSecurityGetMountOptions(mgr, vm->def);
11836 11837 11838 11839 11840 11841 11842 11843 11844 11845 11846 11847 11848

    if (!mount_options &&
        VIR_STRDUP(mount_options, "") < 0)
        goto cleanup;

    /*
     * tmpfs is limited to 64kb, since we only have device nodes in there
     * and don't want to DOS the entire OS RAM usage
     */
    if (virAsprintf(&opts,
                    "mode=755,size=65536%s", mount_options) < 0)
        goto cleanup;

11849
    if (virFileSetupDev(data->path, opts) < 0)
11850 11851
        goto cleanup;

11852
    if (qemuDomainPopulateDevices(cfg, vm, data) < 0)
11853 11854 11855 11856 11857 11858 11859 11860 11861 11862
        goto cleanup;

    ret = 0;
 cleanup:
    VIR_FREE(opts);
    VIR_FREE(mount_options);
    return ret;
}


11863
static int
11864
qemuDomainSetupDisk(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
11865
                    virDomainDiskDefPtr disk,
11866
                    const struct qemuDomainCreateDeviceData *data)
11867 11868 11869 11870 11871
{
    virStorageSourcePtr next;
    char *dst = NULL;
    int ret = -1;

11872
    for (next = disk->src; virStorageSourceIsBacking(next); next = next->backingStore) {
11873
        if (!next->path || !virStorageSourceIsLocalStorage(next)) {
11874 11875 11876 11877
            /* Not creating device. Just continue. */
            continue;
        }

11878
        if (qemuDomainCreateDevice(next->path, data, false) < 0)
11879 11880 11881
            goto cleanup;
    }

11882
    /* qemu-pr-helper might require access to /dev/mapper/control. */
11883
    if (disk->src->pr &&
11884 11885 11886
        qemuDomainCreateDevice(DEVICE_MAPPER_CONTROL_PATH, data, true) < 0)
        goto cleanup;

11887 11888 11889 11890 11891 11892 11893 11894
    ret = 0;
 cleanup:
    VIR_FREE(dst);
    return ret;
}


static int
11895
qemuDomainSetupAllDisks(virQEMUDriverConfigPtr cfg,
11896
                        virDomainObjPtr vm,
11897
                        const struct qemuDomainCreateDeviceData *data)
11898 11899 11900 11901 11902
{
    size_t i;
    VIR_DEBUG("Setting up disks");

    for (i = 0; i < vm->def->ndisks; i++) {
11903
        if (qemuDomainSetupDisk(cfg,
11904
                                vm->def->disks[i],
11905
                                data) < 0)
11906 11907 11908 11909 11910 11911 11912 11913
            return -1;
    }

    VIR_DEBUG("Setup all disks");
    return 0;
}


11914
static int
11915
qemuDomainSetupHostdev(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
11916
                       virDomainHostdevDefPtr dev,
11917
                       const struct qemuDomainCreateDeviceData *data)
11918 11919
{
    int ret = -1;
11920 11921
    char **path = NULL;
    size_t i, npaths = 0;
11922

11923
    if (qemuDomainGetHostdevPath(NULL, dev, false, &npaths, &path, NULL) < 0)
11924 11925
        goto cleanup;

11926
    for (i = 0; i < npaths; i++) {
11927
        if (qemuDomainCreateDevice(path[i], data, false) < 0)
11928
            goto cleanup;
11929 11930 11931 11932
    }

    ret = 0;
 cleanup:
11933 11934
    for (i = 0; i < npaths; i++)
        VIR_FREE(path[i]);
11935 11936 11937 11938 11939 11940
    VIR_FREE(path);
    return ret;
}


static int
11941
qemuDomainSetupAllHostdevs(virQEMUDriverConfigPtr cfg,
11942
                           virDomainObjPtr vm,
11943
                           const struct qemuDomainCreateDeviceData *data)
11944 11945 11946 11947 11948
{
    size_t i;

    VIR_DEBUG("Setting up hostdevs");
    for (i = 0; i < vm->def->nhostdevs; i++) {
11949
        if (qemuDomainSetupHostdev(cfg,
11950
                                   vm->def->hostdevs[i],
11951
                                   data) < 0)
11952 11953 11954 11955 11956 11957 11958
            return -1;
    }
    VIR_DEBUG("Setup all hostdevs");
    return 0;
}


M
Michal Privoznik 已提交
11959 11960 11961
static int
qemuDomainSetupMemory(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
                      virDomainMemoryDefPtr mem,
11962
                      const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
11963 11964 11965 11966
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

11967
    return qemuDomainCreateDevice(mem->nvdimmPath, data, false);
M
Michal Privoznik 已提交
11968 11969 11970 11971 11972 11973
}


static int
qemuDomainSetupAllMemories(virQEMUDriverConfigPtr cfg,
                           virDomainObjPtr vm,
11974
                           const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
11975 11976 11977 11978 11979 11980 11981
{
    size_t i;

    VIR_DEBUG("Setting up memories");
    for (i = 0; i < vm->def->nmems; i++) {
        if (qemuDomainSetupMemory(cfg,
                                  vm->def->mems[i],
11982
                                  data) < 0)
M
Michal Privoznik 已提交
11983 11984 11985 11986 11987 11988 11989
            return -1;
    }
    VIR_DEBUG("Setup all memories");
    return 0;
}


11990 11991 11992 11993 11994
static int
qemuDomainSetupChardev(virDomainDefPtr def ATTRIBUTE_UNUSED,
                       virDomainChrDefPtr dev,
                       void *opaque)
{
11995
    const struct qemuDomainCreateDeviceData *data = opaque;
11996
    const char *path = NULL;
11997

11998
    if (!(path = virDomainChrSourceDefGetPath(dev->source)))
11999 12000
        return 0;

12001 12002 12003 12004 12005 12006
    /* 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);
12007 12008 12009 12010
}


static int
12011
qemuDomainSetupAllChardevs(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12012
                           virDomainObjPtr vm,
12013
                           const struct qemuDomainCreateDeviceData *data)
12014 12015 12016 12017 12018 12019
{
    VIR_DEBUG("Setting up chardevs");

    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuDomainSetupChardev,
12020
                               (void *)data) < 0)
12021 12022 12023 12024 12025 12026 12027
        return -1;

    VIR_DEBUG("Setup all chardevs");
    return 0;
}


12028
static int
12029
qemuDomainSetupTPM(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12030
                   virDomainObjPtr vm,
12031
                   const struct qemuDomainCreateDeviceData *data)
12032 12033 12034 12035 12036 12037 12038 12039 12040 12041 12042
{
    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,
12043
                                   data, false) < 0)
12044 12045 12046
            return -1;
        break;

