qemu_domain.c 445.7 KB
Newer Older
1
/*
2
 * qemu_domain.c: QEMU domain private state
3
 *
4
 * Copyright (C) 2006-2019 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24
 */

#include <config.h>

#include "qemu_domain.h"
J
John Ferlan 已提交
25
#include "qemu_alias.h"
26
#include "qemu_block.h"
27
#include "qemu_cgroup.h"
28
#include "qemu_command.h"
29
#include "qemu_process.h"
30
#include "qemu_capabilities.h"
31
#include "qemu_migration.h"
32
#include "qemu_migration_params.h"
33
#include "qemu_security.h"
34
#include "qemu_extdevice.h"
35
#include "viralloc.h"
36
#include "virlog.h"
37
#include "virerror.h"
38
#include "c-ctype.h"
39
#include "cpu/cpu.h"
40
#include "viruuid.h"
E
Eric Blake 已提交
41
#include "virfile.h"
42
#include "domain_addr.h"
43
#include "domain_event.h"
44
#include "virtime.h"
45
#include "virnetdevopenvswitch.h"
46
#include "virstoragefile.h"
47
#include "virstring.h"
48
#include "virthreadjob.h"
49
#include "viratomic.h"
50
#include "virprocess.h"
51
#include "vircrypto.h"
52
#include "virrandom.h"
53
#include "virsystemd.h"
54
#include "secret_util.h"
55
#include "logging/log_manager.h"
56
#include "locking/domain_lock.h"
57
#include "virdomainsnapshotobjlist.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 QEMU_QXL_VGAMEM_DEFAULT 16 * 1024

78 79
#define VIR_FROM_THIS VIR_FROM_QEMU

80 81
VIR_LOG_INIT("qemu.qemu_domain");

82 83
#define QEMU_NAMESPACE_HREF "http://libvirt.org/schemas/domain/qemu/1.0"

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

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

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

115 116
VIR_ENUM_IMPL(qemuDomainNamespace,
              QEMU_DOMAIN_NS_LAST,
117 118 119 120
              "mount",
);


121 122
#define PROC_MOUNTS "/proc/mounts"
#define DEVPREFIX "/dev/"
123
#define DEV_VFIO "/dev/vfio/vfio"
124
#define DEVICE_MAPPER_CONTROL_PATH "/dev/mapper/control"
125
#define DEV_SEV "/dev/sev"
126

127

128
struct _qemuDomainLogContext {
129 130
    virObject parent;

131
    int writefd;
132
    int readfd; /* Only used if manager == NULL */
133
    off_t pos;
134
    ino_t inode; /* Only used if manager != NULL */
135
    char *path;
136
    virLogManagerPtr manager;
137 138
};

139
static virClassPtr qemuDomainLogContextClass;
140
static virClassPtr qemuDomainSaveCookieClass;
141 142

static void qemuDomainLogContextDispose(void *obj);
143
static void qemuDomainSaveCookieDispose(void *obj);
144

145 146 147 148 149 150 151 152

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


153
static int
154
qemuDomainOnceInit(void)
155
{
156
    if (!VIR_CLASS_NEW(qemuDomainLogContext, virClassForObject()))
157 158
        return -1;

159
    if (!VIR_CLASS_NEW(qemuDomainSaveCookie, virClassForObject()))
160 161
        return -1;

162 163 164
    return 0;
}

165
VIR_ONCE_GLOBAL_INIT(qemuDomain);
166 167 168 169 170 171 172 173 174 175 176 177 178

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
179
const char *
180
qemuDomainAsyncJobPhaseToString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
181 182 183 184 185
                                int phase ATTRIBUTE_UNUSED)
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
186 187
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
188 189
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
190
    case QEMU_ASYNC_JOB_SNAPSHOT:
191
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
192
    case QEMU_ASYNC_JOB_NONE:
M
Marc Hartmayer 已提交
193
        ATTRIBUTE_FALLTHROUGH;
194 195
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
196 197 198 199 200 201
    }

    return "none";
}

int
202
qemuDomainAsyncJobPhaseFromString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
203 204 205 206 207 208 209 210
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
211 212
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
213 214
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
215
    case QEMU_ASYNC_JOB_SNAPSHOT:
216
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
217
    case QEMU_ASYNC_JOB_NONE:
M
Marc Hartmayer 已提交
218
        ATTRIBUTE_FALLTHROUGH;
219 220
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
221 222 223 224 225 226 227 228
    }

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

229

230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261
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;
}


262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277
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;
        }
    }
}


278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297
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);
298
    virObjectEventStateQueue(driver->domainEventState, event);
299 300 301
}


302 303 304 305 306 307 308 309
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

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

310
    if (virCondInit(&priv->job.asyncCond) < 0) {
311
        virCondDestroy(&priv->job.cond);
312 313 314
        return -1;
    }

315 316 317 318 319 320
    return 0;
}

static void
qemuDomainObjResetJob(qemuDomainObjPrivatePtr priv)
{
321
    qemuDomainJobObjPtr job = &priv->job;
322 323

    job->active = QEMU_JOB_NONE;
324
    job->owner = 0;
325
    job->ownerAPI = NULL;
J
Jiri Denemark 已提交
326
    job->started = 0;
327 328
}

329 330 331 332 333 334 335 336 337 338 339 340 341

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

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


342 343 344
static void
qemuDomainObjResetAsyncJob(qemuDomainObjPrivatePtr priv)
{
345
    qemuDomainJobObjPtr job = &priv->job;
346 347

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
348
    job->asyncOwner = 0;
349
    job->asyncOwnerAPI = NULL;
J
Jiri Denemark 已提交
350
    job->asyncStarted = 0;
J
Jiri Denemark 已提交
351
    job->phase = 0;
352
    job->mask = QEMU_JOB_DEFAULT_MASK;
353
    job->abortJob = false;
354
    job->spiceMigration = false;
355
    job->spiceMigrated = false;
356 357
    job->dumpCompleted = false;
    VIR_FREE(job->error);
J
Jiri Denemark 已提交
358
    VIR_FREE(job->current);
359 360
    qemuMigrationParamsFree(job->migParams);
    job->migParams = NULL;
361
    job->apiFlags = 0;
362 363
}

364 365
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
366
                        qemuDomainJobObjPtr job)
367 368 369 370 371
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
372
    job->owner = priv->job.owner;
373
    job->asyncJob = priv->job.asyncJob;
374
    job->asyncOwner = priv->job.asyncOwner;
J
Jiri Denemark 已提交
375
    job->phase = priv->job.phase;
376
    VIR_STEAL_PTR(job->migParams, priv->job.migParams);
377
    job->apiFlags = priv->job.apiFlags;
378 379 380 381 382

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

383 384 385
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
J
Jiri Denemark 已提交
386
    VIR_FREE(priv->job.current);
387
    VIR_FREE(priv->job.completed);
388 389
    virCondDestroy(&priv->job.cond);
    virCondDestroy(&priv->job.asyncCond);
390 391
}

392
static bool
393
qemuDomainTrackJob(qemuDomainJob job)
394 395 396 397
{
    return (QEMU_DOMAIN_TRACK_JOBS & JOB_MASK(job)) != 0;
}

398

J
Jiri Denemark 已提交
399 400 401 402 403 404 405 406 407 408 409
int
qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->started)
        return 0;

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

410 411 412 413 414 415
    if (now < jobInfo->started) {
        VIR_WARN("Async job starts in the future");
        jobInfo->started = 0;
        return 0;
    }

J
Jiri Denemark 已提交
416 417 418 419
    jobInfo->timeElapsed = now - jobInfo->started;
    return 0;
}

420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436
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;
    }

437 438
    jobInfo->stats.mig.downtime = now - jobInfo->stopped;
    jobInfo->stats.mig.downtime_set = true;
439 440 441
    return 0;
}

442 443 444 445 446 447 448 449
static virDomainJobType
qemuDomainJobStatusToType(qemuDomainJobStatus status)
{
    switch (status) {
    case QEMU_DOMAIN_JOB_STATUS_NONE:
        break;

    case QEMU_DOMAIN_JOB_STATUS_ACTIVE:
450
    case QEMU_DOMAIN_JOB_STATUS_MIGRATING:
451
    case QEMU_DOMAIN_JOB_STATUS_QEMU_COMPLETED:
452
    case QEMU_DOMAIN_JOB_STATUS_POSTCOPY:
453
    case QEMU_DOMAIN_JOB_STATUS_PAUSED:
454 455 456 457 458 459 460 461 462 463 464 465 466 467 468
        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
469 470 471 472
int
qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                        virDomainJobInfoPtr info)
{
473
    info->type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
474 475
    info->timeElapsed = jobInfo->timeElapsed;

476 477 478 479 480 481 482 483 484 485 486 487 488
    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
489

490 491 492 493 494 495
    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;

496
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
497 498 499 500 501
        info->memTotal = jobInfo->stats.dump.total;
        info->memProcessed = jobInfo->stats.dump.completed;
        info->memRemaining = info->memTotal - info->memProcessed;
        break;

502 503 504
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
        break;
    }
J
Jiri Denemark 已提交
505 506 507 508 509 510 511 512

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

    return 0;
}

513 514 515 516 517 518

static int
qemuDomainMigrationJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                                   int *type,
                                   virTypedParameterPtr *params,
                                   int *nparams)
J
Jiri Denemark 已提交
519
{
520
    qemuMonitorMigrationStats *stats = &jobInfo->stats.mig;
521
    qemuDomainMirrorStatsPtr mirrorStats = &jobInfo->mirrorStats;
J
Jiri Denemark 已提交
522 523 524
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;
525 526
    unsigned long long mirrorRemaining = mirrorStats->total -
                                         mirrorStats->transferred;
J
Jiri Denemark 已提交
527

528 529 530 531 532
    if (virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_OPERATION,
                             jobInfo->operation) < 0)
        goto error;

J
Jiri Denemark 已提交
533 534 535 536 537
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED,
                                jobInfo->timeElapsed) < 0)
        goto error;

538 539 540 541 542 543 544
    if (jobInfo->timeDeltaSet &&
        jobInfo->timeElapsed > jobInfo->timeDelta &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED_NET,
                                jobInfo->timeElapsed - jobInfo->timeDelta) < 0)
        goto error;

545
    if (stats->downtime_set &&
J
Jiri Denemark 已提交
546 547
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME,
548
                                stats->downtime) < 0)
J
Jiri Denemark 已提交
549 550
        goto error;

551
    if (stats->downtime_set &&
552
        jobInfo->timeDeltaSet &&
553
        stats->downtime > jobInfo->timeDelta &&
554 555
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME_NET,
556
                                stats->downtime - jobInfo->timeDelta) < 0)
557 558
        goto error;

559
    if (stats->setup_time_set &&
560 561
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_SETUP_TIME,
562
                                stats->setup_time) < 0)
563 564
        goto error;

J
Jiri Denemark 已提交
565 566
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_TOTAL,
567
                                stats->ram_total +
568 569
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
570 571
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_PROCESSED,
572
                                stats->ram_transferred +
573 574
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
575 576
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_REMAINING,
577
                                stats->ram_remaining +
578 579
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
580 581 582 583
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
584
                                stats->ram_total) < 0 ||
J
Jiri Denemark 已提交
585 586
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
587
                                stats->ram_transferred) < 0 ||
J
Jiri Denemark 已提交
588 589
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
590
                                stats->ram_remaining) < 0)
J
Jiri Denemark 已提交
591 592
        goto error;

593
    if (stats->ram_bps &&
594 595
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_BPS,
596
                                stats->ram_bps) < 0)
597 598
        goto error;

599
    if (stats->ram_duplicate_set) {
J
Jiri Denemark 已提交
600 601
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_CONSTANT,
602
                                    stats->ram_duplicate) < 0 ||
J
Jiri Denemark 已提交
603 604
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL,
605
                                    stats->ram_normal) < 0 ||
J
Jiri Denemark 已提交
606 607
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL_BYTES,
608
                                    stats->ram_normal_bytes) < 0)
J
Jiri Denemark 已提交
609 610 611
            goto error;
    }

612 613 614 615 616
    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,
617 618 619 620
                                stats->ram_iteration) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_POSTCOPY_REQS,
                                stats->ram_postcopy_reqs) < 0)
621 622
        goto error;

623 624 625 626 627 628
    if (stats->ram_page_size > 0 &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PAGE_SIZE,
                                stats->ram_page_size) < 0)
        goto error;

629 630 631 632 633
    /* 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
634 635
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_TOTAL,
636 637
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
638 639
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_PROCESSED,
640 641
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
642 643
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_REMAINING,
644 645
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
646 647
        goto error;

648
    if (stats->disk_bps &&
649 650
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_BPS,
651
                                stats->disk_bps) < 0)
652 653
        goto error;

654
    if (stats->xbzrle_set) {
J
Jiri Denemark 已提交
655 656
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE,
657
                                    stats->xbzrle_cache_size) < 0 ||
J
Jiri Denemark 已提交
658 659
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_BYTES,
660
                                    stats->xbzrle_bytes) < 0 ||
J
Jiri Denemark 已提交
661 662
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_PAGES,
663
                                    stats->xbzrle_pages) < 0 ||
J
Jiri Denemark 已提交
664 665
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE_MISSES,
666
                                    stats->xbzrle_cache_miss) < 0 ||
J
Jiri Denemark 已提交
667 668
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_OVERFLOW,
669
                                    stats->xbzrle_overflow) < 0)
J
Jiri Denemark 已提交
670 671 672
            goto error;
    }

673 674 675 676 677 678
    if (stats->cpu_throttle_percentage &&
        virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_AUTO_CONVERGE_THROTTLE,
                             stats->cpu_throttle_percentage) < 0)
        goto error;

679
 done:
680
    *type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
681 682 683 684 685 686 687 688 689 690
    *params = par;
    *nparams = npar;
    return 0;

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


691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733
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;
}


734 735 736 737 738 739 740 741
int
qemuDomainJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                          int *type,
                          virTypedParameterPtr *params,
                          int *nparams)
{
    switch (jobInfo->statsType) {
    case QEMU_DOMAIN_JOB_STATS_TYPE_MIGRATION:
742
    case QEMU_DOMAIN_JOB_STATS_TYPE_SAVEDUMP:
743 744
        return qemuDomainMigrationJobInfoToParams(jobInfo, type, params, nparams);

745
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
746 747
        return qemuDomainDumpJobInfoToParams(jobInfo, type, params, nparams);

748
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
749 750 751 752 753 754
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid job statistics type"));
        break;

    default:
        virReportEnumRangeError(qemuDomainJobStatsType, jobInfo->statsType);
755 756 757 758 759 760 761
        break;
    }

    return -1;
}


J
John Ferlan 已提交
762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783
/* 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:
784 785
 * @driver: qemu driver data
 * @vm: Pointer to the vm object
J
John Ferlan 已提交
786 787 788 789 790
 *
 * 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
 */
791
int
M
Martin Kletzander 已提交
792 793
qemuDomainWriteMasterKeyFile(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
J
John Ferlan 已提交
794 795 796 797
{
    char *path;
    int fd = -1;
    int ret = -1;
M
Martin Kletzander 已提交
798
    qemuDomainObjPrivatePtr priv = vm->privateData;
J
John Ferlan 已提交
799

800 801 802 803
    /* Only gets filled in if we have the capability */
    if (!priv->masterKey)
        return 0;

J
John Ferlan 已提交
804 805 806 807 808 809 810 811 812 813 814 815 816 817 818
    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;
    }

819
    if (qemuSecurityDomainSetPathLabel(driver, vm, path, false) < 0)
M
Martin Kletzander 已提交
820 821
        goto cleanup;

J
John Ferlan 已提交
822 823 824 825 826 827 828 829 830 831
    ret = 0;

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

    return ret;
}


832 833 834 835 836 837 838 839 840
static void
qemuDomainMasterKeyFree(qemuDomainObjPrivatePtr priv)
{
    if (!priv->masterKey)
        return;

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

J
John Ferlan 已提交
841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938
/* 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 */
939
    qemuDomainMasterKeyFree(priv);
J
John Ferlan 已提交
940 941 942 943 944 945 946 947 948 949

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

    VIR_FREE(path);
}


/* qemuDomainMasterKeyCreate:
950
 * @vm: Pointer to the domain object
J
John Ferlan 已提交
951 952 953 954 955 956 957 958
 *
 * 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
959
qemuDomainMasterKeyCreate(virDomainObjPtr vm)
J
John Ferlan 已提交
960
{
M
Martin Kletzander 已提交
961 962
    qemuDomainObjPrivatePtr priv = vm->privateData;

J
John Ferlan 已提交
963 964 965 966
    /* If we don't have the capability, then do nothing. */
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

967
    if (VIR_ALLOC_N(priv->masterKey, QEMU_DOMAIN_MASTER_KEY_LEN) < 0)
968
        return -1;
J
John Ferlan 已提交
969 970
    priv->masterKeyLen = QEMU_DOMAIN_MASTER_KEY_LEN;

971
    if (virRandomBytes(priv->masterKey, priv->masterKeyLen) < 0) {
972 973 974 975
        VIR_DISPOSE_N(priv->masterKey, priv->masterKeyLen);
        return -1;
    }

J
John Ferlan 已提交
976 977 978 979
    return 0;
}


980
static void
981
qemuDomainSecretPlainClear(qemuDomainSecretPlainPtr secret)
982
{
983 984
    VIR_FREE(secret->username);
    VIR_DISPOSE_N(secret->secret, secret->secretlen);
985 986 987
}


J
John Ferlan 已提交
988
static void
989
qemuDomainSecretAESClear(qemuDomainSecretAESPtr secret,
990
                         bool keepAlias)
J
John Ferlan 已提交
991
{
992
    if (!keepAlias)
993
        VIR_FREE(secret->alias);
994

995 996 997
    VIR_FREE(secret->username);
    VIR_FREE(secret->iv);
    VIR_FREE(secret->ciphertext);
J
John Ferlan 已提交
998 999 1000
}


1001 1002 1003
static void
qemuDomainSecretInfoClear(qemuDomainSecretInfoPtr secinfo,
                          bool keepAlias)
1004
{
1005
    if (!secinfo)
1006 1007
        return;

1008
    switch ((qemuDomainSecretInfoType) secinfo->type) {
J
John Ferlan 已提交
1009
    case VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN:
1010
        qemuDomainSecretPlainClear(&secinfo->s.plain);
J
John Ferlan 已提交
1011 1012
        break;

1013
    case VIR_DOMAIN_SECRET_INFO_TYPE_AES:
1014
        qemuDomainSecretAESClear(&secinfo->s.aes, keepAlias);
J
John Ferlan 已提交
1015 1016 1017 1018 1019
        break;

    case VIR_DOMAIN_SECRET_INFO_TYPE_LAST:
        break;
    }
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029
}


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

    qemuDomainSecretInfoClear(*secinfo, false);
J
John Ferlan 已提交
1030

1031 1032 1033 1034
    VIR_FREE(*secinfo);
}


1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
/**
 * qemuDomainSecretInfoDestroy:
 * @secinfo: object to destroy
 *
 * Removes any data unnecessary for further use, but keeps alias allocated.
 */
void
qemuDomainSecretInfoDestroy(qemuDomainSecretInfoPtr secinfo)
{
    qemuDomainSecretInfoClear(secinfo, true);
}


1048
static virClassPtr qemuDomainDiskPrivateClass;
1049
static void qemuDomainDiskPrivateDispose(void *obj);
1050 1051 1052 1053

static int
qemuDomainDiskPrivateOnceInit(void)
{
1054
    if (!VIR_CLASS_NEW(qemuDomainDiskPrivate, virClassForObject()))
1055
        return -1;
1056 1057

    return 0;
1058 1059
}

1060
VIR_ONCE_GLOBAL_INIT(qemuDomainDiskPrivate);
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075

static virObjectPtr
qemuDomainDiskPrivateNew(void)
{
    qemuDomainDiskPrivatePtr priv;

    if (qemuDomainDiskPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}

1076 1077 1078 1079 1080
static void
qemuDomainDiskPrivateDispose(void *obj)
{
    qemuDomainDiskPrivatePtr priv = obj;

1081
    virObjectUnref(priv->migrSource);
1082
    VIR_FREE(priv->qomName);
1083
    VIR_FREE(priv->nodeCopyOnRead);
1084
    virObjectUnref(priv->blockjob);
1085
}
1086

1087 1088 1089 1090 1091 1092
static virClassPtr qemuDomainStorageSourcePrivateClass;
static void qemuDomainStorageSourcePrivateDispose(void *obj);

static int
qemuDomainStorageSourcePrivateOnceInit(void)
{
1093
    if (!VIR_CLASS_NEW(qemuDomainStorageSourcePrivate, virClassForObject()))
1094
        return -1;
1095 1096

    return 0;
1097 1098
}

1099
VIR_ONCE_GLOBAL_INIT(qemuDomainStorageSourcePrivate);
1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125

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);
}


1126 1127 1128 1129 1130 1131
static virClassPtr qemuDomainVcpuPrivateClass;
static void qemuDomainVcpuPrivateDispose(void *obj);

static int
qemuDomainVcpuPrivateOnceInit(void)
{
1132
    if (!VIR_CLASS_NEW(qemuDomainVcpuPrivate, virClassForObject()))
1133
        return -1;
1134 1135

    return 0;
1136 1137
}

1138
VIR_ONCE_GLOBAL_INIT(qemuDomainVcpuPrivate);
1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155

static virObjectPtr
qemuDomainVcpuPrivateNew(void)
{
    qemuDomainVcpuPrivatePtr priv;

    if (qemuDomainVcpuPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}


static void
1156
qemuDomainVcpuPrivateDispose(void *obj)
1157
{
1158 1159 1160 1161
    qemuDomainVcpuPrivatePtr priv = obj;

    VIR_FREE(priv->type);
    VIR_FREE(priv->alias);
1162 1163 1164 1165
    return;
}


1166 1167
static virClassPtr qemuDomainChrSourcePrivateClass;
static void qemuDomainChrSourcePrivateDispose(void *obj);
1168 1169

static int
1170
qemuDomainChrSourcePrivateOnceInit(void)
1171
{
1172
    if (!VIR_CLASS_NEW(qemuDomainChrSourcePrivate, virClassForObject()))
1173
        return -1;
1174 1175

    return 0;
1176 1177
}

1178
VIR_ONCE_GLOBAL_INIT(qemuDomainChrSourcePrivate);
1179 1180

static virObjectPtr
1181
qemuDomainChrSourcePrivateNew(void)
1182
{
1183
    qemuDomainChrSourcePrivatePtr priv;
1184

1185
    if (qemuDomainChrSourcePrivateInitialize() < 0)
1186 1187
        return NULL;

1188
    if (!(priv = virObjectNew(qemuDomainChrSourcePrivateClass)))
1189 1190 1191 1192 1193 1194 1195
        return NULL;

    return (virObjectPtr) priv;
}


static void
1196
qemuDomainChrSourcePrivateDispose(void *obj)
1197
{
1198
    qemuDomainChrSourcePrivatePtr priv = obj;
1199 1200 1201 1202 1203

    qemuDomainSecretInfoFree(&priv->secinfo);
}


J
Ján Tomko 已提交
1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215
static virClassPtr qemuDomainVsockPrivateClass;
static void qemuDomainVsockPrivateDispose(void *obj);

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

    return 0;
}

1216
VIR_ONCE_GLOBAL_INIT(qemuDomainVsockPrivate);
J
Ján Tomko 已提交
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228

static virObjectPtr
qemuDomainVsockPrivateNew(void)
{
    qemuDomainVsockPrivatePtr priv;

    if (qemuDomainVsockPrivateInitialize() < 0)
        return NULL;

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

1229 1230
    priv->vhostfd = -1;

J
Ján Tomko 已提交
1231 1232 1233 1234 1235 1236 1237
    return (virObjectPtr) priv;
}


static void
qemuDomainVsockPrivateDispose(void *obj ATTRIBUTE_UNUSED)
{
1238 1239 1240
    qemuDomainVsockPrivatePtr priv = obj;

    VIR_FORCE_CLOSE(priv->vhostfd);
J
Ján Tomko 已提交
1241 1242 1243
}


1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255
static virClassPtr qemuDomainGraphicsPrivateClass;
static void qemuDomainGraphicsPrivateDispose(void *obj);

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

    return 0;
}

1256
VIR_ONCE_GLOBAL_INIT(qemuDomainGraphicsPrivate);
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278

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);
1279
    qemuDomainSecretInfoFree(&priv->secinfo);
1280 1281 1282
}


1283 1284
/* qemuDomainSecretPlainSetup:
 * @secinfo: Pointer to secret info
J
John Ferlan 已提交
1285
 * @usageType: The virSecretUsageType
1286 1287
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1288 1289 1290 1291 1292 1293
 *
 * Taking a secinfo, fill in the plaintext information
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1294
qemuDomainSecretPlainSetup(qemuDomainSecretInfoPtr secinfo,
J
John Ferlan 已提交
1295
                           virSecretUsageType usageType,
1296 1297
                           const char *username,
                           virSecretLookupTypeDefPtr seclookupdef)
1298
{
1299 1300 1301 1302 1303 1304 1305
    virConnectPtr conn;
    int ret = -1;

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

1306
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN;
1307
    if (VIR_STRDUP(secinfo->s.plain.username, username) < 0)
1308
        goto cleanup;
1309

1310 1311 1312 1313 1314 1315 1316
    ret = virSecretGetSecretString(conn, seclookupdef, usageType,
                                   &secinfo->s.plain.secret,
                                   &secinfo->s.plain.secretlen);

 cleanup:
    virObjectUnref(conn);
    return ret;
1317 1318 1319
}


1320 1321 1322 1323
/* 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
1324
 * @usageType: The virSecretUsageType
1325 1326
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1327
 * @isLuks: True/False for is for luks (alias generation)
1328 1329 1330 1331 1332 1333
 *
 * Taking a secinfo, fill in the AES specific information using the
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1334
qemuDomainSecretAESSetup(qemuDomainObjPrivatePtr priv,
1335 1336
                         qemuDomainSecretInfoPtr secinfo,
                         const char *srcalias,
J
John Ferlan 已提交
1337
                         virSecretUsageType usageType,
1338
                         const char *username,
1339 1340
                         virSecretLookupTypeDefPtr seclookupdef,
                         bool isLuks)
1341
{
1342
    virConnectPtr conn;
1343 1344 1345 1346 1347 1348 1349 1350
    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;

1351 1352 1353 1354
    conn = virGetConnectSecret();
    if (!conn)
        return -1;

1355
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_AES;
1356
    if (VIR_STRDUP(secinfo->s.aes.username, username) < 0)
1357
        goto cleanup;
1358

1359
    if (!(secinfo->s.aes.alias = qemuDomainGetSecretAESAlias(srcalias, isLuks)))
1360
        goto cleanup;
1361

1362 1363 1364
    if (VIR_ALLOC_N(raw_iv, ivlen) < 0)
        goto cleanup;

1365
    /* Create a random initialization vector */
1366
    if (virRandomBytes(raw_iv, ivlen) < 0)
1367
        goto cleanup;
1368 1369 1370 1371 1372 1373

    /* 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
1374
    if (virSecretGetSecretString(conn, seclookupdef, usageType,
1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397
                                 &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);
1398
    virObjectUnref(conn);
1399 1400 1401 1402
    return ret;
}


1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418
/**
 * 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;
}


1419
/* qemuDomainSecretInfoNewPlain:
1420
 * @usageType: Secret usage type
1421 1422
 * @username: username
 * @lookupDef: lookup def describing secret
1423
 *
1424
 * Helper function to create a secinfo to be used for secinfo consumers. This
1425
 * sets up a 'plain' (unencrypted) secret for legacy consumers.
1426 1427 1428 1429 1430
 *
 * Returns @secinfo on success, NULL on failure. Caller is responsible
 * to eventually free @secinfo.
 */
static qemuDomainSecretInfoPtr
1431
qemuDomainSecretInfoNewPlain(virSecretUsageType usageType,
1432
                             const char *username,
1433
                             virSecretLookupTypeDefPtr lookupDef)
1434 1435 1436 1437 1438 1439
{
    qemuDomainSecretInfoPtr secinfo = NULL;

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

1440 1441 1442
    if (qemuDomainSecretPlainSetup(secinfo, usageType, username, lookupDef) < 0) {
        qemuDomainSecretInfoFree(&secinfo);
        return NULL;
1443 1444 1445 1446 1447 1448
    }

    return secinfo;
}


1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491
/* 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;
}


1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502
/**
 * 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.
 */
1503
qemuDomainSecretInfoPtr
1504
qemuDomainSecretInfoTLSNew(qemuDomainObjPrivatePtr priv,
1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517
                           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;

1518 1519 1520
    return qemuDomainSecretInfoNew(priv, srcAlias,
                                   VIR_SECRET_USAGE_TYPE_TLS, NULL,
                                   &seclookupdef, false);
1521 1522 1523
}


1524 1525 1526
void
qemuDomainSecretDiskDestroy(virDomainDiskDefPtr disk)
{
1527 1528
    qemuDomainStorageSourcePrivatePtr srcPriv;
    virStorageSourcePtr n;
1529

1530 1531 1532 1533 1534 1535
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if ((srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(n))) {
            qemuDomainSecretInfoDestroy(srcPriv->secinfo);
            qemuDomainSecretInfoDestroy(srcPriv->encinfo);
        }
    }
1536 1537 1538
}


1539
bool
1540
qemuDomainStorageSourceHasAuth(virStorageSourcePtr src)
1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552
{
    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;
}


1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564
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;
}


1565 1566 1567 1568 1569 1570
/**
 * 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
1571
 *
1572 1573 1574 1575
 * 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.
1576
 *
1577
 * Returns 0 on success; -1 on error while reporting an libvirt error.
1578
 */
1579
static int
1580
qemuDomainSecretStorageSourcePrepare(qemuDomainObjPrivatePtr priv,
1581 1582 1583
                                     virStorageSourcePtr src,
                                     const char *authalias,
                                     const char *encalias)
1584
{
1585
    qemuDomainStorageSourcePrivatePtr srcPriv;
1586
    bool iscsiHasPS = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET);
1587
    bool hasAuth = qemuDomainStorageSourceHasAuth(src);
1588 1589 1590 1591
    bool hasEnc = qemuDomainDiskHasEncryptionSecret(src);

    if (!hasAuth && !hasEnc)
        return 0;
1592

1593
    if (!(src->privateData = qemuDomainStorageSourcePrivateNew()))
1594 1595
        return -1;

1596
    srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
1597

1598
    if (hasAuth) {
J
John Ferlan 已提交
1599
        virSecretUsageType usageType = VIR_SECRET_USAGE_TYPE_ISCSI;
1600

1601
        if (src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD)
J
John Ferlan 已提交
1602
            usageType = VIR_SECRET_USAGE_TYPE_CEPH;
1603

1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618
        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;
1619 1620
    }

1621
    if (hasEnc) {
1622
        if (!(srcPriv->encinfo =
1623 1624 1625 1626
              qemuDomainSecretInfoNew(priv, encalias,
                                      VIR_SECRET_USAGE_TYPE_VOLUME, NULL,
                                      &src->encryption->secrets[0]->seclookupdef,
                                      true)))
1627
              return -1;
1628 1629
    }

1630 1631 1632 1633
    return 0;
}


1634 1635 1636
void
qemuDomainSecretHostdevDestroy(virDomainHostdevDefPtr hostdev)
{
1637
    qemuDomainStorageSourcePrivatePtr srcPriv;
1638

1639 1640 1641
    if (virHostdevIsSCSIDevice(hostdev)) {
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1642

1643 1644 1645 1646 1647 1648
        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);
        }
    }
1649 1650 1651 1652
}


/* qemuDomainSecretHostdevPrepare:
J
John Ferlan 已提交
1653
 * @priv: pointer to domain private object
1654 1655 1656 1657 1658 1659 1660
 * @hostdev: Pointer to a hostdev definition
 *
 * For the right host device, generate the qemuDomainSecretInfo structure.
 *
 * Returns 0 on success, -1 on failure
 */
int
1661
qemuDomainSecretHostdevPrepare(qemuDomainObjPrivatePtr priv,
1662 1663
                               virDomainHostdevDefPtr hostdev)
{
1664
    if (virHostdevIsSCSIDevice(hostdev)) {
1665 1666
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1667
        virStorageSourcePtr src = iscsisrc->src;
1668 1669

        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI &&
1670
            src->auth) {
1671 1672
            if (qemuDomainSecretStorageSourcePrepare(priv, src,
                                                     hostdev->info->alias, NULL) < 0)
1673 1674 1675 1676 1677 1678 1679 1680
                return -1;
        }
    }

    return 0;
}


1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705
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
1706
qemuDomainSecretChardevPrepare(virQEMUDriverConfigPtr cfg,
1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721
                               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)))
1722
            return -1;
1723

1724
        chrSourcePriv->secinfo =
1725
            qemuDomainSecretInfoTLSNew(priv, charAlias,
1726
                                       cfg->chardevTLSx509secretUUID);
1727
        VIR_FREE(charAlias);
1728 1729 1730

        if (!chrSourcePriv->secinfo)
            return -1;
1731 1732 1733 1734 1735 1736
    }

    return 0;
}


1737 1738 1739 1740 1741 1742 1743 1744 1745
static void
qemuDomainSecretGraphicsDestroy(virDomainGraphicsDefPtr graphics)
{
    qemuDomainGraphicsPrivatePtr gfxPriv = QEMU_DOMAIN_GRAPHICS_PRIVATE(graphics);

    if (!gfxPriv)
        return;

    VIR_FREE(gfxPriv->tlsAlias);
1746
    qemuDomainSecretInfoFree(&gfxPriv->secinfo);
1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769
}


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;

1770 1771 1772 1773 1774 1775 1776
    if (cfg->vncTLSx509secretUUID) {
        gfxPriv->secinfo = qemuDomainSecretInfoTLSNew(priv, gfxPriv->tlsAlias,
                                                      cfg->vncTLSx509secretUUID);
        if (!gfxPriv->secinfo)
            return -1;
    }

1777 1778 1779 1780
    return 0;
}


1781 1782 1783
/* qemuDomainSecretDestroy:
 * @vm: Domain object
 *
1784
 * Removes all unnecessary data which was needed to generate 'secret' objects.
1785 1786 1787 1788 1789 1790 1791 1792
 */
void
qemuDomainSecretDestroy(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++)
        qemuDomainSecretDiskDestroy(vm->def->disks[i]);
1793 1794 1795

    for (i = 0; i < vm->def->nhostdevs; i++)
        qemuDomainSecretHostdevDestroy(vm->def->hostdevs[i]);
1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821

    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);
1822 1823 1824

    for (i = 0; i < vm->def->ngraphics; i++)
        qemuDomainSecretGraphicsDestroy(vm->def->graphics[i]);
1825 1826 1827 1828
}


/* qemuDomainSecretPrepare:
1829
 * @driver: Pointer to driver object
1830 1831 1832
 * @vm: Domain object
 *
 * For any objects that may require an auth/secret setup, create a
1833
 * qemuDomainSecretInfo and save it in the appropriate place within
1834 1835 1836 1837 1838 1839 1840
 * 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
1841
qemuDomainSecretPrepare(virQEMUDriverPtr driver,
1842 1843
                        virDomainObjPtr vm)
{
J
John Ferlan 已提交
1844
    qemuDomainObjPrivatePtr priv = vm->privateData;
1845
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1846
    size_t i;
1847
    int ret = -1;
1848

1849
    /* disk secrets are prepared when preparing disks */
1850

1851
    for (i = 0; i < vm->def->nhostdevs; i++) {
1852
        if (qemuDomainSecretHostdevPrepare(priv,
J
John Ferlan 已提交
1853
                                           vm->def->hostdevs[i]) < 0)
1854
            goto cleanup;
1855 1856
    }

1857
    for (i = 0; i < vm->def->nserials; i++) {
1858
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1859 1860 1861 1862 1863 1864
                                           vm->def->serials[i]->info.alias,
                                           vm->def->serials[i]->source) < 0)
            goto cleanup;
    }

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

    for (i = 0; i < vm->def->nchannels; i++) {
1872
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1873 1874 1875 1876 1877 1878
                                           vm->def->channels[i]->info.alias,
                                           vm->def->channels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nconsoles; i++) {
1879
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1880 1881 1882 1883 1884 1885 1886 1887
                                           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 &&
1888
            qemuDomainSecretChardevPrepare(cfg, priv,
1889 1890 1891 1892 1893 1894
                                           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 &&
1895
            qemuDomainSecretChardevPrepare(cfg, priv,
1896 1897 1898 1899 1900 1901
                                           vm->def->rngs[i]->info.alias,
                                           vm->def->rngs[i]->source.chardev) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nredirdevs; i++) {
1902
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1903 1904 1905 1906 1907
                                           vm->def->redirdevs[i]->info.alias,
                                           vm->def->redirdevs[i]->source) < 0)
            goto cleanup;
    }

1908 1909 1910 1911 1912
    for (i = 0; i < vm->def->ngraphics; i++) {
        if (qemuDomainSecretGraphicsPrepare(cfg, priv, vm->def->graphics[i]) < 0)
            goto cleanup;
    }

1913 1914 1915 1916 1917
    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
1918 1919 1920
}


1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947
/* 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
1948 1949
qemuDomainSetPrivatePaths(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
1950
{
1951 1952
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainObjPrivatePtr priv = vm->privateData;
1953
    char *domname = virDomainDefGetShortName(vm->def);
1954
    int ret = -1;
1955

M
Martin Kletzander 已提交
1956 1957 1958
    if (!domname)
        goto cleanup;

1959
    if (!priv->libDir &&
M
Martin Kletzander 已提交
1960
        virAsprintf(&priv->libDir, "%s/domain-%s", cfg->libDir, domname) < 0)
1961
        goto cleanup;
1962

1963
    if (!priv->channelTargetDir &&
M
Martin Kletzander 已提交
1964 1965
        virAsprintf(&priv->channelTargetDir, "%s/domain-%s",
                    cfg->channelTargetDir, domname) < 0)
1966
        goto cleanup;
1967

1968 1969 1970
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
M
Martin Kletzander 已提交
1971
    VIR_FREE(domname);
1972
    return ret;
1973 1974 1975
}


