qemu_domain.c 397.4 KB
Newer Older
1
/*
2
 * qemu_domain.c: QEMU domain private state
3
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

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

61 62 63 64 65
#ifdef MAJOR_IN_MKDEV
# include <sys/mkdev.h>
#elif MAJOR_IN_SYSMACROS
# include <sys/sysmacros.h>
#endif
66
#include <sys/time.h>
67
#include <fcntl.h>
68 69 70
#if defined(HAVE_SYS_MOUNT_H)
# include <sys/mount.h>
#endif
71 72 73
#ifdef WITH_SELINUX
# include <selinux/selinux.h>
#endif
74

75
#include <libxml/xpathInternals.h>
76
#include "dosname.h"
77 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 86 87 88 89
VIR_ENUM_IMPL(qemuDomainJob, QEMU_JOB_LAST,
              "none",
              "query",
              "destroy",
              "suspend",
              "modify",
90
              "abort",
91
              "migration operation",
92 93 94 95
              "none",   /* async job is never stored in job.active */
              "async nested",
);

96 97 98 99 100 101
VIR_ENUM_IMPL(qemuDomainAgentJob, QEMU_AGENT_JOB_LAST,
              "none",
              "query",
              "modify",
);

102 103 104 105 106 107
VIR_ENUM_IMPL(qemuDomainAsyncJob, QEMU_ASYNC_JOB_LAST,
              "none",
              "migration out",
              "migration in",
              "save",
              "dump",
108
              "snapshot",
109
              "start",
110 111
);

112 113 114 115 116
VIR_ENUM_IMPL(qemuDomainNamespace, QEMU_DOMAIN_NS_LAST,
              "mount",
);


117 118
#define PROC_MOUNTS "/proc/mounts"
#define DEVPREFIX "/dev/"
119
#define DEV_VFIO "/dev/vfio/vfio"
120
#define DEVICE_MAPPER_CONTROL_PATH "/dev/mapper/control"
121

122

123
struct _qemuDomainLogContext {
124 125
    virObject parent;

126
    int writefd;
127
    int readfd; /* Only used if manager == NULL */
128
    off_t pos;
129
    ino_t inode; /* Only used if manager != NULL */
130
    char *path;
131
    virLogManagerPtr manager;
132 133
};

134
static virClassPtr qemuDomainLogContextClass;
135
static virClassPtr qemuDomainSaveCookieClass;
136 137

static void qemuDomainLogContextDispose(void *obj);
138
static void qemuDomainSaveCookieDispose(void *obj);
139 140

static int
141
qemuDomainOnceInit(void)
142
{
143
    if (!VIR_CLASS_NEW(qemuDomainLogContext, virClassForObject()))
144 145
        return -1;

146
    if (!VIR_CLASS_NEW(qemuDomainSaveCookie, virClassForObject()))
147 148
        return -1;

149 150 151
    return 0;
}

152
VIR_ONCE_GLOBAL_INIT(qemuDomain)
153 154 155 156 157 158 159 160 161 162 163 164 165

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
166
const char *
167
qemuDomainAsyncJobPhaseToString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
168 169 170 171 172
                                int phase ATTRIBUTE_UNUSED)
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
173 174
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
175 176
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
177
    case QEMU_ASYNC_JOB_SNAPSHOT:
178
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
179
    case QEMU_ASYNC_JOB_NONE:
M
Marc Hartmayer 已提交
180
        ATTRIBUTE_FALLTHROUGH;
181 182
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
183 184 185 186 187 188
    }

    return "none";
}

int
189
qemuDomainAsyncJobPhaseFromString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
190 191 192 193 194 195 196 197
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
198 199
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
200 201
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
202
    case QEMU_ASYNC_JOB_SNAPSHOT:
203
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
204
    case QEMU_ASYNC_JOB_NONE:
M
Marc Hartmayer 已提交
205
        ATTRIBUTE_FALLTHROUGH;
206 207
    case QEMU_ASYNC_JOB_LAST:
        break;
J
Jiri Denemark 已提交
208 209 210 211 212 213 214 215
    }

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

216

217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248
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;
}


249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
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;
        }
    }
}


265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
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);
285
    virObjectEventStateQueue(driver->domainEventState, event);
286 287 288
}


289 290 291 292 293 294 295 296
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

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

297
    if (virCondInit(&priv->job.asyncCond) < 0) {
298
        virCondDestroy(&priv->job.cond);
299 300 301
        return -1;
    }

302 303 304 305 306 307
    return 0;
}

static void
qemuDomainObjResetJob(qemuDomainObjPrivatePtr priv)
{
308
    qemuDomainJobObjPtr job = &priv->job;
309 310

    job->active = QEMU_JOB_NONE;
311
    job->owner = 0;
312
    job->ownerAPI = NULL;
J
Jiri Denemark 已提交
313
    job->started = 0;
314 315
}

316 317 318 319 320 321 322 323 324 325 326 327 328

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

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


329 330 331
static void
qemuDomainObjResetAsyncJob(qemuDomainObjPrivatePtr priv)
{
332
    qemuDomainJobObjPtr job = &priv->job;
333 334

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
335
    job->asyncOwner = 0;
336
    job->asyncOwnerAPI = NULL;
J
Jiri Denemark 已提交
337
    job->asyncStarted = 0;
J
Jiri Denemark 已提交
338
    job->phase = 0;
339
    job->mask = QEMU_JOB_DEFAULT_MASK;
340
    job->abortJob = false;
341
    job->spiceMigration = false;
342
    job->spiceMigrated = false;
343 344
    job->dumpCompleted = false;
    VIR_FREE(job->error);
J
Jiri Denemark 已提交
345
    VIR_FREE(job->current);
346 347
    qemuMigrationParamsFree(job->migParams);
    job->migParams = NULL;
348
    job->apiFlags = 0;
349 350
}

351 352
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
353
                        qemuDomainJobObjPtr job)
354 355 356 357 358
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
359
    job->owner = priv->job.owner;
360
    job->asyncJob = priv->job.asyncJob;
361
    job->asyncOwner = priv->job.asyncOwner;
J
Jiri Denemark 已提交
362
    job->phase = priv->job.phase;
363
    VIR_STEAL_PTR(job->migParams, priv->job.migParams);
364
    job->apiFlags = priv->job.apiFlags;
365 366 367 368 369