12047
    case VIR_DOMAIN_TPM_TYPE_EMULATOR:
12048 12049 12050 12051 12052 12053 12054 12055 12056 12057
    case VIR_DOMAIN_TPM_TYPE_LAST:
        /* nada */
        break;
    }

    VIR_DEBUG("Setup TPM");
    return 0;
}


12058
static int
12059
qemuDomainSetupGraphics(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12060
                        virDomainGraphicsDefPtr gfx,
12061
                        const struct qemuDomainCreateDeviceData *data)
12062
{
12063
    const char *rendernode = virDomainGraphicsGetRenderNode(gfx);
12064

12065
    if (!rendernode)
12066 12067
        return 0;

12068
    return qemuDomainCreateDevice(rendernode, data, false);
12069 12070 12071 12072
}


static int
12073
qemuDomainSetupAllGraphics(virQEMUDriverConfigPtr cfg,
12074
                           virDomainObjPtr vm,
12075
                           const struct qemuDomainCreateDeviceData *data)
12076 12077 12078 12079 12080
{
    size_t i;

    VIR_DEBUG("Setting up graphics");
    for (i = 0; i < vm->def->ngraphics; i++) {
12081
        if (qemuDomainSetupGraphics(cfg,
12082
                                    vm->def->graphics[i],
12083
                                    data) < 0)
12084 12085 12086 12087 12088 12089 12090 12091
            return -1;
    }

    VIR_DEBUG("Setup all graphics");
    return 0;
}


12092
static int
12093
qemuDomainSetupInput(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12094
                     virDomainInputDefPtr input,
12095
                     const struct qemuDomainCreateDeviceData *data)
12096
{
J
Ján Tomko 已提交
12097
    const char *path = virDomainInputDefGetPath(input);
12098

J
Ján Tomko 已提交
12099 12100
    if (path && qemuDomainCreateDevice(path, data, false) < 0)
        return -1;
12101

J
Ján Tomko 已提交
12102
    return 0;
12103 12104 12105 12106
}


static int
12107
qemuDomainSetupAllInputs(virQEMUDriverConfigPtr cfg,
12108
                         virDomainObjPtr vm,
12109
                         const struct qemuDomainCreateDeviceData *data)
12110 12111 12112
{
    size_t i;

12113
    VIR_DEBUG("Setting up inputs");
12114
    for (i = 0; i < vm->def->ninputs; i++) {
12115
        if (qemuDomainSetupInput(cfg,
12116
                                 vm->def->inputs[i],
12117
                                 data) < 0)
12118 12119
            return -1;
    }
12120
    VIR_DEBUG("Setup all inputs");
12121 12122 12123 12124
    return 0;
}


12125
static int
12126
qemuDomainSetupRNG(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12127
                   virDomainRNGDefPtr rng,
12128
                   const struct qemuDomainCreateDeviceData *data)
12129 12130 12131
{
    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
12132
        if (qemuDomainCreateDevice(rng->source.file, data, false) < 0)
12133 12134 12135 12136 12137 12138 12139 12140 12141 12142 12143 12144 12145
            return -1;

    case VIR_DOMAIN_RNG_BACKEND_EGD:
    case VIR_DOMAIN_RNG_BACKEND_LAST:
        /* nada */
        break;
    }

    return 0;
}


static int
12146
qemuDomainSetupAllRNGs(virQEMUDriverConfigPtr cfg,
12147
                       virDomainObjPtr vm,
12148
                       const struct qemuDomainCreateDeviceData *data)
12149 12150 12151 12152 12153
{
    size_t i;

    VIR_DEBUG("Setting up RNGs");
    for (i = 0; i < vm->def->nrngs; i++) {
12154
        if (qemuDomainSetupRNG(cfg,
12155
                               vm->def->rngs[i],
12156
                               data) < 0)
12157 12158 12159 12160 12161 12162 12163 12164
            return -1;
    }

    VIR_DEBUG("Setup all RNGs");
    return 0;
}


12165 12166 12167 12168 12169 12170 12171 12172 12173 12174 12175 12176 12177 12178 12179 12180 12181 12182 12183 12184 12185 12186 12187 12188 12189 12190 12191 12192 12193 12194 12195 12196 12197 12198 12199 12200 12201 12202
static int
qemuDomainSetupLoader(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
                      virDomainObjPtr vm,
                      const struct qemuDomainCreateDeviceData *data)
{
    virDomainLoaderDefPtr loader = vm->def->os.loader;
    int ret = -1;

    VIR_DEBUG("Setting up loader");

    if (loader) {
        switch ((virDomainLoader) loader->type) {
        case VIR_DOMAIN_LOADER_TYPE_ROM:
            if (qemuDomainCreateDevice(loader->path, data, false) < 0)
                goto cleanup;
            break;

        case VIR_DOMAIN_LOADER_TYPE_PFLASH:
            if (qemuDomainCreateDevice(loader->path, data, false) < 0)
                goto cleanup;

            if (loader->nvram &&
                qemuDomainCreateDevice(loader->nvram, data, false) < 0)
                goto cleanup;
            break;

        case VIR_DOMAIN_LOADER_TYPE_LAST:
            break;
        }
    }

    VIR_DEBUG("Setup loader");
    ret = 0;
 cleanup:
    return ret;
}


12203
int
12204 12205
qemuDomainBuildNamespace(virQEMUDriverConfigPtr cfg,
                         virSecurityManagerPtr mgr,
12206 12207
                         virDomainObjPtr vm)
{
12208
    struct qemuDomainCreateDeviceData data;
12209
    char *devPath = NULL;
12210
    char **devMountsPath = NULL, **devMountsSavePath = NULL;
12211 12212 12213 12214 12215 12216 12217 12218
    size_t ndevMountsPath = 0, i;
    int ret = -1;

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT)) {
        ret = 0;
        goto cleanup;
    }

12219
    if (qemuDomainGetPreservedMounts(cfg, vm,
12220 12221
                                     &devMountsPath, &devMountsSavePath,
                                     &ndevMountsPath) < 0)
12222 12223
        goto cleanup;

12224 12225 12226 12227 12228 12229 12230 12231 12232 12233 12234 12235 12236
    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;
    }

12237
    data.path = devPath;
12238 12239
    data.devMountsPath = devMountsPath;
    data.ndevMountsPath = ndevMountsPath;
12240

12241 12242 12243
    if (virProcessSetupPrivateMountNS() < 0)
        goto cleanup;

12244
    if (qemuDomainSetupDev(cfg, mgr, vm, &data) < 0)
12245 12246
        goto cleanup;

12247
    if (qemuDomainSetupAllDisks(cfg, vm, &data) < 0)
12248 12249
        goto cleanup;