1976
static void *
1977
qemuDomainObjPrivateAlloc(void *opaque)
1978 1979 1980 1981 1982 1983
{
    qemuDomainObjPrivatePtr priv;

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

1984 1985 1986
    if (qemuDomainObjInitJob(priv) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to init qemu driver mutexes"));
1987
        goto error;
1988
    }
1989

1990
    if (!(priv->devs = virChrdevAlloc()))
1991 1992
        goto error;

1993
    priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
1994
    priv->driver = opaque;
1995

1996
    return priv;
1997

1998
 error:
1999 2000
    VIR_FREE(priv);
    return NULL;
2001 2002
}

2003 2004 2005 2006 2007 2008 2009 2010 2011
/**
 * 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)
2012
{
2013 2014 2015
    virStringListFree(priv->qemuDevices);
    priv->qemuDevices = NULL;

2016
    virCgroupFree(&priv->cgroup);
2017 2018 2019 2020 2021

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

    VIR_FREE(priv->machineName);
2022

2023
    virObjectUnref(priv->qemuCaps);
2024
    priv->qemuCaps = NULL;
2025

2026
    VIR_FREE(priv->pidfile);
2027

2028 2029 2030
    VIR_FREE(priv->libDir);
    VIR_FREE(priv->channelTargetDir);

2031 2032
    priv->memPrealloc = false;

2033 2034 2035 2036 2037 2038 2039
    /* remove automatic pinning data */
    virBitmapFree(priv->autoNodeset);
    priv->autoNodeset = NULL;
    virBitmapFree(priv->autoCpuset);
    priv->autoCpuset = NULL;

    /* remove address data */
2040
    virDomainPCIAddressSetFree(priv->pciaddrs);
2041
    priv->pciaddrs = NULL;
J
Ján Tomko 已提交
2042
    virDomainUSBAddressSetFree(priv->usbaddrs);
2043 2044 2045 2046 2047 2048 2049 2050
    priv->usbaddrs = NULL;

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

    /* clear previously used namespaces */
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
2051

2052 2053
    priv->rememberOwner = false;

2054
    priv->reconnectBlockjobs = VIR_TRISTATE_BOOL_ABSENT;
2055
    priv->allowReboot = VIR_TRISTATE_BOOL_ABSENT;
2056 2057 2058

    virBitmapFree(priv->migrationCaps);
    priv->migrationCaps = NULL;
2059 2060 2061

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
2062 2063 2064 2065 2066 2067 2068 2069 2070 2071
}


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

    qemuDomainObjPrivateDataClear(priv);

2072
    virObjectUnref(priv->monConfig);
2073
    qemuDomainObjFreeJob(priv);
2074
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
2075
    VIR_FREE(priv->origname);
2076

2077
    virChrdevFree(priv->devs);
2078

2079 2080
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
2081
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
2082 2083
        qemuMonitorClose(priv->mon);
    }
D
Daniel P. Berrange 已提交
2084 2085 2086 2087
    if (priv->agent) {
        VIR_ERROR(_("Unexpected QEMU agent still active during domain deletion"));
        qemuAgentClose(priv->agent);
    }
2088
    VIR_FREE(priv->cleanupCallbacks);
2089 2090

    qemuDomainSecretInfoFree(&priv->migSecinfo);
2091
    qemuDomainMasterKeyFree(priv);
2092

2093 2094 2095 2096
    VIR_FREE(priv);
}


2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117
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;
}


2118 2119 2120 2121
static int
qemuStorageSourcePrivateDataParse(xmlXPathContextPtr ctxt,
                                  virStorageSourcePtr src)
{
2122 2123 2124 2125 2126
    qemuDomainStorageSourcePrivatePtr priv;
    char *authalias = NULL;
    char *encalias = NULL;
    int ret = -1;

2127 2128
    src->nodestorage = virXPathString("string(./nodenames/nodename[@type='storage']/@name)", ctxt);
    src->nodeformat = virXPathString("string(./nodenames/nodename[@type='format']/@name)", ctxt);
2129
    src->tlsAlias = virXPathString("string(./objects/TLSx509/@alias)", ctxt);
2130

2131 2132 2133
    if (src->pr)
        src->pr->mgralias = virXPathString("string(./reservations/@mgralias)", ctxt);

2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150
    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;
    }

2151
    if (virStorageSourcePrivateDataParseRelPath(ctxt, src) < 0)
2152
        goto cleanup;
2153

2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175
    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);
2176 2177 2178 2179 2180 2181 2182
}


static int
qemuStorageSourcePrivateDataFormat(virStorageSourcePtr src,
                                   virBufferPtr buf)
{
2183
    VIR_AUTOCLEAN(virBuffer) tmp = VIR_BUFFER_INITIALIZER;
2184 2185 2186
    qemuDomainStorageSourcePrivatePtr srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
    int ret = -1;

2187 2188 2189 2190 2191 2192 2193 2194 2195
    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");
    }

2196 2197 2198
    if (src->pr)
        virBufferAsprintf(buf, "<reservations mgralias='%s'/>\n", src->pr->mgralias);

2199
    if (virStorageSourcePrivateDataFormatRelPath(src, buf) < 0)
2200
        goto cleanup;
2201

2202 2203 2204 2205 2206 2207 2208
    virBufferSetChildIndent(&tmp, buf);

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

2209 2210 2211
    if (src->tlsAlias)
        virBufferAsprintf(&tmp, "<TLSx509 alias='%s'/>\n", src->tlsAlias);

2212 2213 2214 2215 2216 2217 2218
    if (virXMLFormatElement(buf, "objects", NULL, &tmp) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    return ret;
2219 2220 2221
}


2222 2223 2224 2225 2226 2227 2228
static int
qemuDomainDiskPrivateParse(xmlXPathContextPtr ctxt,
                           virDomainDiskDefPtr disk)
{
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);

    priv->qomName = virXPathString("string(./qom/@name)", ctxt);
2229
    priv->nodeCopyOnRead = virXPathString("string(./nodenames/nodename[@type='copyOnRead']/@name)", ctxt);
2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242

    return 0;
}


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

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

2243 2244 2245 2246 2247 2248 2249 2250 2251
    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");
    }

2252 2253 2254 2255
    return 0;
}


2256 2257
static void
qemuDomainObjPrivateXMLFormatVcpus(virBufferPtr buf,
2258
                                   virDomainDefPtr def)
2259 2260
{
    size_t i;
2261 2262 2263
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    pid_t tid;
2264 2265 2266 2267

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

2268 2269 2270 2271 2272 2273 2274 2275 2276
    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);
    }
2277 2278 2279 2280 2281 2282

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


2283
static int
2284
qemuDomainObjPrivateXMLFormatAutomaticPlacement(virBufferPtr buf,
2285
                                                qemuDomainObjPrivatePtr priv)
2286 2287
{
    char *nodeset = NULL;
2288
    char *cpuset = NULL;
2289 2290
    int ret = -1;

2291
    if (!priv->autoNodeset && !priv->autoCpuset)
2292 2293
        return 0;

2294 2295
    if (priv->autoNodeset &&
        !((nodeset = virBitmapFormat(priv->autoNodeset))))
2296 2297
        goto cleanup;

2298 2299 2300 2301 2302 2303 2304 2305
    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");
2306 2307 2308 2309 2310

    ret = 0;

 cleanup:
    VIR_FREE(nodeset);
2311
    VIR_FREE(cpuset);
2312 2313 2314 2315
    return ret;
}


2316 2317 2318 2319 2320 2321 2322 2323 2324 2325
static int
qemuDomainObjPrivateXMLFormatBlockjobs(virBufferPtr buf,
                                       virDomainObjPtr vm)
{
    virBuffer attrBuf = VIR_BUFFER_INITIALIZER;
    bool bj = qemuDomainHasBlockjob(vm, false);

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

2326
    return virXMLFormatElement(buf, "blockjobs", &attrBuf, NULL);
2327 2328 2329
}


2330
void
2331 2332 2333 2334 2335 2336 2337 2338 2339
qemuDomainObjPrivateXMLFormatAllowReboot(virBufferPtr buf,
                                         virTristateBool allowReboot)
{
    virBufferAsprintf(buf, "<allowReboot value='%s'/>\n",
                      virTristateBoolTypeToString(allowReboot));

}


2340 2341 2342 2343 2344 2345 2346 2347 2348
static void
qemuDomainObjPrivateXMLFormatPR(virBufferPtr buf,
                                qemuDomainObjPrivatePtr priv)
{
    if (priv->prDaemonRunning)
        virBufferAddLit(buf, "<prDaemon/>\n");
}


2349 2350
static int
qemuDomainObjPrivateXMLFormatNBDMigrationSource(virBufferPtr buf,
2351 2352
                                                virStorageSourcePtr src,
                                                virDomainXMLOptionPtr xmlopt)
2353
{
2354 2355
    VIR_AUTOCLEAN(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    VIR_AUTOCLEAN(virBuffer) childBuf = VIR_BUFFER_INITIALIZER;
2356 2357 2358 2359 2360 2361 2362 2363
    int ret = -1;

    virBufferSetChildIndent(&childBuf, buf);

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

2364
    if (virDomainDiskSourceFormat(&childBuf, src, "source", 0, false,
2365
                                  VIR_DOMAIN_DEF_FORMAT_STATUS, xmlopt) < 0)
2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378
        goto cleanup;

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

    ret = 0;

 cleanup:
    return ret;
}


static int
2379 2380 2381
qemuDomainObjPrivateXMLFormatNBDMigration(virBufferPtr buf,
                                          virDomainObjPtr vm)
{
2382
    qemuDomainObjPrivatePtr priv = vm->privateData;
2383 2384
    VIR_AUTOCLEAN(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    VIR_AUTOCLEAN(virBuffer) childBuf = VIR_BUFFER_INITIALIZER;
2385 2386 2387
    size_t i;
    virDomainDiskDefPtr disk;
    qemuDomainDiskPrivatePtr diskPriv;
2388
    int ret = -1;
2389 2390 2391 2392

    for (i = 0; i < vm->def->ndisks; i++) {
        disk = vm->def->disks[i];
        diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
2393 2394 2395 2396

        virBufferSetChildIndent(&childBuf, buf);

        virBufferAsprintf(&attrBuf, " dev='%s' migrating='%s'",
2397
                          disk->dst, diskPriv->migrating ? "yes" : "no");
2398 2399 2400

        if (diskPriv->migrSource &&
            qemuDomainObjPrivateXMLFormatNBDMigrationSource(&childBuf,
2401 2402
                                                            diskPriv->migrSource,
                                                            priv->driver->xmlopt) < 0)
2403 2404 2405 2406
            goto cleanup;

        if (virXMLFormatElement(buf, "disk", &attrBuf, &childBuf) < 0)
            goto cleanup;
2407
    }
2408 2409 2410 2411 2412

    ret = 0;

 cleanup:
    return ret;
2413 2414 2415
}


2416
static int
2417 2418 2419 2420
qemuDomainObjPrivateXMLFormatJob(virBufferPtr buf,
                                 virDomainObjPtr vm,
                                 qemuDomainObjPrivatePtr priv)
{
2421 2422
    VIR_AUTOCLEAN(virBuffer) attrBuf = VIR_BUFFER_INITIALIZER;
    VIR_AUTOCLEAN(virBuffer) childBuf = VIR_BUFFER_INITIALIZER;
2423
    qemuDomainJob job = priv->job.active;
2424
    int ret = -1;
2425 2426

    if (!qemuDomainTrackJob(job))
2427
        job = QEMU_JOB_NONE;
2428

2429 2430
    if (job == QEMU_JOB_NONE &&
        priv->job.asyncJob == QEMU_ASYNC_JOB_NONE)
2431 2432 2433
        return 0;

    virBufferSetChildIndent(&childBuf, buf);
2434

2435
    virBufferAsprintf(&attrBuf, " type='%s' async='%s'",
2436 2437
                      qemuDomainJobTypeToString(job),
                      qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
2438

2439
    if (priv->job.phase) {
2440
        virBufferAsprintf(&attrBuf, " phase='%s'",
2441 2442 2443
                          qemuDomainAsyncJobPhaseToString(priv->job.asyncJob,
                                                          priv->job.phase));
    }
2444

2445 2446 2447
    if (priv->job.asyncJob != QEMU_ASYNC_JOB_NONE)
        virBufferAsprintf(&attrBuf, " flags='0x%lx'", priv->job.apiFlags);

2448 2449 2450
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_OUT &&
        qemuDomainObjPrivateXMLFormatNBDMigration(&childBuf, vm) < 0)
        goto cleanup;
2451

2452 2453 2454
    if (priv->job.migParams)
        qemuMigrationParamsFormat(&childBuf, priv->job.migParams);

2455 2456 2457 2458 2459 2460 2461
    if (virXMLFormatElement(buf, "job", &attrBuf, &childBuf) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    return ret;
2462 2463 2464
}


2465
static int
2466 2467
qemuDomainObjPrivateXMLFormat(virBufferPtr buf,
                              virDomainObjPtr vm)
2468
{
2469
    qemuDomainObjPrivatePtr priv = vm->privateData;
2470 2471 2472 2473
    const char *monitorpath;

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
2474
        switch (priv->monConfig->type) {
2475
        case VIR_DOMAIN_CHR_TYPE_UNIX:
2476
            monitorpath = priv->monConfig->data.nix.path;
2477 2478 2479
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
2480
            monitorpath = priv->monConfig->data.file.path;
2481 2482 2483
            break;
        }

2484
        virBufferEscapeString(buf, "<monitor path='%s'", monitorpath);
J
Ján Tomko 已提交
2485
        virBufferAddLit(buf, " json='1'");
2486
        virBufferAsprintf(buf, " type='%s'/>\n",
2487
                          virDomainChrTypeToString(priv->monConfig->type));
2488 2489
    }

2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500
    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");
    }

2501
    qemuDomainObjPrivateXMLFormatVcpus(buf, vm->def);
2502

2503
    if (priv->qemuCaps) {
2504
        size_t i;
2505 2506
        virBufferAddLit(buf, "<qemuCaps>\n");
        virBufferAdjustIndent(buf, 2);
2507
        for (i = 0; i < QEMU_CAPS_LAST; i++) {
2508
            if (virQEMUCapsGet(priv->qemuCaps, i)) {
2509
                virBufferAsprintf(buf, "<flag name='%s'/>\n",
2510
                                  virQEMUCapsTypeToString(i));
2511 2512
            }
        }
2513 2514
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</qemuCaps>\n");
2515 2516
    }

2517
    if (priv->lockState)
2518
        virBufferAsprintf(buf, "<lockstate>%s</lockstate>\n", priv->lockState);
2519

2520 2521
    if (qemuDomainObjPrivateXMLFormatJob(buf, vm, priv) < 0)
        return -1;
2522

2523
    if (priv->fakeReboot)
2524
        virBufferAddLit(buf, "<fakereboot/>\n");
2525

2526 2527
    if (priv->qemuDevices && *priv->qemuDevices) {
        char **tmp = priv->qemuDevices;
2528 2529
        virBufferAddLit(buf, "<devices>\n");
        virBufferAdjustIndent(buf, 2);
2530
        while (*tmp) {
2531
            virBufferAsprintf(buf, "<device alias='%s'/>\n", *tmp);
2532 2533
            tmp++;
        }
2534 2535
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</devices>\n");
2536 2537
    }

2538
    if (qemuDomainObjPrivateXMLFormatAutomaticPlacement(buf, priv) < 0)
2539
        return -1;
2540

2541 2542 2543 2544 2545
    /* Various per-domain paths */
    virBufferEscapeString(buf, "<libDir path='%s'/>\n", priv->libDir);
    virBufferEscapeString(buf, "<channelTargetDir path='%s'/>\n",
                          priv->channelTargetDir);

2546
    virCPUDefFormatBufFull(buf, priv->origCPU, NULL);
2547

2548
    if (priv->chardevStdioLogd)
2549
        virBufferAddLit(buf, "<chardevStdioLogd/>\n");
2550

2551 2552 2553
    if (priv->rememberOwner)
        virBufferAddLit(buf, "<rememberOwner/>\n");

2554 2555
    qemuDomainObjPrivateXMLFormatAllowReboot(buf, priv->allowReboot);

2556 2557
    qemuDomainObjPrivateXMLFormatPR(buf, priv);

2558 2559 2560
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV))
        virBufferAsprintf(buf, "<nodename index='%llu'/>\n", priv->nodenameindex);

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

2564 2565 2566
    if (qemuDomainObjPrivateXMLFormatBlockjobs(buf, vm) < 0)
        return -1;

2567 2568 2569
    return 0;
}

2570 2571 2572 2573

static int
qemuDomainObjPrivateXMLParseVcpu(xmlNodePtr node,
                                 unsigned int idx,
2574
                                 virDomainDefPtr def)
2575
{
2576
    virDomainVcpuDefPtr vcpu;
2577
    char *idstr;
2578
    char *pidstr;
2579
    unsigned int tmp;
2580 2581
    int ret = -1;

2582 2583
    idstr = virXMLPropString(node, "id");

2584 2585
    if (idstr &&
        (virStrToLong_uip(idstr, NULL, 10, &idx) < 0)) {
2586
        virReportError(VIR_ERR_INTERNAL_ERROR,
2587 2588 2589 2590 2591 2592
                       _("cannot parse vcpu index '%s'"), idstr);
        goto cleanup;
    }
    if (!(vcpu = virDomainDefGetVcpu(def, idx))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("invalid vcpu index '%u'"), idx);
2593
        goto cleanup;
2594 2595
    }

2596 2597 2598
    if (!(pidstr = virXMLPropString(node, "pid")))
        goto cleanup;

2599
    if (virStrToLong_uip(pidstr, NULL, 10, &tmp) < 0)
2600 2601
        goto cleanup;

2602 2603
    QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid = tmp;

2604 2605 2606
    ret = 0;

 cleanup:
2607
    VIR_FREE(idstr);
2608 2609 2610 2611 2612
    VIR_FREE(pidstr);
    return ret;
}


2613 2614 2615 2616 2617 2618 2619
static int
qemuDomainObjPrivateXMLParseAutomaticPlacement(xmlXPathContextPtr ctxt,
                                               qemuDomainObjPrivatePtr priv,
                                               virQEMUDriverPtr driver)
{
    virCapsPtr caps = NULL;
    char *nodeset;
2620
    char *cpuset;
2621 2622
    int nodesetSize = 0;
    size_t i;
2623 2624 2625
    int ret = -1;

    nodeset = virXPathString("string(./numad/@nodeset)", ctxt);
2626
    cpuset = virXPathString("string(./numad/@cpuset)", ctxt);
2627

2628
    if (!nodeset && !cpuset)
2629 2630 2631 2632 2633
        return 0;

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

2634 2635 2636 2637 2638 2639
    /* 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);

2640
    if (nodeset &&
2641
        virBitmapParse(nodeset, &priv->autoNodeset, nodesetSize) < 0)
2642 2643
        goto cleanup;

2644 2645 2646 2647 2648 2649 2650 2651 2652 2653
    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;
    }
2654 2655 2656 2657 2658 2659

    ret = 0;

 cleanup:
    virObjectUnref(caps);
    VIR_FREE(nodeset);
2660
    VIR_FREE(cpuset);
2661 2662 2663 2664 2665

    return ret;
}


2666 2667 2668 2669
static int
qemuDomainObjPrivateXMLParseBlockjobs(qemuDomainObjPrivatePtr priv,
                                      xmlXPathContextPtr ctxt)
{
2670
    VIR_AUTOFREE(char *) active = NULL;
2671 2672 2673 2674 2675 2676 2677 2678 2679 2680
    int tmp;

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

    return 0;
}


2681
int
2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705
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;
}


2706 2707 2708 2709 2710 2711 2712 2713
static void
qemuDomainObjPrivateXMLParsePR(xmlXPathContextPtr ctxt,
                               bool *prDaemonRunning)
{
    *prDaemonRunning = virXPathBoolean("boolean(./prDaemon)", ctxt) > 0;
}


2714 2715 2716
static int
qemuDomainObjPrivateXMLParseJobNBDSource(xmlNodePtr node,
                                         xmlXPathContextPtr ctxt,
2717 2718
                                         virDomainDiskDefPtr disk,
                                         virDomainXMLOptionPtr xmlopt)
2719
{
2720
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
2721
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
2722 2723
    VIR_AUTOFREE(char *) format = NULL;
    VIR_AUTOFREE(char *) type = NULL;
2724
    VIR_AUTOUNREF(virStorageSourcePtr) migrSource = NULL;
2725
    xmlNodePtr sourceNode;
2726 2727 2728

    ctxt->node = node;

2729 2730
    if (!(ctxt->node = virXPathNode("./migrationSource", ctxt)))
        return 0;
2731 2732 2733 2734

    if (!(type = virXMLPropString(ctxt->node, "type"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
                       _("missing storage source type"));
2735
        return -1;
2736 2737 2738 2739
    }

    if (!(format = virXMLPropString(ctxt->node, "format"))) {
        virReportError(VIR_ERR_XML_ERROR, "%s",
2740
                       _("missing storage source format"));
2741
        return -1;
2742 2743
    }

2744
    if (!(migrSource = virDomainStorageSourceParseBase(type, format, NULL)))
2745
        return -1;
2746

2747 2748 2749 2750 2751
    /* newer libvirt uses the <source> subelement instead of formatting the
     * source directly into <migrationSource> */
    if ((sourceNode = virXPathNode("./source", ctxt)))
        ctxt->node = sourceNode;

2752
    if (virDomainStorageSourceParse(ctxt->node, ctxt, migrSource,
2753
                                    VIR_DOMAIN_DEF_PARSE_STATUS, xmlopt) < 0)
2754
        return -1;
2755 2756

    VIR_STEAL_PTR(diskPriv->migrSource, migrSource);
2757
    return 0;
2758 2759 2760
}


2761 2762 2763 2764 2765 2766
static int
qemuDomainObjPrivateXMLParseJobNBD(virDomainObjPtr vm,
                                   qemuDomainObjPrivatePtr priv,
                                   xmlXPathContextPtr ctxt)
{
    xmlNodePtr *nodes = NULL;
2767
    char *dst = NULL;
2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783
    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;

2784 2785
            if ((dst = virXMLPropString(nodes[i], "dev")) &&
                (disk = virDomainDiskByName(vm->def, dst, false))) {
2786
                QEMU_DOMAIN_DISK_PRIVATE(disk)->migrating = true;
2787 2788

                if (qemuDomainObjPrivateXMLParseJobNBDSource(nodes[i], ctxt,
2789 2790
                                                             disk,
                                                             priv->driver->xmlopt) < 0)
2791 2792 2793
                    goto cleanup;
            }

2794 2795 2796 2797 2798 2799 2800 2801
            VIR_FREE(dst);
        }
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
2802
    VIR_FREE(dst);
2803 2804 2805 2806
    return ret;
}


2807 2808 2809 2810 2811
static int
qemuDomainObjPrivateXMLParseJob(virDomainObjPtr vm,
                                qemuDomainObjPrivatePtr priv,
                                xmlXPathContextPtr ctxt)
{
2812
    VIR_XPATH_NODE_AUTORESTORE(ctxt);
2813 2814 2815
    char *tmp = NULL;
    int ret = -1;

2816 2817 2818 2819 2820 2821
    if (!(ctxt->node = virXPathNode("./job[1]", ctxt))) {
        ret = 0;
        goto cleanup;
    }

    if ((tmp = virXPathString("string(@type)", ctxt))) {
2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832
        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;
    }

2833
    if ((tmp = virXPathString("string(@async)", ctxt))) {
2834 2835 2836 2837 2838 2839 2840 2841 2842 2843
        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;

2844
        if ((tmp = virXPathString("string(@phase)", ctxt))) {
2845 2846 2847 2848 2849 2850 2851 2852 2853 2854
            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);
        }
    }

2855 2856 2857 2858 2859
    if (virXPathULongHex("string(@flags)", ctxt, &priv->job.apiFlags) == -2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Invalid job flags"));
        goto cleanup;
    }

2860
    if (qemuDomainObjPrivateXMLParseJobNBD(vm, priv, ctxt) < 0)
2861 2862
        goto cleanup;

2863 2864 2865
    if (qemuMigrationParamsParse(ctxt, &priv->job.migParams) < 0)
        goto cleanup;

2866 2867 2868 2869 2870 2871 2872 2873
    ret = 0;

 cleanup:
    VIR_FREE(tmp);
    return ret;
}


2874
static int
2875
qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt,
2876
                             virDomainObjPtr vm,
2877
                             virDomainDefParserConfigPtr config)
2878
{
2879
    qemuDomainObjPrivatePtr priv = vm->privateData;
2880
    virQEMUDriverPtr driver = config->priv;
2881
    char *monitorpath;
2882
    char *tmp = NULL;
2883 2884
    int n;
    size_t i;
2885
    xmlNodePtr *nodes = NULL;
2886
    xmlNodePtr node = NULL;
2887
    virQEMUCapsPtr qemuCaps = NULL;
2888

2889
    if (!(priv->monConfig = virDomainChrSourceDefNew(NULL)))
2890 2891 2892 2893
        goto error;

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
2894 2895
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("no monitor path"));
2896 2897 2898 2899 2900
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
2901
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
2902
    else
2903
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
2904 2905
    VIR_FREE(tmp);

2906
    switch (priv->monConfig->type) {
2907
    case VIR_DOMAIN_CHR_TYPE_PTY:
2908
        priv->monConfig->data.file.path = monitorpath;
2909 2910
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
2911
        priv->monConfig->data.nix.path = monitorpath;
2912 2913 2914
        break;
    default:
        VIR_FREE(monitorpath);
2915 2916 2917
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported monitor type '%s'"),
                       virDomainChrTypeToString(priv->monConfig->type));
2918 2919 2920
        goto error;
    }

2921 2922 2923 2924
    if ((node = virXPathNode("./namespaces", ctxt))) {
        xmlNodePtr next;

        for (next = node->children; next; next = next->next) {
2925
            int ns = qemuDomainNamespaceTypeFromString((const char *)next->name);
2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944

            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;
    }

2945 2946
    priv->rememberOwner = virXPathBoolean("count(./rememberOwner) > 0", ctxt);

2947
    if ((n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes)) < 0)
2948 2949
        goto error;

2950
    for (i = 0; i < n; i++) {
2951
        if (qemuDomainObjPrivateXMLParseVcpu(nodes[i], i, vm->def) < 0)
2952
            goto error;
2953
    }
2954
    VIR_FREE(nodes);
2955

2956
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
2957 2958
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse qemu capabilities flags"));
2959 2960 2961
        goto error;
    }
    if (n > 0) {
2962
        if (!(qemuCaps = virQEMUCapsNew()))
2963 2964
            goto error;

2965
        for (i = 0; i < n; i++) {
2966 2967
            char *str = virXMLPropString(nodes[i], "name");
            if (str) {
2968
                int flag = virQEMUCapsTypeFromString(str);
2969
                if (flag < 0) {
2970 2971
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Unknown qemu capabilities flag %s"), str);
2972
                    VIR_FREE(str);
2973 2974
                    goto error;
                }
2975
                VIR_FREE(str);
2976
                virQEMUCapsSet(qemuCaps, flag);
2977 2978 2979
            }
        }

M
Marc Hartmayer 已提交
2980
        VIR_STEAL_PTR(priv->qemuCaps, qemuCaps);
2981 2982 2983
    }
    VIR_FREE(nodes);

2984
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
2985

2986
    if (qemuDomainObjPrivateXMLParseJob(vm, priv, ctxt) < 0)
2987 2988
        goto error;

2989 2990
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011
    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);

3012
    if (qemuDomainObjPrivateXMLParseAutomaticPlacement(ctxt, priv, driver) < 0)
3013 3014
        goto error;

3015 3016 3017 3018 3019 3020 3021 3022 3023
    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;

3024 3025 3026
    if (virCPUDefParseXML(ctxt, "./cpu", VIR_CPU_TYPE_GUEST, &priv->origCPU) < 0)
        goto error;

3027 3028 3029
    priv->chardevStdioLogd = virXPathBoolean("boolean(./chardevStdioLogd)",
                                             ctxt) == 1;

3030 3031
    qemuDomainObjPrivateXMLParseAllowReboot(ctxt, &priv->allowReboot);

3032 3033
    qemuDomainObjPrivateXMLParsePR(ctxt, &priv->prDaemonRunning);

3034 3035 3036
    if (qemuDomainObjPrivateXMLParseBlockjobs(priv, ctxt) < 0)
        goto error;

3037 3038 3039 3040 3041 3042 3043 3044
    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;
    }

3045 3046
    priv->memPrealloc = virXPathBoolean("boolean(./memPrealloc)", ctxt) == 1;

3047 3048
    return 0;

3049
 error:
3050
    VIR_FREE(nodes);
3051
    VIR_FREE(tmp);
3052 3053
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
3054
    virObjectUnref(priv->monConfig);
3055
    priv->monConfig = NULL;
3056
    virStringListFree(priv->qemuDevices);
3057
    priv->qemuDevices = NULL;
3058
    virObjectUnref(qemuCaps);
3059 3060 3061 3062
    return -1;
}


3063 3064 3065 3066 3067 3068 3069 3070 3071
static void *
qemuDomainObjPrivateXMLGetParseOpaque(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return priv->qemuCaps;
}


3072 3073 3074
virDomainXMLPrivateDataCallbacks virQEMUDriverPrivateDataCallbacks = {
    .alloc = qemuDomainObjPrivateAlloc,
    .free = qemuDomainObjPrivateFree,
3075
    .diskNew = qemuDomainDiskPrivateNew,
3076 3077
    .diskParse = qemuDomainDiskPrivateParse,
    .diskFormat = qemuDomainDiskPrivateFormat,
3078
    .vcpuNew = qemuDomainVcpuPrivateNew,
3079
    .chrSourceNew = qemuDomainChrSourcePrivateNew,
J
Ján Tomko 已提交
3080
    .vsockNew = qemuDomainVsockPrivateNew,
3081
    .graphicsNew = qemuDomainGraphicsPrivateNew,
3082 3083
    .parse = qemuDomainObjPrivateXMLParse,
    .format = qemuDomainObjPrivateXMLFormat,
3084
    .getParseOpaque = qemuDomainObjPrivateXMLGetParseOpaque,
3085 3086
    .storageParse = qemuStorageSourcePrivateDataParse,
    .storageFormat = qemuStorageSourcePrivateDataFormat,
3087 3088 3089
};


3090 3091 3092 3093 3094 3095
static void
qemuDomainXmlNsDefFree(qemuDomainXmlNsDefPtr def)
{
    if (!def)
        return;

3096 3097 3098
    virStringListFreeCount(def->args, def->num_args);
    virStringListFreeCount(def->env_name, def->num_env);
    virStringListFreeCount(def->env_value, def->num_env);
3099 3100
    virStringListFreeCount(def->capsadd, def->ncapsadd);
    virStringListFreeCount(def->capsdel, def->ncapsdel);
3101

3102 3103 3104 3105
    VIR_FREE(def);
}


3106 3107 3108
static void
qemuDomainDefNamespaceFree(void *nsdata)
{
3109
    qemuDomainXmlNsDefPtr cmd = nsdata;
3110

3111
    qemuDomainXmlNsDefFree(cmd);
3112 3113
}

3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143

static int
qemuDomainDefNamespaceParseCommandlineArgs(qemuDomainXmlNsDefPtr nsdef,
                                           xmlXPathContextPtr ctxt)
{
    VIR_AUTOFREE(xmlNodePtr *) nodes = NULL;
    ssize_t nnodes;
    size_t i;

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

    if (nnodes == 0)
        return 0;

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

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

    return 0;
}


3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199
static int
qemuDomainDefNamespaceParseCommandlineEnvNameValidate(const char *envname)
{
    if (!c_isalpha(envname[0]) && envname[0] != '_') {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Invalid environment name, it must begin with a letter or underscore"));
        return -1;
    }

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

    return 0;
}


static int
qemuDomainDefNamespaceParseCommandlineEnv(qemuDomainXmlNsDefPtr nsdef,
                                          xmlXPathContextPtr ctxt)
{
    VIR_AUTOFREE(xmlNodePtr *) nodes = NULL;
    ssize_t nnodes;
    size_t i;

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

    if (nnodes == 0)
        return 0;

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

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

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

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

    return 0;
}


3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243
static int
qemuDomainDefNamespaceParseCaps(qemuDomainXmlNsDefPtr nsdef,
                                xmlXPathContextPtr ctxt)
{
    VIR_AUTOFREE(xmlNodePtr *) nodesadd = NULL;
    ssize_t nnodesadd;
    VIR_AUTOFREE(xmlNodePtr *) nodesdel = NULL;
    ssize_t nnodesdel;
    size_t i;

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

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

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

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

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

    return 0;
}


3244
static int
P
Philipp Hahn 已提交
3245 3246
qemuDomainDefNamespaceParse(xmlDocPtr xml ATTRIBUTE_UNUSED,
                            xmlNodePtr root ATTRIBUTE_UNUSED,
3247 3248 3249
                            xmlXPathContextPtr ctxt,
                            void **data)
{
3250 3251
    qemuDomainXmlNsDefPtr nsdata = NULL;
    int ret = -1;
3252

P
Philipp Hahn 已提交
3253
    if (xmlXPathRegisterNs(ctxt, BAD_CAST "qemu", BAD_CAST QEMU_NAMESPACE_HREF) < 0) {
3254 3255 3256
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to register xml namespace '%s'"),
                       QEMU_NAMESPACE_HREF);
3257 3258 3259
        return -1;
    }

3260
    if (VIR_ALLOC(nsdata) < 0)
3261 3262
        return -1;

3263
    if (qemuDomainDefNamespaceParseCommandlineArgs(nsdata, ctxt) < 0 ||
3264 3265
        qemuDomainDefNamespaceParseCommandlineEnv(nsdata, ctxt) < 0 ||
        qemuDomainDefNamespaceParseCaps(nsdata, ctxt) < 0)
3266
        goto cleanup;
3267

3268 3269
    if (nsdata->num_args > 0 || nsdata->num_env > 0 ||
        nsdata->ncapsadd > 0 || nsdata->ncapsdel > 0)
3270
        VIR_STEAL_PTR(*data, nsdata);
3271

3272
    ret = 0;
3273

3274 3275 3276
 cleanup:
    qemuDomainDefNamespaceFree(nsdata);
    return ret;
3277 3278
}

3279 3280 3281 3282

static void
qemuDomainDefNamespaceFormatXMLCommandline(virBufferPtr buf,
                                           qemuDomainXmlNsDefPtr cmd)
3283
{
3284
    size_t i;
3285 3286

    if (!cmd->num_args && !cmd->num_env)
3287
        return;
3288

3289 3290 3291
    virBufferAddLit(buf, "<qemu:commandline>\n");
    virBufferAdjustIndent(buf, 2);

3292
    for (i = 0; i < cmd->num_args; i++)
3293
        virBufferEscapeString(buf, "<qemu:arg value='%s'/>\n",
3294 3295
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
3296
        virBufferAsprintf(buf, "<qemu:env name='%s'", cmd->env_name[i]);
3297 3298 3299 3300 3301
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }

3302 3303
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:commandline>\n");
3304 3305 3306
}


3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329
static void
qemuDomainDefNamespaceFormatXMLCaps(virBufferPtr buf,
                                    qemuDomainXmlNsDefPtr xmlns)
{
    size_t i;

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

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

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

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

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


3330 3331 3332 3333 3334 3335 3336
static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainXmlNsDefPtr cmd = nsdata;

    qemuDomainDefNamespaceFormatXMLCommandline(buf, cmd);
3337
    qemuDomainDefNamespaceFormatXMLCaps(buf, cmd);
3338

3339 3340 3341 3342 3343 3344 3345 3346 3347 3348
    return 0;
}

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


3349 3350 3351 3352 3353 3354
virDomainXMLNamespace virQEMUDriverDomainXMLNamespace = {
    .parse = qemuDomainDefNamespaceParse,
    .free = qemuDomainDefNamespaceFree,
    .format = qemuDomainDefNamespaceFormatXML,
    .href = qemuDomainDefNamespaceHref,
};
3355

3356

P
Pavel Hrdina 已提交
3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375
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;
}


3376
static int
3377 3378
qemuDomainDefAddDefaultDevices(virDomainDefPtr def,
                               virQEMUCapsPtr qemuCaps)
3379
{
3380
    bool addDefaultUSB = true;
3381
    int usbModel = -1; /* "default for machinetype" */
3382
    int pciRoot;       /* index within def->controllers */
3383
    bool addImplicitSATA = false;
3384
    bool addPCIRoot = false;
L
Laine Stump 已提交
3385
    bool addPCIeRoot = false;
3386
    bool addDefaultMemballoon = true;
3387 3388
    bool addDefaultUSBKBD = false;
    bool addDefaultUSBMouse = false;
3389
    bool addPanicDevice = false;
3390
    int ret = -1;
3391

P
Pavel Hrdina 已提交
3392 3393 3394 3395
    /* add implicit input devices */
    if (qemuDomainDefAddImplicitInputDevice(def) < 0)
        goto cleanup;

3396 3397 3398 3399
    /* 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
3400
        if (STREQ(def->os.machine, "isapc")) {
3401
            addDefaultUSB = false;
3402
            break;
3403
        }
3404
        if (qemuDomainIsQ35(def)) {
3405 3406
            addPCIeRoot = true;
            addImplicitSATA = true;
3407

3408 3409 3410
            /* 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.
3411
             */
3412 3413 3414
            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))
3415 3416
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
            else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_USB_EHCI1))
3417 3418 3419
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_ICH9_EHCI1;
            else
                addDefaultUSB = false;
3420
            break;
L
Laine Stump 已提交
3421
        }
3422
        if (qemuDomainIsI440FX(def))
3423
            addPCIRoot = true;
3424 3425
        break;

3426 3427 3428 3429 3430 3431 3432
    case VIR_ARCH_ARMV6L:
        addDefaultUSB = false;
        addDefaultMemballoon = false;
        if (STREQ(def->os.machine, "versatilepb"))
            addPCIRoot = true;
        break;