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

370 371 372
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
J
Jiri Denemark 已提交
373
    VIR_FREE(priv->job.current);
374
    VIR_FREE(priv->job.completed);
375 376
    virCondDestroy(&priv->job.cond);
    virCondDestroy(&priv->job.asyncCond);
377 378
}

379
static bool
380
qemuDomainTrackJob(qemuDomainJob job)
381 382 383 384
{
    return (QEMU_DOMAIN_TRACK_JOBS & JOB_MASK(job)) != 0;
}

385

J
Jiri Denemark 已提交
386 387 388 389 390 391 392 393 394 395 396
int
qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->started)
        return 0;

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

397 398 399 400 401 402
    if (now < jobInfo->started) {
        VIR_WARN("Async job starts in the future");
        jobInfo->started = 0;
        return 0;
    }

J
Jiri Denemark 已提交
403 404 405 406
    jobInfo->timeElapsed = now - jobInfo->started;
    return 0;
}

407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423
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;
    }

424 425
    jobInfo->stats.mig.downtime = now - jobInfo->stopped;
    jobInfo->stats.mig.downtime_set = true;
426 427 428
    return 0;
}

429 430 431 432 433 434 435 436
static virDomainJobType
qemuDomainJobStatusToType(qemuDomainJobStatus status)
{
    switch (status) {
    case QEMU_DOMAIN_JOB_STATUS_NONE:
        break;

    case QEMU_DOMAIN_JOB_STATUS_ACTIVE:
437
    case QEMU_DOMAIN_JOB_STATUS_MIGRATING:
438
    case QEMU_DOMAIN_JOB_STATUS_QEMU_COMPLETED:
439
    case QEMU_DOMAIN_JOB_STATUS_POSTCOPY:
440
    case QEMU_DOMAIN_JOB_STATUS_PAUSED:
441 442 443 444 445 446 447 448 449 450 451 452 453 454 455
        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
456 457 458 459
int
qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                        virDomainJobInfoPtr info)
{
460
    info->type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
461 462
    info->timeElapsed = jobInfo->timeElapsed;

463 464 465 466 467 468 469 470 471 472 473 474 475
    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
476

477 478 479 480 481 482
    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;

483
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
484 485 486 487 488
        info->memTotal = jobInfo->stats.dump.total;
        info->memProcessed = jobInfo->stats.dump.completed;
        info->memRemaining = info->memTotal - info->memProcessed;
        break;

489 490 491
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
        break;
    }
J
Jiri Denemark 已提交
492 493 494 495 496 497 498 499

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

    return 0;
}

500 501 502 503 504 505

static int
qemuDomainMigrationJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                                   int *type,
                                   virTypedParameterPtr *params,
                                   int *nparams)
J
Jiri Denemark 已提交
506
{
507
    qemuMonitorMigrationStats *stats = &jobInfo->stats.mig;
508
    qemuDomainMirrorStatsPtr mirrorStats = &jobInfo->mirrorStats;
J
Jiri Denemark 已提交
509 510 511
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;
512 513
    unsigned long long mirrorRemaining = mirrorStats->total -
                                         mirrorStats->transferred;
J
Jiri Denemark 已提交
514

515 516 517 518 519
    if (virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_OPERATION,
                             jobInfo->operation) < 0)
        goto error;

J
Jiri Denemark 已提交
520 521 522 523 524
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED,
                                jobInfo->timeElapsed) < 0)
        goto error;

525 526 527 528 529 530 531
    if (jobInfo->timeDeltaSet &&
        jobInfo->timeElapsed > jobInfo->timeDelta &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED_NET,
                                jobInfo->timeElapsed - jobInfo->timeDelta) < 0)
        goto error;

532
    if (stats->downtime_set &&
J
Jiri Denemark 已提交
533 534
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME,
535
                                stats->downtime) < 0)
J
Jiri Denemark 已提交
536 537
        goto error;

538
    if (stats->downtime_set &&
539
        jobInfo->timeDeltaSet &&
540
        stats->downtime > jobInfo->timeDelta &&
541 542
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME_NET,
543
                                stats->downtime - jobInfo->timeDelta) < 0)
544 545
        goto error;

546
    if (stats->setup_time_set &&
547 548
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_SETUP_TIME,
549
                                stats->setup_time) < 0)
550 551
        goto error;

J
Jiri Denemark 已提交
552 553
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_TOTAL,
554
                                stats->ram_total +
555 556
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
557 558
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_PROCESSED,
559
                                stats->ram_transferred +
560 561
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
562 563
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_REMAINING,
564
                                stats->ram_remaining +
565 566
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
567 568 569 570
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
571
                                stats->ram_total) < 0 ||
J
Jiri Denemark 已提交
572 573
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
574
                                stats->ram_transferred) < 0 ||
J
Jiri Denemark 已提交
575 576
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
577
                                stats->ram_remaining) < 0)
J
Jiri Denemark 已提交
578 579
        goto error;

580
    if (stats->ram_bps &&
581 582
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_BPS,
583
                                stats->ram_bps) < 0)
584 585
        goto error;

586
    if (stats->ram_duplicate_set) {
J
Jiri Denemark 已提交
587 588
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_CONSTANT,
589
                                    stats->ram_duplicate) < 0 ||
J
Jiri Denemark 已提交
590 591
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL,
592
                                    stats->ram_normal) < 0 ||
J
Jiri Denemark 已提交
593 594
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL_BYTES,
595
                                    stats->ram_normal_bytes) < 0)
J
Jiri Denemark 已提交
596 597 598
            goto error;
    }

599 600 601 602 603 604 605 606
    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,
                                stats->ram_iteration) < 0)
        goto error;

607 608 609 610 611 612
    if (stats->ram_page_size > 0 &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PAGE_SIZE,
                                stats->ram_page_size) < 0)
        goto error;

613 614 615 616 617
    /* 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
618 619
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_TOTAL,
620 621
                                stats->disk_total +
                                mirrorStats->total) < 0 ||
J
Jiri Denemark 已提交
622 623
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_PROCESSED,
624 625
                                stats->disk_transferred +
                                mirrorStats->transferred) < 0 ||
J
Jiri Denemark 已提交
626 627
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_REMAINING,
628 629
                                stats->disk_remaining +
                                mirrorRemaining) < 0)
J
Jiri Denemark 已提交
630 631
        goto error;

632
    if (stats->disk_bps &&
633 634
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_BPS,
635
                                stats->disk_bps) < 0)
636 637
        goto error;