12250
    if (qemuDomainSetupAllHostdevs(cfg, vm, &data) < 0)
12251 12252
        goto cleanup;

12253
    if (qemuDomainSetupAllMemories(cfg, vm, &data) < 0)
12254 12255
        goto cleanup;

12256
    if (qemuDomainSetupAllChardevs(cfg, vm, &data) < 0)
12257 12258
        goto cleanup;

12259
    if (qemuDomainSetupTPM(cfg, vm, &data) < 0)
12260 12261
        goto cleanup;

12262
    if (qemuDomainSetupAllGraphics(cfg, vm, &data) < 0)
12263 12264
        goto cleanup;

12265
    if (qemuDomainSetupAllInputs(cfg, vm, &data) < 0)
12266 12267
        goto cleanup;

12268
    if (qemuDomainSetupAllRNGs(cfg, vm, &data) < 0)
12269 12270
        goto cleanup;

12271 12272 12273
    if (qemuDomainSetupLoader(cfg, vm, &data) < 0)
        goto cleanup;

12274 12275
    /* Save some mount points because we want to share them with the host */
    for (i = 0; i < ndevMountsPath; i++) {
12276 12277
        struct stat sb;

12278 12279 12280
        if (devMountsSavePath[i] == devPath)
            continue;

12281 12282 12283 12284 12285 12286 12287
        if (stat(devMountsPath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsPath[i]);
            goto cleanup;
        }

12288 12289 12290
        /* At this point, devMountsPath is either:
         * a file (regular or special), or
         * a directory. */
12291
        if ((S_ISDIR(sb.st_mode) && virFileMakePath(devMountsSavePath[i]) < 0) ||
12292
            (!S_ISDIR(sb.st_mode) && virFileTouch(devMountsSavePath[i], sb.st_mode) < 0)) {
12293 12294 12295 12296 12297 12298
            virReportSystemError(errno,
                                 _("Failed to create %s"),
                                 devMountsSavePath[i]);
            goto cleanup;
        }

12299
        if (virFileMoveMount(devMountsPath[i], devMountsSavePath[i]) < 0)
12300 12301 12302
            goto cleanup;
    }

12303
    if (virFileMoveMount(devPath, "/dev") < 0)
12304 12305 12306
        goto cleanup;

    for (i = 0; i < ndevMountsPath; i++) {
12307 12308
        struct stat sb;

12309 12310 12311
        if (devMountsSavePath[i] == devPath)
            continue;

12312 12313 12314 12315
        if (stat(devMountsSavePath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsSavePath[i]);
12316 12317 12318
            goto cleanup;
        }

12319 12320 12321 12322 12323 12324 12325 12326 12327 12328 12329 12330 12331 12332 12333
        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;
            }
        }

12334
        if (virFileMoveMount(devMountsSavePath[i], devMountsPath[i]) < 0)
12335 12336 12337 12338 12339
            goto cleanup;
    }

    ret = 0;
 cleanup:
12340 12341 12342 12343 12344 12345 12346
    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]);
    }
12347
    virStringListFreeCount(devMountsPath, ndevMountsPath);
12348
    virStringListFreeCount(devMountsSavePath, ndevMountsPath);
12349 12350 12351 12352 12353 12354 12355 12356 12357 12358 12359
    return ret;
}


int
qemuDomainCreateNamespace(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

12360 12361
    if (virBitmapIsBitSet(cfg->namespaces, QEMU_DOMAIN_NS_MOUNT) &&
        qemuDomainEnableNamespace(vm, QEMU_DOMAIN_NS_MOUNT) < 0)
M
Michal Privoznik 已提交
12362
        goto cleanup;
12363 12364 12365 12366 12367 12368 12369 12370

    ret = 0;
 cleanup:
    virObjectUnref(cfg);
    return ret;
}


12371 12372 12373 12374 12375 12376 12377 12378 12379
void
qemuDomainDestroyNamespace(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                           virDomainObjPtr vm)
{
    if (qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        qemuDomainDisableNamespace(vm, QEMU_DOMAIN_NS_MOUNT);
}


12380 12381 12382 12383 12384 12385 12386 12387 12388 12389 12390 12391 12392 12393 12394 12395 12396 12397 12398 12399 12400 12401 12402 12403 12404 12405 12406 12407 12408
bool
qemuDomainNamespaceAvailable(qemuDomainNamespace ns ATTRIBUTE_UNUSED)
{
#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__) */
}


12409 12410 12411 12412
struct qemuDomainAttachDeviceMknodData {
    virQEMUDriverPtr driver;
    virDomainObjPtr vm;
    const char *file;
12413
    const char *target;
12414 12415
    struct stat sb;
    void *acl;
12416 12417 12418
#ifdef WITH_SELINUX
    char *tcon;
#endif
12419 12420 12421
};


12422 12423
/* Our way of creating devices is highly linux specific */
#if defined(__linux__)
12424 12425 12426 12427 12428 12429
static int
qemuDomainAttachDeviceMknodHelper(pid_t pid ATTRIBUTE_UNUSED,
                                  void *opaque)
{
    struct qemuDomainAttachDeviceMknodData *data = opaque;
    int ret = -1;
12430
    bool delDevice = false;
12431
    bool isLink = S_ISLNK(data->sb.st_mode);
12432
    bool isDev = S_ISCHR(data->sb.st_mode) || S_ISBLK(data->sb.st_mode);
12433
    bool isReg = S_ISREG(data->sb.st_mode) || S_ISFIFO(data->sb.st_mode) || S_ISSOCK(data->sb.st_mode);
12434
    bool isDir = S_ISDIR(data->sb.st_mode);
12435

12436
    qemuSecurityPostFork(data->driver->securityManager);
12437 12438 12439 12440 12441 12442 12443

    if (virFileMakeParentPath(data->file) < 0) {
        virReportSystemError(errno,
                             _("Unable to create %s"), data->file);
        goto cleanup;
    }

12444 12445
    if (isLink) {
        VIR_DEBUG("Creating symlink %s -> %s", data->file, data->target);
12446 12447 12448 12449 12450 12451 12452 12453 12454 12455 12456 12457