3433
    case VIR_ARCH_ARMV7L:
3434
    case VIR_ARCH_AARCH64:
3435 3436
        addDefaultUSB = false;
        addDefaultMemballoon = false;
3437
        if (qemuDomainIsARMVirt(def))
3438 3439
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
        break;
3440

3441
    case VIR_ARCH_PPC64:
3442
    case VIR_ARCH_PPC64LE:
3443 3444 3445
        addPCIRoot = true;
        addDefaultUSBKBD = true;
        addDefaultUSBMouse = true;
3446 3447 3448
        /* For pSeries guests, the firmware provides the same
         * functionality as the pvpanic device, so automatically
         * add the definition if not already present */
3449
        if (qemuDomainIsPSeries(def))
3450
            addPanicDevice = true;
3451 3452
        break;

3453 3454 3455 3456 3457 3458 3459
    case VIR_ARCH_ALPHA:
    case VIR_ARCH_PPC:
    case VIR_ARCH_PPCEMB:
    case VIR_ARCH_SH4:
    case VIR_ARCH_SH4EB:
        addPCIRoot = true;
        break;
3460

3461 3462 3463
    case VIR_ARCH_RISCV32:
    case VIR_ARCH_RISCV64:
        addDefaultUSB = false;
3464 3465
        if (qemuDomainIsRISCVVirt(def))
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
3466 3467
        break;

3468 3469 3470
    case VIR_ARCH_S390:
    case VIR_ARCH_S390X:
        addDefaultUSB = false;
3471
        addPanicDevice = true;
3472
        addPCIRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_ZPCI);
3473
        break;
3474 3475 3476 3477 3478 3479

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

3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499
    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:
3500 3501 3502 3503
    default:
        break;
    }

3504
    if (addDefaultUSB &&
3505 3506
        virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_USB, 0) < 0 &&
        virDomainDefAddUSBController(def, 0, usbModel) < 0)
3507
        goto cleanup;
3508

3509 3510 3511
    if (addImplicitSATA &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_SATA, 0, -1) < 0)
3512
        goto cleanup;
3513

3514 3515
    pciRoot = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

3516 3517 3518
    /* NB: any machine that sets addPCIRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
     */
3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533
    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;
        }
    }
3534

3535 3536 3537
    /* 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
3538 3539 3540
     *
     * NB: any machine that sets addPCIeRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
3541 3542
     */
    if (addPCIeRoot) {
3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555
        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;
        }
3556
    }
3557

3558
    if (addDefaultMemballoon && !def->memballoon) {
3559 3560
        virDomainMemballoonDefPtr memballoon;
        if (VIR_ALLOC(memballoon) < 0)
3561
            goto cleanup;
3562 3563 3564 3565 3566

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

3567 3568 3569 3570
    if (STRPREFIX(def->os.machine, "s390-virtio") &&
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_S390) && def->memballoon)
        def->memballoon->model = VIR_DOMAIN_MEMBALLOON_MODEL_NONE;

3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590
    if (addDefaultUSBMouse) {
        bool hasUSBTablet = false;
        size_t j;

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

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

3591 3592 3593 3594 3595
    if (addDefaultUSBKBD &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_KBD,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
3596
        goto cleanup;
3597 3598 3599 3600 3601 3602

    if (addDefaultUSBMouse &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
3603
        goto cleanup;
3604

D
Dmitry Andreev 已提交
3605 3606 3607 3608
    if (addPanicDevice) {
        size_t j;
        for (j = 0; j < def->npanics; j++) {
            if (def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT ||
3609 3610 3611 3612
                (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
3613 3614
                break;
        }
3615

D
Dmitry Andreev 已提交
3616 3617 3618 3619 3620 3621 3622 3623 3624
        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;
            }
        }
3625 3626
    }

3627 3628 3629 3630 3631 3632
    ret = 0;
 cleanup:
    return ret;
}


A
Andrea Bolognani 已提交
3633 3634 3635
/**
 * qemuDomainDefEnableDefaultFeatures:
 * @def: domain definition
3636
 * @qemuCaps: QEMU capabilities
A
Andrea Bolognani 已提交
3637 3638 3639 3640 3641
 *
 * Make sure that features that should be enabled by default are actually
 * enabled and configure default values related to those features.
 */
static void
3642 3643
qemuDomainDefEnableDefaultFeatures(virDomainDefPtr def,
                                   virQEMUCapsPtr qemuCaps)
A
Andrea Bolognani 已提交
3644
{
3645
    /* The virt machine type always uses GIC: if the relevant information
3646 3647
     * was not included in the domain XML, we need to choose a suitable
     * GIC version ourselves */
3648
    if ((def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ABSENT &&
3649
         qemuDomainIsARMVirt(def)) ||
3650 3651 3652
        (def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ON &&
         def->gic_version == VIR_GIC_VERSION_NONE)) {
        virGICVersion version;
3653 3654 3655 3656 3657

        VIR_DEBUG("Looking for usable GIC version in domain capabilities");
        for (version = VIR_GIC_VERSION_LAST - 1;
             version > VIR_GIC_VERSION_NONE;
             version--) {
3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672

            /* 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;
            }

3673 3674 3675 3676 3677 3678 3679 3680
            if (virQEMUCapsSupportsGICVersion(qemuCaps,
                                              def->virtType,
                                              version)) {
                VIR_DEBUG("Using GIC version %s",
                          virGICVersionTypeToString(version));
                def->gic_version = version;
                break;
            }
3681 3682
        }

3683 3684 3685 3686 3687 3688 3689
        /* 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;
        }

3690 3691 3692
        /* 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;
3693
    }
A
Andrea Bolognani 已提交
3694 3695 3696
}


3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716
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;
}


3717
static int
3718 3719 3720 3721 3722
qemuDomainRecheckInternalPaths(virDomainDefPtr def,
                               virQEMUDriverConfigPtr cfg,
                               unsigned int flags)
{
    size_t i = 0;
3723
    size_t j = 0;
3724 3725 3726 3727

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

3728 3729 3730 3731 3732 3733
        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
3734
             * the auto-generated socket based on config option from qemu.conf
3735 3736 3737 3738
             * 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 &&
3739
                !glisten->autoGenerated &&
3740 3741 3742 3743 3744 3745 3746
                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;
                }
3747
            }
3748 3749
        }
    }
3750 3751

    return 0;
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 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 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824
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;
}


3825 3826 3827 3828 3829 3830
static int
qemuDomainDefCPUPostParse(virDomainDefPtr def)
{
    if (!def->cpu)
        return 0;

3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884
    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;
        }
    }

3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917
    /* 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;
}


3918 3919 3920 3921 3922 3923 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
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;
}


3950 3951 3952 3953 3954 3955 3956 3957
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)))
3958
        return 1;
3959 3960 3961 3962 3963

    return 0;
}


3964 3965
static int
qemuDomainDefPostParse(virDomainDefPtr def,
3966
                       virCapsPtr caps ATTRIBUTE_UNUSED,
3967
                       unsigned int parseFlags,
3968
                       void *opaque,
3969
                       void *parseOpaque)
3970 3971
{
    virQEMUDriverPtr driver = opaque;
3972
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3973 3974 3975
    /* 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. */
3976
    virQEMUCapsPtr qemuCaps = parseOpaque;
3977 3978 3979 3980 3981
    int ret = -1;

    if (def->os.bootloader || def->os.bootloaderArgs) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bootloader is not supported by QEMU"));
3982
        goto cleanup;
3983 3984
    }

3985 3986 3987
    if (!def->os.machine) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing machine type"));
3988
        goto cleanup;
3989 3990
    }

3991 3992
    if (qemuDomainNVRAMPathGenerate(cfg, def) < 0)
        goto cleanup;
3993

3994 3995 3996
    if (qemuDomainDefAddDefaultDevices(def, qemuCaps) < 0)
        goto cleanup;

3997 3998 3999
    if (qemuCanonicalizeMachine(def, qemuCaps) < 0)
        goto cleanup;

4000
    qemuDomainDefEnableDefaultFeatures(def, qemuCaps);
A
Andrea Bolognani 已提交
4001

4002 4003
    if (qemuDomainRecheckInternalPaths(def, cfg, parseFlags) < 0)
        goto cleanup;
4004

4005
    if (qemuSecurityVerify(driver->securityManager, def) < 0)
4006 4007
        goto cleanup;

4008 4009 4010
    if (qemuDomainDefVcpusPostParse(def) < 0)
        goto cleanup;

4011 4012 4013
    if (qemuDomainDefCPUPostParse(def) < 0)
        goto cleanup;

4014 4015 4016
    if (qemuDomainDefTsegPostParse(def, qemuCaps) < 0)
        goto cleanup;

4017 4018
    ret = 0;
 cleanup:
4019
    virObjectUnref(cfg);
4020
    return ret;
4021 4022
}

4023

4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053
/**
 * 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;
}


4054 4055 4056
#define QEMU_MAX_VCPUS_WITHOUT_EIM 255


4057
static int
4058 4059
qemuDomainDefValidateFeatures(const virDomainDef *def,
                              virQEMUCapsPtr qemuCaps)
4060
{
4061
    size_t i;
4062

4063 4064 4065 4066 4067
    for (i = 0; i < VIR_DOMAIN_FEATURE_LAST; i++) {
        const char *featureName = virDomainFeatureTypeToString(i);

        switch ((virDomainFeature) i) {
        case VIR_DOMAIN_FEATURE_IOAPIC:
4068
            if (def->features[i] != VIR_DOMAIN_IOAPIC_NONE &&
4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080
                !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:
4081
        case VIR_DOMAIN_FEATURE_HTM:
4082
        case VIR_DOMAIN_FEATURE_NESTED_HV:
4083
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094
                !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;

4095 4096
        case VIR_DOMAIN_FEATURE_GIC:
            if (def->features[i] == VIR_TRISTATE_SWITCH_ON &&
4097
                !qemuDomainIsARMVirt(def)) {
4098 4099 4100 4101 4102 4103 4104 4105 4106 4107
                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;

4108 4109
        case VIR_DOMAIN_FEATURE_SMM:
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
4110
                !virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT)) {
4111 4112 4113 4114 4115 4116
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("smm is not available with this QEMU binary"));
                return -1;
            }
            break;

4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129
        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:
4130
        case VIR_DOMAIN_FEATURE_MSRS:
4131 4132 4133
        case VIR_DOMAIN_FEATURE_LAST:
            break;
        }
4134 4135 4136 4137 4138 4139
    }

    return 0;
}


4140
static int
M
Marc-André Lureau 已提交
4141 4142
qemuDomainDefValidateMemory(const virDomainDef *def,
                            virQEMUCapsPtr qemuCaps)
4143 4144
{
    const long system_page_size = virGetSystemPageSizeKB();
4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162
    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;
    }
4163

M
Marc-André Lureau 已提交
4164 4165 4166
    if (mem->source == VIR_DOMAIN_MEMORY_SOURCE_MEMFD &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_MEMORY_MEMFD_HUGETLB)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4167
                       _("hugepages is not supported with memfd memory source"));
M
Marc-André Lureau 已提交
4168 4169 4170
        return -1;
    }

4171 4172
    /* We can't guarantee any other mem.access
     * if no guest NUMA nodes are defined. */
4173
    if (mem->hugepages[0].size != system_page_size &&
4174
        virDomainNumaGetNodeCount(def->numa) == 0 &&
4175 4176
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_DEFAULT &&
        mem->access != VIR_DOMAIN_MEMORY_ACCESS_PRIVATE) {
4177 4178 4179
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("memory access mode '%s' not supported "
                         "without guest numa node"),
4180
                       virDomainMemoryAccessTypeToString(mem->access));
4181 4182 4183 4184 4185 4186 4187
        return -1;
    }

    return 0;
}


4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210
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;
}


4211 4212
static int
qemuDomainDefValidate(const virDomainDef *def,
4213
                      virCapsPtr caps ATTRIBUTE_UNUSED,
4214
                      void *opaque)
4215
{
4216
    virQEMUDriverPtr driver = opaque;
4217
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4218 4219
    virQEMUCapsPtr qemuCaps = NULL;
    int ret = -1;
4220
    size_t i;
4221

4222
    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
4223 4224 4225
                                            def->emulator)))
        goto cleanup;

4226 4227 4228
    if (def->mem.min_guarantee) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Parameter 'min_guarantee' not supported by QEMU."));
4229
        goto cleanup;
4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249
    }

    /* 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;
4250 4251
    }

M
Michal Privoznik 已提交
4252 4253 4254 4255 4256
    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. */

4257
        if (!qemuDomainIsQ35(def)) {
M
Michal Privoznik 已提交
4258 4259
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot is supported with q35 machine types only"));
4260
            goto cleanup;
M
Michal Privoznik 已提交
4261 4262 4263 4264 4265 4266 4267 4268
        }

        /* 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"));
4269
            goto cleanup;
M
Michal Privoznik 已提交
4270 4271
        }

4272 4273 4274
        /* SMM will be enabled by qemuFirmwareFillDomain() if needed. */
        if (def->os.firmware == VIR_DOMAIN_OS_DEF_FIRMWARE_NONE &&
            def->features[VIR_DOMAIN_FEATURE_SMM] != VIR_TRISTATE_SWITCH_ON) {
M
Michal Privoznik 已提交
4275 4276
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot requires SMM feature enabled"));
4277
            goto cleanup;
M
Michal Privoznik 已提交
4278 4279 4280
        }
    }

4281 4282 4283 4284 4285
    /* 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;
4286
        unsigned int granularity;
4287 4288 4289 4290 4291 4292

        /* 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)) {
4293 4294 4295
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CPU topology doesn't match maximum vcpu count"));
            goto cleanup;
4296
        }
4297 4298 4299 4300 4301 4302 4303 4304 4305 4306

        /* 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;
        }
4307 4308
    }

4309
    if (qemuDomainValidateCpuCount(def, qemuCaps) < 0)
4310 4311
        goto cleanup;

4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329
    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
4330
    if (def->nresctrls &&
4331 4332 4333 4334 4335 4336
        def->virtType != VIR_DOMAIN_VIRT_KVM) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("cachetune is only supported for KVM domains"));
        goto cleanup;
    }

4337
    if (qemuDomainDefValidateFeatures(def, qemuCaps) < 0)
4338 4339
        goto cleanup;

M
Marc-André Lureau 已提交
4340
    if (qemuDomainDefValidateMemory(def, qemuCaps) < 0)
4341 4342
        goto cleanup;

4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354
    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;
            }
        }
    }

4355 4356 4357 4358
    ret = 0;

 cleanup:
    virObjectUnref(qemuCaps);
4359
    virObjectUnref(cfg);
4360
    return ret;
4361 4362 4363
}


4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387
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;
}


4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434
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;
}


4435 4436 4437
static int
qemuDomainChrSerialTargetTypeToAddressType(int targetType)
{
4438
    switch ((virDomainChrSerialTargetType)targetType) {
4439 4440 4441 4442 4443 4444
    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;
4445 4446
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
        return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO;
4447
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
4448
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
4449 4450 4451 4452 4453 4454 4455 4456 4457
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        break;
    }

    return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE;
}


4458 4459 4460 4461 4462 4463 4464 4465 4466 4467
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;
4468 4469
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
4470
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
4471
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
4472
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
4473 4474 4475
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
4476 4477 4478 4479 4480 4481 4482 4483 4484
    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;
}


4485
static int
4486
qemuDomainChrTargetDefValidate(const virDomainChrDef *chr)
4487
{
4488 4489
    int expected;

4490
    switch ((virDomainChrDeviceType)chr->deviceType) {
4491 4492 4493
    case VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL:

        /* Validate target type */
4494
        switch ((virDomainChrSerialTargetType)chr->targetType) {
4495
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA:
4496 4497
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI:
4498
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
4499

4500
            expected = qemuDomainChrSerialTargetTypeToAddressType(chr->targetType);
4501 4502

            if (chr->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
4503 4504 4505 4506 4507
                chr->info.type != expected) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Target type '%s' requires address type '%s'"),
                               virDomainChrSerialTargetTypeToString(chr->targetType),
                               virDomainDeviceAddressTypeToString(expected));
4508 4509 4510 4511
                return -1;
            }
            break;

4512
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
4513
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
4514 4515 4516 4517 4518 4519 4520 4521 4522
            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;

4523 4524 4525 4526
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            break;
        }
4527 4528 4529 4530 4531 4532

        /* 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:
4533
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
4534
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
4535 4536
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
4537
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_16550A:
4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553

            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;
        }
4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567
        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;
}


4568
static int
4569
qemuDomainChrDefValidate(const virDomainChrDef *dev,
4570
                         const virDomainDef *def)
4571
{
4572
    if (qemuDomainChrSourceDefValidate(dev->source) < 0)
4573 4574
        return -1;

4575
    if (qemuDomainChrTargetDefValidate(dev) < 0)
4576 4577
        return -1;

4578
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL &&
4579
        (ARCH_IS_S390(def->os.arch) || qemuDomainIsPSeries(def))) {
4580 4581 4582 4583 4584
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("parallel ports are not supported"));
            return -1;
    }

4585 4586 4587
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL) {
        bool isCompatible = true;

4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598
        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;
            }
        }

4599 4600 4601 4602 4603 4604
        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;
        }

4605 4606 4607 4608 4609 4610 4611
        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;
        }

4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622
        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;
        }
    }

4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638
    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
4639 4640
qemuDomainRNGDefValidate(const virDomainRNGDef *def,
                         virQEMUCapsPtr qemuCaps)
4641
{
4642 4643
    bool modelIsSupported = false;

4644 4645 4646 4647
    if (def->backend == VIR_DOMAIN_RNG_BACKEND_EGD &&
        qemuDomainChrSourceDefValidate(def->source.chardev) < 0)
        return -1;

4648 4649 4650 4651 4652
    switch ((virDomainRNGModel) def->model) {
    case VIR_DOMAIN_RNG_MODEL_VIRTIO:
        modelIsSupported = virQEMUCapsGet(qemuCaps,
                                          QEMU_CAPS_DEVICE_VIRTIO_RNG);
        break;
4653 4654 4655 4656 4657 4658
    case VIR_DOMAIN_RNG_MODEL_VIRTIO_TRANSITIONAL:
    case VIR_DOMAIN_RNG_MODEL_VIRTIO_NON_TRANSITIONAL:
        modelIsSupported = (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_RNG) &&
                            (virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_PCI_TRANSITIONAL) ||
                             virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_PCI_DISABLE_LEGACY)));
        break;
4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669
    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;
    }

4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683
    return 0;
}


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

    return 0;
}


4684 4685 4686 4687 4688 4689 4690 4691 4692
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
4693
                           _("%s model of watchdog can go only on PCI bus"),
4694 4695 4696 4697 4698 4699 4700 4701 4702
                           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
4703
                           _("%s model of watchdog can go only on ISA bus"),
4704 4705 4706 4707 4708 4709 4710 4711
                           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
4712
                           _("%s model of watchdog is virtual and cannot go on any bus."),
4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731
                           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;
}


4732
static int
4733
qemuDomainDeviceDefValidateNetwork(const virDomainNetDef *net)
4734
{
4735 4736
    bool hasIPv4 = false;
    bool hasIPv6 = false;
4737
    size_t i;
4738

4739 4740 4741 4742 4743 4744 4745 4746 4747 4748
    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];
4749

4750
            if (VIR_SOCKET_ADDR_VALID(&net->guestIP.ips[i]->peer)) {
4751
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4752 4753
                               _("Invalid attempt to set peer IP for guest"));
                return -1;
4754
            }
4755

4756 4757
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ip->address, AF_INET)) {
                if (hasIPv4) {
4758
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4759 4760 4761
                                   _("Only one IPv4 address per "
                                     "interface is allowed"));
                    return -1;
4762
                }
4763
                hasIPv4 = true;
4764

4765 4766
                if (ip->prefix > 0 &&
                    (ip->prefix < 4 || ip->prefix > 27)) {
4767
                    virReportError(VIR_ERR_XML_ERROR, "%s",
4768
                                   _("invalid prefix, must be in range of 4-27"));
4769
                    return -1;
4770
                }
4771
            }
4772

4773 4774 4775 4776 4777 4778
            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;
4779
                }
4780
                hasIPv6 = true;
4781

4782 4783 4784 4785 4786
                if (ip->prefix > 120) {
                    virReportError(VIR_ERR_XML_ERROR, "%s",
                                   _("prefix too long"));
                    return -1;
                }
4787
            }
4788
        }
4789 4790 4791 4792 4793 4794 4795
    } 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;
    }
4796

4797
    if (virDomainNetIsVirtioModel(net)) {
4798 4799 4800 4801 4802 4803 4804 4805 4806
        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;
4807
        }
4808
    }
4809

4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828
    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;
}


4829
static int
4830 4831 4832
qemuDomainMdevDefVFIOPCIValidate(const virDomainHostdevSubsysMediatedDev *dev,
                                 const virDomainDef *def,
                                 virQEMUCapsPtr qemuCaps)
4833
{
4834
    if (dev->display == VIR_TRISTATE_SWITCH_ABSENT)
4835 4836 4837 4838 4839 4840 4841 4842 4843
        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;
    }

4844
    if (dev->model != VIR_MDEV_MODEL_TYPE_VFIO_PCI) {
4845 4846 4847 4848 4849 4850 4851
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("<hostdev> attribute 'display' is only supported"
                         " with model='vfio-pci'"));

        return -1;
    }

4852
    if (dev->display == VIR_TRISTATE_SWITCH_ON) {
4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864
        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;
}


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
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;
}


4891 4892 4893 4894 4895 4896 4897 4898 4899
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:
4900
        return qemuDomainMdevDefVFIOAPValidate(def);
4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913
    case VIR_MDEV_MODEL_TYPE_VFIO_CCW:
        break;
    case VIR_MDEV_MODEL_TYPE_LAST:
    default:
        virReportEnumRangeError(virMediatedDeviceModelType,
                                mdevsrc->model);
        return -1;
    }

    return 0;
}


4914 4915
static int
qemuDomainDeviceDefValidateHostdev(const virDomainHostdevDef *hostdev,
4916 4917
                                   const virDomainDef *def,
                                   virQEMUCapsPtr qemuCaps)
4918
{
4919 4920
    const virDomainHostdevSubsysMediatedDev *mdevsrc;

4921 4922 4923 4924 4925 4926 4927 4928 4929
    /* 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;
    }

4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947
    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;
        }
    }

4948 4949 4950 4951
    return 0;
}


4952 4953 4954
static int
qemuDomainDeviceDefValidateVideo(const virDomainVideoDef *video)
{
4955
    switch ((virDomainVideoType) video->type) {
4956 4957
    case VIR_DOMAIN_VIDEO_TYPE_NONE:
        return 0;
4958 4959 4960
    case VIR_DOMAIN_VIDEO_TYPE_XEN:
    case VIR_DOMAIN_VIDEO_TYPE_VBOX:
    case VIR_DOMAIN_VIDEO_TYPE_PARALLELS:
4961
    case VIR_DOMAIN_VIDEO_TYPE_GOP:
4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002
    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);
5003 5004
            return -1;
        }
5005 5006 5007 5008 5009 5010 5011
        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;
            }
5012

5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026
            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)"));
5027 5028 5029 5030 5031 5032 5033 5034
            return -1;
        }
    }

    return 0;
}


5035
int
5036 5037
qemuDomainValidateStorageSource(virStorageSourcePtr src,
                                virQEMUCapsPtr qemuCaps)
5038
{
5039 5040
    int actualType = virStorageSourceGetActualType(src);

5041 5042 5043 5044 5045 5046
    if (src->format == VIR_STORAGE_FILE_COW) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                      _("'cow' storage format is not supported"));
        return -1;
    }

5047 5048 5049 5050 5051 5052 5053
    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;
    }

5054 5055 5056 5057 5058 5059 5060
    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;
    }

5061 5062 5063 5064 5065 5066 5067 5068 5069 5070
    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;
    }

5071 5072 5073 5074 5075
    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",
5076
                       _("LUKS encrypted QCOW2 images are not supported by this QEMU"));
5077 5078 5079
        return -1;
    }

5080
    if (src->format == VIR_STORAGE_FILE_FAT &&
5081
        actualType != VIR_STORAGE_TYPE_VOLUME &&
5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103
        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;
        }
    }

5104 5105 5106 5107 5108
    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;
5109 5110
    }

5111 5112 5113 5114 5115 5116 5117 5118 5119
    /* 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;
    }

5120 5121 5122 5123
    return 0;
}


5124
int
5125 5126
qemuDomainDeviceDefValidateDisk(const virDomainDiskDef *disk,
                                virQEMUCapsPtr qemuCaps)
5127
{
5128
    const char *driverName = virDomainDiskGetDriver(disk);
5129 5130
    virStorageSourcePtr n;

5131 5132 5133 5134 5135 5136
    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;
5137 5138
    }

5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153
    if (disk->copy_on_read == VIR_TRISTATE_SWITCH_ON) {
        if (disk->src->readonly) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("copy_on_read is not compatible with read-only disk '%s'"),
                           disk->dst);
            return -1;
        }

        if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM ||
            disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("copy_on_read is not supported with removable disk '%s'"),
                           disk->dst);
            return -1;
        }
5154 5155
    }

5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182
    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;
    }

5183 5184 5185 5186 5187 5188 5189
    if (driverName && STRNEQ(driverName, "qemu")) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("unsupported driver name '%s' for disk '%s'"),
                       driverName, disk->dst);
        return -1;
    }

5190
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
5191
        if (qemuDomainValidateStorageSource(n, qemuCaps) < 0)
5192 5193 5194
            return -1;
    }

P
Peter Krempa 已提交
5195 5196 5197 5198 5199 5200 5201 5202
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
        disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("disk type 'virtio' of '%s' does not support ejectable media"),
                       disk->dst);
        return -1;
    }

5203 5204 5205 5206
    return 0;
}


5207 5208 5209 5210
static int
qemuDomainDeviceDefValidateControllerAttributes(const virDomainControllerDef *controller)
{
    if (!(controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
5211 5212 5213
          (controller->model == VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI ||
           controller->model == VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL ||
           controller->model == VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL))) {
5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233
        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;
        }
5234 5235 5236 5237 5238
        if (controller->iothread) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'iothread' is only supported for virtio-scsi controller"));
            return -1;
        }
5239 5240 5241 5242 5243 5244
    }

    return 0;
}


5245 5246 5247 5248 5249 5250
/**
 * @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
5251
 * Returns true if acceptable, false otherwise with error message set.
5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266
 */
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:
5267 5268
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL:
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL:
5269 5270 5271 5272 5273 5274 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
        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;
5302 5303 5304 5305 5306 5307
    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;
5308 5309 5310 5311 5312 5313
    }

    return true;
}


5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339
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;
}


5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356
/* 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;

5357 5358
    if (controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI &&
5359 5360
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
       virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
5361 5362
                       _("virtio-scsi IOThreads only available for virtio "
                         "pci and virtio ccw controllers"));
5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383
       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:
5384 5385
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_TRANSITIONAL:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_NON_TRANSITIONAL:
5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396
            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:
5397
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT:
5398 5399 5400 5401 5402 5403 5404 5405
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LAST:
            break;
    }

    return 0;
}


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
/**
 * 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;
5438 5439
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCIE_PCI_BRIDGE:
        return QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE;
5440 5441 5442 5443 5444
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE:
        return 0;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_LAST:
    default:
        return -1;
5445 5446
    }

5447
    return -1;
5448 5449 5450
}


5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467
#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));


5468 5469 5470 5471 5472 5473 5474 5475 5476
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);
5477
    int cap = virDomainControllerPCIModelNameToQEMUCaps(pciopts->modelName);
5478 5479 5480 5481 5482 5483 5484 5485 5486 5487

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

5488 5489 5490 5491 5492 5493 5494 5495 5496
    /* 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:
5497
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601
        /* 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;

5602 5603 5604 5605 5606 5607 5608
    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;

5609 5610 5611 5612 5613 5614 5615
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

5616 5617 5618 5619 5620 5621 5622 5623 5624
    /* 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:
5625
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659
        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;
    }

5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686
    /* 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:
5687
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700
        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;
    }

5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720
    /* 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:
5721
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735
        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;
    }

5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752
    /* 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:
5753
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766
        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;
    }

5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797
    /* 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:
5798
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811
        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;
    }

5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828
    /* 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:
5829
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5830 5831 5832 5833 5834 5835 5836 5837 5838 5839 5840 5841 5842
        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;
    }

5843 5844 5845 5846 5847 5848 5849 5850 5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863
    /* 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:
5864
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880
        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);
    }

5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906
    /* 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;
    }

5907
    return 0;
5908 5909 5910
}


5911 5912 5913 5914 5915
#undef virReportControllerInvalidValue
#undef virReportControllerInvalidOption
#undef virReportControllerMissingOption


5916 5917 5918 5919 5920 5921 5922 5923 5924 5925 5926 5927 5928 5929 5930 5931 5932 5933
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;
}


5934
static int
5935 5936 5937
qemuDomainDeviceDefValidateController(const virDomainControllerDef *controller,
                                      const virDomainDef *def,
                                      virQEMUCapsPtr qemuCaps)
5938
{
5939 5940
    int ret = 0;

5941
    if (!qemuDomainCheckCCWS390AddressSupport(def, &controller->info, qemuCaps,
5942 5943 5944
                                              "controller"))
        return -1;

5945 5946 5947 5948
    if (controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
        !qemuDomainCheckSCSIControllerModel(qemuCaps, controller->model))
        return -1;

5949 5950 5951
    if (qemuDomainDeviceDefValidateControllerAttributes(controller) < 0)
        return -1;

5952
    switch ((virDomainControllerType)controller->type) {
5953
    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
5954 5955 5956
        ret = qemuDomainDeviceDefValidateControllerIDE(controller, def);
        break;

5957
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
5958 5959 5960
        ret = qemuDomainDeviceDefValidateControllerSCSI(controller, def);
        break;

5961
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
5962 5963
        ret = qemuDomainDeviceDefValidateControllerPCI(controller, def,
                                                       qemuCaps);
5964 5965
        break;

5966
    case VIR_DOMAIN_CONTROLLER_TYPE_SATA:
5967 5968 5969 5970 5971
        ret = qemuDomainDeviceDefValidateControllerSATA(controller, def,
                                                        qemuCaps);
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
5972 5973 5974
    case VIR_DOMAIN_CONTROLLER_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_CONTROLLER_TYPE_CCID:
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
5975
    case VIR_DOMAIN_CONTROLLER_TYPE_XENBUS:
5976 5977 5978 5979
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
    }

5980
    return ret;
5981 5982 5983
}


5984
static int
5985 5986
qemuDomainDeviceDefValidateVsock(const virDomainVsockDef *vsock,
                                 const virDomainDef *def,
5987 5988 5989 5990 5991 5992 5993 5994
                                 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;
    }
5995

5996
    if (!qemuDomainCheckCCWS390AddressSupport(def, &vsock->info, qemuCaps,
5997 5998 5999
                                              "vsock"))
        return -1;

6000 6001
    return 0;
}
6002

6003 6004 6005 6006 6007 6008 6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027

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;
    }
6028 6029 6030 6031
    return 0;
}


6032 6033 6034 6035 6036 6037 6038 6039 6040
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++) {
6041
        if (def->graphics[i]->type == VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS) {
6042 6043 6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083
            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;
}


6084 6085 6086 6087 6088
static int
qemuDomainDeviceDefValidateInput(const virDomainInputDef *input,
                                 const virDomainDef *def ATTRIBUTE_UNUSED,
                                 virQEMUCapsPtr qemuCaps)
{
6089 6090 6091 6092
    const char *baseName;
    int cap;
    int ccwCap;

6093 6094 6095
    if (input->bus != VIR_DOMAIN_INPUT_BUS_VIRTIO)
        return 0;

6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112 6113 6114 6115 6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127
    /* Only type=passthrough supports model=virtio-(non-)transitional */
    switch ((virDomainInputModel)input->model) {
    case VIR_DOMAIN_INPUT_MODEL_VIRTIO_TRANSITIONAL:
    case VIR_DOMAIN_INPUT_MODEL_VIRTIO_NON_TRANSITIONAL:
        switch ((virDomainInputType)input->type) {
        case VIR_DOMAIN_INPUT_TYPE_MOUSE:
        case VIR_DOMAIN_INPUT_TYPE_TABLET:
        case VIR_DOMAIN_INPUT_TYPE_KBD:
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("virtio (non-)transitional models are not "
                             "supported for input type=%s"),
                           virDomainInputTypeToString(input->type));
            return -1;
        case VIR_DOMAIN_INPUT_TYPE_PASSTHROUGH:
            break;
        case VIR_DOMAIN_INPUT_TYPE_LAST:
        default:
            virReportEnumRangeError(virDomainInputType,
                                    input->type);
            return -1;
        }
        break;
    case VIR_DOMAIN_INPUT_MODEL_VIRTIO:
    case VIR_DOMAIN_INPUT_MODEL_DEFAULT:
        break;
    case VIR_DOMAIN_INPUT_MODEL_LAST:
    default:
        virReportEnumRangeError(virDomainInputModel,
                                input->model);
        return -1;
    }

6128 6129
    switch ((virDomainInputType)input->type) {
    case VIR_DOMAIN_INPUT_TYPE_MOUSE:
6130 6131 6132
        baseName = "virtio-mouse";
        cap = QEMU_CAPS_VIRTIO_MOUSE;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_MOUSE_CCW;
6133 6134
        break;
    case VIR_DOMAIN_INPUT_TYPE_TABLET:
6135 6136 6137
        baseName = "virtio-tablet";
        cap = QEMU_CAPS_VIRTIO_TABLET;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_TABLET_CCW;
6138 6139
        break;
    case VIR_DOMAIN_INPUT_TYPE_KBD:
6140 6141 6142
        baseName = "virtio-keyboard";
        cap = QEMU_CAPS_VIRTIO_KEYBOARD;
        ccwCap = QEMU_CAPS_DEVICE_VIRTIO_KEYBOARD_CCW;
6143 6144
        break;
    case VIR_DOMAIN_INPUT_TYPE_PASSTHROUGH:
6145 6146 6147
        baseName = "virtio-input-host";
        cap = QEMU_CAPS_VIRTIO_INPUT_HOST;
        ccwCap = QEMU_CAPS_LAST;
6148 6149 6150 6151 6152 6153 6154 6155
        break;
    case VIR_DOMAIN_INPUT_TYPE_LAST:
    default:
        virReportEnumRangeError(virDomainInputType,
                                input->type);
        return -1;
    }

6156 6157 6158 6159 6160 6161 6162 6163 6164
    if (!virQEMUCapsGet(qemuCaps, cap) ||
        (input->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW &&
         !virQEMUCapsGet(qemuCaps, ccwCap))) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("%s is not supported by this QEMU binary"),
                       baseName);
        return -1;
    }

6165 6166 6167 6168
    return 0;
}


6169 6170 6171 6172 6173 6174 6175 6176 6177
static int
qemuDomainDeviceDefValidateMemballoon(const virDomainMemballoonDef *memballoon,
                                      virQEMUCapsPtr qemuCaps)
{
    if (!memballoon ||
        memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_NONE) {
        return 0;
    }

6178 6179 6180
    if (memballoon->model != VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO &&
        memballoon->model != VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO_TRANSITIONAL &&
        memballoon->model != VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO_NON_TRANSITIONAL) {
6181 6182 6183 6184 6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197
        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;
}


6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221
static int
qemuDomainDeviceDefValidateIOMMU(const virDomainIOMMUDef *iommu,
                                 const virDomainDef *def,
                                 virQEMUCapsPtr qemuCaps)
{
    switch (iommu->model) {
    case VIR_DOMAIN_IOMMU_MODEL_INTEL:
        if (!qemuDomainIsQ35(def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("IOMMU device: '%s' is only supported with "
                             "Q35 machines"),
                           virDomainIOMMUModelTypeToString(iommu->model));
            return -1;
        }
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_INTEL_IOMMU) &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_IOMMU)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("IOMMU device: '%s' is not supported with "
                             "this QEMU binary"),
                           virDomainIOMMUModelTypeToString(iommu->model));
            return -1;
        }
        break;

6222
    case VIR_DOMAIN_IOMMU_MODEL_SMMUV3:
6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236
        if (!qemuDomainIsARMVirt(def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("IOMMU device: '%s' is only supported with "
                             "ARM Virt machines"),
                           virDomainIOMMUModelTypeToString(iommu->model));
            return -1;
        }
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_VIRT_IOMMU)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("IOMMU device: '%s' is not supported with "
                             "this QEMU binary"),
                           virDomainIOMMUModelTypeToString(iommu->model));
            return -1;
        }
6237 6238
        break;

6239 6240 6241
    case VIR_DOMAIN_IOMMU_MODEL_LAST:
    default:
        virReportEnumRangeError(virDomainIOMMUModel, iommu->model);
6242
        return -1;
6243 6244
    }

6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277 6278
    /* These capability checks ensure we're not trying to use features
     * of Intel IOMMU that the QEMU binary does not support, but they
     * also make sure we report an error when trying to use features
     * that are not implemented by SMMUv3 */

    if (iommu->intremap != VIR_TRISTATE_SWITCH_ABSENT &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_INTEL_IOMMU_INTREMAP)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("iommu: interrupt remapping is not supported "
                         "with this QEMU binary"));
        return -1;
    }
    if (iommu->caching_mode != VIR_TRISTATE_SWITCH_ABSENT &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_INTEL_IOMMU_CACHING_MODE))  {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("iommu: caching mode is not supported "
                         "with this QEMU binary"));
        return -1;
    }
    if (iommu->eim != VIR_TRISTATE_SWITCH_ABSENT &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_INTEL_IOMMU_EIM))  {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("iommu: eim is not supported "
                         "with this QEMU binary"));
        return -1;
    }
    if (iommu->iotlb != VIR_TRISTATE_SWITCH_ABSENT &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_INTEL_IOMMU_DEVICE_IOTLB)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("iommu: device IOTLB is not supported "
                         "with this QEMU binary"));
        return -1;
    }

6279 6280 6281 6282
    return 0;
}


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
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;

6308 6309
    switch ((virDomainDeviceAddressType) info->type) {
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI:
6310 6311
        return qemuDomainDeviceDefValidateZPCIAddress(info, qemuCaps);