638
    if (stats->xbzrle_set) {
J
Jiri Denemark 已提交
639 640
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE,
641
                                    stats->xbzrle_cache_size) < 0 ||
J
Jiri Denemark 已提交
642 643
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_BYTES,
644
                                    stats->xbzrle_bytes) < 0 ||
J
Jiri Denemark 已提交
645 646
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_PAGES,
647
                                    stats->xbzrle_pages) < 0 ||
J
Jiri Denemark 已提交
648 649
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE_MISSES,
650
                                    stats->xbzrle_cache_miss) < 0 ||
J
Jiri Denemark 已提交
651 652
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_OVERFLOW,
653
                                    stats->xbzrle_overflow) < 0)
J
Jiri Denemark 已提交
654 655 656
            goto error;
    }

657 658 659 660 661 662
    if (stats->cpu_throttle_percentage &&
        virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_AUTO_CONVERGE_THROTTLE,
                             stats->cpu_throttle_percentage) < 0)
        goto error;

663
 done:
664
    *type = qemuDomainJobStatusToType(jobInfo->status);
J
Jiri Denemark 已提交
665 666 667 668 669 670 671 672 673 674
    *params = par;
    *nparams = npar;
    return 0;

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


675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
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;
}


718 719 720 721 722 723 724 725
int
qemuDomainJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                          int *type,
                          virTypedParameterPtr *params,
                          int *nparams)
{
    switch (jobInfo->statsType) {
    case QEMU_DOMAIN_JOB_STATS_TYPE_MIGRATION:
726
    case QEMU_DOMAIN_JOB_STATS_TYPE_SAVEDUMP:
727 728
        return qemuDomainMigrationJobInfoToParams(jobInfo, type, params, nparams);

729
    case QEMU_DOMAIN_JOB_STATS_TYPE_MEMDUMP:
730 731
        return qemuDomainDumpJobInfoToParams(jobInfo, type, params, nparams);

732
    case QEMU_DOMAIN_JOB_STATS_TYPE_NONE:
733 734 735 736 737 738
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid job statistics type"));
        break;

    default:
        virReportEnumRangeError(qemuDomainJobStatsType, jobInfo->statsType);
739 740 741 742 743 744 745
        break;
    }

    return -1;
}


J
John Ferlan 已提交
746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767
/* 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:
768 769
 * @driver: qemu driver data
 * @vm: Pointer to the vm object
J
John Ferlan 已提交
770 771 772 773 774
 *
 * 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
 */
775
int
M
Martin Kletzander 已提交
776 777
qemuDomainWriteMasterKeyFile(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
J
John Ferlan 已提交
778 779 780 781
{
    char *path;
    int fd = -1;
    int ret = -1;
M
Martin Kletzander 已提交
782
    qemuDomainObjPrivatePtr priv = vm->privateData;
J
John Ferlan 已提交
783

784 785 786 787
    /* Only gets filled in if we have the capability */
    if (!priv->masterKey)
        return 0;

J
John Ferlan 已提交
788 789 790 791 792 793 794 795 796 797 798 799 800 801 802
    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;
    }

803
    if (qemuSecurityDomainSetPathLabel(driver->securityManager,
804
                                       vm->def, path, false) < 0)
M
Martin Kletzander 已提交
805 806
        goto cleanup;

J
John Ferlan 已提交
807 808 809 810 811 812 813 814 815 816
    ret = 0;

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

    return ret;
}


817 818 819 820 821 822 823 824 825
static void
qemuDomainMasterKeyFree(qemuDomainObjPrivatePtr priv)
{
    if (!priv->masterKey)
        return;

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

J
John Ferlan 已提交
826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923
/* 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 */
924
    qemuDomainMasterKeyFree(priv);
J
John Ferlan 已提交
925 926 927 928 929 930 931 932 933 934

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

    VIR_FREE(path);
}


/* qemuDomainMasterKeyCreate:
935
 * @vm: Pointer to the domain object
J
John Ferlan 已提交
936 937 938 939 940 941 942 943
 *
 * 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
944
qemuDomainMasterKeyCreate(virDomainObjPtr vm)
J
John Ferlan 已提交
945
{
M
Martin Kletzander 已提交
946 947
    qemuDomainObjPrivatePtr priv = vm->privateData;

J
John Ferlan 已提交
948 949 950 951
    /* If we don't have the capability, then do nothing. */
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

952
    if (VIR_ALLOC_N(priv->masterKey, QEMU_DOMAIN_MASTER_KEY_LEN) < 0)
953
        return -1;
J
John Ferlan 已提交
954 955
    priv->masterKeyLen = QEMU_DOMAIN_MASTER_KEY_LEN;

956
    if (virRandomBytes(priv->masterKey, priv->masterKeyLen) < 0) {
957 958 959 960
        VIR_DISPOSE_N(priv->masterKey, priv->masterKeyLen);
        return -1;
    }

J
John Ferlan 已提交
961 962 963 964
    return 0;
}


965
static void
966
qemuDomainSecretPlainClear(qemuDomainSecretPlainPtr secret)
967
{
968 969
    VIR_FREE(secret->username);
    VIR_DISPOSE_N(secret->secret, secret->secretlen);
970 971 972
}


J
John Ferlan 已提交
973
static void
974
qemuDomainSecretAESClear(qemuDomainSecretAESPtr secret,
975
                         bool keepAlias)
J
John Ferlan 已提交
976
{
977
    if (!keepAlias)
978
        VIR_FREE(secret->alias);
979

980 981 982
    VIR_FREE(secret->username);
    VIR_FREE(secret->iv);
    VIR_FREE(secret->ciphertext);
J
John Ferlan 已提交
983 984 985
}


986 987 988
static void
qemuDomainSecretInfoClear(qemuDomainSecretInfoPtr secinfo,
                          bool keepAlias)
989
{
990
    if (!secinfo)
991 992
        return;

993
    switch ((qemuDomainSecretInfoType) secinfo->type) {
J
John Ferlan 已提交
994
    case VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN:
995
        qemuDomainSecretPlainClear(&secinfo->s.plain);
J
John Ferlan 已提交
996 997
        break;

998
    case VIR_DOMAIN_SECRET_INFO_TYPE_AES:
999
        qemuDomainSecretAESClear(&secinfo->s.aes, keepAlias);
J
John Ferlan 已提交
1000 1001 1002 1003 1004
        break;

    case VIR_DOMAIN_SECRET_INFO_TYPE_LAST:
        break;
    }
1005 1006 1007 1008 1009 1010 1011 1012 1013 1014
}