        /* 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;
        }

12458
        if (symlink(data->target, data->file) < 0) {
12459 12460 12461 12462
            virReportSystemError(errno,
                                 _("Unable to create symlink %s (pointing to %s)"),
                                 data->file, data->target);
            goto cleanup;
12463 12464
        } else {
            delDevice = true;
12465
        }
12466
    } else if (isDev) {
12467 12468 12469 12470 12471 12472 12473 12474 12475 12476 12477 12478 12479 12480 12481 12482
        VIR_DEBUG("Creating dev %s (%d,%d)",
                  data->file, major(data->sb.st_rdev), minor(data->sb.st_rdev));
        if (mknod(data->file, data->sb.st_mode, data->sb.st_rdev) < 0) {
            /* Because we are not removing devices on hotunplug, or
             * we might be creating part of backing chain that
             * already exist due to a different disk plugged to
             * domain, accept EEXIST. */
            if (errno != EEXIST) {
                virReportSystemError(errno,
                                     _("Unable to create device %s"),
                                     data->file);
                goto cleanup;
            }
        } else {
            delDevice = true;
        }
12483
    } else if (isReg || isDir) {
12484 12485 12486 12487 12488
        /* 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 &&
12489
            errno != ENOENT && errno != EINVAL) {
12490 12491 12492 12493 12494
            virReportSystemError(errno,
                                 _("Unable to umount %s"),
                                 data->file);
            goto cleanup;
        }
12495 12496
        if ((isReg && virFileTouch(data->file, data->sb.st_mode) < 0) ||
            (isDir && virFileMakePathWithMode(data->file, data->sb.st_mode) < 0))
12497 12498 12499 12500
            goto cleanup;
        delDevice = true;
        /* Just create the file here so that code below sets
         * proper owner and mode. Move the mount only after that. */
12501 12502 12503 12504 12505
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       data->file, data->sb.st_mode);
        goto cleanup;
12506 12507
    }

12508 12509 12510 12511 12512 12513 12514
    if (lchown(data->file, data->sb.st_uid, data->sb.st_gid) < 0) {
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             data->file);
        goto cleanup;
    }

12515 12516 12517 12518 12519 12520 12521 12522 12523
    /* 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;
    }

12524 12525 12526
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileSetACLs(data->file, data->acl) < 0 &&
12527 12528 12529 12530 12531 12532
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to set ACLs on %s"), data->file);
        goto cleanup;
    }

12533
# ifdef WITH_SELINUX
12534
    if (data->tcon &&
12535
        lsetfilecon_raw(data->file, (VIR_SELINUX_CTX_CONST char *)data->tcon) < 0) {
12536 12537 12538 12539 12540 12541 12542 12543 12544
        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;
        }
    }
12545
# endif
12546

12547
    /* Finish mount process started earlier. */
12548
    if ((isReg || isDir) &&
12549 12550 12551
        virFileMoveMount(data->target, data->file) < 0)
        goto cleanup;

12552 12553
    ret = 0;
 cleanup:
12554 12555 12556 12557 12558 12559
    if (ret < 0 && delDevice) {
        if (isDir)
            virFileDeleteTree(data->file);
        else
            unlink(data->file);
    }
12560
# ifdef WITH_SELINUX
12561
    freecon(data->tcon);
12562
# endif
12563 12564 12565 12566 12567 12568
    virFileFreeACLs(&data->acl);
    return ret;
}


static int
12569 12570 12571
qemuDomainAttachDeviceMknodRecursive(virQEMUDriverPtr driver,
                                     virDomainObjPtr vm,
                                     const char *file,
12572 12573
                                     char * const *devMountsPath,
                                     size_t ndevMountsPath,
12574
                                     unsigned int ttl)
12575
{
12576
    virQEMUDriverConfigPtr cfg = NULL;
12577 12578
    struct qemuDomainAttachDeviceMknodData data;
    int ret = -1;
12579 12580
    char *target = NULL;
    bool isLink;
12581
    bool isReg;
12582
    bool isDir;
12583

12584 12585 12586 12587 12588 12589 12590
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             file);
        return ret;
    }

12591 12592 12593 12594 12595 12596
    memset(&data, 0, sizeof(data));

    data.driver = driver;
    data.vm = vm;
    data.file = file;

12597
    if (lstat(file, &data.sb) < 0) {
12598 12599 12600 12601 12602
        virReportSystemError(errno,
                             _("Unable to access %s"), file);
        return ret;
    }

12603
    isLink = S_ISLNK(data.sb.st_mode);
12604
    isReg = S_ISREG(data.sb.st_mode) || S_ISFIFO(data.sb.st_mode) || S_ISSOCK(data.sb.st_mode);
12605
    isDir = S_ISDIR(data.sb.st_mode);
12606

12607
    if ((isReg || isDir) && STRPREFIX(file, DEVPREFIX)) {
12608 12609 12610 12611 12612 12613 12614 12615 12616
        cfg = virQEMUDriverGetConfig(driver);
        if (!(target = qemuDomainGetPreservedMountPath(cfg, vm, file)))
            goto cleanup;

        if (virFileBindMountDevice(file, target) < 0)
            goto cleanup;

        data.target = target;
    } else if (isLink) {
12617 12618 12619 12620 12621 12622 12623 12624 12625 12626 12627 12628 12629 12630 12631 12632 12633 12634 12635 12636 12637 12638
        if (virFileReadLink(file, &target) < 0) {
            virReportSystemError(errno,
                                 _("unable to resolve symlink %s"),
                                 file);
            return ret;
        }

        if (IS_RELATIVE_FILE_NAME(target)) {
            char *c = NULL, *tmp = NULL, *fileTmp = NULL;

            if (VIR_STRDUP(fileTmp, file) < 0)
                goto cleanup;

            if ((c = strrchr(fileTmp, '/')))
                *(c + 1) = '\0';

            if (virAsprintf(&tmp, "%s%s", fileTmp, target) < 0) {
                VIR_FREE(fileTmp);
                goto cleanup;
            }
            VIR_FREE(fileTmp);
            VIR_FREE(target);
12639
            VIR_STEAL_PTR(target, tmp);
12640 12641 12642 12643 12644 12645 12646 12647
        }

        data.target = target;
    }

    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileGetACLs(file, &data.acl) < 0 &&
12648 12649 12650
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to get ACLs on %s"), file);
12651
        goto cleanup;
12652 12653
    }

12654
# ifdef WITH_SELINUX
12655
    if (lgetfilecon_raw(file, &data.tcon) < 0 &&
12656 12657 12658 12659 12660
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"), file);
        goto cleanup;
    }
12661
# endif
12662

12663
    if (STRPREFIX(file, DEVPREFIX)) {
12664 12665 12666 12667 12668 12669 12670 12671 12672 12673 12674 12675
        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;
12676

12677 12678 12679 12680 12681 12682
            if (virProcessRunInMountNamespace(vm->pid,
                                              qemuDomainAttachDeviceMknodHelper,
                                              &data) < 0) {
                qemuSecurityPostFork(driver->securityManager);
                goto cleanup;
            }
12683
            qemuSecurityPostFork(driver->securityManager);
12684 12685 12686
        } else {
            VIR_DEBUG("Skipping dev %s because of %s mount point",
                      file, devMountsPath[i]);
12687
        }
12688 12689
    }