6312 6313 6314 6315 6316 6317
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE:
        /* Address validation might happen before we have had a chance to
         * automatically assign addresses to devices for which the user
         * didn't specify one themselves */
        break;

6318 6319 6320 6321 6322 6323 6324 6325 6326 6327 6328 6329 6330 6331
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO: {
        virDomainDeviceSpaprVioAddressPtr addr = &(info->addr.spaprvio);

        if (addr->has_reg && addr->reg > 0xffffffff) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("spapr-vio reg='0x%llx' exceeds maximum "
                             "possible value (0xffffffff)"),
                           addr->reg);
            return -1;
        }

        break;
        }

6332 6333 6334 6335 6336 6337 6338 6339 6340 6341 6342 6343 6344 6345 6346 6347 6348 6349
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCID:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_USB:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_MMIO:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_ISA:
    case VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DIMM:
        /* No validation for these address types yet */
        break;

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

6350 6351 6352 6353
    return 0;
}


6354 6355 6356
static int
qemuDomainDeviceDefValidate(const virDomainDeviceDef *dev,
                            const virDomainDef *def,
6357
                            void *opaque)
6358
{
6359
    int ret = 0;
6360 6361 6362 6363 6364 6365
    virQEMUDriverPtr driver = opaque;
    virQEMUCapsPtr qemuCaps = NULL;

    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
                                            def->emulator)))
        return -1;
6366

6367 6368 6369
    if ((ret = qemuDomainDeviceDefValidateAddress(dev, qemuCaps)) < 0)
        goto cleanup;

6370
    switch ((virDomainDeviceType)dev->type) {
6371 6372 6373 6374 6375 6376 6377 6378 6379 6380 6381 6382 6383
    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:
6384
        ret = qemuDomainRNGDefValidate(dev->data.rng, qemuCaps);
6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395
        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:
6396 6397
        ret = qemuDomainDeviceDefValidateHostdev(dev->data.hostdev, def,
                                                 qemuCaps);
6398 6399 6400 6401 6402 6403 6404
        break;

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

    case VIR_DOMAIN_DEVICE_DISK:
6405
        ret = qemuDomainDeviceDefValidateDisk(dev->data.disk, qemuCaps);
6406 6407
        break;

6408
    case VIR_DOMAIN_DEVICE_CONTROLLER:
6409 6410
        ret = qemuDomainDeviceDefValidateController(dev->data.controller, def,
                                                    qemuCaps);
6411 6412
        break;

6413
    case VIR_DOMAIN_DEVICE_VSOCK:
6414
        ret = qemuDomainDeviceDefValidateVsock(dev->data.vsock, def, qemuCaps);
6415 6416
        break;

6417 6418 6419 6420
    case VIR_DOMAIN_DEVICE_TPM:
        ret = qemuDomainDeviceDefValidateTPM(dev->data.tpm, def);
        break;

6421 6422 6423 6424 6425
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainDeviceDefValidateGraphics(dev->data.graphics, def,
                                                  qemuCaps);
        break;

6426 6427 6428 6429
    case VIR_DOMAIN_DEVICE_INPUT:
        ret = qemuDomainDeviceDefValidateInput(dev->data.input, def, qemuCaps);
        break;

6430 6431 6432 6433
    case VIR_DOMAIN_DEVICE_MEMBALLOON:
        ret = qemuDomainDeviceDefValidateMemballoon(dev->data.memballoon, qemuCaps);
        break;

6434 6435 6436 6437
    case VIR_DOMAIN_DEVICE_IOMMU:
        ret = qemuDomainDeviceDefValidateIOMMU(dev->data.iommu, def, qemuCaps);
        break;

6438 6439 6440 6441 6442 6443
    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:
6444
    case VIR_DOMAIN_DEVICE_MEMORY:
6445 6446 6447 6448
    case VIR_DOMAIN_DEVICE_PANIC:
    case VIR_DOMAIN_DEVICE_NONE:
    case VIR_DOMAIN_DEVICE_LAST:
        break;
6449 6450
    }

6451
 cleanup:
6452
    virObjectUnref(qemuCaps);
6453 6454 6455 6456
    return ret;
}


6457 6458 6459 6460 6461 6462 6463 6464 6465
/**
 * 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.
 */
6466
static int
6467 6468
qemuDomainDefaultNetModel(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
6469
{
6470
    if (ARCH_IS_S390(def->os.arch))
6471
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
6472

S
Stefan Schallenberg 已提交
6473 6474
    if (def->os.arch == VIR_ARCH_ARMV6L ||
        def->os.arch == VIR_ARCH_ARMV7L ||
6475
        def->os.arch == VIR_ARCH_AARCH64) {
6476
        if (STREQ(def->os.machine, "versatilepb"))
6477
            return VIR_DOMAIN_NET_MODEL_SMC91C111;
6478

6479
        if (qemuDomainIsARMVirt(def))
6480
            return VIR_DOMAIN_NET_MODEL_VIRTIO;
6481

6482 6483
        /* Incomplete. vexpress (and a few others) use this, but not all
         * arm boards */
6484
        return VIR_DOMAIN_NET_MODEL_LAN9118;
6485 6486
    }

6487 6488
    /* virtio is a sensible default for RISC-V virt guests */
    if (qemuDomainIsRISCVVirt(def))
6489
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
6490

6491 6492 6493
    /* In all other cases the model depends on the capabilities. If they were
     * not provided don't report any default. */
    if (!qemuCaps)
6494
        return VIR_DOMAIN_NET_MODEL_UNKNOWN;
6495

6496 6497 6498 6499
    /* 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))
6500
        return VIR_DOMAIN_NET_MODEL_RTL8139;
6501
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_E1000))
6502
        return VIR_DOMAIN_NET_MODEL_E1000;
6503
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_NET))
6504
        return VIR_DOMAIN_NET_MODEL_VIRTIO;
6505 6506 6507

    /* We've had no luck detecting support for any network device,
     * but we have to return something: might as well be rtl8139 */
6508
    return VIR_DOMAIN_NET_MODEL_RTL8139;
6509
}
6510

6511 6512

/*
6513 6514 6515 6516 6517 6518 6519 6520 6521 6522 6523 6524 6525 6526 6527 6528
 * 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.
6529
 */
6530
static int
6531 6532 6533
qemuDomainChrDefDropDefaultPath(virDomainChrDefPtr chr,
                                virQEMUDriverPtr driver)
{
6534 6535 6536 6537
    virQEMUDriverConfigPtr cfg;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *regexp = NULL;
    int ret = -1;
6538

6539 6540 6541 6542 6543
    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;
6544 6545
    }

6546 6547 6548 6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562
    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);
6563
    virObjectUnref(cfg);
6564
    return ret;
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
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;
}


6618
#define QEMU_USB_XHCI_MAXPORTS 15
6619 6620


6621 6622 6623
static int
qemuDomainControllerDefPostParse(virDomainControllerDefPtr cont,
                                 const virDomainDef *def,
6624 6625
                                 virQEMUCapsPtr qemuCaps,
                                 unsigned int parseFlags)
6626
{
6627 6628
    switch ((virDomainControllerType)cont->type) {
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
6629 6630 6631
        /* Set the default SCSI controller model if not already set */
        if (qemuDomainSetSCSIControllerModel(def, cont, qemuCaps) < 0)
            return -1;
6632
        break;
6633

6634
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
6635
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT && qemuCaps) {
6636
            /* Pick a suitable default model for the USB controller if none
6637 6638
             * has been selected by the user and we have the qemuCaps for
             * figuring out which contollers are supported.
6639 6640 6641 6642 6643 6644 6645
             *
             * 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() */
6646 6647 6648 6649 6650 6651 6652 6653 6654 6655 6656

            /* 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;
                }
6657
            } else if (ARCH_IS_PPC64(def->os.arch)) {
6658 6659
                /* To not break migration we need to set default USB controller
                 * for ppc64 to pci-ohci if we cannot change ABI of the VM.
6660 6661
                 * The nec-usb-xhci or qemu-xhci controller is used as default
                 * only for newly defined domains or devices. */
6662
                if ((parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE) &&
6663 6664 6665
                    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) &&
6666 6667 6668
                    virQEMUCapsGet(qemuCaps, QEMU_CAPS_NEC_USB_XHCI)) {
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
                } else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_OHCI)) {
6669
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_PCI_OHCI;
6670 6671 6672
                } else {
                    /* Explicitly fallback to legacy USB controller for PPC64. */
                    cont->model = -1;
6673
                }
6674
            } else if (def->os.arch == VIR_ARCH_AARCH64) {
6675 6676 6677
                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))
6678
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
6679 6680 6681 6682 6683 6684 6685 6686 6687 6688 6689
            }
        }
        /* 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;
        }
6690 6691 6692
        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) {
6693
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
6694 6695 6696
                           _("'%s' controller only supports up to '%u' ports"),
                           virDomainControllerModelUSBTypeToString(cont->model),
                           QEMU_USB_XHCI_MAXPORTS);
6697 6698 6699
            return -1;
        }
        break;
6700

6701
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
6702 6703 6704 6705 6706 6707 6708

        /* 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) {
6709 6710 6711 6712 6713 6714
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("pci-root and pcie-root controllers "
                             "should have index 0"));
            return -1;
        }

6715
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS &&
6716
            !qemuDomainIsI440FX(def)) {
6717 6718 6719 6720 6721 6722
            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 &&
6723
            !qemuDomainIsQ35(def)) {
6724 6725 6726 6727 6728 6729
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("pcie-expander-bus controllers are only supported "
                             "on q35-based machinetypes"));
            return -1;
        }

6730 6731 6732
        /* 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
6733 6734
         * from 0 .. size-1.
         */
6735 6736
        if (cont->opts.pciopts.numaNode >= 0 &&
            cont->opts.pciopts.numaNode >=
6737
            (int)virDomainNumaGetNodeCount(def->numa)) {
6738 6739 6740 6741 6742 6743 6744 6745 6746 6747
            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;
        }
6748 6749 6750 6751 6752 6753 6754
        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:
6755
    case VIR_DOMAIN_CONTROLLER_TYPE_XENBUS:
6756 6757
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
6758 6759 6760 6761 6762
    }

    return 0;
}

6763 6764 6765 6766 6767 6768
static int
qemuDomainChrDefPostParse(virDomainChrDefPtr chr,
                          const virDomainDef *def,
                          virQEMUDriverPtr driver,
                          unsigned int parseFlags)
{
6769 6770 6771 6772 6773 6774 6775 6776 6777 6778 6779 6780 6781 6782
    /* 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)) {
6783
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
6784
        } else if (qemuDomainIsARMVirt(def) || qemuDomainIsRISCVVirt(def)) {
6785
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
6786 6787
        } else if (ARCH_IS_S390(def->os.arch)) {
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
6788 6789 6790
        }
    }

6791 6792 6793
    /* Set the default target model */
    if (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL &&
        chr->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_NONE) {
6794
        switch ((virDomainChrSerialTargetType)chr->targetType) {
6795 6796 6797 6798 6799 6800 6801 6802 6803
        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;
6804 6805 6806
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY;
            break;
6807
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
6808 6809 6810 6811 6812
            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;
            }
6813
            break;
6814 6815 6816
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE;
            break;
6817 6818 6819 6820 6821 6822 6823
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            /* Nothing to do */
            break;
        }
    }

6824 6825 6826 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836 6837 6838 6839
    /* 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;
}
6840

6841

6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857 6858 6859 6860 6861 6862 6863 6864 6865 6866 6867 6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882 6883 6884 6885 6886 6887 6888 6889 6890 6891 6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906 6907 6908 6909 6910 6911 6912 6913 6914 6915 6916 6917 6918 6919 6920 6921 6922
/**
 * 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;
}


6923 6924
static int
qemuDomainDeviceDiskDefPostParse(virDomainDiskDefPtr disk,
6925
                                 virQEMUCapsPtr qemuCaps,
6926
                                 unsigned int parseFlags)
6927 6928
{
    /* set default disk types and drivers */
6929 6930 6931
    if (!virDomainDiskGetDriver(disk) &&
        virDomainDiskSetDriver(disk, "qemu") < 0)
        return -1;
6932

6933 6934
    /* default disk format for drives */
    if (virDomainDiskGetFormat(disk) == VIR_STORAGE_FILE_NONE &&
6935
        virDomainDiskGetType(disk) != VIR_STORAGE_TYPE_VOLUME)
6936
        virDomainDiskSetFormat(disk, VIR_STORAGE_FILE_RAW);
6937

6938 6939 6940 6941
    /* default disk format for mirrored drive */
    if (disk->mirror &&
        disk->mirror->format == VIR_STORAGE_FILE_NONE)
        disk->mirror->format = VIR_STORAGE_FILE_RAW;
6942

6943 6944 6945 6946
    if (qemuDomainDeviceDiskDefPostParseRestoreSecAlias(disk, qemuCaps,
                                                        parseFlags) < 0)
        return -1;

6947 6948 6949 6950 6951 6952 6953
    /* 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;

6954 6955 6956 6957
    return 0;
}


6958 6959 6960 6961 6962 6963
static int
qemuDomainDeviceNetDefPostParse(virDomainNetDefPtr net,
                                const virDomainDef *def,
                                virQEMUCapsPtr qemuCaps)
{
    if (net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
6964 6965
        !virDomainNetGetModelString(net))
        net->model = qemuDomainDefaultNetModel(def, qemuCaps);
6966 6967 6968 6969 6970

    return 0;
}


6971 6972 6973 6974 6975 6976 6977
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;
6978 6979 6980
        else if (qemuDomainIsARMVirt(def) ||
                 qemuDomainIsRISCVVirt(def) ||
                 ARCH_IS_S390(def->os.arch))
6981 6982 6983 6984 6985 6986 6987 6988 6989 6990 6991 6992 6993 6994
            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;
}


6995 6996 6997 6998 6999 7000 7001 7002 7003 7004 7005 7006 7007 7008 7009 7010 7011
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
7012 7013 7014 7015 7016 7017 7018 7019 7020 7021
static int
qemuDomainVsockDefPostParse(virDomainVsockDefPtr vsock)
{
    if (vsock->model == VIR_DOMAIN_VSOCK_MODEL_DEFAULT)
        vsock->model = VIR_DOMAIN_VSOCK_MODEL_VIRTIO;

    return 0;
}


7022 7023 7024 7025 7026 7027 7028
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) &&
7029
        mdevsrc->model == VIR_MDEV_MODEL_TYPE_VFIO_PCI &&
7030 7031 7032 7033 7034 7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051
        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;
}


7052 7053
static int
qemuDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
7054
                             const virDomainDef *def,
7055
                             virCapsPtr caps ATTRIBUTE_UNUSED,
7056
                             unsigned int parseFlags,
7057
                             void *opaque,
7058
                             void *parseOpaque)
7059
{
7060
    virQEMUDriverPtr driver = opaque;
7061 7062 7063
    /* 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. */
7064
    virQEMUCapsPtr qemuCaps = parseOpaque;
7065
    int ret = -1;
7066

7067 7068 7069 7070
    switch ((virDomainDeviceType) dev->type) {
    case VIR_DOMAIN_DEVICE_NET:
        ret = qemuDomainDeviceNetDefPostParse(dev->data.net, def, qemuCaps);
        break;
7071

7072
    case VIR_DOMAIN_DEVICE_DISK:
7073
        ret = qemuDomainDeviceDiskDefPostParse(dev->data.disk, qemuCaps,
7074
                                               parseFlags);
7075
        break;
7076

7077 7078 7079
    case VIR_DOMAIN_DEVICE_VIDEO:
        ret = qemuDomainDeviceVideoDefPostParse(dev->data.video, def);
        break;
7080

7081 7082 7083
    case VIR_DOMAIN_DEVICE_PANIC:
        ret = qemuDomainDevicePanicDefPostParse(dev->data.panic, def);
        break;
7084

7085 7086 7087 7088
    case VIR_DOMAIN_DEVICE_CONTROLLER:
        ret = qemuDomainControllerDefPostParse(dev->data.controller, def,
                                               qemuCaps, parseFlags);
        break;
7089

7090 7091 7092
    case VIR_DOMAIN_DEVICE_SHMEM:
        ret = qemuDomainShmemDefPostParse(dev->data.shmem);
        break;
7093

7094 7095 7096
    case VIR_DOMAIN_DEVICE_CHR:
        ret = qemuDomainChrDefPostParse(dev->data.chr, def, driver, parseFlags);
        break;
7097

J
Ján Tomko 已提交
7098 7099 7100 7101
    case VIR_DOMAIN_DEVICE_VSOCK:
        ret = qemuDomainVsockDefPostParse(dev->data.vsock);
        break;

7102 7103 7104 7105
    case VIR_DOMAIN_DEVICE_HOSTDEV:
        ret = qemuDomainHostdevDefPostParse(dev->data.hostdev, qemuCaps);
        break;

7106 7107 7108 7109 7110 7111 7112 7113 7114 7115 7116 7117 7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131 7132 7133
    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;
    }
7134 7135

    return ret;
7136 7137 7138
}


7139 7140
static int
qemuDomainDefAssignAddresses(virDomainDef *def,
7141
                             virCapsPtr caps ATTRIBUTE_UNUSED,
7142
                             unsigned int parseFlags ATTRIBUTE_UNUSED,
7143
                             void *opaque,
7144
                             void *parseOpaque)
7145 7146
{
    virQEMUDriverPtr driver = opaque;
7147 7148 7149
    /* 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. */
7150
    virQEMUCapsPtr qemuCaps = parseOpaque;
7151
    bool newDomain = parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE;
7152

7153 7154 7155 7156 7157 7158 7159
    /* 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;

7160 7161 7162 7163 7164 7165 7166 7167 7168 7169 7170 7171 7172 7173
    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,
7174
                                                def->emulator)))
7175
        return 1;
7176

7177 7178 7179 7180 7181 7182 7183 7184
    return 0;
}


static void
qemuDomainPostParseDataFree(void *parseOpaque)
{
    virQEMUCapsPtr qemuCaps = parseOpaque;
7185 7186 7187 7188 7189

    virObjectUnref(qemuCaps);
}


7190
virDomainDefParserConfig virQEMUDriverDomainDefParserConfig = {
7191
    .domainPostParseBasicCallback = qemuDomainDefPostParseBasic,
7192 7193
    .domainPostParseDataAlloc = qemuDomainPostParseDataAlloc,
    .domainPostParseDataFree = qemuDomainPostParseDataFree,
7194
    .devicesPostParseCallback = qemuDomainDeviceDefPostParse,
7195
    .domainPostParseCallback = qemuDomainDefPostParse,
7196
    .assignAddressesCallback = qemuDomainDefAssignAddresses,
7197
    .domainValidateCallback = qemuDomainDefValidate,
7198 7199
    .deviceValidateCallback = qemuDomainDeviceDefValidate,

7200
    .features = VIR_DOMAIN_DEF_FEATURE_MEMORY_HOTPLUG |
7201
                VIR_DOMAIN_DEF_FEATURE_OFFLINE_VCPUPIN |
7202
                VIR_DOMAIN_DEF_FEATURE_INDIVIDUAL_VCPUS |
7203
                VIR_DOMAIN_DEF_FEATURE_USER_ALIAS |
7204 7205
                VIR_DOMAIN_DEF_FEATURE_FW_AUTOSELECT |
                VIR_DOMAIN_DEF_FEATURE_NET_MODEL_STRING,
7206 7207 7208
};


7209
static void
7210
qemuDomainObjSaveJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
7211
{
7212 7213 7214
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    if (virDomainObjIsActive(obj)) {
7215
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj, driver->caps) < 0)
7216
            VIR_WARN("Failed to save status on vm %s", obj->def->name);
7217
    }
7218

7219
    virObjectUnref(cfg);
7220 7221
}

J
Jiri Denemark 已提交
7222
void
7223
qemuDomainObjSetJobPhase(virQEMUDriverPtr driver,
J
Jiri Denemark 已提交
7224 7225 7226 7227
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
7228
    unsigned long long me = virThreadSelfID();
J
Jiri Denemark 已提交
7229 7230 7231 7232

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

7233 7234 7235 7236 7237
    VIR_DEBUG("Setting '%s' phase to '%s'",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob, phase));

    if (priv->job.asyncOwner && me != priv->job.asyncOwner) {
7238
        VIR_WARN("'%s' async job is owned by thread %llu",
7239 7240 7241 7242
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }

J
Jiri Denemark 已提交
7243
    priv->job.phase = phase;
7244
    priv->job.asyncOwner = me;
J
Jiri Denemark 已提交
7245 7246 7247
    qemuDomainObjSaveJob(driver, obj);
}

7248
void
7249 7250
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
7251 7252 7253
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

7254 7255 7256 7257 7258 7259 7260
    if (!priv->job.asyncJob)
        return;

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

void
7261
qemuDomainObjDiscardAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
7262 7263 7264 7265 7266 7267
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
7268
    qemuDomainObjSaveJob(driver, obj);
7269 7270
}

7271 7272 7273 7274 7275 7276 7277 7278 7279
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()) {
7280
        VIR_WARN("'%s' async job is owned by thread %llu",
7281 7282 7283 7284 7285 7286
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }
    priv->job.asyncOwner = 0;
}

7287
static bool
7288
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
7289
{
7290 7291 7292
    return !priv->job.asyncJob ||
           job == QEMU_JOB_NONE ||
           (priv->job.mask & JOB_MASK(job)) != 0;
7293 7294
}

7295
bool
7296
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
7297 7298 7299 7300
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311
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));
}

7312 7313 7314
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

7315 7316 7317 7318 7319 7320
/**
 * qemuDomainObjBeginJobInternal:
 * @driver: qemu driver
 * @obj: domain object
 * @job: qemuDomainJob to start
 * @asyncJob: qemuDomainAsyncJob to start
7321
 * @nowait: don't wait trying to acquire @job
7322 7323 7324 7325
 *
 * 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
7326 7327 7328 7329 7330
 * 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.
7331 7332 7333 7334 7335
 *
 * Returns: 0 on success,
 *         -2 if unable to start job because of timeout or
 *            maxQueuedJobs limit,
 *         -1 otherwise.
7336
 */
7337
static int ATTRIBUTE_NONNULL(1)
7338
qemuDomainObjBeginJobInternal(virQEMUDriverPtr driver,
7339
                              virDomainObjPtr obj,
7340
                              qemuDomainJob job,
7341
                              qemuDomainAgentJob agentJob,
7342 7343
                              qemuDomainAsyncJob asyncJob,
                              bool nowait)
7344 7345
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
7346
    unsigned long long now;
7347
    unsigned long long then;
7348
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
J
Jiri Denemark 已提交
7349
    bool async = job == QEMU_JOB_ASYNC;
7350
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
7351
    const char *blocker = NULL;
7352
    const char *agentBlocker = NULL;
J
Jiri Denemark 已提交
7353
    int ret = -1;
J
Jiri Denemark 已提交
7354
    unsigned long long duration = 0;
7355
    unsigned long long agentDuration = 0;
J
Jiri Denemark 已提交
7356
    unsigned long long asyncDuration = 0;
7357

7358 7359 7360 7361 7362 7363
    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
7364
              qemuDomainJobTypeToString(priv->job.active),
7365
              qemuDomainAgentJobTypeToString(priv->job.agentActive),
J
Jiri Denemark 已提交
7366
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
7367

7368 7369
    if (virTimeMillisNow(&now) < 0) {
        virObjectUnref(cfg);
7370
        return -1;
7371 7372
    }

7373
    priv->jobs_queued++;
J
Jiri Denemark 已提交
7374
    then = now + QEMU_JOB_WAIT_TIME;
7375

7376
 retry:
7377 7378
    if ((!async && job != QEMU_JOB_DESTROY) &&
        cfg->maxQueuedJobs &&
7379
        priv->jobs_queued > cfg->maxQueuedJobs) {
7380 7381 7382
        goto error;
    }

7383
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
7384 7385 7386
        if (nowait)
            goto cleanup;

7387
        VIR_DEBUG("Waiting for async job (vm=%p name=%s)", obj, obj->def->name);
7388
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->parent.lock, then) < 0)
7389 7390 7391
            goto error;
    }

7392
    while (!qemuDomainObjCanSetJob(priv, job, agentJob)) {
7393 7394 7395
        if (nowait)
            goto cleanup;

7396
        VIR_DEBUG("Waiting for job (vm=%p name=%s)", obj, obj->def->name);
7397
        if (virCondWaitUntil(&priv->job.cond, &obj->parent.lock, then) < 0)
7398
            goto error;
7399
    }
7400 7401 7402

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
7403
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
7404 7405
        goto retry;

J
Jiri Denemark 已提交
7406 7407
    ignore_value(virTimeMillisNow(&now));

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
    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;
7448
    }
7449

7450 7451
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
7452

7453
    virObjectUnref(cfg);
7454
    return 0;
7455

7456
 error:
J
Jiri Denemark 已提交
7457 7458 7459
    ignore_value(virTimeMillisNow(&now));
    if (priv->job.active && priv->job.started)
        duration = now - priv->job.started;
7460 7461
    if (priv->job.agentActive && priv->job.agentStarted)
        agentDuration = now - priv->job.agentStarted;
J
Jiri Denemark 已提交
7462 7463 7464
    if (priv->job.asyncJob && priv->job.asyncStarted)
        asyncDuration = now - priv->job.asyncStarted;

7465 7466 7467 7468
    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)",
7469
             qemuDomainJobTypeToString(job),
7470
             qemuDomainAgentJobTypeToString(agentJob),
7471 7472 7473
             qemuDomainAsyncJobTypeToString(asyncJob),
             obj->def->name,
             qemuDomainJobTypeToString(priv->job.active),
7474
             qemuDomainAgentJobTypeToString(priv->job.agentActive),
7475
             qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
7476
             priv->job.owner, NULLSTR(priv->job.ownerAPI),
7477
             priv->job.agentOwner, NULLSTR(priv->job.agentOwnerAPI),
J
Jiri Denemark 已提交
7478
             priv->job.asyncOwner, NULLSTR(priv->job.asyncOwnerAPI),
7479
             priv->job.apiFlags,
7480
             duration / 1000, agentDuration / 1000, asyncDuration / 1000);
7481

7482 7483 7484 7485 7486 7487 7488 7489 7490
    if (job) {
        if (nested || qemuDomainNestedJobAllowed(priv, job))
            blocker = priv->job.ownerAPI;
        else
            blocker = priv->job.asyncOwnerAPI;
    }

    if (agentJob)
        agentBlocker = priv->job.agentOwnerAPI;
7491

7492
    if (errno == ETIMEDOUT) {
7493 7494 7495 7496 7497 7498
        if (blocker && agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by monitor=%s agent=%s)"),
                           blocker, agentBlocker);
        } else if (blocker) {
7499
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
7500 7501
                           _("cannot acquire state change "
                             "lock (held by monitor=%s)"),
7502
                           blocker);
7503 7504 7505 7506 7507
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by agent=%s)"),
                           agentBlocker);
7508 7509 7510 7511
        } else {
            virReportError(VIR_ERR_OPERATION_TIMEOUT, "%s",
                           _("cannot acquire state change lock"));
        }
7512 7513 7514
        ret = -2;
    } else if (cfg->maxQueuedJobs &&
               priv->jobs_queued > cfg->maxQueuedJobs) {
7515 7516 7517 7518 7519 7520 7521
        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) {
7522
            virReportError(VIR_ERR_OPERATION_FAILED,
7523 7524
                           _("cannot acquire state change "
                             "lock (held by monitor=%s) "
7525 7526
                             "due to max_queued limit"),
                           blocker);
7527 7528 7529 7530 7531 7532
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_FAILED,
                           _("cannot acquire state change "
                             "lock (held by agent=%s) "
                             "due to max_queued limit"),
                           agentBlocker);
7533 7534 7535 7536 7537
        } else {
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("cannot acquire state change lock "
                             "due to max_queued limit"));
        }
7538 7539
        ret = -2;
    } else {
7540
        virReportSystemError(errno, "%s", _("cannot acquire job mutex"));
7541
    }
J
Jiri Denemark 已提交
7542 7543

 cleanup:
7544
    priv->jobs_queued--;
7545
    virObjectUnref(cfg);
7546
    return ret;
7547 7548 7549
}

/*
7550
 * obj must be locked before calling
7551 7552 7553 7554
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
7555
 * Successful calls must be followed by EndJob eventually
7556
 */
7557
int qemuDomainObjBeginJob(virQEMUDriverPtr driver,
7558
                          virDomainObjPtr obj,
7559
                          qemuDomainJob job)
7560
{
7561
    if (qemuDomainObjBeginJobInternal(driver, obj, job,
7562
                                      QEMU_AGENT_JOB_NONE,
7563
                                      QEMU_ASYNC_JOB_NONE, false) < 0)
7564 7565 7566
        return -1;
    else
        return 0;
7567 7568
}

7569 7570 7571 7572 7573 7574 7575 7576 7577 7578 7579 7580 7581 7582 7583 7584 7585 7586 7587 7588 7589 7590 7591 7592 7593 7594 7595 7596 7597 7598 7599 7600 7601 7602 7603 7604 7605
/**
 * 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);
}

7606
int qemuDomainObjBeginAsyncJob(virQEMUDriverPtr driver,
7607
                               virDomainObjPtr obj,
7608
                               qemuDomainAsyncJob asyncJob,
7609 7610
                               virDomainJobOperation operation,
                               unsigned long apiFlags)
7611
{
7612 7613
    qemuDomainObjPrivatePtr priv;

7614
    if (qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_ASYNC,
7615
                                      QEMU_AGENT_JOB_NONE,
7616
                                      asyncJob, false) < 0)
7617
        return -1;
7618 7619 7620

    priv = obj->privateData;
    priv->job.current->operation = operation;
7621
    priv->job.apiFlags = apiFlags;
7622
    return 0;
7623 7624
}

7625
int
7626 7627
qemuDomainObjBeginNestedJob(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
7628
                            qemuDomainAsyncJob asyncJob)
7629 7630 7631 7632 7633
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (asyncJob != priv->job.asyncJob) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
7634 7635
                       _("unexpected async job %d type expected %d"),
                       asyncJob, priv->job.asyncJob);
7636 7637 7638 7639
        return -1;
    }

    if (priv->job.asyncOwner != virThreadSelfID()) {
7640
        VIR_WARN("This thread doesn't seem to be the async job owner: %llu",
7641 7642 7643 7644 7645
                 priv->job.asyncOwner);
    }

    return qemuDomainObjBeginJobInternal(driver, obj,
                                         QEMU_JOB_ASYNC_NESTED,
7646
                                         QEMU_AGENT_JOB_NONE,
7647 7648
                                         QEMU_ASYNC_JOB_NONE,
                                         false);
7649 7650
}

7651 7652 7653 7654 7655 7656 7657 7658 7659 7660 7661 7662 7663 7664 7665 7666 7667 7668 7669
/**
 * 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,
7670
                                         QEMU_AGENT_JOB_NONE,
7671 7672
                                         QEMU_ASYNC_JOB_NONE, true);
}
7673

7674
/*
7675
 * obj must be locked and have a reference before calling
7676 7677 7678 7679
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 */
7680 7681
void
qemuDomainObjEndJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
7682 7683
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
7684
    qemuDomainJob job = priv->job.active;
7685

7686 7687
    priv->jobs_queued--;

7688
    VIR_DEBUG("Stopping job: %s (async=%s vm=%p name=%s)",
7689
              qemuDomainJobTypeToString(job),
7690 7691
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
7692

7693
    qemuDomainObjResetJob(priv);
7694 7695
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
7696 7697 7698 7699 7700 7701 7702 7703 7704 7705 7706 7707 7708 7709 7710 7711 7712 7713 7714 7715 7716 7717 7718 7719 7720 7721 7722 7723 7724 7725 7726 7727 7728 7729 7730 7731 7732 7733 7734 7735 7736 7737 7738 7739 7740 7741 7742
    /* 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);
7743 7744
}

7745
void
7746
qemuDomainObjEndAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
7747 7748
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
7749

7750 7751
    priv->jobs_queued--;

7752 7753 7754
    VIR_DEBUG("Stopping async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
7755

7756
    qemuDomainObjResetAsyncJob(priv);
7757
    qemuDomainObjSaveJob(driver, obj);
7758 7759 7760
    virCondBroadcast(&priv->job.asyncCond);
}

7761 7762 7763 7764 7765
void
qemuDomainObjAbortAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

7766 7767 7768
    VIR_DEBUG("Requesting abort of async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
7769

7770 7771
    priv->job.abortJob = true;
    virDomainObjBroadcast(obj);
7772 7773
}

7774 7775 7776 7777
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU monitor API call
7778 7779 7780
 * Must have already either called qemuDomainObjBeginJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active; may not be used for nested async jobs.
7781 7782 7783
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
7784
static int
7785
qemuDomainObjEnterMonitorInternal(virQEMUDriverPtr driver,
7786
                                  virDomainObjPtr obj,
7787
                                  qemuDomainAsyncJob asyncJob)
7788 7789 7790
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

7791
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
7792 7793 7794
        int ret;
        if ((ret = qemuDomainObjBeginNestedJob(driver, obj, asyncJob)) < 0)
            return ret;
7795
        if (!virDomainObjIsActive(obj)) {
7796 7797
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
7798
            qemuDomainObjEndJob(driver, obj);
7799 7800
            return -1;
        }
7801 7802 7803
    } 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");
7804 7805
    }

7806 7807
    VIR_DEBUG("Entering monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
7808
    virObjectLock(priv->mon);
7809
    virObjectRef(priv->mon);
7810
    ignore_value(virTimeMillisNow(&priv->monStart));
7811
    virObjectUnlock(obj);
7812 7813

    return 0;
7814 7815
}

7816
static void ATTRIBUTE_NONNULL(1)
7817
qemuDomainObjExitMonitorInternal(virQEMUDriverPtr driver,
7818
                                 virDomainObjPtr obj)
7819 7820
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
7821
    bool hasRefs;
7822

7823
    hasRefs = virObjectUnref(priv->mon);
7824

7825
    if (hasRefs)
7826
        virObjectUnlock(priv->mon);
7827

7828
    virObjectLock(obj);
7829 7830
    VIR_DEBUG("Exited monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
7831

7832
    priv->monStart = 0;
7833
    if (!hasRefs)
7834
        priv->mon = NULL;
7835

J
Jiri Denemark 已提交
7836 7837
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjEndJob(driver, obj);
7838 7839
}

7840
void qemuDomainObjEnterMonitor(virQEMUDriverPtr driver,
7841
                               virDomainObjPtr obj)
7842
{
7843
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, obj,
7844
                                                   QEMU_ASYNC_JOB_NONE));
7845 7846
}

7847
/* obj must NOT be locked before calling
7848 7849
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
7850 7851 7852 7853 7854 7855
 *
 * 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.
7856
 */
7857 7858
int qemuDomainObjExitMonitor(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
7859
{
7860
    qemuDomainObjExitMonitorInternal(driver, obj);
7861
    if (!virDomainObjIsActive(obj)) {
7862
        if (virGetLastErrorCode() == VIR_ERR_OK)
7863 7864
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
7865 7866 7867
        return -1;
    }
    return 0;
7868
}
7869 7870

/*
7871
 * obj must be locked before calling
7872 7873
 *
 * To be called immediately before any QEMU monitor API call.
7874
 * Must have already either called qemuDomainObjBeginJob()
7875 7876 7877 7878 7879
 * 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
7880 7881 7882
 * 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).
7883 7884
 */
int
7885
qemuDomainObjEnterMonitorAsync(virQEMUDriverPtr driver,
7886
                               virDomainObjPtr obj,
7887
                               qemuDomainAsyncJob asyncJob)
7888
{
7889
    return qemuDomainObjEnterMonitorInternal(driver, obj, asyncJob);
7890 7891
}

D
Daniel P. Berrange 已提交
7892

7893 7894 7895 7896
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU agent API call.
7897 7898 7899
 * Must have already called qemuDomainObjBeginAgentJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active.
7900 7901 7902
 *
 * To be followed with qemuDomainObjExitAgent() once complete
 */
7903
qemuAgentPtr
7904
qemuDomainObjEnterAgent(virDomainObjPtr obj)
D
Daniel P. Berrange 已提交
7905 7906
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
7907
    qemuAgentPtr agent = priv->agent;
D
Daniel P. Berrange 已提交
7908

7909 7910
    VIR_DEBUG("Entering agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
7911 7912 7913

    virObjectLock(agent);
    virObjectRef(agent);
7914
    virObjectUnlock(obj);
7915 7916

    return agent;
D
Daniel P. Berrange 已提交
7917 7918
}

7919 7920 7921 7922 7923 7924

/* obj must NOT be locked before calling
 *
 * Should be paired with an earlier qemuDomainObjEnterAgent() call
 */
void
7925
qemuDomainObjExitAgent(virDomainObjPtr obj, qemuAgentPtr agent)
D
Daniel P. Berrange 已提交
7926
{
7927 7928
    virObjectUnlock(agent);
    virObjectUnref(agent);
7929
    virObjectLock(obj);
D
Daniel P. Berrange 已提交
7930

7931 7932
    VIR_DEBUG("Exited agent (agent=%p vm=%p name=%s)",
              agent, obj, obj->def->name);
D
Daniel P. Berrange 已提交
7933 7934
}

7935
void qemuDomainObjEnterRemote(virDomainObjPtr obj)
7936
{
7937 7938
    VIR_DEBUG("Entering remote (vm=%p name=%s)",
              obj, obj->def->name);
7939
    virObjectUnlock(obj);
7940 7941
}

7942 7943 7944 7945

int
qemuDomainObjExitRemote(virDomainObjPtr obj,
                        bool checkActive)
7946
{
7947
    virObjectLock(obj);
7948 7949
    VIR_DEBUG("Exited remote (vm=%p name=%s)",
              obj, obj->def->name);
7950 7951 7952 7953 7954 7955 7956 7957 7958

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

    return 0;
7959
}
7960 7961


7962 7963 7964 7965 7966 7967 7968 7969 7970 7971 7972 7973 7974 7975 7976 7977 7978 7979 7980
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;
}