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

    qemuDomainSecretInfoClear(*secinfo, false);
J
John Ferlan 已提交
1015

1016 1017 1018 1019
    VIR_FREE(*secinfo);
}


1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032
/**
 * qemuDomainSecretInfoDestroy:
 * @secinfo: object to destroy
 *
 * Removes any data unnecessary for further use, but keeps alias allocated.
 */
void
qemuDomainSecretInfoDestroy(qemuDomainSecretInfoPtr secinfo)
{
    qemuDomainSecretInfoClear(secinfo, true);
}


1033
static virClassPtr qemuDomainDiskPrivateClass;
1034
static void qemuDomainDiskPrivateDispose(void *obj);
1035 1036 1037 1038

static int
qemuDomainDiskPrivateOnceInit(void)
{
1039
    if (!VIR_CLASS_NEW(qemuDomainDiskPrivate, virClassForObject()))
1040
        return -1;
1041 1042

    return 0;
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
}

VIR_ONCE_GLOBAL_INIT(qemuDomainDiskPrivate)

static virObjectPtr
qemuDomainDiskPrivateNew(void)
{
    qemuDomainDiskPrivatePtr priv;

    if (qemuDomainDiskPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}

1061 1062 1063 1064 1065 1066
static void
qemuDomainDiskPrivateDispose(void *obj)
{
    qemuDomainDiskPrivatePtr priv = obj;

    VIR_FREE(priv->blockJobError);
1067
    virStorageSourceFree(priv->migrSource);
1068
}
1069

1070 1071 1072 1073 1074 1075
static virClassPtr qemuDomainStorageSourcePrivateClass;
static void qemuDomainStorageSourcePrivateDispose(void *obj);

static int
qemuDomainStorageSourcePrivateOnceInit(void)
{
1076
    if (!VIR_CLASS_NEW(qemuDomainStorageSourcePrivate, virClassForObject()))
1077
        return -1;
1078 1079

    return 0;
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108
}

VIR_ONCE_GLOBAL_INIT(qemuDomainStorageSourcePrivate)

virObjectPtr
qemuDomainStorageSourcePrivateNew(void)
{
    qemuDomainStorageSourcePrivatePtr priv;

    if (qemuDomainStorageSourcePrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}


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

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


1109 1110 1111 1112 1113 1114
static virClassPtr qemuDomainVcpuPrivateClass;
static void qemuDomainVcpuPrivateDispose(void *obj);

static int
qemuDomainVcpuPrivateOnceInit(void)
{
1115
    if (!VIR_CLASS_NEW(qemuDomainVcpuPrivate, virClassForObject()))
1116
        return -1;
1117 1118

    return 0;
1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138
}

VIR_ONCE_GLOBAL_INIT(qemuDomainVcpuPrivate)

static virObjectPtr
qemuDomainVcpuPrivateNew(void)
{
    qemuDomainVcpuPrivatePtr priv;

    if (qemuDomainVcpuPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}


static void
1139
qemuDomainVcpuPrivateDispose(void *obj)
1140
{
1141 1142 1143 1144
    qemuDomainVcpuPrivatePtr priv = obj;

    VIR_FREE(priv->type);
    VIR_FREE(priv->alias);
1145 1146 1147 1148
    return;
}


1149 1150
static virClassPtr qemuDomainChrSourcePrivateClass;
static void qemuDomainChrSourcePrivateDispose(void *obj);
1151 1152

static int
1153
qemuDomainChrSourcePrivateOnceInit(void)
1154
{
1155
    if (!VIR_CLASS_NEW(qemuDomainChrSourcePrivate, virClassForObject()))
1156
        return -1;
1157 1158

    return 0;
1159 1160
}

1161
VIR_ONCE_GLOBAL_INIT(qemuDomainChrSourcePrivate)
1162 1163

static virObjectPtr
1164
qemuDomainChrSourcePrivateNew(void)
1165
{
1166
    qemuDomainChrSourcePrivatePtr priv;
1167

1168
    if (qemuDomainChrSourcePrivateInitialize() < 0)
1169 1170
        return NULL;

1171
    if (!(priv = virObjectNew(qemuDomainChrSourcePrivateClass)))
1172 1173 1174 1175 1176 1177 1178
        return NULL;

    return (virObjectPtr) priv;
}


static void
1179
qemuDomainChrSourcePrivateDispose(void *obj)
1180
{
1181
    qemuDomainChrSourcePrivatePtr priv = obj;
1182 1183 1184 1185 1186

    qemuDomainSecretInfoFree(&priv->secinfo);
}


J
Ján Tomko 已提交
1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211
static virClassPtr qemuDomainVsockPrivateClass;
static void qemuDomainVsockPrivateDispose(void *obj);

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

    return 0;
}

VIR_ONCE_GLOBAL_INIT(qemuDomainVsockPrivate)

static virObjectPtr
qemuDomainVsockPrivateNew(void)
{
    qemuDomainVsockPrivatePtr priv;

    if (qemuDomainVsockPrivateInitialize() < 0)
        return NULL;

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

1212 1213
    priv->vhostfd = -1;

J
Ján Tomko 已提交
1214 1215 1216 1217 1218 1219 1220
    return (virObjectPtr) priv;
}


static void
qemuDomainVsockPrivateDispose(void *obj ATTRIBUTE_UNUSED)
{
1221 1222 1223
    qemuDomainVsockPrivatePtr priv = obj;

    VIR_FORCE_CLOSE(priv->vhostfd);
J
Ján Tomko 已提交
1224 1225 1226
}


1227 1228
/* qemuDomainSecretPlainSetup:
 * @secinfo: Pointer to secret info
J
John Ferlan 已提交
1229
 * @usageType: The virSecretUsageType
1230 1231
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1232 1233 1234 1235 1236 1237
 *
 * Taking a secinfo, fill in the plaintext information
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1238
qemuDomainSecretPlainSetup(qemuDomainSecretInfoPtr secinfo,
J
John Ferlan 已提交
1239
                           virSecretUsageType usageType,
1240 1241
                           const char *username,
                           virSecretLookupTypeDefPtr seclookupdef)
1242
{
1243 1244 1245 1246 1247 1248 1249
    virConnectPtr conn;
    int ret = -1;