12690
    if (isLink &&
12691 12692 12693
        qemuDomainAttachDeviceMknodRecursive(driver, vm, target,
                                             devMountsPath, ndevMountsPath,
                                             ttl -1) < 0)
12694
        goto cleanup;
12695 12696 12697

    ret = 0;
 cleanup:
12698
# ifdef WITH_SELINUX
12699
    freecon(data.tcon);
12700
# endif
12701
    virFileFreeACLs(&data.acl);
12702 12703
    if (isReg && target)
        umount(target);
12704
    VIR_FREE(target);
12705
    virObjectUnref(cfg);
12706
    return ret;
12707 12708 12709
}


12710 12711 12712 12713 12714 12715 12716 12717 12718 12719 12720 12721 12722 12723 12724 12725 12726 12727 12728 12729
#else /* !defined(__linux__) */


static int
qemuDomainAttachDeviceMknodRecursive(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                                     virDomainObjPtr vm ATTRIBUTE_UNUSED,
                                     const char *file ATTRIBUTE_UNUSED,
                                     char * const *devMountsPath ATTRIBUTE_UNUSED,
                                     size_t ndevMountsPath ATTRIBUTE_UNUSED,
                                     unsigned int ttl ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Namespaces are not supported on this platform."));
    return -1;
}


#endif /* !defined(__linux__) */


12730 12731 12732
static int
qemuDomainAttachDeviceMknod(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
12733 12734 12735
                            const char *file,
                            char * const *devMountsPath,
                            size_t ndevMountsPath)
12736 12737 12738
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

12739 12740 12741
    return qemuDomainAttachDeviceMknodRecursive(driver, vm, file,
                                                devMountsPath, ndevMountsPath,
                                                symloop_max);
12742 12743 12744
}


12745 12746 12747 12748 12749 12750 12751 12752 12753 12754 12755 12756 12757 12758 12759 12760 12761 12762
static int
qemuDomainDetachDeviceUnlinkHelper(pid_t pid ATTRIBUTE_UNUSED,
                                   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
12763
qemuDomainDetachDeviceUnlink(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
12764
                             virDomainObjPtr vm,
12765 12766 12767
                             const char *file,
                             char * const *devMountsPath,
                             size_t ndevMountsPath)
12768
{
12769 12770
    int ret = -1;
    size_t i;
12771

12772 12773 12774 12775 12776 12777 12778 12779 12780 12781 12782 12783 12784 12785 12786 12787 12788 12789 12790
    if (STRPREFIX(file, DEVPREFIX)) {
        for (i = 0; i < ndevMountsPath; i++) {
            if (STREQ(devMountsPath[i], "/dev"))
                continue;
            if (STRPREFIX(file, devMountsPath[i]))
                break;
        }

        if (i == ndevMountsPath) {
            if (virProcessRunInMountNamespace(vm->pid,
                                              qemuDomainDetachDeviceUnlinkHelper,
                                              (void *)file) < 0)
                goto cleanup;
        }
    }

    ret = 0;
 cleanup:
    return ret;
12791 12792 12793
}


12794 12795 12796 12797
static int
qemuDomainNamespaceMknodPaths(virDomainObjPtr vm,
                              const char **paths,
                              size_t npaths)
12798
{
12799 12800 12801
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    virQEMUDriverConfigPtr cfg;
12802 12803
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0;
12804
    int ret = -1;
12805
    size_t i;
12806

12807 12808
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
12809 12810
        return 0;

12811 12812 12813 12814 12815 12816
    cfg = virQEMUDriverGetConfig(driver);
    if (qemuDomainGetPreservedMounts(cfg, vm,
                                     &devMountsPath, NULL,
                                     &ndevMountsPath) < 0)
        goto cleanup;

12817 12818 12819 12820 12821 12822 12823 12824 12825 12826 12827 12828 12829 12830 12831 12832
    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;
}


12833 12834 12835 12836 12837 12838 12839 12840 12841 12842
static int
qemuDomainNamespaceMknodPath(virDomainObjPtr vm,
                             const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceMknodPaths(vm, paths, 1);
}


12843 12844 12845 12846 12847 12848 12849 12850 12851 12852 12853 12854 12855
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;

12856 12857
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
12858 12859 12860 12861 12862 12863 12864 12865 12866 12867 12868 12869 12870 12871 12872
        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;
    }

12873
    ret = 0;
12874 12875 12876 12877 12878 12879 12880
 cleanup:
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    virObjectUnref(cfg);
    return ret;
}


12881 12882 12883 12884 12885 12886 12887 12888 12889 12890
static int
qemuDomainNamespaceUnlinkPath(virDomainObjPtr vm,
                              const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceUnlinkPaths(vm, paths, 1);
}


12891
int
12892
qemuDomainNamespaceSetupDisk(virDomainObjPtr vm,
12893 12894 12895
                             virStorageSourcePtr src)
{
    virStorageSourcePtr next;
12896
    const char **paths = NULL;
12897
    size_t npaths = 0;
12898
    char *dmPath = NULL;
12899 12900
    int ret = -1;

12901
    for (next = src; virStorageSourceIsBacking(next); next = next->backingStore) {
12902 12903
        if (virStorageSourceIsEmpty(next) ||
            !virStorageSourceIsLocalStorage(next)) {
12904 12905 12906 12907
            /* Not creating device. Just continue. */
            continue;
        }

12908
        if (VIR_APPEND_ELEMENT_COPY(paths, npaths, next->path) < 0)
12909 12910 12911
            goto cleanup;
    }

12912
    /* qemu-pr-helper might require access to /dev/mapper/control. */
12913
    if (src->pr &&
12914 12915 12916 12917
        (VIR_STRDUP(dmPath, DEVICE_MAPPER_CONTROL_PATH) < 0 ||
         VIR_APPEND_ELEMENT_COPY(paths, npaths, dmPath) < 0))
        goto cleanup;

12918
    if (qemuDomainNamespaceMknodPaths(vm, paths, npaths) < 0)
12919
        goto cleanup;
12920

12921 12922
    ret = 0;
 cleanup:
12923
    VIR_FREE(dmPath);
12924
    VIR_FREE(paths);
12925 12926 12927 12928 12929
    return ret;
}


int
12930
qemuDomainNamespaceTeardownDisk(virDomainObjPtr vm ATTRIBUTE_UNUSED,
12931
                                virStorageSourcePtr src ATTRIBUTE_UNUSED)
12932 12933 12934 12935 12936 12937 12938 12939 12940
{
    /* 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;
}
12941 12942 12943


int
12944
qemuDomainNamespaceSetupHostdev(virDomainObjPtr vm,
12945 12946 12947
                                virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
12948
    char **paths = NULL;
12949
    size_t i, npaths = 0;
12950