7981 7982 7983 7984 7985 7986
virDomainDefPtr
qemuDomainDefCopy(virQEMUDriverPtr driver,
                  virDomainDefPtr src,
                  unsigned int flags)
{
    virDomainDefPtr ret = NULL;
7987
    char *xml;
7988

7989
    if (!(xml = qemuDomainDefFormatXML(driver, src, flags)))
7990
        return NULL;
7991

7992
    ret = qemuDomainDefFromXML(driver, xml);
7993 7994 7995 7996 7997

    VIR_FREE(xml);
    return ret;
}

7998 7999 8000 8001 8002 8003 8004

static int
qemuDomainDefFormatBufInternal(virQEMUDriverPtr driver,
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags,
                               virBuffer *buf)
8005
{
8006
    int ret = -1;
8007
    virDomainDefPtr copy = NULL;
8008
    virCapsPtr caps = NULL;
8009
    virQEMUCapsPtr qemuCaps = NULL;
8010

8011 8012
    virCheckFlags(VIR_DOMAIN_XML_COMMON_FLAGS | VIR_DOMAIN_XML_UPDATE_CPU, -1);

8013 8014
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;
8015

8016 8017 8018
    if (!(flags & (VIR_DOMAIN_XML_UPDATE_CPU | VIR_DOMAIN_XML_MIGRATABLE)))
        goto format;

8019
    if (!(copy = virDomainDefCopy(def, caps, driver->xmlopt, NULL,
8020 8021 8022 8023 8024
                                  flags & VIR_DOMAIN_XML_MIGRATABLE)))
        goto cleanup;

    def = copy;

8025
    /* Update guest CPU requirements according to host CPU */
8026
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) &&
8027
        def->cpu &&
8028 8029
        (def->cpu->mode != VIR_CPU_MODE_CUSTOM ||
         def->cpu->model)) {
8030 8031 8032 8033 8034 8035 8036 8037
        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)
8038 8039 8040
            goto cleanup;
    }

8041
    if ((flags & VIR_DOMAIN_XML_MIGRATABLE)) {
8042
        size_t i;
8043
        int toremove = 0;
8044
        virDomainControllerDefPtr usb = NULL, pci = NULL;
8045 8046 8047 8048 8049 8050 8051 8052 8053 8054 8055 8056 8057 8058 8059

        /* 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];
            }
        }
8060 8061 8062 8063 8064 8065 8066 8067 8068

        /* 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'/>.
8069
         */
8070
        if (qemuDomainIsI440FX(def) &&
8071
            usb && usb->idx == 0 &&
8072
            (usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT ||
8073 8074
             usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_PIIX3_UHCI) &&
            !virDomainDeviceAliasIsUserAlias(usb->info.alias)) {
8075 8076
            VIR_DEBUG("Removing default USB controller from domain '%s'"
                      " for migration compatibility", def->name);
8077
            toremove++;
8078 8079 8080 8081 8082 8083 8084 8085 8086 8087 8088 8089 8090 8091 8092 8093 8094
        } 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 &&
8095 8096 8097
            pci->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT &&
            !virDomainDeviceAliasIsUserAlias(pci->info.alias) &&
            !pci->opts.pciopts.pcihole64) {
L
Laine Stump 已提交
8098
            VIR_DEBUG("Removing default pci-root from domain '%s'"
8099
                      " for migration compatibility", def->name);
8100
            toremove++;
8101 8102 8103 8104
        } else {
            pci = NULL;
        }

8105
        if (toremove) {
8106 8107 8108
            virDomainControllerDefPtr *controllers = def->controllers;
            int ncontrollers = def->ncontrollers;

8109
            if (VIR_ALLOC_N(def->controllers, ncontrollers - toremove) < 0) {
8110
                def->controllers = controllers;
8111 8112 8113 8114 8115
                goto cleanup;
            }

            def->ncontrollers = 0;
            for (i = 0; i < ncontrollers; i++) {
8116
                if (controllers[i] != usb && controllers[i] != pci)
8117 8118
                    def->controllers[def->ncontrollers++] = controllers[i];
            }
8119 8120 8121 8122

            VIR_FREE(controllers);
            virDomainControllerDefFree(pci);
            virDomainControllerDefFree(usb);
8123
        }
8124

8125 8126 8127 8128 8129 8130 8131 8132 8133
        /* 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;
            }
        }

8134 8135 8136 8137
        for (i = 0; i < def->nchannels; i++) {
            if (qemuDomainChrDefDropDefaultPath(def->channels[i], driver) < 0)
                goto cleanup;
        }
8138

8139 8140 8141 8142 8143 8144 8145 8146 8147
        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) {
8148
                switch ((virDomainChrSerialTargetType)serial->targetType) {
8149
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
8150
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
8151 8152 8153 8154 8155 8156
                    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:
8157
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
8158 8159 8160 8161 8162 8163 8164 8165
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
                    /* Nothing to do */
                    break;
                }
            }
        }

8166 8167 8168 8169 8170 8171 8172 8173 8174
        /* 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;
        }
8175 8176
    }

8177 8178
 format:
    ret = virDomainDefFormatInternal(def, caps,
8179
                                     virDomainDefFormatConvertXMLFlags(flags),
8180
                                     buf, driver->xmlopt);
8181

8182
 cleanup:
8183
    virDomainDefFree(copy);
8184
    virObjectUnref(caps);
8185
    virObjectUnref(qemuCaps);
8186 8187
    return ret;
}
8188

8189 8190 8191 8192 8193 8194 8195 8196 8197 8198 8199 8200 8201 8202 8203 8204

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)
8205 8206 8207
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

8208
    if (qemuDomainDefFormatBufInternal(driver, def, origCPU, flags, &buf) < 0)
8209 8210 8211 8212 8213
        return NULL;

    return virBufferContentAndReset(&buf);
}

8214 8215 8216 8217 8218 8219 8220 8221 8222 8223

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


8224
char *qemuDomainFormatXML(virQEMUDriverPtr driver,
8225
                          virDomainObjPtr vm,
8226
                          unsigned int flags)
8227 8228
{
    virDomainDefPtr def;
8229 8230
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCPUDefPtr origCPU = NULL;
8231

8232
    if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef) {
8233
        def = vm->newDef;
8234
    } else {
8235
        def = vm->def;
8236
        origCPU = priv->origCPU;
8237
    }
8238

8239
    return qemuDomainDefFormatXMLInternal(driver, def, origCPU, flags);
8240 8241
}

8242
char *
8243
qemuDomainDefFormatLive(virQEMUDriverPtr driver,
8244
                        virDomainDefPtr def,
8245
                        virCPUDefPtr origCPU,
8246 8247
                        bool inactive,
                        bool compatible)
8248 8249 8250 8251 8252
{
    unsigned int flags = QEMU_DOMAIN_FORMAT_LIVE_FLAGS;

    if (inactive)
        flags |= VIR_DOMAIN_XML_INACTIVE;
8253 8254
    if (compatible)
        flags |= VIR_DOMAIN_XML_MIGRATABLE;
8255

8256
    return qemuDomainDefFormatXMLInternal(driver, def, origCPU, flags);
8257 8258
}

8259

8260
void qemuDomainObjTaint(virQEMUDriverPtr driver,
8261
                        virDomainObjPtr obj,
8262
                        virDomainTaintFlags taint,
8263
                        qemuDomainLogContextPtr logCtxt)
8264
{
8265
    virErrorPtr orig_err = NULL;
8266
    bool closeLog = false;
8267 8268
    char *timestamp = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
8269

8270 8271 8272 8273 8274 8275 8276 8277 8278 8279 8280 8281 8282 8283 8284
    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();
8285

8286 8287 8288 8289 8290 8291 8292 8293 8294
    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;
8295
        }
8296 8297 8298 8299 8300 8301 8302 8303 8304 8305 8306 8307 8308
        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)
8309
        virObjectUnref(logCtxt);
8310 8311 8312
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
8313 8314 8315 8316
    }
}


8317
void qemuDomainObjCheckTaint(virQEMUDriverPtr driver,
8318
                             virDomainObjPtr obj,
8319
                             qemuDomainLogContextPtr logCtxt)
8320
{
8321
    size_t i;
8322
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8323
    qemuDomainObjPrivatePtr priv = obj->privateData;
8324
    bool custom_hypervisor_feat = false;
8325

8326
    if (virQEMUDriverIsPrivileged(driver) &&
8327 8328 8329
        (!cfg->clearEmulatorCapabilities ||
         cfg->user == 0 ||
         cfg->group == 0))
8330
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
8331

8332
    if (priv->hookRun)
8333
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOOK, logCtxt);
8334

8335
    if (obj->def->namespaceData) {
8336 8337
        qemuDomainXmlNsDefPtr qemuxmlns = obj->def->namespaceData;
        if (qemuxmlns->num_args || qemuxmlns->num_env)
8338
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logCtxt);
8339 8340 8341 8342
        if (qemuxmlns->ncapsadd > 0 || qemuxmlns->ncapsdel > 0)
            custom_hypervisor_feat = true;
    }

8343 8344
    if (custom_hypervisor_feat ||
        (cfg->capabilityfilters && *cfg->capabilityfilters)) {
8345 8346
        qemuDomainObjTaint(driver, obj,
                           VIR_DOMAIN_TAINT_CUSTOM_HYPERVISOR_FEATURE, logCtxt);
8347 8348
    }

8349
    if (obj->def->cpu && obj->def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
8350
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOST_CPU, logCtxt);
8351

8352
    for (i = 0; i < obj->def->ndisks; i++)
8353
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logCtxt);
8354

8355 8356
    for (i = 0; i < obj->def->nhostdevs; i++)
        qemuDomainObjCheckHostdevTaint(driver, obj, obj->def->hostdevs[i],
8357
                                       logCtxt);
8358

8359
    for (i = 0; i < obj->def->nnets; i++)
8360
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logCtxt);
8361

8362
    if (obj->def->os.dtb)
8363
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_DTB, logCtxt);
8364

8365
    virObjectUnref(cfg);
8366 8367 8368
}


8369
void qemuDomainObjCheckDiskTaint(virQEMUDriverPtr driver,
8370
                                 virDomainObjPtr obj,
8371
                                 virDomainDiskDefPtr disk,
8372
                                 qemuDomainLogContextPtr logCtxt)
8373
{
8374
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8375

8376 8377
    if (disk->rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
8378
                           logCtxt);
8379

8380 8381
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
        virStorageSourceGetActualType(disk->src) == VIR_STORAGE_TYPE_BLOCK &&
8382
        disk->src->path && virFileIsCDROM(disk->src->path) == 1)
8383
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CDROM_PASSTHROUGH,
8384
                           logCtxt);
8385

8386
    virObjectUnref(cfg);
8387 8388 8389
}


8390 8391 8392
void qemuDomainObjCheckHostdevTaint(virQEMUDriverPtr driver,
                                    virDomainObjPtr obj,
                                    virDomainHostdevDefPtr hostdev,
8393
                                    qemuDomainLogContextPtr logCtxt)
8394
{
8395
    if (!virHostdevIsSCSIDevice(hostdev))
8396
        return;
8397

8398 8399
    if (hostdev->source.subsys.u.scsi.rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
8400 8401 8402
}


8403
void qemuDomainObjCheckNetTaint(virQEMUDriverPtr driver,
8404
                                virDomainObjPtr obj,
8405
                                virDomainNetDefPtr net,
8406
                                qemuDomainLogContextPtr logCtxt)
8407
{
8408 8409 8410 8411 8412 8413
    /* 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)
8414
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logCtxt);
8415
}
8416 8417


8418 8419 8420 8421 8422 8423 8424
qemuDomainLogContextPtr qemuDomainLogContextNew(virQEMUDriverPtr driver,
                                                virDomainObjPtr vm,
                                                qemuDomainLogContextMode mode)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainLogContextPtr ctxt = NULL;

8425
    if (qemuDomainInitialize() < 0)
8426 8427 8428 8429
        goto cleanup;

    if (!(ctxt = virObjectNew(qemuDomainLogContextClass)))
        goto cleanup;
8430

8431
    VIR_DEBUG("Context new %p stdioLogD=%d", ctxt, cfg->stdioLogD);
8432 8433 8434
    ctxt->writefd = -1;
    ctxt->readfd = -1;

8435 8436 8437
    if (virAsprintf(&ctxt->path, "%s/%s.log", cfg->logDir, vm->def->name) < 0)
        goto error;

8438 8439 8440 8441
    if (cfg->stdioLogD) {
        ctxt->manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver));
        if (!ctxt->manager)
            goto error;
8442

8443 8444 8445 8446
        ctxt->writefd = virLogManagerDomainOpenLogFile(ctxt->manager,
                                                       "qemu",
                                                       vm->def->uuid,
                                                       vm->def->name,
8447
                                                       ctxt->path,
8448 8449 8450 8451 8452 8453
                                                       0,
                                                       &ctxt->inode,
                                                       &ctxt->pos);
        if (ctxt->writefd < 0)
            goto error;
    } else {
8454
        if ((ctxt->writefd = open(ctxt->path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
8455
            virReportSystemError(errno, _("failed to create logfile %s"),
8456
                                 ctxt->path);
8457 8458
            goto error;
        }
8459
        if (virSetCloseExec(ctxt->writefd) < 0) {
8460
            virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
8461
                                 ctxt->path);
8462 8463 8464
            goto error;
        }

8465 8466 8467 8468 8469 8470 8471
        /* 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"),
8472
                                 ctxt->path);
8473 8474 8475 8476
            goto error;
        }

        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START) {
8477
            if ((ctxt->readfd = open(ctxt->path, O_RDONLY, S_IRUSR | S_IWUSR)) < 0) {
8478
                virReportSystemError(errno, _("failed to open logfile %s"),
8479
                                     ctxt->path);
8480 8481 8482 8483
                goto error;
            }
            if (virSetCloseExec(ctxt->readfd) < 0) {
                virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
8484
                                     ctxt->path);
8485 8486 8487 8488
                goto error;
            }
        }

8489 8490
        if ((ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END)) < 0) {
            virReportSystemError(errno, _("failed to seek in log file %s"),
8491
                                 ctxt->path);
8492 8493 8494 8495
            goto error;
        }
    }

8496
 cleanup:
8497 8498 8499 8500
    virObjectUnref(cfg);
    return ctxt;

 error:
8501
    virObjectUnref(ctxt);
8502 8503
    ctxt = NULL;
    goto cleanup;
8504 8505 8506 8507 8508 8509 8510 8511 8512 8513 8514 8515 8516 8517
}


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;
8518 8519
    if (!ctxt->manager &&
        lseek(ctxt->writefd, 0, SEEK_END) < 0) {
8520
        virReportSystemError(errno, "%s",
8521
                             _("Unable to seek to end of domain logfile"));
8522 8523 8524 8525 8526 8527 8528 8529 8530 8531 8532 8533 8534 8535 8536 8537 8538 8539 8540 8541
        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)
{
8542 8543 8544 8545
    VIR_DEBUG("Context read %p manager=%p inode=%llu pos=%llu",
              ctxt, ctxt->manager,
              (unsigned long long)ctxt->inode,
              (unsigned long long)ctxt->pos);
8546
    char *buf;
8547 8548 8549
    size_t buflen;
    if (ctxt->manager) {
        buf = virLogManagerDomainReadLogFile(ctxt->manager,
8550
                                             ctxt->path,
8551 8552 8553 8554 8555 8556 8557 8558 8559
                                             ctxt->inode,
                                             ctxt->pos,
                                             1024 * 128,
                                             0);
        if (!buf)
            return -1;
        buflen = strlen(buf);
    } else {
        ssize_t got;
8560

8561
        buflen = 1024 * 128;
8562

8563 8564
        /* Best effort jump to start of messages */
        ignore_value(lseek(ctxt->readfd, ctxt->pos, SEEK_SET));
8565

8566 8567
        if (VIR_ALLOC_N(buf, buflen) < 0)
            return -1;
8568

8569 8570 8571 8572 8573 8574 8575 8576 8577 8578 8579 8580 8581
        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;
    }
8582 8583 8584

    *msg = buf;

8585
    return buflen;
8586 8587 8588
}


8589 8590 8591 8592 8593 8594 8595 8596 8597 8598 8599 8600 8601 8602 8603 8604 8605 8606 8607 8608 8609 8610 8611 8612 8613 8614 8615 8616 8617 8618 8619 8620 8621 8622 8623 8624 8625 8626 8627 8628 8629 8630 8631 8632 8633 8634 8635 8636 8637 8638 8639 8640 8641 8642 8643 8644 8645 8646 8647 8648 8649 8650
/**
 * 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;
}


8651 8652 8653 8654 8655 8656 8657 8658
int qemuDomainLogContextGetWriteFD(qemuDomainLogContextPtr ctxt)
{
    return ctxt->writefd;
}


void qemuDomainLogContextMarkPosition(qemuDomainLogContextPtr ctxt)
{
8659 8660
    if (ctxt->manager)
        virLogManagerDomainGetLogFilePosition(ctxt->manager,
8661
                                              ctxt->path,
8662 8663 8664 8665 8666
                                              0,
                                              &ctxt->inode,
                                              &ctxt->pos);
    else
        ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END);
8667 8668 8669
}


8670 8671 8672 8673 8674 8675
virLogManagerPtr qemuDomainLogContextGetManager(qemuDomainLogContextPtr ctxt)
{
    return ctxt->manager;
}


8676 8677
/* Locate an appropriate 'qemu-img' binary.  */
const char *
8678
qemuFindQemuImgBinary(virQEMUDriverPtr driver)
8679
{
8680 8681
    if (!driver->qemuImgBinary)
        virReportError(VIR_ERR_INTERNAL_ERROR,
8682
                       "%s", _("unable to find qemu-img"));
8683 8684 8685 8686 8687 8688

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
8689
                                virDomainMomentObjPtr snapshot,
8690
                                virCapsPtr caps,
8691
                                virDomainXMLOptionPtr xmlopt,
8692
                                const char *snapshotDir)
8693 8694 8695 8696 8697 8698
{
    char *newxml = NULL;
    int ret = -1;
    char *snapDir = NULL;
    char *snapFile = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
8699 8700
    unsigned int flags = VIR_DOMAIN_SNAPSHOT_FORMAT_SECURE |
        VIR_DOMAIN_SNAPSHOT_FORMAT_INTERNAL;
8701
    virDomainSnapshotDefPtr def = virDomainSnapshotObjGetDef(snapshot);
8702

8703
    if (virDomainSnapshotGetCurrent(vm->snapshots) == snapshot)
8704
        flags |= VIR_DOMAIN_SNAPSHOT_FORMAT_CURRENT;
8705
    virUUIDFormat(vm->def->uuid, uuidstr);
8706
    newxml = virDomainSnapshotDefFormat(uuidstr, def, caps, xmlopt, flags);
8707
    if (newxml == NULL)
8708 8709
        return -1;

8710
    if (virAsprintf(&snapDir, "%s/%s", snapshotDir, vm->def->name) < 0)
8711 8712 8713 8714 8715 8716 8717
        goto cleanup;
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
        goto cleanup;
    }

8718
    if (virAsprintf(&snapFile, "%s/%s.xml", snapDir, def->parent.name) < 0)
8719 8720
        goto cleanup;

J
Ján Tomko 已提交
8721
    ret = virXMLSaveFile(snapFile, NULL, "snapshot-edit", newxml);
8722

8723
 cleanup:
8724 8725 8726 8727 8728 8729 8730 8731
    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.  */
8732
static int
8733
qemuDomainSnapshotForEachQcow2Raw(virQEMUDriverPtr driver,
8734 8735 8736 8737 8738
                                  virDomainDefPtr def,
                                  const char *name,
                                  const char *op,
                                  bool try_all,
                                  int ndisks)
8739 8740
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
8741
    size_t i;
8742 8743 8744 8745 8746 8747 8748 8749 8750
    bool skipped = false;

    qemuimgarg[0] = qemuFindQemuImgBinary(driver);
    if (qemuimgarg[0] == NULL) {
        /* qemuFindQemuImgBinary set the error */
        return -1;
    }

    qemuimgarg[2] = op;
8751
    qemuimgarg[3] = name;
8752

8753
    for (i = 0; i < ndisks; i++) {
8754
        /* FIXME: we also need to handle LVM here */
8755
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
8756 8757 8758
            int format = virDomainDiskGetFormat(def->disks[i]);

            if (format > 0 && format != VIR_STORAGE_FILE_QCOW2) {
8759 8760 8761 8762 8763
                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",
8764
                             def->disks[i]->dst);
8765 8766
                    skipped = true;
                    continue;
8767 8768 8769 8770 8771
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
8772
                }
8773 8774 8775 8776
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("Disk device '%s' does not support"
                                 " snapshotting"),
                               def->disks[i]->dst);
8777 8778 8779
                return -1;
            }

8780
            qemuimgarg[4] = virDomainDiskGetSource(def->disks[i]);
8781 8782 8783 8784

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
8785
                             def->disks[i]->dst);
8786 8787
                    skipped = true;
                    continue;
8788 8789 8790 8791 8792
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
8793 8794 8795 8796 8797 8798 8799 8800 8801
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

8802 8803 8804
/* 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
8805
qemuDomainSnapshotForEachQcow2(virQEMUDriverPtr driver,
8806
                               virDomainObjPtr vm,
8807
                               virDomainMomentObjPtr snap,
8808 8809 8810 8811 8812 8813
                               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.  */
8814
    virDomainDefPtr def = snap->def->dom;
8815 8816 8817

    if (!def)
        def = vm->def;
8818
    return qemuDomainSnapshotForEachQcow2Raw(driver, def, snap->def->name,
8819 8820 8821
                                             op, try_all, def->ndisks);
}

8822 8823
/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
8824
qemuDomainSnapshotDiscard(virQEMUDriverPtr driver,
8825
                          virDomainObjPtr vm,
8826
                          virDomainMomentObjPtr snap,
8827
                          bool update_parent,
8828 8829 8830 8831 8832
                          bool metadata_only)
{
    char *snapFile = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;
8833
    virDomainMomentObjPtr parentsnap = NULL;
8834
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8835 8836 8837 8838 8839 8840 8841 8842 8843

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
                goto cleanup;
        } else {
            priv = vm->privateData;
8844
            qemuDomainObjEnterMonitor(driver, vm);
8845
            /* we continue on even in the face of error */
8846
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
8847
            ignore_value(qemuDomainObjExitMonitor(driver, vm));
8848 8849 8850
        }
    }

8851
    if (virAsprintf(&snapFile, "%s/%s/%s.xml", cfg->snapshotDir,
8852
                    vm->def->name, snap->def->name) < 0)
8853 8854
        goto cleanup;

8855 8856
    if (snap == virDomainSnapshotGetCurrent(vm->snapshots)) {
        virDomainSnapshotSetCurrent(vm->snapshots, NULL);
8857
        if (update_parent && snap->def->parent_name) {
8858
            parentsnap = virDomainSnapshotFindByName(vm->snapshots,
8859
                                                     snap->def->parent_name);
8860 8861
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
8862
                         snap->def->parent_name);
8863
            } else {
8864
                virDomainSnapshotSetCurrent(vm->snapshots, parentsnap);
8865
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap, driver->caps,
8866
                                                    driver->xmlopt,
8867
                                                    cfg->snapshotDir) < 0) {
8868
                    VIR_WARN("failed to set parent snapshot '%s' as current",
8869
                             snap->def->parent_name);
8870
                    virDomainSnapshotSetCurrent(vm->snapshots, NULL);
8871 8872 8873 8874 8875 8876 8877
                }
            }
        }
    }

    if (unlink(snapFile) < 0)
        VIR_WARN("Failed to unlink %s", snapFile);
8878
    if (update_parent)
8879
        virDomainMomentDropParent(snap);
8880
    virDomainSnapshotObjListRemove(vm->snapshots, snap);
8881 8882 8883

    ret = 0;

8884
 cleanup:
8885
    VIR_FREE(snapFile);
8886
    virObjectUnref(cfg);
8887 8888 8889 8890
    return ret;
}

/* Hash iterator callback to discard multiple snapshots.  */
8891 8892 8893
int qemuDomainMomentDiscardAll(void *payload,
                               const void *name ATTRIBUTE_UNUSED,
                               void *data)
8894
{
8895 8896
    virDomainMomentObjPtr moment = payload;
    virQEMUMomentRemovePtr curr = data;
8897 8898
    int err;

8899 8900 8901 8902
    if (!curr->found && curr->current == moment)
        curr->found = true;
    err = curr->momentDiscard(curr->driver, curr->vm, moment, false,
                              curr->metadata_only);
8903 8904
    if (err && !curr->err)
        curr->err = err;
8905
    return 0;
8906 8907 8908
}

int
8909
qemuDomainSnapshotDiscardAllMetadata(virQEMUDriverPtr driver,
8910 8911
                                     virDomainObjPtr vm)
{
8912 8913 8914
    virQEMUMomentRemove rem = {
        .driver = driver,
        .vm = vm,
8915 8916
        .metadata_only = true,
        .momentDiscard = qemuDomainSnapshotDiscard,
8917
    };
8918

8919
    virDomainSnapshotForEach(vm->snapshots, qemuDomainMomentDiscardAll, &rem);
8920
    virDomainSnapshotObjListRemoveAll(vm->snapshots);
8921 8922 8923 8924

    return rem.err;
}

8925

8926 8927 8928
static void
qemuDomainRemoveInactiveCommon(virQEMUDriverPtr driver,
                               virDomainObjPtr vm)
8929
{
8930
    virQEMUDriverConfigPtr cfg;
8931
    VIR_AUTOFREE(char *) snapDir = NULL;
8932 8933

    cfg = virQEMUDriverGetConfig(driver);
8934

8935 8936 8937 8938
    /* 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);
8939 8940
    } else if (virAsprintf(&snapDir, "%s/%s", cfg->snapshotDir,
                           vm->def->name) < 0) {
8941
        VIR_WARN("unable to remove snapshot directory %s/%s",
8942
                 cfg->snapshotDir, vm->def->name);
8943 8944
    } else if (rmdir(snapDir) < 0 && errno != ENOENT) {
        VIR_WARN("unable to remove snapshot directory %s", snapDir);
8945
    }
8946
    qemuExtDevicesCleanupHost(driver, vm->def);
8947

8948
    virObjectUnref(cfg);
8949 8950 8951
}


8952 8953 8954 8955 8956 8957 8958 8959 8960 8961 8962 8963 8964 8965 8966 8967 8968 8969 8970 8971
/**
 * 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);
}


8972 8973 8974 8975 8976 8977 8978 8979 8980 8981 8982 8983 8984 8985 8986 8987 8988 8989 8990 8991 8992 8993
/**
 * 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);
}


8994 8995 8996 8997 8998 8999 9000 9001 9002 9003 9004 9005 9006 9007 9008 9009 9010
/**
 * 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);
9011 9012

    if (haveJob)
9013
        qemuDomainObjEndJob(driver, vm);
9014
}
9015

9016

9017 9018 9019 9020 9021 9022 9023 9024 9025 9026 9027 9028 9029 9030 9031 9032 9033 9034 9035 9036 9037
/**
 * 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);
}


9038
void
9039
qemuDomainSetFakeReboot(virQEMUDriverPtr driver,
9040 9041 9042 9043
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
9044
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
9045 9046

    if (priv->fakeReboot == value)
9047
        goto cleanup;
9048 9049 9050

    priv->fakeReboot = value;

9051
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
9052
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
9053

9054
 cleanup:
9055
    virObjectUnref(cfg);
9056
}
M
Michal Privoznik 已提交
9057

9058
static void
9059 9060
qemuDomainCheckRemoveOptionalDisk(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
9061
                                  size_t diskIndex)
9062 9063
{
    char uuid[VIR_UUID_STRING_BUFLEN];
9064
    virObjectEventPtr event = NULL;
9065
    virDomainDiskDefPtr disk = vm->def->disks[diskIndex];
9066
    const char *src = virDomainDiskGetSource(disk);
9067 9068 9069 9070 9071

    virUUIDFormat(vm->def->uuid, uuid);

    VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
              "due to inaccessible source '%s'",
9072
              disk->dst, vm->def->name, uuid, src);
9073 9074 9075 9076

    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM ||
        disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY) {

9077
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
9078 9079
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
9080
        virDomainDiskEmptySource(disk);
9081 9082
        /* keeping the old startup policy would be invalid for new images */
        disk->startupPolicy = VIR_DOMAIN_STARTUP_POLICY_DEFAULT;
9083
    } else {
9084
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
9085 9086
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_DROP_MISSING_ON_START);
9087 9088
        virDomainDiskRemove(vm->def, diskIndex);
        virDomainDiskDefFree(disk);
9089 9090
    }

9091
    virObjectEventStateQueue(driver->domainEventState, event);
9092 9093
}

9094 9095 9096 9097 9098 9099 9100 9101 9102 9103 9104 9105 9106 9107 9108 9109 9110

/**
 * 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
9111 9112
qemuDomainCheckDiskStartupPolicy(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
9113
                                 size_t diskIndex,
9114 9115
                                 bool cold_boot)
{
9116
    int startupPolicy = vm->def->disks[diskIndex]->startupPolicy;
9117
    int device = vm->def->disks[diskIndex]->device;
9118

9119
    switch ((virDomainStartupPolicy) startupPolicy) {
9120
        case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
9121 9122 9123 9124 9125 9126
            /* 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)
9127
                return -1;
9128 9129
            break;

9130
        case VIR_DOMAIN_STARTUP_POLICY_DEFAULT:
9131
        case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
9132
            return -1;
9133 9134

        case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
9135
            if (cold_boot)
9136
                return -1;
9137 9138 9139 9140 9141 9142 9143
            break;

        case VIR_DOMAIN_STARTUP_POLICY_LAST:
            /* this should never happen */
            break;
    }

9144
    qemuDomainCheckRemoveOptionalDisk(driver, vm, diskIndex);
9145
    virResetLastError();
9146 9147 9148
    return 0;
}

9149

9150 9151 9152 9153 9154 9155 9156 9157 9158 9159

/*
 * The vm must be locked when any of the following cleanup functions is
 * called.
 */
int
qemuDomainCleanupAdd(virDomainObjPtr vm,
                     qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
9160
    size_t i;
9161 9162 9163 9164 9165 9166 9167 9168 9169 9170

    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,
9171
                     priv->ncleanupCallbacks, 1) < 0)
9172 9173 9174 9175 9176 9177 9178 9179 9180 9181 9182
        return -1;

    priv->cleanupCallbacks[priv->ncleanupCallbacks++] = cb;
    return 0;
}

void
qemuDomainCleanupRemove(virDomainObjPtr vm,
                        qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
9183
    size_t i;
9184 9185 9186 9187

    VIR_DEBUG("vm=%s, cb=%p", vm->def->name, cb);

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
9188 9189 9190
        if (priv->cleanupCallbacks[i] == cb)
            VIR_DELETE_ELEMENT_INPLACE(priv->cleanupCallbacks,
                                       i, priv->ncleanupCallbacks);
9191 9192 9193 9194 9195 9196 9197 9198
    }

    VIR_SHRINK_N(priv->cleanupCallbacks,
                 priv->ncleanupCallbacks_max,
                 priv->ncleanupCallbacks_max - priv->ncleanupCallbacks);
}

void
9199
qemuDomainCleanupRun(virQEMUDriverPtr driver,
9200 9201 9202
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
9203
    size_t i;
9204 9205 9206 9207

    VIR_DEBUG("driver=%p, vm=%s", driver, vm->def->name);

    /* run cleanup callbacks in reverse order */
9208 9209
    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[priv->ncleanupCallbacks - (i + 1)])
9210 9211 9212 9213 9214 9215 9216
            priv->cleanupCallbacks[i](driver, vm);
    }

    VIR_FREE(priv->cleanupCallbacks);
    priv->ncleanupCallbacks = 0;
    priv->ncleanupCallbacks_max = 0;
}
9217

9218 9219 9220
static void
qemuDomainGetImageIds(virQEMUDriverConfigPtr cfg,
                      virDomainObjPtr vm,
9221
                      virStorageSourcePtr src,
9222
                      virStorageSourcePtr parentSrc,
9223 9224 9225 9226 9227 9228 9229 9230 9231 9232 9233 9234 9235 9236 9237 9238 9239 9240
                      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;
    }

9241 9242
    if (vm && (vmlabel = virDomainDefGetSecurityLabelDef(vm->def, "dac")) &&
        vmlabel->label)
9243 9244
        virParseOwnershipIds(vmlabel->label, uid, gid);

9245 9246 9247 9248 9249
    if (parentSrc &&
        (disklabel = virStorageSourceGetSecurityLabelDef(parentSrc, "dac")) &&
        disklabel->label)
        virParseOwnershipIds(disklabel->label, uid, gid);

9250
    if ((disklabel = virStorageSourceGetSecurityLabelDef(src, "dac")) &&
9251
        disklabel->label)
9252 9253 9254 9255
        virParseOwnershipIds(disklabel->label, uid, gid);
}


9256 9257 9258
int
qemuDomainStorageFileInit(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
9259 9260
                          virStorageSourcePtr src,
                          virStorageSourcePtr parent)
9261 9262 9263 9264 9265 9266
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    uid_t uid;
    gid_t gid;
    int ret = -1;

9267
    qemuDomainGetImageIds(cfg, vm, src, parent, &uid, &gid);
9268 9269 9270 9271 9272 9273 9274 9275 9276 9277 9278 9279

    if (virStorageFileInitAs(src, uid, gid) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
}


9280 9281 9282 9283 9284
char *
qemuDomainStorageAlias(const char *device, int depth)
{
    char *alias;

9285
    device = qemuAliasDiskDriveSkipPrefix(device);
9286 9287 9288 9289 9290 9291 9292 9293 9294

    if (!depth)
        ignore_value(VIR_STRDUP(alias, device));
    else
        ignore_value(virAsprintf(&alias, "%s.%d", device, depth));
    return alias;
}


9295 9296 9297 9298 9299 9300 9301 9302 9303 9304 9305 9306 9307
/**
 * 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.
 */
9308
int
9309
qemuDomainDetermineDiskChain(virQEMUDriverPtr driver,
9310
                             virDomainObjPtr vm,
9311
                             virDomainDiskDefPtr disk,
9312
                             virStorageSourcePtr disksrc,
9313
                             bool report_broken)
9314
{
9315
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
9316 9317
    virStorageSourcePtr src; /* iterator for the backing chain declared in XML */
    virStorageSourcePtr n; /* iterator for the backing chain detected from disk */
9318
    qemuDomainObjPrivatePtr priv = vm->privateData;
9319
    int ret = -1;
9320 9321
    uid_t uid;
    gid_t gid;
9322

9323 9324 9325
    if (!disksrc)
        disksrc = disk->src;

9326
    if (virStorageSourceIsEmpty(disksrc)) {
9327
        ret = 0;
9328
        goto cleanup;
9329
    }
9330

9331 9332
    /* There is no need to check the backing chain for disks without backing
     * support */
9333 9334 9335
    if (virStorageSourceIsLocalStorage(disksrc) &&
        disksrc->format > VIR_STORAGE_FILE_NONE &&
        disksrc->format < VIR_STORAGE_FILE_BACKING) {
9336

9337
        if (!virFileExists(disksrc->path)) {
9338
            if (report_broken)
9339
                virStorageFileReportBrokenChain(errno, disksrc, disksrc);
9340 9341 9342 9343

            goto cleanup;
        }

9344
        /* terminate the chain for such images as the code below would do */
9345
        if (!disksrc->backingStore &&
9346
            !(disksrc->backingStore = virStorageSourceNew()))
9347 9348
            goto cleanup;

9349 9350 9351
        /* 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 &&
9352 9353 9354 9355
            disksrc->format == VIR_STORAGE_FILE_RAW &&
            virStorageSourceIsBlockLocal(disksrc) &&
            virFileIsCDROM(disksrc->path) == 1)
            disksrc->hostcdrom = true;
9356

9357 9358 9359 9360
        ret = 0;
        goto cleanup;
    }

9361
    src = disksrc;
9362 9363
    /* skip to the end of the chain if there is any */
    while (virStorageSourceHasBacking(src)) {
9364 9365
        if (report_broken) {
            int rv = virStorageFileSupportsAccess(src);
9366

9367
            if (rv < 0)
9368
                goto cleanup;
9369

9370
            if (rv > 0) {
9371
                if (qemuDomainStorageFileInit(driver, vm, src, disksrc) < 0)
9372 9373 9374
                    goto cleanup;

                if (virStorageFileAccess(src, F_OK) < 0) {
9375
                    virStorageFileReportBrokenChain(errno, src, disksrc);
9376 9377 9378 9379
                    virStorageFileDeinit(src);
                    goto cleanup;
                }

9380
                virStorageFileDeinit(src);
9381 9382
            }
        }
9383
        src = src->backingStore;
9384 9385 9386 9387 9388 9389 9390
    }

    /* 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;
9391
    }
9392

9393
    qemuDomainGetImageIds(cfg, vm, src, disksrc, &uid, &gid);
9394

9395
    if (virStorageFileGetMetadata(src, uid, gid, report_broken) < 0)
9396 9397
        goto cleanup;

9398
    for (n = src->backingStore; virStorageSourceIsBacking(n); n = n->backingStore) {
9399 9400 9401
        if (qemuDomainValidateStorageSource(n, priv->qemuCaps) < 0)
            goto cleanup;

9402 9403
        if (qemuDomainPrepareDiskSourceData(disk, n, cfg, priv->qemuCaps) < 0)
            goto cleanup;
9404 9405 9406 9407

        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV) &&
            qemuDomainPrepareStorageSourceBlockdev(disk, n, priv, cfg) < 0)
            goto cleanup;
9408
    }
9409

9410
    ret = 0;
9411

9412
 cleanup:
9413 9414
    virObjectUnref(cfg);
    return ret;
9415
}
9416

9417

9418 9419 9420 9421 9422 9423 9424 9425 9426 9427 9428 9429 9430 9431 9432 9433
/**
 * 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,
9434
                              virQEMUCapsPtr qemuCaps,
9435 9436
                              char **backendAlias)
{
9437 9438
    qemuDomainDiskPrivatePtr priv = QEMU_DOMAIN_DISK_PRIVATE(disk);
    const char *nodename = NULL;
9439 9440
    *backendAlias = NULL;