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

1250
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN;
1251
    if (VIR_STRDUP(secinfo->s.plain.username, username) < 0)
1252
        goto cleanup;
1253

1254 1255 1256 1257 1258 1259 1260
    ret = virSecretGetSecretString(conn, seclookupdef, usageType,
                                   &secinfo->s.plain.secret,
                                   &secinfo->s.plain.secretlen);

 cleanup:
    virObjectUnref(conn);
    return ret;
1261 1262 1263
}


1264 1265 1266 1267
/* 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
1268
 * @usageType: The virSecretUsageType
1269 1270
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1271
 * @isLuks: True/False for is for luks (alias generation)
1272 1273 1274 1275 1276 1277
 *
 * Taking a secinfo, fill in the AES specific information using the
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
1278
qemuDomainSecretAESSetup(qemuDomainObjPrivatePtr priv,
1279 1280
                         qemuDomainSecretInfoPtr secinfo,
                         const char *srcalias,
J
John Ferlan 已提交
1281
                         virSecretUsageType usageType,
1282
                         const char *username,
1283 1284
                         virSecretLookupTypeDefPtr seclookupdef,
                         bool isLuks)
1285
{
1286
    virConnectPtr conn;
1287 1288 1289 1290 1291 1292 1293 1294
    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;

1295 1296 1297 1298
    conn = virGetConnectSecret();
    if (!conn)
        return -1;

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

1303
    if (!(secinfo->s.aes.alias = qemuDomainGetSecretAESAlias(srcalias, isLuks)))
1304
        goto cleanup;
1305

1306 1307 1308
    if (VIR_ALLOC_N(raw_iv, ivlen) < 0)
        goto cleanup;

1309
    /* Create a random initialization vector */
1310
    if (virRandomBytes(raw_iv, ivlen) < 0)
1311
        goto cleanup;
1312 1313 1314 1315 1316 1317

    /* 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
1318
    if (virSecretGetSecretString(conn, seclookupdef, usageType,
1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341
                                 &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);
1342
    virObjectUnref(conn);
1343 1344 1345 1346
    return ret;
}


1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362
/**
 * 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;
}


1363
/* qemuDomainSecretInfoNewPlain:
1364
 * @usageType: Secret usage type
1365 1366
 * @username: username
 * @lookupDef: lookup def describing secret
1367
 *
1368
 * Helper function to create a secinfo to be used for secinfo consumers. This
1369
 * sets up a 'plain' (unencrypted) secret for legacy consumers.
1370 1371 1372 1373 1374
 *
 * Returns @secinfo on success, NULL on failure. Caller is responsible
 * to eventually free @secinfo.
 */
static qemuDomainSecretInfoPtr
1375
qemuDomainSecretInfoNewPlain(virSecretUsageType usageType,
1376
                             const char *username,
1377
                             virSecretLookupTypeDefPtr lookupDef)
1378 1379 1380 1381 1382 1383
{
    qemuDomainSecretInfoPtr secinfo = NULL;

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

1384 1385 1386
    if (qemuDomainSecretPlainSetup(secinfo, usageType, username, lookupDef) < 0) {
        qemuDomainSecretInfoFree(&secinfo);
        return NULL;
1387 1388 1389 1390 1391 1392
    }

    return secinfo;
}


1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435
/* 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;
}


1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446
/**
 * 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.
 */
1447
qemuDomainSecretInfoPtr
1448
qemuDomainSecretInfoTLSNew(qemuDomainObjPrivatePtr priv,
1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461
                           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;

1462 1463 1464
    return qemuDomainSecretInfoNew(priv, srcAlias,
                                   VIR_SECRET_USAGE_TYPE_TLS, NULL,
                                   &seclookupdef, false);
1465 1466 1467
}


1468 1469 1470
/* qemuDomainSecretDiskDestroy:
 * @disk: Pointer to a disk definition
 *
1471
 * Clears unnecessary data associated with disk secret objects.
1472 1473 1474 1475
 */
void
qemuDomainSecretDiskDestroy(virDomainDiskDefPtr disk)
{
1476 1477
    qemuDomainStorageSourcePrivatePtr srcPriv;
    virStorageSourcePtr n;
1478

1479 1480 1481 1482 1483 1484
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
        if ((srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(n))) {
            qemuDomainSecretInfoDestroy(srcPriv->secinfo);
            qemuDomainSecretInfoDestroy(srcPriv->encinfo);
        }
    }
1485 1486 1487
}


1488
bool
1489
qemuDomainStorageSourceHasAuth(virStorageSourcePtr src)
1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501
{
    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;
}


1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513
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;
}


1514 1515 1516 1517 1518 1519
/**
 * 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
1520
 *
1521 1522 1523 1524
 * 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.
1525
 *
1526
 * Returns 0 on success; -1 on error while reporting an libvirt error.
1527
 */
1528
static int
1529
qemuDomainSecretStorageSourcePrepare(qemuDomainObjPrivatePtr priv,
1530 1531 1532
                                     virStorageSourcePtr src,
                                     const char *authalias,
                                     const char *encalias)
1533
{
1534
    qemuDomainStorageSourcePrivatePtr srcPriv;
1535
    bool iscsiHasPS = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_ISCSI_PASSWORD_SECRET);
1536
    bool hasAuth = qemuDomainStorageSourceHasAuth(src);
1537 1538 1539 1540
    bool hasEnc = qemuDomainDiskHasEncryptionSecret(src);

    if (!hasAuth && !hasEnc)
        return 0;
1541

1542
    if (!(src->privateData = qemuDomainStorageSourcePrivateNew()))
1543 1544
        return -1;

1545
    srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
1546

1547
    if (hasAuth) {
J
John Ferlan 已提交
1548
        virSecretUsageType usageType = VIR_SECRET_USAGE_TYPE_ISCSI;
1549

1550
        if (src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD)
J
John Ferlan 已提交
1551
            usageType = VIR_SECRET_USAGE_TYPE_CEPH;
1552

1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
        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;
1568 1569
    }

1570
    if (hasEnc) {
1571
        if (!(srcPriv->encinfo =
1572 1573 1574 1575
              qemuDomainSecretInfoNew(priv, encalias,
                                      VIR_SECRET_USAGE_TYPE_VOLUME, NULL,
                                      &src->encryption->secrets[0]->seclookupdef,
                                      true)))
1576
              return -1;
1577 1578
    }