12951
    if (qemuDomainGetHostdevPath(NULL, hostdev, false, &npaths, &paths, NULL) < 0)
12952 12953
        goto cleanup;

12954
    if (qemuDomainNamespaceMknodPaths(vm, (const char **)paths, npaths) < 0)
12955 12956
        goto cleanup;

12957 12958
    ret = 0;
 cleanup:
12959
    for (i = 0; i < npaths; i++)
12960 12961
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
12962 12963 12964 12965 12966
    return ret;
}


int
12967
qemuDomainNamespaceTeardownHostdev(virDomainObjPtr vm,
12968 12969 12970
                                   virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
12971
    char **paths = NULL;
12972
    size_t i, npaths = 0;
12973

12974
    if (qemuDomainGetHostdevPath(vm->def, hostdev, true,
12975
                                 &npaths, &paths, NULL) < 0)
12976 12977
        goto cleanup;

12978
    if (qemuDomainNamespaceUnlinkPaths(vm, (const char **)paths, npaths) < 0)
12979 12980
        goto cleanup;

12981 12982
    ret = 0;
 cleanup:
12983
    for (i = 0; i < npaths; i++)
12984 12985
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
12986 12987
    return ret;
}
12988 12989


M
Michal Privoznik 已提交
12990
int
12991
qemuDomainNamespaceSetupMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
12992 12993 12994 12995 12996
                               virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

12997 12998
    if (qemuDomainNamespaceMknodPath(vm, mem->nvdimmPath) < 0)
        return -1;
12999

13000
    return 0;
M
Michal Privoznik 已提交
13001 13002 13003 13004
}


int
13005
qemuDomainNamespaceTeardownMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
13006 13007 13008 13009 13010
                                  virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

13011 13012
    if (qemuDomainNamespaceUnlinkPath(vm, mem->nvdimmPath) < 0)
        return -1;
13013

13014
    return 0;
M
Michal Privoznik 已提交
13015 13016 13017
}


13018
int
13019
qemuDomainNamespaceSetupChardev(virDomainObjPtr vm,
13020 13021 13022 13023
                                virDomainChrDefPtr chr)
{
    const char *path;

13024
    if (!(path = virDomainChrSourceDefGetPath(chr->source)))
13025 13026
        return 0;

13027 13028 13029 13030
    /* 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;
13031

13032 13033
    if (qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
13034

13035
    return 0;
13036 13037 13038 13039
}


int
13040
qemuDomainNamespaceTeardownChardev(virDomainObjPtr vm,
13041 13042 13043 13044 13045 13046 13047 13048 13049
                                   virDomainChrDefPtr chr)
{
    const char *path = NULL;

    if (chr->source->type != VIR_DOMAIN_CHR_TYPE_DEV)
        return 0;

    path = chr->source->data.file.path;

13050 13051
    if (qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
13052

13053
    return 0;
13054
}
13055 13056 13057


int
13058
qemuDomainNamespaceSetupRNG(virDomainObjPtr vm,
13059 13060 13061 13062 13063 13064 13065 13066 13067 13068 13069
                            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:
13070
        break;
13071 13072
    }

13073 13074
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
13075

13076
    return 0;
13077 13078 13079 13080
}


int
13081
qemuDomainNamespaceTeardownRNG(virDomainObjPtr vm,
13082 13083 13084 13085 13086 13087 13088 13089 13090 13091 13092
                               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:
13093
        break;
13094 13095
    }

13096 13097
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
13098

13099
    return 0;
13100 13101 13102 13103 13104 13105 13106 13107 13108 13109 13110 13111
}


int
qemuDomainNamespaceSetupInput(virDomainObjPtr vm,
                              virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

13112 13113 13114
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
    return 0;
13115 13116 13117 13118 13119 13120 13121 13122 13123 13124 13125 13126
}


int
qemuDomainNamespaceTeardownInput(virDomainObjPtr vm,
                                 virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

13127 13128
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
13129

13130
    return 0;
13131
}
13132 13133 13134 13135 13136 13137 13138 13139 13140 13141 13142 13143 13144 13145 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


/**
 * 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;
}
13175 13176 13177 13178 13179 13180 13181 13182 13183 13184 13185 13186 13187 13188 13189 13190 13191 13192 13193 13194 13195 13196


/**
 * qemuDomainDiskBackingStoreGetName:
 *
 * Creates a name using the indexed syntax (vda[1])for the given backing store
 * entry for a disk.
 */
char *
qemuDomainDiskBackingStoreGetName(virDomainDiskDefPtr disk,
                                  virStorageSourcePtr src ATTRIBUTE_UNUSED,
                                  unsigned int idx)
{
    char *ret = NULL;

    if (idx)
        ignore_value(virAsprintf(&ret, "%s[%d]", disk->dst, idx));
    else
        ignore_value(VIR_STRDUP(ret, disk->dst));

    return ret;
}
13197 13198 13199 13200 13201 13202 13203 13204 13205 13206 13207 13208 13209 13210 13211 13212 13213 13214 13215 13216 13217 13218 13219 13220 13221 13222 13223


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
13224
                       _("failed to find disk '%s'"), target);
13225 13226 13227
        goto cleanup;
    }

13228 13229 13230 13231
    if (idx == 0)
        src = disk->src;
    else
        src = virStorageFileChainLookup(disk->src, NULL, NULL, idx, NULL);
13232 13233 13234 13235 13236

 cleanup:
    VIR_FREE(target);
    return src;
}
13237 13238 13239 13240 13241 13242 13243 13244


static void
qemuDomainSaveCookieDispose(void *obj)
{
    qemuDomainSaveCookiePtr cookie = obj;

    VIR_DEBUG("cookie=%p", cookie);
13245 13246

    virCPUDefFree(cookie->cpu);
13247 13248 13249 13250 13251 13252
}


qemuDomainSaveCookiePtr
qemuDomainSaveCookieNew(virDomainObjPtr vm ATTRIBUTE_UNUSED)
{
13253
    qemuDomainObjPrivatePtr priv = vm->privateData;
13254 13255 13256 13257 13258 13259 13260 13261
    qemuDomainSaveCookiePtr cookie = NULL;

    if (qemuDomainInitialize() < 0)
        goto error;

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
        goto error;

13262 13263 13264 13265
    if (priv->origCPU && !(cookie->cpu = virCPUDefCopy(vm->def->cpu)))
        goto error;