9441 9442 9443 9444 9445 9446 9447 9448 9449 9450 9451 9452 9453 9454 9455 9456
    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)
9457 9458 9459 9460 9461 9462
        return -1;

    return 0;
}


9463 9464 9465
typedef enum {
    /* revoke access to the image instead of allowing it */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE = 1 << 0,
9466 9467
    /* operate on full backing chain rather than single image */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN = 1 << 1,
9468 9469 9470 9471
    /* force permissions to read-only/read-write when allowing */
    /* currently does not properly work with QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN */
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_ONLY = 1 << 2,
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_WRITE = 1 << 3,
9472
    /* don't revoke permissions when modification has failed */
9473
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE = 1 << 4,
9474
    /* VM already has access to the source and we are just modifying it */
9475
    QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS = 1 << 5,
9476 9477 9478
} qemuDomainStorageSourceAccessFlags;


9479
/**
9480
 * qemuDomainStorageSourceAccessModify:
9481 9482 9483
 * @driver: qemu driver struct
 * @vm: domain object
 * @src: Source to prepare
9484
 * @flags: bitwise or of qemuDomainStorageSourceAccessFlags
9485 9486
 *
 * Setup the locks, cgroups and security permissions on a disk source and its
9487
 * backing chain.
9488 9489 9490
 *
 * Returns 0 on success and -1 on error. Reports libvirt error.
 */
9491
static int
9492 9493 9494
qemuDomainStorageSourceAccessModify(virQEMUDriverPtr driver,
                                    virDomainObjPtr vm,
                                    virStorageSourcePtr src,
9495
                                    qemuDomainStorageSourceAccessFlags flags)
9496 9497 9498 9499 9500
{
    VIR_AUTOUNREF(virQEMUDriverConfigPtr) cfg = virQEMUDriverGetConfig(driver);
    const char *srcstr = NULLSTR(src->path);
    int ret = -1;
    virErrorPtr orig_err = NULL;
9501
    bool chain = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
9502 9503 9504
    bool force_ro = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_ONLY;
    bool force_rw = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_WRITE;
    bool revoke = flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE;
9505
    int rc;
9506
    bool was_readonly = src->readonly;
9507 9508 9509 9510
    bool revoke_cgroup = false;
    bool revoke_label = false;
    bool revoke_namespace = false;
    bool revoke_lockspace = false;
9511

9512 9513 9514 9515
    VIR_DEBUG("src='%s' readonly=%d force_ro=%d force_rw=%d revoke=%d chain=%d",
              NULLSTR(src->path), src->readonly, force_ro, force_rw, revoke, chain);

    if (force_ro)
9516
        src->readonly = true;
9517

9518
    if (force_rw)
9519 9520
        src->readonly = false;

9521
    /* just tear down the disk access */
9522
    if (revoke) {
9523
        virErrorPreserveLast(&orig_err);
9524 9525 9526 9527
        revoke_cgroup = true;
        revoke_label = true;
        revoke_namespace = true;
        revoke_lockspace = true;
9528
        ret = 0;
9529
        goto revoke;
9530 9531 9532
    }

    if (virDomainLockImageAttach(driver->lockManager, cfg->uri, vm, src) < 0)
9533 9534 9535
        goto revoke;

    revoke_lockspace = true;
9536

9537 9538 9539 9540
    /* When modifying access of existing @src namespace does not need update */
    if (!(flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS)) {
        if (qemuDomainNamespaceSetupDisk(vm, src) < 0)
            goto revoke;
9541

9542 9543
        revoke_namespace = true;
    }
9544

9545
    if (qemuSecuritySetImageLabel(driver, vm, src, chain) < 0)
9546 9547 9548
        goto revoke;

    revoke_label = true;
9549

9550 9551 9552 9553 9554 9555
    if (chain)
        rc = qemuSetupImageChainCgroup(vm, src);
    else
        rc = qemuSetupImageCgroup(vm, src);

    if (rc < 0)
9556 9557 9558
        goto revoke;

    revoke_cgroup = true;
9559 9560 9561 9562

    ret = 0;
    goto cleanup;

9563
 revoke:
9564 9565 9566
    if (flags & QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE)
        goto cleanup;

9567 9568 9569 9570 9571
    if (revoke_cgroup) {
        if (chain)
            rc = qemuTeardownImageChainCgroup(vm, src);
        else
            rc = qemuTeardownImageCgroup(vm, src);
9572

9573 9574 9575 9576 9577 9578 9579 9580 9581 9582 9583 9584 9585 9586 9587 9588 9589 9590
        if (rc < 0)
            VIR_WARN("Unable to tear down cgroup access on %s", srcstr);
    }

    if (revoke_label) {
        if (qemuSecurityRestoreImageLabel(driver, vm, src, chain) < 0)
            VIR_WARN("Unable to restore security label on %s", srcstr);
    }

    if (revoke_namespace) {
        if (qemuDomainNamespaceTeardownDisk(vm, src) < 0)
            VIR_WARN("Unable to remove /dev entry for %s", srcstr);
    }

    if (revoke_lockspace) {
        if (virDomainLockImageDetach(driver->lockManager, vm, src) < 0)
            VIR_WARN("Unable to release lock on %s", srcstr);
    }
9591 9592

 cleanup:
9593
    src->readonly = was_readonly;
9594 9595 9596 9597 9598 9599
    virErrorRestore(&orig_err);

    return ret;
}


9600 9601 9602 9603 9604
int
qemuDomainStorageSourceChainAccessAllow(virQEMUDriverPtr driver,
                                        virDomainObjPtr vm,
                                        virStorageSourcePtr src)
{
9605
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
9606 9607

    return qemuDomainStorageSourceAccessModify(driver, vm, src, flags);
9608 9609 9610 9611 9612 9613 9614 9615
}


int
qemuDomainStorageSourceChainAccessRevoke(virQEMUDriverPtr driver,
                                         virDomainObjPtr vm,
                                         virStorageSourcePtr src)
{
9616 9617
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE |
                                               QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_CHAIN;
9618 9619

    return qemuDomainStorageSourceAccessModify(driver, vm, src, flags);
9620 9621 9622
}


9623
/**
9624
 * qemuDomainStorageSourceAccessRevoke:
9625 9626 9627 9628 9629
 *
 * Revoke access to a single backing chain element. This restores the labels,
 * removes cgroup ACLs for devices and removes locks.
 */
void
9630 9631 9632
qemuDomainStorageSourceAccessRevoke(virQEMUDriverPtr driver,
                                    virDomainObjPtr vm,
                                    virStorageSourcePtr elem)
9633
{
9634
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_REVOKE;
9635

9636
    ignore_value(qemuDomainStorageSourceAccessModify(driver, vm, elem, flags));
9637 9638 9639 9640
}


/**
9641
 * qemuDomainStorageSourceAccessAllow:
9642 9643 9644 9645 9646
 * @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
9647 9648 9649
 *
 * 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
9650 9651 9652 9653 9654
 * 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.
 */
9655
int
9656 9657 9658 9659 9660
qemuDomainStorageSourceAccessAllow(virQEMUDriverPtr driver,
                                   virDomainObjPtr vm,
                                   virStorageSourcePtr elem,
                                   bool readonly,
                                   bool newSource)
9661
{
9662
    qemuDomainStorageSourceAccessFlags flags = QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_SKIP_REVOKE;
9663

9664
    if (readonly)
9665 9666 9667
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_ONLY;
    else
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_FORCE_READ_WRITE;
9668

9669
    if (!newSource)
9670
        flags |= QEMU_DOMAIN_STORAGE_SOURCE_ACCESS_MODIFY_ACCESS;
9671

9672
    return qemuDomainStorageSourceAccessModify(driver, vm, elem, flags);
9673 9674 9675
}


9676 9677 9678 9679 9680 9681 9682 9683 9684 9685
/*
 * 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)
{
9686 9687 9688 9689 9690 9691 9692 9693 9694 9695
#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; \
        } \
9696 9697
    } while (0)

9698 9699 9700 9701 9702 9703 9704 9705 9706 9707 9708 9709
#define CHECK_STREQ_NULLABLE(field, field_name) \
    do { \
        if (!disk->field) \
            break; \
        if (STRNEQ_NULLABLE(disk->field, orig_disk->field)) { \
            virReportError(VIR_ERR_OPERATION_UNSUPPORTED, \
                           _("cannot modify field '%s' of the disk"), \
                           field_name); \
            return false; \
        } \
    } while (0)

9710
    CHECK_EQ(device, "device", false);
9711 9712 9713 9714
    CHECK_EQ(bus, "bus", false);
    if (STRNEQ(disk->dst, orig_disk->dst)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
9715
                       "target");
9716 9717 9718 9719
        return false;
    }
    CHECK_EQ(tray_status, "tray", true);
    CHECK_EQ(removable, "removable", true);
9720 9721 9722 9723 9724 9725 9726 9727 9728 9729 9730 9731 9732 9733 9734 9735 9736 9737 9738 9739 9740 9741 9742 9743 9744 9745 9746 9747 9748 9749 9750 9751 9752 9753 9754 9755 9756 9757 9758 9759 9760 9761 9762 9763 9764 9765 9766 9767 9768 9769 9770 9771 9772 9773

    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);
9774 9775
    CHECK_STREQ_NULLABLE(blkdeviotune.group_name,
                         "blkdeviotune group name");
9776

9777 9778 9779 9780 9781 9782 9783 9784
    CHECK_STREQ_NULLABLE(serial,
                         "serial");
    CHECK_STREQ_NULLABLE(wwn,
                         "wwn");
    CHECK_STREQ_NULLABLE(vendor,
                         "vendor");
    CHECK_STREQ_NULLABLE(product,
                         "product");
9785

9786 9787 9788 9789 9790 9791 9792
    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);
9793
    /* "snapshot" is a libvirt internal field and thus can be changed */
9794
    /* startupPolicy is allowed to be updated. Therefore not checked here. */
9795
    CHECK_EQ(transient, "transient", true);
9796 9797 9798 9799 9800 9801 9802 9803 9804 9805 9806 9807 9808

    /* 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;
    }

9809
    /* device alias is checked already in virDomainDefCompatibleDevice */
9810

9811
    CHECK_EQ(info.bootIndex, "boot order", true);
9812 9813 9814 9815 9816
    CHECK_EQ(rawio, "rawio", true);
    CHECK_EQ(sgio, "sgio", true);
    CHECK_EQ(discard, "discard", true);
    CHECK_EQ(iothread, "iothread", true);

9817 9818
    CHECK_STREQ_NULLABLE(domain_name,
                         "backenddomain");
9819

9820 9821 9822 9823 9824 9825
    /* 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);

9826 9827 9828 9829 9830 9831 9832 9833
    if (!virStoragePRDefIsEqual(disk->src->pr,
                                orig_disk->src->pr)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "reservations");
        return false;
    }

9834
#undef CHECK_EQ
9835
#undef CHECK_STREQ_NULLABLE
9836 9837 9838 9839

    return true;
}

9840 9841 9842
bool
qemuDomainDiskBlockJobIsActive(virDomainDiskDefPtr disk)
{
9843 9844
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

9845 9846 9847 9848 9849 9850 9851 9852
    if (disk->mirror) {
        virReportError(VIR_ERR_BLOCK_COPY_ACTIVE,
                       _("disk '%s' already in active block job"),
                       disk->dst);

        return true;
    }

9853
    if (diskPriv->blockjob &&
9854
        qemuBlockJobIsRunning(diskPriv->blockjob)) {
9855 9856 9857 9858 9859 9860 9861 9862 9863 9864
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("disk '%s' already in active block job"),
                       disk->dst);
        return true;
    }

    return false;
}


9865 9866 9867 9868 9869 9870 9871 9872 9873 9874 9875 9876 9877 9878 9879
/**
 * 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++) {
9880 9881 9882
        virDomainDiskDefPtr disk = vm->def->disks[i];
        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

9883 9884
        if (!copy_only && diskPriv->blockjob &&
            qemuBlockJobIsRunning(diskPriv->blockjob))
9885 9886
            return true;

9887
        if (disk->mirror && disk->mirrorJob == VIR_DOMAIN_BLOCK_JOB_TYPE_COPY)
9888 9889 9890 9891 9892 9893 9894
            return true;
    }

    return false;
}


9895 9896
int
qemuDomainUpdateDeviceList(virQEMUDriverPtr driver,
9897 9898
                           virDomainObjPtr vm,
                           int asyncJob)
9899 9900 9901
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char **aliases;
9902
    int rc;
9903

9904 9905
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
9906 9907 9908 9909
    rc = qemuMonitorGetDeviceAliases(priv->mon, &aliases);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
    if (rc < 0)
9910 9911
        return -1;

9912
    virStringListFree(priv->qemuDevices);
9913 9914 9915
    priv->qemuDevices = aliases;
    return 0;
}
9916

9917 9918 9919 9920 9921 9922 9923 9924 9925 9926 9927 9928 9929 9930 9931 9932 9933 9934 9935

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);

9936 9937
    if (qemuDomainObjExitMonitor(driver, vm) < 0) {
        virHashFree(meminfo);
9938
        return -1;
9939
    }
9940 9941 9942 9943 9944 9945 9946 9947 9948 9949 9950 9951 9952 9953 9954 9955 9956 9957 9958 9959 9960 9961 9962 9963 9964 9965 9966 9967

    /* 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;
}


9968 9969 9970 9971
static bool
qemuDomainABIStabilityCheck(const virDomainDef *src,
                            const virDomainDef *dst)
{
9972 9973
    size_t i;

9974 9975 9976 9977 9978 9979 9980 9981 9982
    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;
    }

9983 9984 9985 9986 9987 9988 9989 9990 9991 9992 9993 9994 9995
    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;
        }
    }

9996 9997 9998 9999 10000 10001 10002 10003 10004
    return true;
}


virDomainABIStability virQEMUDriverDomainABIStability = {
    .domain = qemuDomainABIStabilityCheck,
};


10005 10006 10007 10008 10009 10010 10011 10012 10013 10014 10015 10016 10017 10018 10019 10020 10021 10022 10023 10024
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;
}


10025 10026 10027
#define COPY_FLAGS (VIR_DOMAIN_XML_SECURE | \
                    VIR_DOMAIN_XML_MIGRATABLE)

10028 10029 10030 10031 10032 10033 10034 10035 10036
bool
qemuDomainDefCheckABIStability(virQEMUDriverPtr driver,
                               virDomainDefPtr src,
                               virDomainDefPtr dst)
{
    virDomainDefPtr migratableDefSrc = NULL;
    virDomainDefPtr migratableDefDst = NULL;
    bool ret = false;

10037 10038
    if (!(migratableDefSrc = qemuDomainDefCopy(driver, src, COPY_FLAGS)) ||
        !(migratableDefDst = qemuDomainDefCopy(driver, dst, COPY_FLAGS)))
10039 10040
        goto cleanup;

10041 10042 10043
    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   src, migratableDefSrc,
                                                   dst, migratableDefDst);
10044

10045
 cleanup:
10046 10047 10048 10049
    virDomainDefFree(migratableDefSrc);
    virDomainDefFree(migratableDefDst);
    return ret;
}
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

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


10081
bool
10082
qemuDomainAgentAvailable(virDomainObjPtr vm,
10083 10084
                         bool reportError)
{
10085 10086
    qemuDomainObjPrivatePtr priv = vm->privateData;

10087 10088 10089 10090 10091 10092 10093
    if (virDomainObjGetState(vm, NULL) != VIR_DOMAIN_RUNNING) {
        if (reportError) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
        }
        return false;
    }
10094 10095 10096 10097 10098 10099 10100 10101 10102
    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) {
10103 10104 10105 10106 10107 10108 10109 10110 10111 10112 10113 10114
        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;
10115 10116 10117 10118
        }
    }
    return true;
}
10119

10120

10121
static unsigned long long
10122
qemuDomainGetMemorySizeAlignment(virDomainDefPtr def)
10123
{
10124 10125 10126 10127 10128
    /* 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;

10129 10130 10131 10132 10133 10134 10135
    /* Align memory size. QEMU requires rounding to next 4KiB block.
     * We'll take the "traditional" path and round it to 1MiB*/

    return 1024;
}


10136 10137 10138 10139 10140 10141 10142 10143 10144 10145 10146 10147 10148 10149 10150
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;
}


10151 10152 10153
int
qemuDomainAlignMemorySizes(virDomainDefPtr def)
{
10154 10155
    unsigned long long maxmemkb = virMemoryMaxValue(false) >> 10;
    unsigned long long maxmemcapped = virMemoryMaxValue(true) >> 10;
10156
    unsigned long long initialmem = 0;
10157
    unsigned long long hotplugmem = 0;
10158
    unsigned long long mem;
10159
    unsigned long long align = qemuDomainGetMemorySizeAlignment(def);
10160 10161 10162 10163 10164
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    /* align NUMA cell sizes if relevant */
    for (i = 0; i < ncells; i++) {
10165 10166
        mem = VIR_ROUND_UP(virDomainNumaGetNodeMemorySize(def->numa, i), align);
        initialmem += mem;
10167 10168 10169 10170 10171 10172 10173

        if (mem > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("memory size of NUMA node '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
10174
        virDomainNumaSetNodeMemorySize(def->numa, i, mem);
10175 10176
    }

10177 10178 10179 10180 10181
    /* 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);

10182 10183 10184 10185 10186 10187
    if (initialmem > maxmemcapped) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("initial memory size overflowed after alignment"));
        return -1;
    }

10188
    def->mem.max_memory = VIR_ROUND_UP(def->mem.max_memory, align);
10189 10190 10191 10192 10193
    if (def->mem.max_memory > maxmemkb) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("maximum memory size overflowed after alignment"));
        return -1;
    }
10194

10195
    /* Align memory module sizes */
10196 10197
    for (i = 0; i < def->nmems; i++) {
        align = qemuDomainGetMemoryModuleSizeAlignment(def, def->mems[i]);
10198
        def->mems[i]->size = VIR_ROUND_UP(def->mems[i]->size, align);
10199
        hotplugmem += def->mems[i]->size;
10200 10201 10202 10203 10204 10205 10206

        if (def->mems[i]->size > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("size of memory module '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
10207
    }
10208

10209 10210
    virDomainDefSetMemoryTotal(def, initialmem + hotplugmem);

10211 10212
    return 0;
}
10213 10214 10215 10216 10217 10218 10219 10220 10221 10222 10223


/**
 * 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
10224 10225
qemuDomainMemoryDeviceAlignSize(virDomainDefPtr def,
                                virDomainMemoryDefPtr mem)
10226
{
10227
    mem->size = VIR_ROUND_UP(mem->size, qemuDomainGetMemorySizeAlignment(def));
10228
}
10229 10230 10231 10232 10233 10234 10235 10236 10237 10238 10239 10240 10241


/**
 * qemuDomainGetMonitor:
 * @vm: domain object
 *
 * Returns the monitor pointer corresponding to the domain object @vm.
 */
qemuMonitorPtr
qemuDomainGetMonitor(virDomainObjPtr vm)
{
    return ((qemuDomainObjPrivatePtr) vm->privateData)->mon;
}
10242 10243


10244 10245 10246 10247 10248 10249 10250
/**
 * 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.
 */
10251
virDomainChrDefPtr
10252 10253 10254 10255 10256 10257 10258 10259 10260 10261
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;

10262 10263
        if (STREQ_NULLABLE(channel->target.name, "org.qemu.guest_agent.0"))
            return channel;
10264 10265
    }

10266
    return NULL;
10267
}
10268 10269


10270
static bool
10271
qemuDomainMachineIsQ35(const char *machine,
10272
                       const virArch arch)
10273
{
10274 10275 10276
    if (!ARCH_IS_X86(arch))
        return false;

10277 10278 10279 10280 10281 10282
    if (STREQ(machine, "q35") ||
        STRPREFIX(machine, "pc-q35-")) {
        return true;
    }

    return false;
10283
}
10284 10285


10286
static bool
10287
qemuDomainMachineIsI440FX(const char *machine,
10288
                          const virArch arch)
10289
{
10290 10291 10292
    if (!ARCH_IS_X86(arch))
        return false;

10293 10294 10295 10296 10297 10298 10299 10300 10301
    if (STREQ(machine, "pc") ||
        STRPREFIX(machine, "pc-0.") ||
        STRPREFIX(machine, "pc-1.") ||
        STRPREFIX(machine, "pc-i440fx-") ||
        STRPREFIX(machine, "rhel")) {
        return true;
    }

    return false;
10302 10303 10304
}


10305
static bool
10306
qemuDomainMachineIsS390CCW(const char *machine,
10307
                           const virArch arch)
10308
{
10309 10310 10311
    if (!ARCH_IS_S390(arch))
        return false;

10312 10313 10314 10315
    if (STRPREFIX(machine, "s390-ccw"))
        return true;

    return false;
A
Andrea Bolognani 已提交
10316
}
10317

A
Andrea Bolognani 已提交
10318

10319 10320
/* You should normally avoid this function and use
 * qemuDomainIsARMVirt() instead. */
A
Andrea Bolognani 已提交
10321 10322 10323 10324 10325 10326
bool
qemuDomainMachineIsARMVirt(const char *machine,
                           const virArch arch)
{
    if (arch != VIR_ARCH_ARMV6L &&
        arch != VIR_ARCH_ARMV7L &&
10327
        arch != VIR_ARCH_AARCH64) {
10328
        return false;
10329
    }
10330

10331 10332 10333 10334
    if (STREQ(machine, "virt") ||
        STRPREFIX(machine, "virt-")) {
        return true;
    }
10335

10336
    return false;
10337 10338 10339
}


10340
static bool
A
Andrea Bolognani 已提交
10341 10342
qemuDomainMachineIsRISCVVirt(const char *machine,
                             const virArch arch)
10343
{
A
Andrea Bolognani 已提交
10344 10345
    if (!ARCH_IS_RISCV(arch))
        return false;
10346

10347 10348 10349 10350
    if (STREQ(machine, "virt") ||
        STRPREFIX(machine, "virt-")) {
        return true;
    }
10351

10352
    return false;
A
Andrea Bolognani 已提交
10353 10354 10355
}


10356 10357
/* You should normally avoid this function and use
 * qemuDomainIsPSeries() instead. */
A
Andrea Bolognani 已提交
10358 10359 10360 10361 10362 10363 10364
bool
qemuDomainMachineIsPSeries(const char *machine,
                           const virArch arch)
{
    if (!ARCH_IS_PPC64(arch))
        return false;

10365 10366 10367 10368
    if (STREQ(machine, "pseries") ||
        STRPREFIX(machine, "pseries-")) {
        return true;
    }
10369

10370
    return false;
10371 10372 10373
}


10374 10375
/* You should normally avoid this function and use
 * qemuDomainHasBuiltinIDE() instead. */
10376
bool
10377 10378
qemuDomainMachineHasBuiltinIDE(const char *machine,
                               const virArch arch)
10379
{
10380
    return qemuDomainMachineIsI440FX(machine, arch) ||
A
Andrea Bolognani 已提交
10381 10382 10383
        STREQ(machine, "malta") ||
        STREQ(machine, "sun4u") ||
        STREQ(machine, "g3beige");
10384 10385 10386
}


10387
static bool
10388
qemuDomainMachineNeedsFDC(const char *machine,
10389
                          const virArch arch)
10390 10391
{
    const char *p = STRSKIP(machine, "pc-q35-");
10392

10393 10394 10395
    if (!ARCH_IS_X86(arch))
        return false;

10396 10397 10398 10399 10400 10401 10402 10403 10404
    if (!p)
        return false;

    if (STRPREFIX(p, "1.") ||
        STREQ(p, "2.0") ||
        STREQ(p, "2.1") ||
        STREQ(p, "2.2") ||
        STREQ(p, "2.3")) {
        return false;
10405
    }
10406 10407

    return true;
10408 10409 10410
}


10411
bool
A
Andrea Bolognani 已提交
10412
qemuDomainIsQ35(const virDomainDef *def)
10413
{
10414
    return qemuDomainMachineIsQ35(def->os.machine, def->os.arch);
10415 10416
}

10417

J
Ján Tomko 已提交
10418
bool
A
Andrea Bolognani 已提交
10419
qemuDomainIsI440FX(const virDomainDef *def)
J
Ján Tomko 已提交
10420
{
10421
    return qemuDomainMachineIsI440FX(def->os.machine, def->os.arch);
10422 10423 10424 10425
}


bool
A
Andrea Bolognani 已提交
10426
qemuDomainIsS390CCW(const virDomainDef *def)
10427
{
10428
    return qemuDomainMachineIsS390CCW(def->os.machine, def->os.arch);
10429 10430 10431 10432
}


bool
A
Andrea Bolognani 已提交
10433
qemuDomainIsARMVirt(const virDomainDef *def)
10434
{
A
Andrea Bolognani 已提交
10435
    return qemuDomainMachineIsARMVirt(def->os.machine, def->os.arch);
J
Ján Tomko 已提交
10436 10437 10438
}


10439 10440 10441 10442 10443 10444 10445 10446
bool
qemuDomainIsRISCVVirt(const virDomainDef *def)
{
    return qemuDomainMachineIsRISCVVirt(def->os.machine, def->os.arch);
}


bool
A
Andrea Bolognani 已提交
10447
qemuDomainIsPSeries(const virDomainDef *def)
10448
{
A
Andrea Bolognani 已提交
10449 10450 10451 10452 10453 10454 10455 10456 10457 10458
    return qemuDomainMachineIsPSeries(def->os.machine, def->os.arch);
}


bool
qemuDomainHasPCIRoot(const virDomainDef *def)
{
    int root = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

    if (root < 0)
10459 10460
        return false;

A
Andrea Bolognani 已提交
10461
    if (def->controllers[root]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)
10462 10463 10464 10465 10466 10467
        return false;

    return true;
}


10468
bool
A
Andrea Bolognani 已提交
10469
qemuDomainHasPCIeRoot(const virDomainDef *def)
10470
{
A
Andrea Bolognani 已提交
10471 10472 10473 10474 10475 10476 10477 10478 10479
    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;
10480 10481 10482 10483
}


bool
A
Andrea Bolognani 已提交
10484
qemuDomainHasBuiltinIDE(const virDomainDef *def)
10485
{
10486
    return qemuDomainMachineHasBuiltinIDE(def->os.machine, def->os.arch);
A
Andrea Bolognani 已提交
10487
}
10488 10489


A
Andrea Bolognani 已提交
10490 10491 10492
bool
qemuDomainNeedsFDC(const virDomainDef *def)
{
10493
    return qemuDomainMachineNeedsFDC(def->os.machine, def->os.arch);
10494 10495 10496
}


10497 10498 10499 10500
bool
qemuDomainSupportsPCI(virDomainDefPtr def,
                      virQEMUCapsPtr qemuCaps)
{
10501 10502 10503 10504
    if (def->os.arch != VIR_ARCH_ARMV6L &&
        def->os.arch != VIR_ARCH_ARMV7L &&
        def->os.arch != VIR_ARCH_AARCH64 &&
        !ARCH_IS_RISCV(def->os.arch)) {
10505
        return true;
10506
    }
10507 10508 10509 10510 10511 10512

    if (STREQ(def->os.machine, "versatilepb"))
        return true;

    if ((qemuDomainIsARMVirt(def) ||
         qemuDomainIsRISCVVirt(def)) &&
10513
        virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX)) {
10514
        return true;
10515
    }
10516 10517 10518 10519 10520

    return false;
}


10521 10522 10523 10524 10525 10526 10527 10528 10529 10530 10531 10532 10533 10534 10535 10536 10537 10538 10539 10540 10541 10542 10543 10544 10545 10546 10547 10548 10549 10550 10551 10552 10553 10554 10555 10556 10557 10558 10559
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
10560
    case VIR_DOMAIN_MEMORY_MODEL_NVDIMM:
10561 10562 10563 10564 10565 10566 10567 10568
        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;
        }

10569 10570 10571
        if (virDomainNumaGetNodeCount(def->numa) != 0) {
            if (mem->targetNode == -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
Y
Yuri Chornoivan 已提交
10572
                               _("target NUMA node needs to be specified for "
10573 10574 10575
                                 "memory device"));
                return -1;
            }
10576 10577
        }

10578 10579 10580 10581 10582 10583 10584 10585 10586 10587 10588 10589 10590 10591 10592 10593 10594 10595 10596 10597 10598 10599 10600 10601
        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;
}


10602 10603 10604 10605 10606 10607 10608 10609 10610 10611 10612 10613 10614 10615 10616 10617 10618 10619 10620 10621 10622
/**
 * 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
10623 10624
    bool needPCDimmCap = false;
    bool needNvdimmCap = false;
10625 10626 10627 10628 10629 10630 10631
    size_t i;

    hotplugSpace = def->mem.max_memory - virDomainDefGetMemoryInitial(def);

    if (mem) {
        nmems++;
        hotplugMemory = mem->size;
10632 10633 10634

        if (qemuDomainDefValidateMemoryHotplugDevice(mem, def) < 0)
            return -1;
10635 10636 10637 10638 10639 10640 10641 10642 10643 10644 10645 10646 10647
    }

    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;
    }

10648 10649 10650 10651 10652 10653 10654 10655 10656 10657 10658
    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;
        }
10659 10660 10661 10662 10663 10664 10665 10666 10667
    }

    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;
    }

10668
    for (i = 0; i < def->nmems; i++) {
10669 10670
        hotplugMemory += def->mems[i]->size;

M
Michal Privoznik 已提交
10671 10672 10673 10674 10675 10676 10677 10678 10679 10680 10681 10682 10683 10684
        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;
        }

10685 10686 10687 10688 10689 10690
        /* 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
10691 10692 10693 10694 10695 10696 10697 10698 10699 10700 10701 10702 10703 10704
    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;
    }

10705 10706 10707 10708 10709 10710 10711 10712 10713 10714
    if (hotplugMemory > hotplugSpace) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory device total size exceeds hotplug space"));
        return -1;
    }

    return 0;
}


10715 10716 10717
/**
 * qemuDomainUpdateCurrentMemorySize:
 *
10718 10719
 * In case when the balloon is not present for the domain, the function
 * recalculates the maximum size to reflect possible changes.
10720
 */
10721 10722
void
qemuDomainUpdateCurrentMemorySize(virDomainObjPtr vm)
10723 10724 10725
{
    /* inactive domain doesn't need size update */
    if (!virDomainObjIsActive(vm))
10726
        return;
10727 10728 10729

    /* if no balloning is available, the current size equals to the current
     * full memory size */
10730
    if (!virDomainDefHasMemballoon(vm->def))
10731
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
10732
}
10733 10734


10735 10736 10737 10738 10739 10740 10741 10742
/**
 * ppc64VFIODeviceIsNV2Bridge:
 * @device: string with the PCI device address
 *
 * This function receives a string that represents a PCI device,
 * such as '0004:04:00.0', and tells if the device is a NVLink2
 * bridge.
 */
10743
static bool
10744 10745 10746 10747 10748 10749 10750 10751 10752 10753 10754 10755 10756 10757 10758 10759 10760 10761 10762 10763 10764
ppc64VFIODeviceIsNV2Bridge(const char *device)
{
    const char *nvlink2Files[] = {"ibm,gpu", "ibm,nvlink",
                                  "ibm,nvlink-speed", "memory-region"};
    size_t i;

    for (i = 0; i < ARRAY_CARDINALITY(nvlink2Files); i++) {
        VIR_AUTOFREE(char *) file = NULL;

        if ((virAsprintf(&file, "/sys/bus/pci/devices/%s/of_node/%s",
                         device, nvlink2Files[i])) < 0)
            return false;

        if (!virFileExists(file))
            return false;
    }

    return true;
}


10765 10766 10767 10768 10769 10770 10771 10772 10773 10774 10775 10776 10777 10778 10779 10780
/**
 * getPPC64MemLockLimitBytes:
 * @def: domain definition
 *
 * A PPC64 helper that calculates the memory locking limit in order for
 * the guest to operate properly.
 */
static unsigned long long
getPPC64MemLockLimitBytes(virDomainDefPtr def)
{
    unsigned long long memKB = 0;
    unsigned long long baseLimit = 0;
    unsigned long long memory = 0;
    unsigned long long maxMemory = 0;
    unsigned long long passthroughLimit = 0;
    size_t i, nPCIHostBridges = 0;
10781
    virPCIDeviceAddressPtr pciAddr;
10782
    bool usesVFIO = false;
10783
    bool nvlink2Capable = false;
10784 10785 10786 10787 10788 10789 10790 10791 10792 10793 10794 10795 10796 10797 10798 10799 10800

    for (i = 0; i < def->ncontrollers; i++) {
        virDomainControllerDefPtr cont = def->controllers[i];

        if (!virDomainControllerIsPSeriesPHB(cont))
            continue;

        nPCIHostBridges++;
    }

    for (i = 0; i < def->nhostdevs; i++) {
        virDomainHostdevDefPtr dev = def->hostdevs[i];

        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;
10801 10802 10803 10804 10805 10806 10807 10808 10809 10810 10811

            pciAddr = &dev->source.subsys.u.pci.addr;
            if (virPCIDeviceAddressIsValid(pciAddr, false)) {
                VIR_AUTOFREE(char *) pciAddrStr = NULL;

                pciAddrStr = virPCIDeviceAddressAsString(pciAddr);
                if (ppc64VFIODeviceIsNV2Bridge(pciAddrStr)) {
                    nvlink2Capable = true;
                    break;
                }
            }
10812 10813 10814 10815 10816 10817 10818 10819 10820 10821 10822 10823 10824 10825 10826 10827 10828 10829 10830 10831 10832 10833 10834 10835 10836 10837
        }
    }

    memory = virDomainDefGetMemoryTotal(def);

    if (def->mem.max_memory)
        maxMemory = def->mem.max_memory;
    else
        maxMemory = memory;

    /* baseLimit := maxMemory / 128                                  (a)
     *              + 4 MiB * #PHBs + 8 MiB                          (b)
     *
     * (a) is the hash table
     *
     * (b) is accounting for the 32-bit DMA window - it could be either the
     * KVM accelerated TCE tables for emulated devices, or the VFIO
     * userspace view. The 4 MiB per-PHB (including the default one) covers
     * a 2GiB DMA window: default is 1GiB, but it's possible it'll be
     * increased to help performance. The 8 MiB extra should be plenty for
     * the TCE table index for any reasonable number of PHBs and several
     * spapr-vlan or spapr-vscsi devices (512kB + a tiny bit each) */
    baseLimit = maxMemory / 128 +
                4096 * nPCIHostBridges +
                8192;

10838 10839 10840 10841 10842
    /* NVLink2 support in QEMU is a special case of the passthrough
     * mechanics explained in the usesVFIO case below. The GPU RAM
     * is placed with a gap after maxMemory. The current QEMU
     * implementation puts the NVIDIA RAM above the PCI MMIO, which
     * starts at 32TiB and is the MMIO reserved for the guest main RAM.
10843
     *
10844 10845 10846
     * This window ends at 64TiB, and this is where the GPUs are being
     * placed. The next available window size is at 128TiB, and
     * 64TiB..128TiB will fit all possible NVIDIA GPUs.
10847
     *
10848 10849 10850 10851
     * The same assumption as the most common case applies here:
     * the guest will request a 64-bit DMA window, per PHB, that is
     * big enough to map all its RAM, which is now at 128TiB due
     * to the GPUs.
10852
     *
10853 10854 10855 10856 10857 10858 10859 10860 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
     * Note that the NVIDIA RAM window must be accounted for the TCE
     * table size, but *not* for the main RAM (maxMemory). This gives
     * us the following passthroughLimit for the NVLink2 case:
     *
     * passthroughLimit = maxMemory +
     *                    128TiB/512KiB * #PHBs + 8 MiB */
    if (nvlink2Capable) {
        passthroughLimit = maxMemory +
                           128 * (1ULL<<30) / 512 * nPCIHostBridges +
                           8192;
    } else if (usesVFIO) {
        /* For regular (non-NVLink2 present) VFIO passthrough, the value
         * of passthroughLimit is:
         *
         * passthroughLimit := max( 2 GiB * #PHBs,                       (c)
         *                          memory                               (d)
         *                          + memory * 1/512 * #PHBs + 8 MiB )   (e)
         *
         * (c) is the pre-DDW VFIO DMA window accounting. We're allowing 2
         * GiB rather than 1 GiB
         *
         * (d) is the with-DDW (and memory pre-registration and related
         * features) DMA window accounting - assuming that we only account
         * RAM once, even if mapped to multiple PHBs
         *
         * (e) is the with-DDW userspace view and overhead for the 64-bit
         * DMA window. This is based a bit on expected guest behaviour, but
         * there really isn't a way to completely avoid that. We assume the
         * guest requests a 64-bit DMA window (per PHB) just big enough to
         * map all its RAM. 4 kiB page size gives the 1/512; it will be
         * less with 64 kiB pages, less still if the guest is mapped with
         * hugepages (unlike the default 32-bit DMA window, DDW windows
         * can use large IOMMU pages). 8 MiB is for second and further level
         * overheads, like (b) */
10887 10888 10889
        passthroughLimit = MAX(2 * 1024 * 1024 * nPCIHostBridges,
                               memory +
                               memory / 512 * nPCIHostBridges + 8192);
10890
    }
10891 10892 10893 10894 10895 10896 10897

    memKB = baseLimit + passthroughLimit;

    return memKB << 10;
}


10898
/**
10899
 * qemuDomainGetMemLockLimitBytes:
10900 10901
 * @def: domain definition
 *
10902 10903 10904 10905 10906 10907
 * 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
10908 10909
 */
unsigned long long
10910
qemuDomainGetMemLockLimitBytes(virDomainDefPtr def)
10911
{
10912 10913
    unsigned long long memKB = 0;
    size_t i;
10914

10915 10916 10917 10918 10919 10920
    /* prefer the hard limit */
    if (virMemoryLimitIsSet(def->mem.hard_limit)) {
        memKB = def->mem.hard_limit;
        goto done;
    }

10921 10922 10923 10924 10925 10926 10927
    /* 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;
10928

10929 10930
    if (ARCH_IS_PPC64(def->os.arch) && def->virtType == VIR_DOMAIN_VIRT_KVM)
        return getPPC64MemLockLimitBytes(def);
10931