1579 1580 1581 1582
    return 0;
}


1583 1584 1585 1586 1587 1588 1589 1590
/* qemuDomainSecretHostdevDestroy:
 * @disk: Pointer to a hostdev definition
 *
 * Clear and destroy memory associated with the secret
 */
void
qemuDomainSecretHostdevDestroy(virDomainHostdevDefPtr hostdev)
{
1591
    qemuDomainStorageSourcePrivatePtr srcPriv;
1592

1593 1594 1595
    if (virHostdevIsSCSIDevice(hostdev)) {
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1596

1597 1598 1599 1600 1601 1602
        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);
        }
    }
1603 1604 1605 1606
}


/* qemuDomainSecretHostdevPrepare:
J
John Ferlan 已提交
1607
 * @priv: pointer to domain private object
1608 1609 1610 1611 1612 1613 1614
 * @hostdev: Pointer to a hostdev definition
 *
 * For the right host device, generate the qemuDomainSecretInfo structure.
 *
 * Returns 0 on success, -1 on failure
 */
int
1615
qemuDomainSecretHostdevPrepare(qemuDomainObjPrivatePtr priv,
1616 1617
                               virDomainHostdevDefPtr hostdev)
{
1618
    if (virHostdevIsSCSIDevice(hostdev)) {
1619 1620
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
1621
        virStorageSourcePtr src = iscsisrc->src;
1622 1623

        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI &&
1624
            src->auth) {
1625 1626
            if (qemuDomainSecretStorageSourcePrepare(priv, src,
                                                     hostdev->info->alias, NULL) < 0)
1627 1628 1629 1630 1631 1632 1633 1634
                return -1;
        }
    }

    return 0;
}


1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664
/* qemuDomainSecretChardevDestroy:
 * @disk: Pointer to a chardev definition
 *
 * Clear and destroy memory associated with the secret
 */
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
1665
qemuDomainSecretChardevPrepare(virQEMUDriverConfigPtr cfg,
1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680
                               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)))
1681
            return -1;
1682

1683
        chrSourcePriv->secinfo =
1684
            qemuDomainSecretInfoTLSNew(priv, charAlias,
1685
                                       cfg->chardevTLSx509secretUUID);
1686
        VIR_FREE(charAlias);
1687 1688 1689

        if (!chrSourcePriv->secinfo)
            return -1;
1690 1691 1692 1693 1694 1695
    }

    return 0;
}


1696 1697 1698
/* qemuDomainSecretDestroy:
 * @vm: Domain object
 *
1699
 * Removes all unnecessary data which was needed to generate 'secret' objects.
1700 1701 1702 1703 1704 1705 1706 1707
 */
void
qemuDomainSecretDestroy(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++)
        qemuDomainSecretDiskDestroy(vm->def->disks[i]);
1708 1709 1710

    for (i = 0; i < vm->def->nhostdevs; i++)
        qemuDomainSecretHostdevDestroy(vm->def->hostdevs[i]);
1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736

    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);
1737 1738 1739 1740
}


/* qemuDomainSecretPrepare:
1741
 * @driver: Pointer to driver object
1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752
 * @vm: Domain object
 *
 * For any objects that may require an auth/secret setup, create a
 * qemuDomainSecretInfo and save it in the approriate place within
 * 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
1753
qemuDomainSecretPrepare(virQEMUDriverPtr driver,
1754 1755
                        virDomainObjPtr vm)
{
J
John Ferlan 已提交
1756
    qemuDomainObjPrivatePtr priv = vm->privateData;
1757
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1758
    size_t i;
1759
    int ret = -1;
1760

1761
    /* disk secrets are prepared when preparing disks */
1762

1763
    for (i = 0; i < vm->def->nhostdevs; i++) {
1764
        if (qemuDomainSecretHostdevPrepare(priv,
J
John Ferlan 已提交
1765
                                           vm->def->hostdevs[i]) < 0)
1766
            goto cleanup;
1767 1768
    }

1769
    for (i = 0; i < vm->def->nserials; i++) {
1770
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1771 1772 1773 1774 1775 1776
                                           vm->def->serials[i]->info.alias,
                                           vm->def->serials[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nparallels; i++) {
1777
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1778 1779 1780 1781 1782 1783
                                           vm->def->parallels[i]->info.alias,
                                           vm->def->parallels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nchannels; i++) {
1784
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1785 1786 1787 1788 1789 1790
                                           vm->def->channels[i]->info.alias,
                                           vm->def->channels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nconsoles; i++) {
1791
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1792 1793 1794 1795 1796 1797 1798 1799
                                           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 &&
1800
            qemuDomainSecretChardevPrepare(cfg, priv,
1801 1802 1803 1804 1805 1806
                                           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 &&
1807
            qemuDomainSecretChardevPrepare(cfg, priv,
1808 1809 1810 1811 1812 1813
                                           vm->def->rngs[i]->info.alias,
                                           vm->def->rngs[i]->source.chardev) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nredirdevs; i++) {
1814
        if (qemuDomainSecretChardevPrepare(cfg, priv,
1815 1816 1817 1818 1819 1820 1821 1822 1823 1824
                                           vm->def->redirdevs[i]->info.alias,
                                           vm->def->redirdevs[i]->source) < 0)
            goto cleanup;
    }

    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
1825 1826 1827
}


1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854
/* 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
1855 1856
qemuDomainSetPrivatePaths(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
1857
{
1858 1859
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainObjPrivatePtr priv = vm->privateData;
1860
    char *domname = virDomainDefGetShortName(vm->def);
1861
    int ret = -1;
1862

M
Martin Kletzander 已提交
1863 1864 1865
    if (!domname)
        goto cleanup;

1866
    if (!priv->libDir &&
M
Martin Kletzander 已提交
1867
        virAsprintf(&priv->libDir, "%s/domain-%s", cfg->libDir, domname) < 0)
1868
        goto cleanup;
1869

1870
    if (!priv->channelTargetDir &&
M
Martin Kletzander 已提交
1871 1872
        virAsprintf(&priv->channelTargetDir, "%s/domain-%s",
                    cfg->channelTargetDir, domname) < 0)
1873
        goto cleanup;
1874

1875 1876 1877
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
M
Martin Kletzander 已提交
1878
    VIR_FREE(domname);
1879
    return ret;
1880 1881 1882
}