    VIR_DEBUG("Save cookie %p, cpu=%p", cookie, cookie->cpu);
13266 13267 13268 13269 13270 13271 13272 13273 13274 13275 13276 13277 13278 13279 13280 13281 13282 13283 13284 13285 13286

    return cookie;

 error:
    virObjectUnref(cookie);
    return NULL;
}


static int
qemuDomainSaveCookieParse(xmlXPathContextPtr ctxt ATTRIBUTE_UNUSED,
                          virObjectPtr *obj)
{
    qemuDomainSaveCookiePtr cookie = NULL;

    if (qemuDomainInitialize() < 0)
        goto error;

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
        goto error;

13287 13288 13289 13290
    if (virCPUDefParseXML(ctxt, "./cpu[1]", VIR_CPU_TYPE_GUEST,
                          &cookie->cpu) < 0)
        goto error;

13291 13292 13293 13294 13295 13296 13297 13298 13299 13300
    *obj = (virObjectPtr) cookie;
    return 0;

 error:
    virObjectUnref(cookie);
    return -1;
}


static int
13301 13302
qemuDomainSaveCookieFormat(virBufferPtr buf,
                           virObjectPtr obj)
13303
{
13304 13305 13306
    qemuDomainSaveCookiePtr cookie = (qemuDomainSaveCookiePtr) obj;

    if (cookie->cpu &&
13307
        virCPUDefFormatBufFull(buf, cookie->cpu, NULL) < 0)
13308 13309
        return -1;

13310 13311 13312 13313 13314 13315 13316 13317
    return 0;
}


virSaveCookieCallbacks virQEMUDriverDomainSaveCookie = {
    .parse = qemuDomainSaveCookieParse,
    .format = qemuDomainSaveCookieFormat,
};
13318 13319 13320 13321 13322 13323 13324 13325 13326 13327 13328 13329 13330 13331 13332 13333 13334 13335 13336 13337 13338 13339 13340 13341 13342 13343 13344 13345 13346 13347 13348 13349 13350 13351 13352 13353 13354 13355 13356 13357


/**
 * 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;
}
13358

13359 13360 13361 13362 13363 13364 13365 13366 13367 13368 13369 13370 13371 13372 13373 13374 13375 13376 13377 13378 13379 13380 13381 13382 13383 13384 13385 13386

/**
 * 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;
13387
    int ret = -1;
13388 13389 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 13426 13427 13428 13429 13430 13431 13432 13433 13434

    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);
        VIR_STEAL_PTR(vm->def->cpu, fixedCPU);
    }

    if (fixedOrig) {
        virCPUDefFree(*origCPU);
        VIR_STEAL_PTR(*origCPU, fixedOrig);
    }

    ret = 0;

 cleanup:
    virCPUDefFree(fixedCPU);
    virCPUDefFree(fixedOrig);
    return ret;
}


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
/**
 * qemuDomainUpdateQEMUCaps:
 * @vm: domain object
 * @qemuCapsCache: cache of QEMU capabilities
 *
 * This function updates the used QEMU capabilities of @vm by querying
 * the QEMU capabilities cache.
 *
 * Returns 0 on success, -1 on error.
 */
int
qemuDomainUpdateQEMUCaps(virDomainObjPtr vm,
                         virFileCachePtr qemuCapsCache)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    virObjectUnref(priv->qemuCaps);
    if (!(priv->qemuCaps = virQEMUCapsCacheLookupCopy(qemuCapsCache,
                                                      vm->def->emulator,
                                                      vm->def->os.machine)))
        return -1;
    return 0;
}


13460 13461 13462 13463 13464 13465 13466
char *
qemuDomainGetMachineName(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    char *ret = NULL;

13467
    if (vm->pid > 0) {
13468 13469 13470 13471 13472 13473 13474 13475 13476 13477 13478
        ret = virSystemdGetMachineNameByPID(vm->pid);
        if (!ret)
            virResetLastError();
    }

    if (!ret)
        ret = virDomainGenerateMachineName("qemu", vm->def->id, vm->def->name,
                                           virQEMUDriverIsPrivileged(driver));

    return ret;
}
13479 13480 13481 13482 13483 13484 13485 13486 13487 13488 13489 13490 13491 13492 13493 13494 13495 13496 13497 13498 13499 13500 13501 13502


/* 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,
                                     virDomainDeviceInfo info,
                                     virQEMUCapsPtr qemuCaps,
                                     const char *devicename)
{
    if (info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
        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;
13503
        } else if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW)) {
13504 13505 13506 13507 13508 13509 13510 13511 13512 13513 13514 13515 13516 13517 13518
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CCW address type is not supported by "
                             "this QEMU"));
            return false;
        }
    } else if (info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390) {
        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;
}
13519 13520


13521
/**
13522
 * qemuDomainPrepareDiskSourceData:
13523 13524 13525 13526 13527
 *
 * @disk: Disk config object
 * @src: source to start from
 * @cfg: qemu driver config object
 *
13528 13529 13530
 * Prepares various aspects of a storage source belonging to a disk backing
 * chain. This function should be also called for detected backing chain
 * members.
13531 13532
 */
int
13533 13534 13535 13536
qemuDomainPrepareDiskSourceData(virDomainDiskDefPtr disk,
                                virStorageSourcePtr src,
                                virQEMUDriverConfigPtr cfg,
                                virQEMUCapsPtr qemuCaps)
13537
{
13538
    /* transfer properties valid only for the top level image */
13539 13540
    if (src == disk->src)
        src->detect_zeroes = disk->detect_zeroes;
13541

13542 13543 13544 13545 13546 13547 13548
    if (cfg &&
        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;
    }
13549

13550 13551 13552 13553
    /* transfer properties valid for the full chain */
    src->iomode = disk->iomode;
    src->cachemode = disk->cachemode;
    src->discard = disk->discard;
13554

13555 13556
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY)
        src->floppyimg = true;
13557 13558 13559 13560 13561

    return 0;
}


13562 13563 13564 13565 13566 13567 13568 13569 13570
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;
}


13571 13572
static int
qemuDomainPrepareStorageSourcePR(virStorageSourcePtr src,
13573 13574
                                 qemuDomainObjPrivatePtr priv,
                                 const char *parentalias)
13575 13576 13577 13578 13579
{
    if (!src->pr)
        return 0;

    if (virStoragePRDefIsManaged(src->pr)) {
13580
        VIR_FREE(src->pr->path);
13581 13582
        if (!(src->pr->path = qemuDomainGetManagedPRSocketPath(priv)))
            return -1;
13583 13584 13585 13586 13587
        if (VIR_STRDUP(src->pr->mgralias, qemuDomainGetManagedPRAlias()) < 0)
            return -1;
    } else {
        if (!(src->pr->mgralias = qemuDomainGetUnmanagedPRAlias(parentalias)))
            return -1;
13588 13589 13590 13591 13592 13593
    }

    return 0;
}


13594 13595 13596 13597 13598 13599 13600 13601 13602 13603 13604 13605
/**
 * 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)
13606
{
13607 13608 13609
    if (qemuDomainValidateStorageSource(disk->src, priv->qemuCaps) < 0)
        return -1;