10932 10933 10934 10935 10936 10937 10938 10939 10940 10941 10942 10943 10944 10945 10946 10947 10948 10949
    /* 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.
     */
10950
    for (i = 0; i < def->nhostdevs; i++) {
10951
        virDomainHostdevSubsysPtr subsys = &def->hostdevs[i]->source.subsys;
10952

10953 10954 10955
        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 &&
10956 10957 10958 10959
              subsys->u.pci.backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO))) {
            memKB = virDomainDefGetMemoryTotal(def) + 1024 * 1024;
            goto done;
        }
10960 10961
    }

10962 10963
 done:
    return memKB << 10;
10964
}
10965

10966

10967 10968 10969 10970 10971 10972 10973
/**
 * qemuDomainAdjustMaxMemLock:
 * @vm: domain
 *
 * Adjust the memory locking limit for the QEMU process associated to @vm, in
 * order to comply with VFIO or architecture requirements.
 *
10974 10975 10976 10977
 * 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.
10978 10979 10980 10981 10982 10983 10984 10985 10986
 *
 * Returns: 0 on success, <0 on failure
 */
int
qemuDomainAdjustMaxMemLock(virDomainObjPtr vm)
{
    unsigned long long bytes = 0;
    int ret = -1;

10987 10988 10989
    bytes = qemuDomainGetMemLockLimitBytes(vm->def);

    if (bytes) {
10990 10991 10992 10993 10994 10995 10996 10997 10998 10999 11000 11001 11002 11003
        /* 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;
    }
11004 11005 11006 11007 11008 11009 11010 11011 11012 11013

    /* 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;
}
11014 11015 11016 11017 11018 11019 11020 11021 11022 11023

/**
 * qemuDomainHasVcpuPids:
 * @vm: Domain object
 *
 * Returns true if we were able to successfully detect vCPU pids for the VM.
 */
bool
qemuDomainHasVcpuPids(virDomainObjPtr vm)
{
11024 11025 11026 11027 11028 11029
    size_t i;
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    virDomainVcpuDefPtr vcpu;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
11030

11031 11032 11033 11034 11035
        if (QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid > 0)
            return true;
    }

    return false;
11036
}
11037 11038 11039 11040 11041 11042 11043 11044 11045 11046 11047


/**
 * 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,
11048
                     unsigned int vcpuid)
11049
{
11050 11051
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
    return QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid;
11052
}
11053 11054


11055 11056 11057 11058 11059 11060 11061 11062 11063 11064 11065 11066 11067 11068 11069 11070 11071 11072 11073 11074 11075 11076 11077 11078 11079 11080 11081 11082 11083 11084 11085 11086 11087 11088 11089 11090 11091 11092 11093 11094 11095 11096
/**
 * 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;
}


11097 11098 11099 11100 11101 11102 11103 11104 11105
bool
qemuDomainSupportsNewVcpuHotplug(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
}


11106
/**
11107
 * qemuDomainRefreshVcpuInfo:
11108 11109 11110
 * @driver: qemu driver data
 * @vm: domain object
 * @asyncJob: current asynchronous job type
11111
 * @state: refresh vcpu state
11112
 *
11113 11114
 * Updates vCPU information private data of @vm. Due to historical reasons this
 * function returns success even if some data were not reported by qemu.
11115
 *
11116 11117
 * If @state is true, the vcpu state is refreshed as reported by the monitor.
 *
11118
 * Returns 0 on success and -1 on fatal error.
11119 11120
 */
int
11121 11122
qemuDomainRefreshVcpuInfo(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
11123 11124
                          int asyncJob,
                          bool state)
11125
{
11126
    virDomainVcpuDefPtr vcpu;
11127 11128
    qemuDomainVcpuPrivatePtr vcpupriv;
    qemuMonitorCPUInfoPtr info = NULL;
11129
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
11130
    size_t i, j;
11131
    bool hotplug;
11132
    bool fast;
11133
    bool validTIDs = true;
11134
    int rc;
11135
    int ret = -1;
11136

11137
    hotplug = qemuDomainSupportsNewVcpuHotplug(vm);
11138 11139
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
11140

11141 11142
    VIR_DEBUG("Maxvcpus %zu hotplug %d fast query %d", maxvcpus, hotplug, fast);

11143 11144
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
11145

11146 11147
    rc = qemuMonitorGetCPUInfo(qemuDomainGetMonitor(vm), &info, maxvcpus,
                               hotplug, fast);
11148

11149
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
11150 11151
        goto cleanup;

11152
    if (rc < 0)
11153 11154
        goto cleanup;

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 11182 11183 11184 11185 11186 11187
    /*
     * 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++) {
11188
            if (info[i].tid != 0 && info[i].tid == info[j].tid) {
11189 11190 11191 11192 11193 11194 11195 11196 11197 11198 11199 11200
                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);
11201 11202
    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
11203
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
11204

11205
        if (validTIDs)
11206 11207 11208 11209 11210
            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;
11211
        vcpupriv->node_id = info[i].node_id;
11212 11213 11214 11215 11216 11217
        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;
11218
        vcpupriv->qemu_id = info[i].qemu_id;
11219

11220
        if (hotplug && state) {
11221 11222 11223 11224 11225
            vcpu->online = info[i].online;
            if (info[i].hotpluggable)
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_YES;
            else
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_NO;
11226
        }
11227 11228
    }

11229
    ret = 0;
11230 11231

 cleanup:
11232
    qemuMonitorCPUInfoFree(info, maxvcpus);
11233
    return ret;
11234
}
11235

11236 11237 11238 11239 11240 11241 11242 11243 11244 11245 11246 11247 11248 11249 11250 11251 11252 11253 11254 11255 11256 11257 11258 11259 11260 11261 11262 11263 11264 11265 11266
/**
 * 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;
11267
    qemuDomainVcpuPrivatePtr vcpupriv;
11268
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
11269
    virBitmapPtr haltedmap = NULL;
11270 11271
    size_t i;
    int ret = -1;
11272
    bool fast;
11273 11274 11275 11276 11277

    /* Not supported currently for TCG, see qemuDomainRefreshVcpuInfo */
    if (vm->def->virtType == VIR_DOMAIN_VIRT_QEMU)
        return 0;

11278
    /* The halted state is interresting only on s390(x). On other platforms
11279 11280 11281 11282 11283 11284 11285
     * 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))
11286 11287
        return 0;

11288 11289 11290
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;

11291 11292 11293 11294
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
    haltedmap = qemuMonitorGetCpuHalted(qemuDomainGetMonitor(vm), maxvcpus,
                                        fast);
11295
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || !haltedmap)
11296 11297 11298 11299
        goto cleanup;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
11300
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
11301 11302
        vcpupriv->halted = virTristateBoolFromBool(virBitmapIsBitSet(haltedmap,
                                                                     vcpupriv->qemu_id));
11303 11304 11305 11306 11307
    }

    ret = 0;

 cleanup:
11308
    virBitmapFree(haltedmap);
11309 11310
    return ret;
}
11311 11312 11313 11314 11315 11316

bool
qemuDomainSupportsNicdev(virDomainDefPtr def,
                         virDomainNetDefPtr net)
{
    /* non-virtio ARM nics require legacy -net nic */
S
Stefan Schallenberg 已提交
11317 11318
    if (((def->os.arch == VIR_ARCH_ARMV6L) ||
        (def->os.arch == VIR_ARCH_ARMV7L) ||
11319 11320 11321 11322 11323 11324 11325 11326
        (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;
}

11327 11328 11329 11330
bool
qemuDomainNetSupportsMTU(virDomainNetType type)
{
    switch (type) {
11331 11332
    case VIR_DOMAIN_NET_TYPE_NETWORK:
    case VIR_DOMAIN_NET_TYPE_BRIDGE:
11333 11334
    case VIR_DOMAIN_NET_TYPE_ETHERNET:
    case VIR_DOMAIN_NET_TYPE_VHOSTUSER:
11335 11336
        return true;
    case VIR_DOMAIN_NET_TYPE_USER:
11337 11338 11339 11340 11341 11342 11343 11344 11345 11346 11347 11348
    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
11349

P
Peter Krempa 已提交
11350 11351 11352 11353 11354 11355 11356 11357 11358 11359 11360 11361 11362 11363 11364

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;
}
11365 11366 11367 11368 11369 11370 11371 11372 11373 11374 11375 11376 11377 11378 11379 11380 11381 11382 11383 11384 11385 11386 11387 11388 11389 11390 11391 11392 11393 11394 11395


/**
 * 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;
}
11396 11397 11398 11399 11400 11401


int
qemuDomainPrepareChannel(virDomainChrDefPtr channel,
                         const char *domainChannelTargetDir)
{
S
Scott Garfinkle 已提交
11402 11403 11404 11405 11406 11407
    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) {
11408
        if (virAsprintf(&channel->source->data.nix.path,
11409
                        "%s/%s", domainChannelTargetDir,
S
Scott Garfinkle 已提交
11410 11411
                        channel->target.name) < 0)
            return -1;
11412 11413
    } else {
        /* Generate a unique name */
S
Scott Garfinkle 已提交
11414 11415 11416 11417 11418 11419
        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)
11420 11421 11422 11423 11424
            return -1;
    }

    return 0;
}
11425 11426


11427
/* qemuDomainPrepareChardevSourceTLS:
11428 11429 11430 11431 11432 11433 11434 11435 11436 11437 11438 11439 11440 11441 11442 11443 11444
 * @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;
11445
            source->data.tcp.tlsFromConfig = true;
11446 11447 11448 11449 11450
        }
    }
}


11451
/* qemuDomainPrepareChardevSource:
11452
 * @def: live domain definition
11453
 * @cfg: driver configuration
11454 11455 11456 11457 11458 11459
 *
 * Iterate through all devices that use virDomainChrSourceDefPtr as host
 * interface part.
 */
void
qemuDomainPrepareChardevSource(virDomainDefPtr def,
11460
                               virQEMUDriverConfigPtr cfg)
11461 11462 11463 11464 11465 11466 11467 11468 11469 11470 11471 11472 11473 11474 11475 11476 11477 11478 11479 11480 11481 11482 11483 11484 11485 11486 11487 11488 11489
{
    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);
}


11490 11491 11492 11493 11494 11495 11496 11497 11498 11499 11500 11501 11502 11503 11504 11505 11506 11507 11508 11509 11510 11511 11512 11513
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;
}


11514 11515 11516 11517 11518 11519 11520 11521 11522 11523 11524 11525 11526 11527 11528 11529 11530 11531 11532 11533 11534 11535 11536 11537 11538 11539 11540 11541
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;
}


11542
/* qemuProcessPrepareStorageSourceTLS:
11543
 * @source: source for a disk
11544
 * @cfg: driver configuration
11545
 * @parentAlias: alias of the parent device
11546 11547 11548 11549 11550 11551 11552
 *
 * 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
 */
11553
static int
11554
qemuDomainPrepareStorageSourceTLS(virStorageSourcePtr src,
11555
                                  virQEMUDriverConfigPtr cfg,
11556 11557
                                  const char *parentAlias,
                                  virQEMUCapsPtr qemuCaps)
11558
{
11559 11560 11561 11562 11563 11564 11565 11566
    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;
11567

11568
    case VIR_STORAGE_NET_PROTOCOL_NBD:
11569 11570 11571 11572
        if (qemuProcessPrepareStorageSourceTLSNBD(src, cfg, qemuCaps) < 0)
            return -1;
        break;

11573 11574 11575 11576 11577 11578 11579 11580 11581 11582
    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:
11583 11584 11585 11586 11587 11588
        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;
        }
11589 11590 11591 11592 11593 11594
        break;

    case VIR_STORAGE_NET_PROTOCOL_NONE:
    case VIR_STORAGE_NET_PROTOCOL_LAST:
    default:
        virReportEnumRangeError(virStorageNetProtocol, src->protocol);
11595
        return -1;
11596
    }
11597

11598 11599 11600 11601
    if (src->haveTLS == VIR_TRISTATE_BOOL_YES &&
        !(src->tlsAlias = qemuAliasTLSObjFromSrcAlias(parentAlias)))
        return -1;

11602 11603 11604 11605
    return 0;
}


11606 11607 11608 11609 11610 11611 11612 11613 11614 11615 11616 11617 11618
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);
}


11619 11620 11621 11622 11623 11624 11625 11626 11627 11628 11629 11630 11631 11632 11633 11634 11635 11636 11637 11638 11639 11640 11641 11642 11643 11644 11645 11646 11647 11648 11649 11650 11651
/**
 * 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);
}
11652 11653 11654 11655 11656 11657 11658 11659 11660 11661 11662 11663 11664 11665 11666 11667 11668 11669 11670 11671 11672 11673 11674 11675 11676 11677 11678 11679 11680 11681 11682 11683 11684 11685


/**
 * 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;
        }
    }
}
11686 11687 11688 11689 11690 11691 11692 11693 11694 11695 11696 11697 11698 11699 11700 11701 11702 11703 11704 11705


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;
}
11706 11707 11708 11709 11710 11711 11712 11713 11714 11715 11716 11717


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;
}
11718 11719


11720 11721
/**
 * qemuDomainGetHostdevPath:
11722
 * @def: domain definition
11723
 * @dev: host device definition
11724
 * @teardown: true if device will be removed
11725
 * @npaths: number of items in @path and @perms arrays
11726 11727 11728
 * @path: resulting path to @dev
 * @perms: Optional pointer to VIR_CGROUP_DEVICE_* perms
 *
11729 11730 11731 11732 11733 11734
 * 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.
11735 11736 11737 11738
 *
 * Returns 0 on success, -1 otherwise.
 */
int
11739 11740 11741
qemuDomainGetHostdevPath(virDomainDefPtr def,
                         virDomainHostdevDefPtr dev,
                         bool teardown,
11742 11743 11744
                         size_t *npaths,
                         char ***path,
                         int **perms)
11745 11746 11747 11748 11749 11750
{
    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;
11751
    virDomainHostdevSubsysMediatedDevPtr mdevsrc = &dev->source.subsys.u.mdev;
11752 11753 11754 11755 11756 11757
    virPCIDevicePtr pci = NULL;
    virUSBDevicePtr usb = NULL;
    virSCSIDevicePtr scsi = NULL;
    virSCSIVHostDevicePtr host = NULL;
    char *tmpPath = NULL;
    bool freeTmpPath = false;
11758 11759 11760 11761 11762
    bool includeVFIO = false;
    char **tmpPaths = NULL;
    int *tmpPerms = NULL;
    size_t i, tmpNpaths = 0;
    int perm = 0;
11763

11764
    *npaths = 0;
11765 11766 11767

    switch ((virDomainHostdevMode) dev->mode) {
    case VIR_DOMAIN_HOSTDEV_MODE_SUBSYS:
11768
        switch ((virDomainHostdevSubsysType)dev->source.subsys.type) {
11769 11770 11771 11772 11773 11774 11775 11776 11777 11778 11779 11780
        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;
11781 11782

                perm = VIR_CGROUP_DEVICE_RW;
11783 11784 11785 11786 11787 11788 11789 11790 11791 11792 11793 11794 11795 11796 11797
                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;
                }
11798 11799 11800 11801 11802 11803 11804 11805 11806 11807 11808 11809
            }
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
            if (dev->missing)
                break;
            usb = virUSBDeviceNew(usbsrc->bus,
                                  usbsrc->device,
                                  NULL);
            if (!usb)
                goto cleanup;

11810
            if (!(tmpPath = (char *)virUSBDeviceGetPath(usb)))
11811
                goto cleanup;
11812
            perm = VIR_CGROUP_DEVICE_RW;
11813 11814 11815 11816 11817
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI:
            if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI) {
                virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
11818
                VIR_DEBUG("Not updating /dev for hostdev iSCSI path '%s'", iscsisrc->src->path);
11819 11820 11821 11822 11823 11824 11825 11826 11827 11828 11829 11830 11831
            } 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;

11832
                if (!(tmpPath = (char *)virSCSIDeviceGetPath(scsi)))
11833
                    goto cleanup;
11834 11835
                perm = virSCSIDeviceGetReadonly(scsi) ?
                    VIR_CGROUP_DEVICE_READ : VIR_CGROUP_DEVICE_RW;
11836 11837 11838 11839 11840 11841 11842 11843 11844
            }
            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;

11845
                if (!(tmpPath = (char *)virSCSIVHostDeviceGetPath(host)))
11846
                    goto cleanup;
11847
                perm = VIR_CGROUP_DEVICE_RW;
11848 11849 11850 11851
            }
            break;
        }

11852
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
11853
            if (!(tmpPath = virMediatedDeviceGetIOMMUGroupDev(mdevsrc->uuidstr)))
11854 11855 11856 11857 11858 11859
                goto cleanup;

            freeTmpPath = true;
            includeVFIO = true;
            perm = VIR_CGROUP_DEVICE_RW;
            break;
11860 11861 11862 11863 11864 11865 11866 11867 11868 11869 11870
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
            break;
        }
        break;

    case VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES:
    case VIR_DOMAIN_HOSTDEV_MODE_LAST:
        /* nada */
        break;
    }

11871 11872 11873 11874 11875
    if (tmpPath) {
        size_t toAlloc = 1;

        if (includeVFIO)
            toAlloc = 2;
11876

11877 11878 11879 11880 11881 11882 11883 11884 11885 11886 11887 11888 11889 11890 11891 11892 11893 11894 11895 11896 11897 11898
        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;
    }
11899 11900
    ret = 0;
 cleanup:
11901
    virStringListFreeCount(tmpPaths, tmpNpaths);
11902
    VIR_FREE(tmpPerms);
11903 11904 11905 11906 11907 11908 11909 11910 11911 11912
    virPCIDeviceFree(pci);
    virUSBDeviceFree(usb);
    virSCSIDeviceFree(scsi);
    virSCSIVHostDeviceFree(host);
    if (freeTmpPath)
        VIR_FREE(tmpPath);
    return ret;
}


11913 11914 11915 11916
/**
 * qemuDomainGetPreservedMountPath:
 * @cfg: driver configuration data
 * @vm: domain object
11917
 * @mountpoint: mount point path to convert
11918
 *
11919
 * For given @mountpoint return new path where the mount point
11920 11921 11922 11923 11924 11925 11926 11927
 * 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,
11928
                                const char *mountpoint)
11929 11930 11931
{
    char *path = NULL;
    char *tmp;
11932
    const char *suffix = mountpoint + strlen(DEVPREFIX);
11933
    char *domname = virDomainDefGetShortName(vm->def);
11934 11935
    size_t off;

11936 11937 11938
    if (!domname)
        return NULL;

11939
    if (STREQ(mountpoint, "/dev"))
11940 11941 11942
        suffix = "dev";

    if (virAsprintf(&path, "%s/%s.%s",
11943 11944
                    cfg->stateDir, domname, suffix) < 0)
        goto cleanup;
11945

11946
    /* Now consider that @mountpoint is "/dev/blah/blah2".
11947 11948 11949 11950 11951 11952 11953 11954 11955 11956 11957 11958 11959
     * @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++;
    }

11960 11961
 cleanup:
    VIR_FREE(domname);
11962 11963 11964 11965
    return path;
}


11966 11967 11968 11969 11970 11971 11972 11973 11974 11975 11976 11977
/**
 * 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
11978
qemuDomainGetPreservedMounts(virQEMUDriverConfigPtr cfg,
11979 11980 11981 11982 11983 11984
                             virDomainObjPtr vm,
                             char ***devPath,
                             char ***devSavePath,
                             size_t *ndevPath)
{
    char **paths = NULL, **mounts = NULL;
11985
    size_t i, j, nmounts;
11986 11987 11988 11989 11990 11991 11992 11993 11994 11995 11996

    if (virFileGetMountSubtree(PROC_MOUNTS, "/dev",
                               &mounts, &nmounts) < 0)
        goto error;

    if (!nmounts) {
        if (ndevPath)
            *ndevPath = 0;
        return 0;
    }

11997 11998 11999 12000 12001 12002 12003 12004 12005 12006 12007 12008
    /* 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) {
12009 12010
            char *c = STRSKIP(mounts[j], mounts[i]);

12011
            if (c && (*c == '/' || *c == '\0')) {
12012 12013 12014 12015 12016 12017 12018 12019
                VIR_DEBUG("Dropping path %s because of %s", mounts[j], mounts[i]);
                VIR_DELETE_ELEMENT(mounts, j, nmounts);
            } else {
                j++;
            }
        }
    }

12020 12021 12022 12023
    if (VIR_ALLOC_N(paths, nmounts) < 0)
        goto error;

    for (i = 0; i < nmounts; i++) {
12024
        if (!(paths[i] = qemuDomainGetPreservedMountPath(cfg, vm, mounts[i])))
12025 12026 12027 12028 12029 12030 12031 12032 12033 12034 12035 12036 12037 12038 12039 12040 12041 12042 12043 12044 12045 12046 12047 12048 12049
            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;
}


12050 12051
struct qemuDomainCreateDeviceData {
    const char *path;     /* Path to temp new /dev location */
12052 12053
    char * const *devMountsPath;
    size_t ndevMountsPath;
12054 12055 12056
};


12057
static int
12058
qemuDomainCreateDeviceRecursive(const char *device,
12059
                                const struct qemuDomainCreateDeviceData *data,
12060 12061
                                bool allow_noent,
                                unsigned int ttl)
12062 12063
{
    char *devicePath = NULL;
12064
    char *target = NULL;
12065 12066
    struct stat sb;
    int ret = -1;
12067
    bool isLink = false;
12068
    bool isDev = false;
12069
    bool isReg = false;
12070
    bool isDir = false;
12071
    bool create = false;
12072 12073 12074
#ifdef WITH_SELINUX
    char *tcon = NULL;
#endif
12075

12076 12077 12078 12079 12080 12081 12082
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             device);
        return ret;
    }

12083
    if (lstat(device, &sb) < 0) {
12084 12085
        if (errno == ENOENT && allow_noent) {
            /* Ignore non-existent device. */
12086
            return 0;
12087
        }
12088 12089
        virReportSystemError(errno, _("Unable to stat %s"), device);
        return ret;
12090 12091
    }

12092
    isLink = S_ISLNK(sb.st_mode);
12093
    isDev = S_ISCHR(sb.st_mode) || S_ISBLK(sb.st_mode);
12094
    isReg = S_ISREG(sb.st_mode) || S_ISFIFO(sb.st_mode) || S_ISSOCK(sb.st_mode);
12095
    isDir = S_ISDIR(sb.st_mode);
12096 12097 12098 12099 12100 12101 12102 12103 12104 12105 12106 12107 12108 12109 12110 12111 12112

    /* 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)) {
12113
        size_t i;
12114

12115 12116 12117 12118 12119 12120 12121 12122 12123 12124 12125 12126 12127 12128 12129 12130 12131 12132 12133 12134 12135 12136 12137 12138 12139
        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]);
12140
        }
12141 12142
    }

12143 12144 12145 12146 12147 12148 12149 12150 12151
    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;
        }
12152

12153 12154 12155 12156 12157 12158 12159 12160 12161
        if (create &&
            symlink(target, devicePath) < 0) {
            if (errno == EEXIST) {
                ret = 0;
            } else {
                virReportSystemError(errno,
                                     _("unable to create symlink %s"),
                                     devicePath);
            }
12162 12163 12164
            goto cleanup;
        }

12165 12166 12167 12168 12169 12170 12171 12172 12173 12174
        /* 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;
12175

12176 12177
            if (VIR_STRDUP(devTmp, device) < 0)
                goto cleanup;
12178

12179 12180 12181 12182 12183 12184 12185 12186 12187
            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);
12188
            VIR_STEAL_PTR(target, tmp);
12189 12190
        }

12191
        if (qemuDomainCreateDeviceRecursive(target, data,
12192
                                            allow_noent, ttl - 1) < 0)
12193
            goto cleanup;
12194
    } else if (isDev) {
12195 12196 12197 12198 12199 12200 12201 12202 12203 12204
        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;
12205
        }
12206
    } else if (isReg) {
12207
        if (create &&
12208
            virFileTouch(devicePath, sb.st_mode) < 0)
12209
            goto cleanup;
12210 12211
        /* Just create the file here so that code below sets
         * proper owner and mode. Bind mount only after that. */
12212 12213 12214 12215
    } else if (isDir) {
        if (create &&
            virFileMakePathWithMode(devicePath, sb.st_mode) < 0)
            goto cleanup;
12216 12217 12218 12219 12220
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       device, sb.st_mode);
        goto cleanup;
12221 12222 12223 12224
    }

    if (!create) {
        ret = 0;
12225 12226 12227
        goto cleanup;
    }

12228
    if (lchown(devicePath, sb.st_uid, sb.st_gid) < 0) {
12229 12230 12231 12232 12233 12234
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             devicePath);
        goto cleanup;
    }

12235 12236 12237 12238 12239 12240 12241 12242 12243
    /* 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;
    }

12244 12245 12246
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileCopyACLs(device, devicePath) < 0 &&
12247 12248 12249 12250 12251 12252 12253
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Failed to copy ACLs on device %s"),
                             devicePath);
        goto cleanup;
    }

12254
#ifdef WITH_SELINUX
12255
    if (lgetfilecon_raw(device, &tcon) < 0 &&
12256 12257 12258
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"),
12259
                             device);
12260 12261 12262 12263
        goto cleanup;
    }

    if (tcon &&
12264
        lsetfilecon_raw(devicePath, (VIR_SELINUX_CTX_CONST char *)tcon) < 0) {
12265 12266 12267 12268 12269 12270 12271 12272 12273 12274 12275
        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

12276
    /* Finish mount process started earlier. */
12277
    if ((isReg || isDir) &&
12278 12279 12280
        virFileBindMountDevice(device, devicePath) < 0)
        goto cleanup;

12281 12282
    ret = 0;
 cleanup:
12283
    VIR_FREE(target);
12284
    VIR_FREE(devicePath);
12285 12286 12287
#ifdef WITH_SELINUX
    freecon(tcon);
#endif
12288 12289 12290 12291
    return ret;
}


12292 12293
static int
qemuDomainCreateDevice(const char *device,
12294
                       const struct qemuDomainCreateDeviceData *data,
12295 12296 12297 12298
                       bool allow_noent)
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

12299
    return qemuDomainCreateDeviceRecursive(device, data,
12300 12301 12302
                                           allow_noent, symloop_max);
}

12303 12304

static int
12305
qemuDomainPopulateDevices(virQEMUDriverConfigPtr cfg,
12306
                          virDomainObjPtr vm ATTRIBUTE_UNUSED,
12307
                          const struct qemuDomainCreateDeviceData *data)
12308 12309 12310 12311 12312 12313 12314 12315 12316
{
    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++) {
12317
        if (qemuDomainCreateDevice(devices[i], data, true) < 0)
12318 12319 12320 12321 12322 12323 12324 12325 12326 12327
            goto cleanup;
    }

    ret = 0;
 cleanup:
    return ret;
}


static int
12328 12329
qemuDomainSetupDev(virQEMUDriverConfigPtr cfg,
                   virSecurityManagerPtr mgr,
12330
                   virDomainObjPtr vm,
12331
                   const struct qemuDomainCreateDeviceData *data)
12332 12333 12334 12335 12336 12337 12338
{
    char *mount_options = NULL;
    char *opts = NULL;
    int ret = -1;

    VIR_DEBUG("Setting up /dev/ for domain %s", vm->def->name);

12339
    mount_options = qemuSecurityGetMountOptions(mgr, vm->def);
12340 12341 12342 12343 12344 12345 12346 12347 12348 12349 12350 12351 12352

    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;

12353
    if (virFileSetupDev(data->path, opts) < 0)
12354 12355
        goto cleanup;

12356
    if (qemuDomainPopulateDevices(cfg, vm, data) < 0)
12357 12358 12359 12360 12361 12362 12363 12364 12365 12366
        goto cleanup;

    ret = 0;
 cleanup:
    VIR_FREE(opts);
    VIR_FREE(mount_options);
    return ret;
}


12367
static int
12368
qemuDomainSetupDisk(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12369
                    virDomainDiskDefPtr disk,
12370
                    const struct qemuDomainCreateDeviceData *data)
12371 12372 12373 12374 12375
{
    virStorageSourcePtr next;
    char *dst = NULL;
    int ret = -1;

12376
    for (next = disk->src; virStorageSourceIsBacking(next); next = next->backingStore) {
12377
        if (!next->path || !virStorageSourceIsLocalStorage(next)) {
12378 12379 12380 12381
            /* Not creating device. Just continue. */
            continue;
        }

12382
        if (qemuDomainCreateDevice(next->path, data, false) < 0)
12383 12384 12385
            goto cleanup;
    }

12386
    /* qemu-pr-helper might require access to /dev/mapper/control. */
12387
    if (disk->src->pr &&
12388 12389 12390
        qemuDomainCreateDevice(DEVICE_MAPPER_CONTROL_PATH, data, true) < 0)
        goto cleanup;

12391 12392 12393 12394 12395 12396 12397 12398
    ret = 0;
 cleanup:
    VIR_FREE(dst);
    return ret;
}


static int
12399
qemuDomainSetupAllDisks(virQEMUDriverConfigPtr cfg,
12400
                        virDomainObjPtr vm,
12401
                        const struct qemuDomainCreateDeviceData *data)
12402 12403 12404 12405 12406
{
    size_t i;
    VIR_DEBUG("Setting up disks");

    for (i = 0; i < vm->def->ndisks; i++) {
12407
        if (qemuDomainSetupDisk(cfg,
12408
                                vm->def->disks[i],
12409
                                data) < 0)
12410 12411 12412 12413 12414 12415 12416 12417
            return -1;
    }

    VIR_DEBUG("Setup all disks");
    return 0;
}


12418
static int
12419
qemuDomainSetupHostdev(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12420
                       virDomainHostdevDefPtr dev,
12421
                       const struct qemuDomainCreateDeviceData *data)
12422 12423
{
    int ret = -1;
12424 12425
    char **path = NULL;
    size_t i, npaths = 0;
12426

12427
    if (qemuDomainGetHostdevPath(NULL, dev, false, &npaths, &path, NULL) < 0)
12428 12429
        goto cleanup;

12430
    for (i = 0; i < npaths; i++) {
12431
        if (qemuDomainCreateDevice(path[i], data, false) < 0)
12432
            goto cleanup;
12433 12434 12435 12436
    }

    ret = 0;
 cleanup:
12437 12438
    for (i = 0; i < npaths; i++)
        VIR_FREE(path[i]);
12439 12440 12441 12442 12443 12444
    VIR_FREE(path);
    return ret;
}


static int
12445
qemuDomainSetupAllHostdevs(virQEMUDriverConfigPtr cfg,
12446
                           virDomainObjPtr vm,
12447
                           const struct qemuDomainCreateDeviceData *data)
12448 12449 12450 12451 12452
{
    size_t i;

    VIR_DEBUG("Setting up hostdevs");
    for (i = 0; i < vm->def->nhostdevs; i++) {
12453
        if (qemuDomainSetupHostdev(cfg,
12454
                                   vm->def->hostdevs[i],
12455
                                   data) < 0)
12456 12457 12458 12459 12460 12461 12462
            return -1;
    }
    VIR_DEBUG("Setup all hostdevs");
    return 0;
}


M
Michal Privoznik 已提交
12463 12464 12465
static int
qemuDomainSetupMemory(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
                      virDomainMemoryDefPtr mem,
12466
                      const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
12467 12468 12469 12470
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

12471
    return qemuDomainCreateDevice(mem->nvdimmPath, data, false);
M
Michal Privoznik 已提交
12472 12473 12474 12475 12476 12477
}


static int
qemuDomainSetupAllMemories(virQEMUDriverConfigPtr cfg,
                           virDomainObjPtr vm,
12478
                           const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
12479 12480 12481 12482 12483 12484 12485
{
    size_t i;

    VIR_DEBUG("Setting up memories");
    for (i = 0; i < vm->def->nmems; i++) {
        if (qemuDomainSetupMemory(cfg,
                                  vm->def->mems[i],
12486
                                  data) < 0)
M
Michal Privoznik 已提交
12487 12488 12489 12490 12491 12492 12493
            return -1;
    }
    VIR_DEBUG("Setup all memories");
    return 0;
}


12494 12495 12496 12497 12498
static int
qemuDomainSetupChardev(virDomainDefPtr def ATTRIBUTE_UNUSED,
                       virDomainChrDefPtr dev,
                       void *opaque)
{
12499
    const struct qemuDomainCreateDeviceData *data = opaque;
12500
    const char *path = NULL;
12501

12502
    if (!(path = virDomainChrSourceDefGetPath(dev->source)))
12503 12504
        return 0;

12505 12506 12507 12508 12509 12510
    /* 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);
12511 12512 12513 12514
}


static int
12515
qemuDomainSetupAllChardevs(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12516
                           virDomainObjPtr vm,
12517
                           const struct qemuDomainCreateDeviceData *data)
12518 12519 12520 12521 12522 12523
{
    VIR_DEBUG("Setting up chardevs");

    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuDomainSetupChardev,
12524
                               (void *)data) < 0)
12525 12526 12527 12528 12529 12530 12531
        return -1;

    VIR_DEBUG("Setup all chardevs");
    return 0;
}


12532
static int
12533
qemuDomainSetupTPM(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12534
                   virDomainObjPtr vm,
12535
                   const struct qemuDomainCreateDeviceData *data)
12536 12537 12538 12539 12540 12541 12542 12543 12544 12545 12546
{
    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,
12547
                                   data, false) < 0)
12548 12549 12550
            return -1;
        break;

12551
    case VIR_DOMAIN_TPM_TYPE_EMULATOR:
12552 12553 12554 12555 12556 12557 12558 12559 12560 12561
    case VIR_DOMAIN_TPM_TYPE_LAST:
        /* nada */
        break;
    }

    VIR_DEBUG("Setup TPM");
    return 0;
}


12562
static int
12563
qemuDomainSetupGraphics(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12564
                        virDomainGraphicsDefPtr gfx,
12565
                        const struct qemuDomainCreateDeviceData *data)
12566
{
12567
    const char *rendernode = virDomainGraphicsGetRenderNode(gfx);
12568

12569
    if (!rendernode)
12570 12571
        return 0;

12572
    return qemuDomainCreateDevice(rendernode, data, false);
12573 12574 12575 12576
}


static int
12577
qemuDomainSetupAllGraphics(virQEMUDriverConfigPtr cfg,
12578
                           virDomainObjPtr vm,
12579
                           const struct qemuDomainCreateDeviceData *data)
12580 12581 12582 12583 12584
{
    size_t i;

    VIR_DEBUG("Setting up graphics");
    for (i = 0; i < vm->def->ngraphics; i++) {
12585
        if (qemuDomainSetupGraphics(cfg,
12586
                                    vm->def->graphics[i],
12587
                                    data) < 0)
12588 12589 12590 12591 12592 12593 12594 12595
            return -1;
    }

    VIR_DEBUG("Setup all graphics");
    return 0;
}


12596
static int
12597
qemuDomainSetupInput(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12598
                     virDomainInputDefPtr input,
12599
                     const struct qemuDomainCreateDeviceData *data)
12600
{
J
Ján Tomko 已提交
12601
    const char *path = virDomainInputDefGetPath(input);
12602

J
Ján Tomko 已提交
12603 12604
    if (path && qemuDomainCreateDevice(path, data, false) < 0)
        return -1;
12605

J
Ján Tomko 已提交
12606
    return 0;
12607 12608 12609 12610
}


static int
12611
qemuDomainSetupAllInputs(virQEMUDriverConfigPtr cfg,
12612
                         virDomainObjPtr vm,
12613
                         const struct qemuDomainCreateDeviceData *data)
12614 12615 12616
{
    size_t i;

12617
    VIR_DEBUG("Setting up inputs");
12618
    for (i = 0; i < vm->def->ninputs; i++) {
12619
        if (qemuDomainSetupInput(cfg,
12620
                                 vm->def->inputs[i],
12621
                                 data) < 0)
12622 12623
            return -1;
    }
12624
    VIR_DEBUG("Setup all inputs");
12625 12626 12627 12628
    return 0;
}


12629
static int
12630
qemuDomainSetupRNG(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
12631
                   virDomainRNGDefPtr rng,
12632
                   const struct qemuDomainCreateDeviceData *data)
12633 12634 12635
{
    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
12636
        if (qemuDomainCreateDevice(rng->source.file, data, false) < 0)
12637 12638 12639 12640 12641 12642 12643 12644 12645 12646 12647 12648 12649
            return -1;

    case VIR_DOMAIN_RNG_BACKEND_EGD:
    case VIR_DOMAIN_RNG_BACKEND_LAST:
        /* nada */
        break;
    }

    return 0;
}


static int
12650
qemuDomainSetupAllRNGs(virQEMUDriverConfigPtr cfg,
12651
                       virDomainObjPtr vm,
12652
                       const struct qemuDomainCreateDeviceData *data)
12653 12654 12655 12656 12657
{
    size_t i;

    VIR_DEBUG("Setting up RNGs");
    for (i = 0; i < vm->def->nrngs; i++) {
12658
        if (qemuDomainSetupRNG(cfg,
12659
                               vm->def->rngs[i],
12660
                               data) < 0)
12661 12662 12663 12664 12665 12666 12667 12668
            return -1;
    }

    VIR_DEBUG("Setup all RNGs");
    return 0;
}


12669 12670 12671 12672 12673 12674 12675 12676 12677 12678 12679 12680 12681 12682 12683 12684 12685 12686 12687 12688 12689 12690 12691 12692 12693 12694
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;

12695
        case VIR_DOMAIN_LOADER_TYPE_NONE:
12696 12697 12698 12699 12700 12701 12702 12703 12704 12705 12706 12707
        case VIR_DOMAIN_LOADER_TYPE_LAST:
            break;
        }
    }

    VIR_DEBUG("Setup loader");
    ret = 0;
 cleanup:
    return ret;
}


12708 12709 12710 12711 12712 12713 12714 12715 12716 12717 12718 12719 12720 12721 12722 12723 12724 12725 12726 12727
static int
qemuDomainSetupLaunchSecurity(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
                              virDomainObjPtr vm,
                              const struct qemuDomainCreateDeviceData *data)
{
    virDomainSEVDefPtr sev = vm->def->sev;

    if (!sev || sev->sectype != VIR_DOMAIN_LAUNCH_SECURITY_SEV)
        return 0;