1883
static void *
1884
qemuDomainObjPrivateAlloc(void *opaque)
1885 1886 1887 1888 1889 1890
{
    qemuDomainObjPrivatePtr priv;

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

1891 1892 1893
    if (qemuDomainObjInitJob(priv) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to init qemu driver mutexes"));
1894
        goto error;
1895
    }
1896

1897
    if (!(priv->devs = virChrdevAlloc()))
1898 1899
        goto error;

1900
    priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
1901
    priv->driver = opaque;
1902

1903
    return priv;
1904

1905
 error:
1906 1907
    VIR_FREE(priv);
    return NULL;
1908 1909
}

1910 1911 1912 1913 1914 1915 1916 1917 1918
/**
 * 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)
1919
{
1920 1921 1922 1923 1924 1925 1926 1927 1928
    virStringListFree(priv->qemuDevices);
    priv->qemuDevices = NULL;

    virCgroupFree(&priv->cgroup);

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

    VIR_FREE(priv->machineName);
1929

1930
    virObjectUnref(priv->qemuCaps);
1931
    priv->qemuCaps = NULL;
1932

1933
    VIR_FREE(priv->pidfile);
1934

1935 1936 1937 1938 1939 1940 1941 1942 1943 1944
    VIR_FREE(priv->libDir);
    VIR_FREE(priv->channelTargetDir);

    /* remove automatic pinning data */
    virBitmapFree(priv->autoNodeset);
    priv->autoNodeset = NULL;
    virBitmapFree(priv->autoCpuset);
    priv->autoCpuset = NULL;

    /* remove address data */
1945
    virDomainPCIAddressSetFree(priv->pciaddrs);
1946
    priv->pciaddrs = NULL;
J
Ján Tomko 已提交
1947
    virDomainUSBAddressSetFree(priv->usbaddrs);
1948 1949 1950 1951 1952 1953 1954 1955
    priv->usbaddrs = NULL;

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

    /* clear previously used namespaces */
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
1956 1957

    priv->reconnectBlockjobs = VIR_TRISTATE_BOOL_ABSENT;
1958
    priv->allowReboot = VIR_TRISTATE_BOOL_ABSENT;
1959 1960 1961

    virBitmapFree(priv->migrationCaps);
    priv->migrationCaps = NULL;
1962 1963 1964

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
1965 1966 1967 1968 1969 1970 1971 1972 1973 1974
}


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

    qemuDomainObjPrivateDataClear(priv);

1975
    virDomainChrSourceDefFree(priv->monConfig);
1976
    qemuDomainObjFreeJob(priv);
1977
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
1978
    VIR_FREE(priv->origname);
1979

1980
    virChrdevFree(priv->devs);
1981

1982 1983
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
1984
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
1985 1986
        qemuMonitorClose(priv->mon);
    }
D
Daniel P. Berrange 已提交
1987 1988 1989 1990
    if (priv->agent) {
        VIR_ERROR(_("Unexpected QEMU agent still active during domain deletion"));
        qemuAgentClose(priv->agent);
    }
1991
    VIR_FREE(priv->cleanupCallbacks);
1992 1993

    qemuDomainSecretInfoFree(&priv->migSecinfo);
1994
    qemuDomainMasterKeyFree(priv);
1995

1996 1997 1998 1999
    VIR_FREE(priv);
}


2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
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;
}


2021 2022 2023 2024
static int
qemuStorageSourcePrivateDataParse(xmlXPathContextPtr ctxt,
                                  virStorageSourcePtr src)
{
2025 2026 2027 2028 2029
    qemuDomainStorageSourcePrivatePtr priv;
    char *authalias = NULL;
    char *encalias = NULL;
    int ret = -1;

2030 2031
    src->nodestorage = virXPathString("string(./nodenames/nodename[@type='storage']/@name)", ctxt);
    src->nodeformat = virXPathString("string(./nodenames/nodename[@type='format']/@name)", ctxt);
2032
    src->tlsAlias = virXPathString("string(./objects/TLSx509/@alias)", ctxt);
2033

2034 2035 2036
    if (src->pr)
        src->pr->mgralias = virXPathString("string(./reservations/@mgralias)", ctxt);

2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053
    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;
    }

2054
    if (virStorageSourcePrivateDataParseRelPath(ctxt, src) < 0)
2055
        goto cleanup;
2056

2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078
    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);
2079 2080 2081 2082 2083 2084 2085
}


static int
qemuStorageSourcePrivateDataFormat(virStorageSourcePtr src,
                                   virBufferPtr buf)
{
2086 2087 2088 2089
    virBuffer tmp = VIR_BUFFER_INITIALIZER;
    qemuDomainStorageSourcePrivatePtr srcPriv = QEMU_DOMAIN_STORAGE_SOURCE_PRIVATE(src);
    int ret = -1;

2090 2091 2092 2093 2094 2095 2096 2097 2098
    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");
    }

2099 2100 2101
    if (src->pr)
        virBufferAsprintf(buf, "<reservations mgralias='%s'/>\n", src->pr->mgralias);

2102
    if (virStorageSourcePrivateDataFormatRelPath(src, buf) < 0)
2103
        goto cleanup;
2104

2105 2106 2107 2108 2109 2110 2111
    virBufferSetChildIndent(&tmp, buf);

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

2112 2113 2114
    if (src->tlsAlias)
        virBufferAsprintf(&tmp, "<TLSx509 alias='%s'/>\n", src->tlsAlias);

2115 2116 2117 2118 2119 2120 2121 2122
    if (virXMLFormatElement(buf, "objects", NULL, &tmp) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virBufferFreeAndReset(&tmp);
    return ret;
2123 2124 2125
}


2126 2127
static void
qemuDomainObjPrivateXMLFormatVcpus(virBufferPtr buf,
2128
                                   virDomainDefPtr def)
2129 2130
{
    size_t i;
2131 2132 2133
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    pid_t tid;
2134 2135 2136 2137

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

2138 2139 2140 2141 2142 2143 2144 2145 2146
    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);
    }
2147 2148 2149 2150 2151 2152

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


2153
static int
2154
qemuDomainObjPrivateXMLFormatAutomaticPlacement(virBufferPtr buf,
2155
                                                qemuDomainObjPrivatePtr priv)
2156 2157
{
    char *nodeset = NULL;
2158
    char *cpuset = NULL;
2159 2160
    int ret = -1;