13610
    if (qemuDomainPrepareDiskSourceData(disk, disk->src, cfg, priv->qemuCaps) < 0)
13611 13612
        return -1;

13613 13614 13615
    if (qemuDomainSecretStorageSourcePrepare(priv, disk->src,
                                             disk->info.alias,
                                             disk->info.alias) < 0)
13616 13617
        return -1;

13618
    if (qemuDomainPrepareStorageSourcePR(disk->src, priv, disk->info.alias) < 0)
13619
        return -1;
13620

13621 13622
    if (qemuDomainPrepareStorageSourceTLS(disk->src, cfg, disk->info.alias,
                                          priv->qemuCaps) < 0)
13623 13624
        return -1;

13625 13626 13627 13628
    return 0;
}


13629 13630 13631 13632 13633 13634 13635 13636 13637 13638 13639 13640 13641 13642 13643 13644 13645 13646 13647 13648 13649 13650 13651
static int
qemuDomainPrepareStorageSourceBlockdev(virDomainDiskDefPtr disk,
                                       virStorageSourcePtr src,
                                       qemuDomainObjPrivatePtr priv,
                                       virQEMUDriverConfigPtr cfg)
{
    src->id = qemuDomainStorageIdNew(priv);

    if (virAsprintf(&src->nodestorage, "libvirt-%u-storage", src->id) < 0 ||
        virAsprintf(&src->nodeformat, "libvirt-%u-format", src->id) < 0)
        return -1;

    if (qemuDomainValidateStorageSource(src, priv->qemuCaps) < 0)
        return -1;

    if (qemuDomainPrepareDiskSourceData(disk, src, cfg, priv->qemuCaps) < 0)
        return -1;

    if (qemuDomainSecretStorageSourcePrepare(priv, src,
                                             src->nodestorage,
                                             src->nodeformat) < 0)
        return -1;

13652
    if (qemuDomainPrepareStorageSourcePR(src, priv, src->nodestorage) < 0)
13653 13654
        return -1;

13655
    if (qemuDomainPrepareStorageSourceTLS(src, cfg, src->nodestorage,
13656 13657 13658 13659 13660 13661 13662 13663 13664 13665 13666 13667 13668 13669 13670 13671 13672 13673 13674 13675 13676 13677 13678 13679 13680 13681 13682 13683 13684
                                          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 &&
        !diskPriv->nodeCopyOnRead &&
        virAsprintf(&diskPriv->nodeCopyOnRead, "libvirt-CoR-%s", disk->dst) < 0)
        return -1;

    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if (qemuDomainPrepareStorageSourceBlockdev(disk, n, priv, cfg) < 0)
            return -1;
    }

    return 0;
}


13685 13686 13687 13688 13689 13690 13691
int
qemuDomainPrepareDiskSource(virDomainDiskDefPtr disk,
                            qemuDomainObjPrivatePtr priv,
                            virQEMUDriverConfigPtr cfg)
{
    qemuDomainPrepareDiskCachemode(disk);

13692 13693 13694 13695 13696 13697 13698
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if (qemuDomainPrepareDiskSourceBlockdev(disk, priv, cfg) < 0)
            return -1;
    } else {
        if (qemuDomainPrepareDiskSourceLegacy(disk, priv, cfg) < 0)
            return -1;
    }
13699

13700 13701
    return 0;
}
13702 13703


13704 13705 13706 13707 13708 13709 13710 13711 13712 13713 13714 13715 13716 13717 13718 13719 13720 13721 13722 13723 13724 13725 13726 13727 13728 13729 13730 13731 13732 13733 13734 13735 13736 13737 13738 13739 13740 13741 13742 13743 13744 13745 13746 13747 13748 13749 13750 13751 13752 13753 13754 13755 13756 13757 13758 13759 13760 13761 13762 13763 13764 13765 13766 13767 13768 13769 13770 13771 13772 13773 13774 13775 13776 13777 13778
/**
 * 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;
}


13779 13780 13781 13782 13783 13784 13785 13786 13787 13788
void
qemuProcessEventFree(struct qemuProcessEvent *event)
{
    if (!event)
        return;

    switch (event->eventType) {
    case QEMU_PROCESS_EVENT_GUESTPANIC:
        qemuMonitorEventPanicInfoFree(event->data);
        break;
13789 13790 13791
    case QEMU_PROCESS_EVENT_RDMA_GID_STATUS_CHANGED:
        qemuMonitorEventRdmaGidStatusFree(event->data);
        break;
13792 13793 13794 13795 13796 13797 13798 13799
    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;
13800
    case QEMU_PROCESS_EVENT_PR_DISCONNECT:
13801 13802 13803 13804 13805
    case QEMU_PROCESS_EVENT_LAST:
        break;
    }
    VIR_FREE(event);
}
13806 13807 13808


char *
13809
qemuDomainGetManagedPRSocketPath(qemuDomainObjPrivatePtr priv)
13810 13811 13812
{
    char *ret = NULL;

13813 13814
    ignore_value(virAsprintf(&ret, "%s/%s.sock", priv->libDir,
                             qemuDomainGetManagedPRAlias()));
13815 13816 13817

    return ret;
}
13818 13819 13820 13821 13822 13823 13824 13825 13826 13827 13828 13829 13830 13831 13832 13833 13834 13835 13836 13837 13838 13839 13840 13841 13842 13843 13844


/**
 * 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;
}
13845 13846 13847 13848 13849 13850 13851 13852 13853 13854 13855 13856 13857 13858 13859 13860 13861 13862 13863 13864 13865 13866 13867 13868 13869 13870 13871 13872 13873


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
13874 13875 13876 13877 13878 13879 13880 13881 13882 13883 13884 13885 13886 13887 13888 13889 13890


/* qemuDomainIsUsingNoShutdown:
 * @priv: Domain private data
 *
 * If JSON monitor is enabled, 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.
 *
 * Returns: @true when -no-shutdown either should be or was added to the
 * command line.
 */
bool
qemuDomainIsUsingNoShutdown(qemuDomainObjPrivatePtr priv)
{
    return priv->monJSON && priv->allowReboot == VIR_TRISTATE_BOOL_YES;
}
13891 13892 13893 13894 13895 13896 13897 13898 13899


bool
qemuDomainDiskIsMissingLocalOptional(virDomainDiskDefPtr disk)
{
    return disk->startupPolicy == VIR_DOMAIN_STARTUP_POLICY_OPTIONAL &&
           virStorageSourceIsLocalStorage(disk->src) && disk->src->path &&
           !virFileExists(disk->src->path);
}