    VIR_DEBUG("Setting up launch security");

    if (qemuDomainCreateDevice(DEV_SEV, data, false) < 0)
        return -1;

    VIR_DEBUG("Set up launch security");
    return 0;
}


12728
int
12729 12730
qemuDomainBuildNamespace(virQEMUDriverConfigPtr cfg,
                         virSecurityManagerPtr mgr,
12731 12732
                         virDomainObjPtr vm)
{
12733
    struct qemuDomainCreateDeviceData data;
12734
    char *devPath = NULL;
12735
    char **devMountsPath = NULL, **devMountsSavePath = NULL;
12736 12737 12738 12739 12740 12741 12742 12743
    size_t ndevMountsPath = 0, i;
    int ret = -1;

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT)) {
        ret = 0;
        goto cleanup;
    }

12744
    if (qemuDomainGetPreservedMounts(cfg, vm,
12745 12746
                                     &devMountsPath, &devMountsSavePath,
                                     &ndevMountsPath) < 0)
12747 12748
        goto cleanup;

12749 12750 12751 12752 12753 12754 12755 12756 12757 12758 12759 12760 12761
    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;
    }

12762
    data.path = devPath;
12763 12764
    data.devMountsPath = devMountsPath;
    data.ndevMountsPath = ndevMountsPath;
12765

12766 12767 12768
    if (virProcessSetupPrivateMountNS() < 0)
        goto cleanup;

12769
    if (qemuDomainSetupDev(cfg, mgr, vm, &data) < 0)
12770 12771
        goto cleanup;

12772
    if (qemuDomainSetupAllDisks(cfg, vm, &data) < 0)
12773 12774
        goto cleanup;

12775
    if (qemuDomainSetupAllHostdevs(cfg, vm, &data) < 0)
12776 12777
        goto cleanup;

12778
    if (qemuDomainSetupAllMemories(cfg, vm, &data) < 0)
12779 12780
        goto cleanup;

12781
    if (qemuDomainSetupAllChardevs(cfg, vm, &data) < 0)
12782 12783
        goto cleanup;

12784
    if (qemuDomainSetupTPM(cfg, vm, &data) < 0)
12785 12786
        goto cleanup;

12787
    if (qemuDomainSetupAllGraphics(cfg, vm, &data) < 0)
12788 12789
        goto cleanup;

12790
    if (qemuDomainSetupAllInputs(cfg, vm, &data) < 0)
12791 12792
        goto cleanup;

12793
    if (qemuDomainSetupAllRNGs(cfg, vm, &data) < 0)
12794 12795
        goto cleanup;

12796 12797 12798
    if (qemuDomainSetupLoader(cfg, vm, &data) < 0)
        goto cleanup;

12799 12800 12801
    if (qemuDomainSetupLaunchSecurity(cfg, vm, &data) < 0)
        goto cleanup;

12802 12803
    /* Save some mount points because we want to share them with the host */
    for (i = 0; i < ndevMountsPath; i++) {
12804 12805
        struct stat sb;

12806 12807 12808
        if (devMountsSavePath[i] == devPath)
            continue;

12809 12810 12811 12812 12813 12814 12815
        if (stat(devMountsPath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsPath[i]);
            goto cleanup;
        }

12816 12817 12818
        /* At this point, devMountsPath is either:
         * a file (regular or special), or
         * a directory. */
12819
        if ((S_ISDIR(sb.st_mode) && virFileMakePath(devMountsSavePath[i]) < 0) ||
12820
            (!S_ISDIR(sb.st_mode) && virFileTouch(devMountsSavePath[i], sb.st_mode) < 0)) {
12821 12822 12823 12824 12825 12826
            virReportSystemError(errno,
                                 _("Failed to create %s"),
                                 devMountsSavePath[i]);
            goto cleanup;
        }

12827
        if (virFileMoveMount(devMountsPath[i], devMountsSavePath[i]) < 0)
12828 12829 12830
            goto cleanup;
    }

12831
    if (virFileMoveMount(devPath, "/dev") < 0)
12832 12833 12834
        goto cleanup;

    for (i = 0; i < ndevMountsPath; i++) {
12835 12836
        struct stat sb;

12837 12838 12839
        if (devMountsSavePath[i] == devPath)
            continue;

12840 12841 12842 12843
        if (stat(devMountsSavePath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsSavePath[i]);
12844 12845 12846
            goto cleanup;
        }

12847 12848 12849 12850 12851 12852 12853 12854 12855 12856 12857 12858 12859 12860 12861
        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;
            }
        }

12862
        if (virFileMoveMount(devMountsSavePath[i], devMountsPath[i]) < 0)
12863 12864 12865 12866 12867
            goto cleanup;
    }

    ret = 0;
 cleanup:
12868 12869 12870 12871 12872 12873 12874
    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]);
    }
12875
    virStringListFreeCount(devMountsPath, ndevMountsPath);
12876
    virStringListFreeCount(devMountsSavePath, ndevMountsPath);
12877 12878 12879 12880 12881 12882 12883 12884 12885 12886 12887
    return ret;
}


int
qemuDomainCreateNamespace(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

12888 12889
    if (virBitmapIsBitSet(cfg->namespaces, QEMU_DOMAIN_NS_MOUNT) &&
        qemuDomainEnableNamespace(vm, QEMU_DOMAIN_NS_MOUNT) < 0)
M
Michal Privoznik 已提交
12890
        goto cleanup;
12891 12892 12893 12894 12895 12896 12897 12898

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


12899 12900 12901 12902 12903 12904 12905 12906 12907
void
qemuDomainDestroyNamespace(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                           virDomainObjPtr vm)
{
    if (qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        qemuDomainDisableNamespace(vm, QEMU_DOMAIN_NS_MOUNT);
}


12908 12909 12910 12911 12912 12913 12914 12915 12916 12917 12918 12919 12920 12921 12922 12923 12924 12925 12926 12927 12928 12929 12930 12931 12932 12933 12934 12935 12936
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__) */
}


12937 12938 12939 12940
struct qemuDomainAttachDeviceMknodData {
    virQEMUDriverPtr driver;
    virDomainObjPtr vm;
    const char *file;
12941
    const char *target;
12942 12943
    struct stat sb;
    void *acl;
12944 12945 12946
#ifdef WITH_SELINUX
    char *tcon;
#endif
12947 12948 12949
};


12950 12951
/* Our way of creating devices is highly linux specific */
#if defined(__linux__)
12952 12953 12954 12955 12956 12957
static int
qemuDomainAttachDeviceMknodHelper(pid_t pid ATTRIBUTE_UNUSED,
                                  void *opaque)
{
    struct qemuDomainAttachDeviceMknodData *data = opaque;
    int ret = -1;
12958
    bool delDevice = false;
12959
    bool isLink = S_ISLNK(data->sb.st_mode);
12960
    bool isDev = S_ISCHR(data->sb.st_mode) || S_ISBLK(data->sb.st_mode);
12961
    bool isReg = S_ISREG(data->sb.st_mode) || S_ISFIFO(data->sb.st_mode) || S_ISSOCK(data->sb.st_mode);
12962
    bool isDir = S_ISDIR(data->sb.st_mode);
12963

12964
    qemuSecurityPostFork(data->driver->securityManager);
12965 12966 12967 12968 12969 12970 12971

    if (virFileMakeParentPath(data->file) < 0) {
        virReportSystemError(errno,
                             _("Unable to create %s"), data->file);
        goto cleanup;
    }

12972 12973
    if (isLink) {
        VIR_DEBUG("Creating symlink %s -> %s", data->file, data->target);
12974 12975 12976 12977 12978 12979 12980 12981 12982 12983 12984 12985

        /* 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;
        }

12986
        if (symlink(data->target, data->file) < 0) {
12987 12988 12989 12990
            virReportSystemError(errno,
                                 _("Unable to create symlink %s (pointing to %s)"),
                                 data->file, data->target);
            goto cleanup;
12991 12992
        } else {
            delDevice = true;
12993
        }
12994
    } else if (isDev) {
12995 12996 12997 12998 12999 13000 13001 13002 13003 13004 13005 13006 13007 13008 13009 13010
        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;
        }
13011
    } else if (isReg || isDir) {
13012 13013 13014 13015 13016
        /* 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 &&
13017
            errno != ENOENT && errno != EINVAL) {
13018 13019 13020 13021 13022
            virReportSystemError(errno,
                                 _("Unable to umount %s"),
                                 data->file);
            goto cleanup;
        }
13023 13024
        if ((isReg && virFileTouch(data->file, data->sb.st_mode) < 0) ||
            (isDir && virFileMakePathWithMode(data->file, data->sb.st_mode) < 0))
13025 13026 13027 13028
            goto cleanup;
        delDevice = true;
        /* Just create the file here so that code below sets
         * proper owner and mode. Move the mount only after that. */
13029 13030 13031 13032 13033
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       data->file, data->sb.st_mode);
        goto cleanup;
13034 13035
    }

13036 13037 13038 13039 13040 13041 13042
    if (lchown(data->file, data->sb.st_uid, data->sb.st_gid) < 0) {
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             data->file);
        goto cleanup;
    }

13043 13044 13045 13046 13047 13048 13049 13050 13051
    /* 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;
    }

13052 13053 13054
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileSetACLs(data->file, data->acl) < 0 &&
13055 13056 13057 13058 13059 13060
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to set ACLs on %s"), data->file);
        goto cleanup;
    }

13061
# ifdef WITH_SELINUX
13062
    if (data->tcon &&
13063
        lsetfilecon_raw(data->file, (VIR_SELINUX_CTX_CONST char *)data->tcon) < 0) {
13064 13065 13066 13067 13068 13069 13070 13071 13072
        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;
        }
    }
13073
# endif
13074

13075
    /* Finish mount process started earlier. */
13076
    if ((isReg || isDir) &&
13077 13078 13079
        virFileMoveMount(data->target, data->file) < 0)
        goto cleanup;

13080 13081
    ret = 0;
 cleanup:
13082 13083 13084 13085 13086 13087
    if (ret < 0 && delDevice) {
        if (isDir)
            virFileDeleteTree(data->file);
        else
            unlink(data->file);
    }
13088
# ifdef WITH_SELINUX
13089
    freecon(data->tcon);
13090
# endif
13091 13092 13093 13094 13095 13096
    virFileFreeACLs(&data->acl);
    return ret;
}


static int
13097 13098 13099
qemuDomainAttachDeviceMknodRecursive(virQEMUDriverPtr driver,
                                     virDomainObjPtr vm,
                                     const char *file,
13100 13101
                                     char * const *devMountsPath,
                                     size_t ndevMountsPath,
13102
                                     unsigned int ttl)
13103
{
13104
    virQEMUDriverConfigPtr cfg = NULL;
13105 13106
    struct qemuDomainAttachDeviceMknodData data;
    int ret = -1;
13107 13108
    char *target = NULL;
    bool isLink;
13109
    bool isReg;
13110
    bool isDir;
13111

13112 13113 13114 13115 13116 13117 13118
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             file);
        return ret;
    }

13119 13120 13121 13122 13123 13124
    memset(&data, 0, sizeof(data));

    data.driver = driver;
    data.vm = vm;
    data.file = file;

13125
    if (lstat(file, &data.sb) < 0) {
13126 13127 13128 13129 13130
        virReportSystemError(errno,
                             _("Unable to access %s"), file);
        return ret;
    }

13131
    isLink = S_ISLNK(data.sb.st_mode);
13132
    isReg = S_ISREG(data.sb.st_mode) || S_ISFIFO(data.sb.st_mode) || S_ISSOCK(data.sb.st_mode);
13133
    isDir = S_ISDIR(data.sb.st_mode);
13134

13135
    if ((isReg || isDir) && STRPREFIX(file, DEVPREFIX)) {
13136 13137 13138 13139 13140 13141 13142 13143 13144
        cfg = virQEMUDriverGetConfig(driver);
        if (!(target = qemuDomainGetPreservedMountPath(cfg, vm, file)))
            goto cleanup;

        if (virFileBindMountDevice(file, target) < 0)
            goto cleanup;

        data.target = target;
    } else if (isLink) {
13145 13146 13147 13148 13149 13150 13151 13152 13153 13154 13155 13156 13157 13158 13159 13160 13161 13162 13163 13164 13165 13166
        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);
13167
            VIR_STEAL_PTR(target, tmp);
13168 13169 13170 13171 13172 13173 13174 13175
        }

        data.target = target;
    }

    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileGetACLs(file, &data.acl) < 0 &&
13176 13177 13178
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to get ACLs on %s"), file);
13179
        goto cleanup;
13180 13181
    }

13182
# ifdef WITH_SELINUX
13183
    if (lgetfilecon_raw(file, &data.tcon) < 0 &&
13184 13185 13186 13187 13188
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"), file);
        goto cleanup;
    }
13189
# endif
13190

13191
    if (STRPREFIX(file, DEVPREFIX)) {
13192 13193 13194 13195 13196 13197 13198 13199 13200 13201 13202 13203
        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;
13204

13205 13206 13207 13208 13209 13210
            if (virProcessRunInMountNamespace(vm->pid,
                                              qemuDomainAttachDeviceMknodHelper,
                                              &data) < 0) {
                qemuSecurityPostFork(driver->securityManager);
                goto cleanup;
            }
13211
            qemuSecurityPostFork(driver->securityManager);
13212 13213 13214
        } else {
            VIR_DEBUG("Skipping dev %s because of %s mount point",
                      file, devMountsPath[i]);
13215
        }
13216 13217
    }

13218
    if (isLink &&
13219 13220 13221
        qemuDomainAttachDeviceMknodRecursive(driver, vm, target,
                                             devMountsPath, ndevMountsPath,
                                             ttl -1) < 0)
13222
        goto cleanup;
13223 13224 13225

    ret = 0;
 cleanup:
13226
# ifdef WITH_SELINUX
13227
    freecon(data.tcon);
13228
# endif
13229
    virFileFreeACLs(&data.acl);
13230 13231
    if (isReg && target)
        umount(target);
13232
    VIR_FREE(target);
13233
    virObjectUnref(cfg);
13234
    return ret;
13235 13236 13237
}


13238 13239 13240 13241 13242 13243 13244 13245 13246 13247 13248 13249 13250 13251 13252 13253 13254 13255 13256 13257
#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__) */


13258 13259 13260
static int
qemuDomainAttachDeviceMknod(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
13261 13262 13263
                            const char *file,
                            char * const *devMountsPath,
                            size_t ndevMountsPath)
13264 13265 13266
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

13267 13268 13269
    return qemuDomainAttachDeviceMknodRecursive(driver, vm, file,
                                                devMountsPath, ndevMountsPath,
                                                symloop_max);
13270 13271 13272
}


13273 13274 13275 13276 13277 13278 13279 13280 13281 13282 13283 13284 13285 13286 13287 13288 13289 13290
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
13291
qemuDomainDetachDeviceUnlink(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
13292
                             virDomainObjPtr vm,
13293 13294 13295
                             const char *file,
                             char * const *devMountsPath,
                             size_t ndevMountsPath)
13296
{
13297 13298
    int ret = -1;
    size_t i;
13299

13300 13301 13302 13303 13304 13305 13306 13307 13308 13309 13310 13311 13312 13313 13314 13315 13316 13317 13318
    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;
13319 13320 13321
}


13322 13323 13324 13325
static int
qemuDomainNamespaceMknodPaths(virDomainObjPtr vm,
                              const char **paths,
                              size_t npaths)
13326
{
13327 13328 13329
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    virQEMUDriverConfigPtr cfg;
13330 13331
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0;
13332
    int ret = -1;
13333
    size_t i;
13334

13335 13336
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
13337 13338
        return 0;

13339 13340 13341 13342 13343 13344
    cfg = virQEMUDriverGetConfig(driver);
    if (qemuDomainGetPreservedMounts(cfg, vm,
                                     &devMountsPath, NULL,
                                     &ndevMountsPath) < 0)
        goto cleanup;

13345 13346 13347 13348 13349 13350 13351 13352 13353 13354 13355 13356 13357 13358 13359 13360
    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;
}


13361 13362 13363 13364 13365 13366 13367 13368 13369 13370
static int
qemuDomainNamespaceMknodPath(virDomainObjPtr vm,
                             const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceMknodPaths(vm, paths, 1);
}


13371 13372 13373 13374 13375 13376 13377 13378 13379 13380 13381 13382 13383
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;

13384 13385
    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT) ||
        !npaths)
13386 13387 13388 13389 13390 13391 13392 13393 13394 13395 13396 13397 13398 13399 13400
        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;
    }

13401
    ret = 0;
13402 13403 13404 13405 13406 13407 13408
 cleanup:
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    virObjectUnref(cfg);
    return ret;
}


13409 13410 13411 13412 13413 13414 13415 13416 13417 13418
static int
qemuDomainNamespaceUnlinkPath(virDomainObjPtr vm,
                              const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceUnlinkPaths(vm, paths, 1);
}


13419
int
13420
qemuDomainNamespaceSetupDisk(virDomainObjPtr vm,
13421 13422 13423
                             virStorageSourcePtr src)
{
    virStorageSourcePtr next;
13424
    const char **paths = NULL;
13425
    size_t npaths = 0;
13426
    char *dmPath = NULL;
13427 13428
    int ret = -1;

13429
    for (next = src; virStorageSourceIsBacking(next); next = next->backingStore) {
13430 13431
        if (virStorageSourceIsEmpty(next) ||
            !virStorageSourceIsLocalStorage(next)) {
13432 13433 13434 13435
            /* Not creating device. Just continue. */
            continue;
        }

13436
        if (VIR_APPEND_ELEMENT_COPY(paths, npaths, next->path) < 0)
13437 13438 13439
            goto cleanup;
    }

13440
    /* qemu-pr-helper might require access to /dev/mapper/control. */
13441
    if (src->pr &&
13442 13443 13444 13445
        (VIR_STRDUP(dmPath, DEVICE_MAPPER_CONTROL_PATH) < 0 ||
         VIR_APPEND_ELEMENT_COPY(paths, npaths, dmPath) < 0))
        goto cleanup;

13446
    if (qemuDomainNamespaceMknodPaths(vm, paths, npaths) < 0)
13447
        goto cleanup;
13448

13449 13450
    ret = 0;
 cleanup:
13451
    VIR_FREE(dmPath);
13452
    VIR_FREE(paths);
13453 13454 13455 13456 13457
    return ret;
}


int
13458
qemuDomainNamespaceTeardownDisk(virDomainObjPtr vm ATTRIBUTE_UNUSED,
13459
                                virStorageSourcePtr src ATTRIBUTE_UNUSED)
13460 13461 13462 13463 13464 13465 13466 13467 13468
{
    /* 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;
}
13469 13470 13471


int
13472
qemuDomainNamespaceSetupHostdev(virDomainObjPtr vm,
13473 13474 13475
                                virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
13476
    char **paths = NULL;
13477
    size_t i, npaths = 0;
13478

13479
    if (qemuDomainGetHostdevPath(NULL, hostdev, false, &npaths, &paths, NULL) < 0)
13480 13481
        goto cleanup;

13482
    if (qemuDomainNamespaceMknodPaths(vm, (const char **)paths, npaths) < 0)
13483 13484
        goto cleanup;

13485 13486
    ret = 0;
 cleanup:
13487
    for (i = 0; i < npaths; i++)
13488 13489
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
13490 13491 13492 13493 13494
    return ret;
}


int
13495
qemuDomainNamespaceTeardownHostdev(virDomainObjPtr vm,
13496 13497 13498
                                   virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
13499
    char **paths = NULL;
13500
    size_t i, npaths = 0;
13501

13502
    if (qemuDomainGetHostdevPath(vm->def, hostdev, true,
13503
                                 &npaths, &paths, NULL) < 0)
13504 13505
        goto cleanup;

13506
    if (qemuDomainNamespaceUnlinkPaths(vm, (const char **)paths, npaths) < 0)
13507 13508
        goto cleanup;

13509 13510
    ret = 0;
 cleanup:
13511
    for (i = 0; i < npaths; i++)
13512 13513
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
13514 13515
    return ret;
}
13516 13517


M
Michal Privoznik 已提交
13518
int
13519
qemuDomainNamespaceSetupMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
13520 13521 13522 13523 13524
                               virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

13525 13526
    if (qemuDomainNamespaceMknodPath(vm, mem->nvdimmPath) < 0)
        return -1;
13527

13528
    return 0;
M
Michal Privoznik 已提交
13529 13530 13531 13532
}


int
13533
qemuDomainNamespaceTeardownMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
13534 13535 13536 13537 13538
                                  virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

13539 13540
    if (qemuDomainNamespaceUnlinkPath(vm, mem->nvdimmPath) < 0)
        return -1;
13541

13542
    return 0;
M
Michal Privoznik 已提交
13543 13544 13545
}


13546
int
13547
qemuDomainNamespaceSetupChardev(virDomainObjPtr vm,
13548 13549 13550 13551
                                virDomainChrDefPtr chr)
{
    const char *path;

13552
    if (!(path = virDomainChrSourceDefGetPath(chr->source)))
13553 13554
        return 0;

13555 13556 13557 13558
    /* 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;
13559

13560 13561
    if (qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
13562

13563
    return 0;
13564 13565 13566 13567
}


int
13568
qemuDomainNamespaceTeardownChardev(virDomainObjPtr vm,
13569 13570 13571 13572 13573 13574 13575 13576 13577
                                   virDomainChrDefPtr chr)
{
    const char *path = NULL;

    if (chr->source->type != VIR_DOMAIN_CHR_TYPE_DEV)
        return 0;

    path = chr->source->data.file.path;

13578 13579
    if (qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
13580

13581
    return 0;
13582
}
13583 13584 13585


int
13586
qemuDomainNamespaceSetupRNG(virDomainObjPtr vm,
13587 13588 13589 13590 13591 13592 13593 13594 13595 13596 13597
                            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:
13598
        break;
13599 13600
    }

13601 13602
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
13603

13604
    return 0;
13605 13606 13607 13608
}


int
13609
qemuDomainNamespaceTeardownRNG(virDomainObjPtr vm,
13610 13611 13612 13613 13614 13615 13616 13617 13618 13619 13620
                               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:
13621
        break;
13622 13623
    }

13624 13625
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
13626

13627
    return 0;
13628 13629 13630 13631 13632 13633 13634 13635 13636 13637 13638 13639
}


int
qemuDomainNamespaceSetupInput(virDomainObjPtr vm,
                              virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

13640 13641 13642
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
    return 0;
13643 13644 13645 13646 13647 13648 13649 13650 13651 13652 13653 13654
}


int
qemuDomainNamespaceTeardownInput(virDomainObjPtr vm,
                                 virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

13655 13656
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
13657

13658
    return 0;
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 13685 13686 13687 13688 13689 13690 13691 13692 13693 13694 13695 13696 13697 13698 13699 13700 13701 13702


/**
 * 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;
}
13703 13704 13705 13706 13707 13708 13709 13710 13711 13712 13713 13714 13715 13716 13717 13718 13719 13720 13721 13722 13723 13724


/**
 * 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;
}
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


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
13752
                       _("failed to find disk '%s'"), target);
13753 13754 13755
        goto cleanup;
    }

13756 13757 13758 13759
    if (idx == 0)
        src = disk->src;
    else
        src = virStorageFileChainLookup(disk->src, NULL, NULL, idx, NULL);
13760 13761 13762 13763 13764

 cleanup:
    VIR_FREE(target);
    return src;
}
13765 13766 13767 13768 13769 13770 13771 13772


static void
qemuDomainSaveCookieDispose(void *obj)
{
    qemuDomainSaveCookiePtr cookie = obj;

    VIR_DEBUG("cookie=%p", cookie);
13773 13774

    virCPUDefFree(cookie->cpu);
13775 13776 13777 13778 13779 13780
}


qemuDomainSaveCookiePtr
qemuDomainSaveCookieNew(virDomainObjPtr vm ATTRIBUTE_UNUSED)
{
13781
    qemuDomainObjPrivatePtr priv = vm->privateData;
13782 13783 13784 13785 13786 13787 13788 13789
    qemuDomainSaveCookiePtr cookie = NULL;

    if (qemuDomainInitialize() < 0)
        goto error;

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
        goto error;

13790 13791 13792 13793
    if (priv->origCPU && !(cookie->cpu = virCPUDefCopy(vm->def->cpu)))
        goto error;

    VIR_DEBUG("Save cookie %p, cpu=%p", cookie, cookie->cpu);
13794 13795 13796 13797 13798 13799 13800 13801 13802 13803 13804 13805 13806 13807 13808 13809 13810 13811 13812 13813 13814

    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;

13815 13816 13817 13818
    if (virCPUDefParseXML(ctxt, "./cpu[1]", VIR_CPU_TYPE_GUEST,
                          &cookie->cpu) < 0)
        goto error;

13819 13820 13821 13822 13823 13824 13825 13826 13827 13828
    *obj = (virObjectPtr) cookie;
    return 0;

 error:
    virObjectUnref(cookie);
    return -1;
}


static int
13829 13830
qemuDomainSaveCookieFormat(virBufferPtr buf,
                           virObjectPtr obj)
13831
{
13832 13833 13834
    qemuDomainSaveCookiePtr cookie = (qemuDomainSaveCookiePtr) obj;

    if (cookie->cpu &&
13835
        virCPUDefFormatBufFull(buf, cookie->cpu, NULL) < 0)
13836 13837
        return -1;

13838 13839 13840 13841 13842 13843 13844 13845
    return 0;
}


virSaveCookieCallbacks virQEMUDriverDomainSaveCookie = {
    .parse = qemuDomainSaveCookieParse,
    .format = qemuDomainSaveCookieFormat,
};
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 13874 13875 13876 13877 13878 13879 13880 13881 13882 13883 13884 13885


/**
 * 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;
}
13886

13887 13888 13889 13890 13891 13892 13893 13894 13895 13896 13897 13898 13899 13900 13901 13902 13903 13904 13905 13906 13907 13908 13909 13910 13911 13912 13913 13914

/**
 * 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;
13915
    int ret = -1;
13916 13917 13918 13919 13920 13921 13922 13923 13924 13925 13926 13927 13928 13929 13930 13931 13932 13933 13934 13935 13936 13937 13938 13939 13940 13941 13942 13943 13944 13945 13946 13947 13948 13949 13950 13951 13952 13953 13954 13955 13956 13957 13958 13959 13960 13961 13962

    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;
}


13963 13964 13965 13966 13967 13968 13969 13970 13971 13972 13973 13974 13975 13976 13977 13978 13979 13980 13981 13982 13983 13984 13985 13986 13987
/**
 * 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;
}


13988 13989 13990 13991 13992 13993 13994
char *
qemuDomainGetMachineName(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    char *ret = NULL;

13995
    if (vm->pid > 0) {
13996 13997 13998 13999 14000 14001 14002 14003 14004 14005 14006
        ret = virSystemdGetMachineNameByPID(vm->pid);
        if (!ret)
            virResetLastError();
    }

    if (!ret)
        ret = virDomainGenerateMachineName("qemu", vm->def->id, vm->def->name,
                                           virQEMUDriverIsPrivileged(driver));

    return ret;
}
14007 14008 14009 14010 14011 14012 14013 14014 14015 14016 14017 14018 14019


/* 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,
14020
                                     const virDomainDeviceInfo *info,
14021 14022 14023
                                     virQEMUCapsPtr qemuCaps,
                                     const char *devicename)
{
14024
    if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
14025 14026 14027 14028 14029 14030
        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;
14031
        } else if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW)) {
14032 14033 14034 14035 14036
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CCW address type is not supported by "
                             "this QEMU"));
            return false;
        }
14037
    } else if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390) {
14038 14039 14040 14041 14042 14043 14044 14045 14046
        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;
}
14047 14048


14049
/**
14050
 * qemuDomainPrepareDiskSourceData:
14051 14052 14053 14054 14055
 *
 * @disk: Disk config object
 * @src: source to start from
 * @cfg: qemu driver config object
 *
14056 14057 14058
 * Prepares various aspects of a storage source belonging to a disk backing
 * chain. This function should be also called for detected backing chain
 * members.
14059 14060
 */
int
14061 14062 14063 14064
qemuDomainPrepareDiskSourceData(virDomainDiskDefPtr disk,
                                virStorageSourcePtr src,
                                virQEMUDriverConfigPtr cfg,
                                virQEMUCapsPtr qemuCaps)
14065
{
14066
    /* transfer properties valid only for the top level image */
14067 14068
    if (src == disk->src)
        src->detect_zeroes = disk->detect_zeroes;
14069

14070 14071 14072 14073 14074 14075 14076
    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;
    }
14077

14078 14079 14080 14081
    /* transfer properties valid for the full chain */
    src->iomode = disk->iomode;
    src->cachemode = disk->cachemode;
    src->discard = disk->discard;
14082

14083 14084
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY)
        src->floppyimg = true;
14085 14086 14087 14088 14089

    return 0;
}


14090 14091 14092 14093 14094 14095 14096 14097 14098
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;
}


14099 14100
static int
qemuDomainPrepareStorageSourcePR(virStorageSourcePtr src,
14101 14102
                                 qemuDomainObjPrivatePtr priv,
                                 const char *parentalias)
14103 14104 14105 14106 14107
{
    if (!src->pr)
        return 0;

    if (virStoragePRDefIsManaged(src->pr)) {
14108
        VIR_FREE(src->pr->path);
14109 14110
        if (!(src->pr->path = qemuDomainGetManagedPRSocketPath(priv)))
            return -1;
14111 14112 14113 14114 14115
        if (VIR_STRDUP(src->pr->mgralias, qemuDomainGetManagedPRAlias()) < 0)
            return -1;
    } else {
        if (!(src->pr->mgralias = qemuDomainGetUnmanagedPRAlias(parentalias)))
            return -1;
14116 14117 14118 14119 14120 14121
    }

    return 0;
}


14122 14123 14124 14125 14126 14127 14128 14129 14130 14131 14132 14133
/**
 * 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)
14134
{
14135 14136 14137
    if (qemuDomainValidateStorageSource(disk->src, priv->qemuCaps) < 0)
        return -1;

14138
    if (qemuDomainPrepareDiskSourceData(disk, disk->src, cfg, priv->qemuCaps) < 0)
14139 14140
        return -1;

14141 14142 14143
    if (qemuDomainSecretStorageSourcePrepare(priv, disk->src,
                                             disk->info.alias,
                                             disk->info.alias) < 0)
14144 14145
        return -1;

14146
    if (qemuDomainPrepareStorageSourcePR(disk->src, priv, disk->info.alias) < 0)
14147
        return -1;
14148

14149 14150
    if (qemuDomainPrepareStorageSourceTLS(disk->src, cfg, disk->info.alias,
                                          priv->qemuCaps) < 0)
14151 14152
        return -1;

14153 14154 14155 14156
    return 0;
}


14157 14158 14159 14160 14161 14162 14163 14164 14165 14166 14167 14168 14169 14170 14171 14172 14173 14174 14175 14176 14177 14178 14179
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;

14180
    if (qemuDomainPrepareStorageSourcePR(src, priv, src->nodestorage) < 0)
14181 14182
        return -1;

14183
    if (qemuDomainPrepareStorageSourceTLS(src, cfg, src->nodestorage,
14184 14185 14186 14187 14188 14189 14190 14191 14192 14193 14194 14195 14196 14197 14198 14199 14200 14201 14202 14203 14204 14205 14206 14207 14208 14209 14210 14211 14212
                                          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;
}


14213 14214 14215 14216 14217 14218 14219
int
qemuDomainPrepareDiskSource(virDomainDiskDefPtr disk,
                            qemuDomainObjPrivatePtr priv,
                            virQEMUDriverConfigPtr cfg)
{
    qemuDomainPrepareDiskCachemode(disk);

14220
    /* set default format for storage pool based disks */
14221
    if (disk->src->type == VIR_STORAGE_TYPE_VOLUME &&
14222 14223 14224 14225 14226 14227 14228 14229
        disk->src->format <= VIR_STORAGE_FILE_NONE) {
        int actualType = virStorageSourceGetActualType(disk->src);

        if (actualType == VIR_STORAGE_TYPE_DIR)
            disk->src->format = VIR_STORAGE_FILE_FAT;
        else
            disk->src->format = VIR_STORAGE_FILE_RAW;
    }
14230

14231 14232 14233 14234 14235 14236 14237
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV)) {
        if (qemuDomainPrepareDiskSourceBlockdev(disk, priv, cfg) < 0)
            return -1;
    } else {
        if (qemuDomainPrepareDiskSourceLegacy(disk, priv, cfg) < 0)
            return -1;
    }
14238

14239 14240
    return 0;
}
14241 14242


14243 14244 14245 14246 14247 14248 14249 14250 14251 14252 14253 14254 14255 14256 14257 14258 14259 14260 14261 14262 14263 14264 14265 14266 14267 14268 14269 14270 14271 14272 14273 14274 14275 14276 14277 14278 14279 14280 14281 14282 14283 14284 14285 14286 14287 14288 14289 14290 14291 14292 14293 14294 14295 14296 14297 14298 14299 14300 14301 14302 14303 14304 14305 14306 14307 14308 14309 14310 14311 14312 14313 14314 14315 14316 14317
/**
 * 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;
}


14318 14319 14320 14321 14322 14323 14324 14325 14326 14327
void
qemuProcessEventFree(struct qemuProcessEvent *event)
{
    if (!event)
        return;

    switch (event->eventType) {
    case QEMU_PROCESS_EVENT_GUESTPANIC:
        qemuMonitorEventPanicInfoFree(event->data);
        break;
14328 14329 14330
    case QEMU_PROCESS_EVENT_RDMA_GID_STATUS_CHANGED:
        qemuMonitorEventRdmaGidStatusFree(event->data);
        break;
14331 14332 14333 14334 14335 14336 14337 14338
    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;
14339
    case QEMU_PROCESS_EVENT_PR_DISCONNECT:
14340 14341 14342 14343 14344
    case QEMU_PROCESS_EVENT_LAST:
        break;
    }
    VIR_FREE(event);
}
14345 14346 14347


char *
14348
qemuDomainGetManagedPRSocketPath(qemuDomainObjPrivatePtr priv)
14349 14350 14351
{
    char *ret = NULL;

14352 14353
    ignore_value(virAsprintf(&ret, "%s/%s.sock", priv->libDir,
                             qemuDomainGetManagedPRAlias()));
14354 14355 14356

    return ret;
}
14357 14358 14359 14360 14361 14362 14363 14364 14365 14366 14367 14368 14369 14370 14371 14372 14373 14374 14375 14376 14377 14378 14379 14380 14381 14382 14383


/**
 * 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;
}
14384 14385 14386 14387 14388 14389 14390 14391 14392 14393 14394 14395 14396 14397 14398 14399 14400 14401 14402 14403 14404 14405 14406 14407 14408 14409 14410 14411 14412


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
14413 14414 14415 14416 14417


/* qemuDomainIsUsingNoShutdown:
 * @priv: Domain private data
 *
J
Ján Tomko 已提交
14418 14419
 * We can receive an event when QEMU stops. If we use no-shutdown, then
 * we can watch for this event and do a soft/warm reboot.
J
John Ferlan 已提交
14420 14421 14422 14423 14424 14425 14426
 *
 * Returns: @true when -no-shutdown either should be or was added to the
 * command line.
 */
bool
qemuDomainIsUsingNoShutdown(qemuDomainObjPrivatePtr priv)
{
J
Ján Tomko 已提交
14427
    return priv->allowReboot == VIR_TRISTATE_BOOL_YES;
J
John Ferlan 已提交
14428
}
14429 14430 14431 14432 14433 14434 14435 14436 14437


bool
qemuDomainDiskIsMissingLocalOptional(virDomainDiskDefPtr disk)
{
    return disk->startupPolicy == VIR_DOMAIN_STARTUP_POLICY_OPTIONAL &&
           virStorageSourceIsLocalStorage(disk->src) && disk->src->path &&
           !virFileExists(disk->src->path);
}
14438 14439 14440 14441 14442 14443 14444 14445 14446 14447 14448 14449 14450 14451 14452 14453


int
qemuDomainNVRAMPathGenerate(virQEMUDriverConfigPtr cfg,
                            virDomainDefPtr def)
{
    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) {
        return virAsprintf(&def->os.loader->nvram, "%s/%s_VARS.fd",
                           cfg->nvramDir, def->name);
    }

    return 0;
}
14454 14455 14456 14457 14458 14459 14460 14461 14462 14463 14464 14465 14466 14467 14468 14469 14470 14471 14472 14473 14474 14475 14476 14477 14478 14479 14480 14481 14482 14483 14484 14485 14486 14487


virDomainEventSuspendedDetailType
qemuDomainPausedReasonToSuspendedEvent(virDomainPausedReason reason)
{
    switch (reason) {
    case VIR_DOMAIN_PAUSED_MIGRATION:
        return VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED;

    case VIR_DOMAIN_PAUSED_FROM_SNAPSHOT:
        return VIR_DOMAIN_EVENT_SUSPENDED_FROM_SNAPSHOT;

    case VIR_DOMAIN_PAUSED_POSTCOPY_FAILED:
        return VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY_FAILED;

    case VIR_DOMAIN_PAUSED_POSTCOPY:
        return VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY;

    case VIR_DOMAIN_PAUSED_UNKNOWN:
    case VIR_DOMAIN_PAUSED_USER:
    case VIR_DOMAIN_PAUSED_SAVE:
    case VIR_DOMAIN_PAUSED_DUMP:
    case VIR_DOMAIN_PAUSED_IOERROR:
    case VIR_DOMAIN_PAUSED_WATCHDOG:
    case VIR_DOMAIN_PAUSED_SHUTTING_DOWN:
    case VIR_DOMAIN_PAUSED_SNAPSHOT:
    case VIR_DOMAIN_PAUSED_CRASHED:
    case VIR_DOMAIN_PAUSED_STARTING_UP:
    case VIR_DOMAIN_PAUSED_LAST:
        break;
    }

    return VIR_DOMAIN_EVENT_SUSPENDED_PAUSED;
}