2161
    if (!priv->autoNodeset && !priv->autoCpuset)
2162 2163
        return 0;

2164 2165
    if (priv->autoNodeset &&
        !((nodeset = virBitmapFormat(priv->autoNodeset))))
2166 2167
        goto cleanup;

2168 2169 2170 2171 2172 2173 2174 2175
    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");
2176 2177 2178 2179 2180

    ret = 0;

 cleanup:
    VIR_FREE(nodeset);
2181
    VIR_FREE(cpuset);
2182 2183 2184 2185
    return ret;
}


2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199
static int
qemuDomainObjPrivateXMLFormatBlockjobs(virBufferPtr buf,
                                       virDomainObjPtr vm)
{
    virBuffer attrBuf = VIR_BUFFER_INITIALIZER;
    bool bj = qemuDomainHasBlockjob(vm, false);

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

    return virXMLFormatElement(buf, "blockjobs", &attrBuf, NULL);
}


2200
void
2201 2202 2203 2204 2205 2206 2207 2208 2209
qemuDomainObjPrivateXMLFormatAllowReboot(virBufferPtr buf,
                                         virTristateBool allowReboot)
{
    virBufferAsprintf(buf, "<allowReboot value='%s'/>\n",
                      virTristateBoolTypeToString(allowReboot));

}


2210 2211 2212 2213 2214 2215 2216 2217 2218
static void
qemuDomainObjPrivateXMLFormatPR(virBufferPtr buf,
                                qemuDomainObjPrivatePtr priv)
{
    if (priv->prDaemonRunning)
        virBufferAddLit(buf, "<prDaemon/>\n");
}


2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259
static int
qemuDomainObjPrivateXMLFormatNBDMigrationSource(virBufferPtr buf,
                                                virStorageSourcePtr src)
{
    virBuffer attrBuf = VIR_BUFFER_INITIALIZER;
    virBuffer childBuf = VIR_BUFFER_INITIALIZER;
    virBuffer privateDataBuf = VIR_BUFFER_INITIALIZER;
    int ret = -1;

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

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

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

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

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

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

    ret = 0;

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

    return ret;
}


static int
2260 2261 2262
qemuDomainObjPrivateXMLFormatNBDMigration(virBufferPtr buf,
                                          virDomainObjPtr vm)
{
2263 2264
    virBuffer attrBuf = VIR_BUFFER_INITIALIZER;
    virBuffer childBuf = VIR_BUFFER_INITIALIZER;
2265 2266 2267
    size_t i;
    virDomainDiskDefPtr disk;
    qemuDomainDiskPrivatePtr diskPriv;
2268
    int ret = -1;
2269 2270 2271 2272

    for (i = 0; i < vm->def->ndisks; i++) {
        disk = vm->def->disks[i];
        diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
2273 2274 2275 2276

        virBufferSetChildIndent(&childBuf, buf);

        virBufferAsprintf(&attrBuf, " dev='%s' migrating='%s'",
2277
                          disk->dst, diskPriv->migrating ? "yes" : "no");
2278 2279 2280 2281 2282 2283 2284 2285

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

        if (virXMLFormatElement(buf, "disk", &attrBuf, &childBuf) < 0)
            goto cleanup;
2286
    }
2287 2288 2289 2290 2291 2292 2293 2294

    ret = 0;

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

    return ret;
2295 2296 2297
}


2298
static int
2299 2300 2301 2302
qemuDomainObjPrivateXMLFormatJob(virBufferPtr buf,
                                 virDomainObjPtr vm,
                                 qemuDomainObjPrivatePtr priv)
{
2303 2304
    virBuffer attrBuf = VIR_BUFFER_INITIALIZER;
    virBuffer childBuf = VIR_BUFFER_INITIALIZER;
2305
    qemuDomainJob job = priv->job.active;
2306
    int ret = -1;
2307 2308

    if (!qemuDomainTrackJob(job))
2309
        job = QEMU_JOB_NONE;
2310

2311 2312
    if (job == QEMU_JOB_NONE &&
        priv->job.asyncJob == QEMU_ASYNC_JOB_NONE)
2313 2314 2315
        return 0;

    virBufferSetChildIndent(&childBuf, buf);
2316

2317
    virBufferAsprintf(&attrBuf, " type='%s' async='%s'",
2318 2319
                      qemuDomainJobTypeToString(job),
                      qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
2320

2321
    if (priv->job.phase) {
2322
        virBufferAsprintf(&attrBuf, " phase='%s'",
2323 2324 2325
                          qemuDomainAsyncJobPhaseToString(priv->job.asyncJob,
                                                          priv->job.phase));
    }
2326

2327 2328 2329
    if (priv->job.asyncJob != QEMU_ASYNC_JOB_NONE)
        virBufferAsprintf(&attrBuf, " flags='0x%lx'", priv->job.apiFlags);

2330 2331 2332
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_OUT &&
        qemuDomainObjPrivateXMLFormatNBDMigration(&childBuf, vm) < 0)
        goto cleanup;
2333

2334 2335 2336
    if (priv->job.migParams)
        qemuMigrationParamsFormat(&childBuf, priv->job.migParams);

2337 2338 2339 2340 2341 2342 2343 2344 2345 2346
    if (virXMLFormatElement(buf, "job", &attrBuf, &childBuf) < 0)
        goto cleanup;

    ret = 0;

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

    return ret;
2347 2348 2349
}


2350
static int
2351 2352
qemuDomainObjPrivateXMLFormat(virBufferPtr buf,
                              virDomainObjPtr vm)
2353
{
2354
    qemuDomainObjPrivatePtr priv = vm->privateData;
2355 2356 2357 2358
    const char *monitorpath;

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
2359
        switch (priv->monConfig->type) {
2360
        case VIR_DOMAIN_CHR_TYPE_UNIX:
2361
            monitorpath = priv->monConfig->data.nix.path;
2362 2363 2364
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
2365
            monitorpath = priv->monConfig->data.file.path;
2366 2367 2368
            break;
        }

2369
        virBufferEscapeString(buf, "<monitor path='%s'", monitorpath);
2370 2371
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
2372
        virBufferAsprintf(buf, " type='%s'/>\n",
2373
                          virDomainChrTypeToString(priv->monConfig->type));
2374 2375
    }

2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386
    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");
    }

2387
    qemuDomainObjPrivateXMLFormatVcpus(buf, vm->def);
2388

2389
    if (priv->qemuCaps) {
2390
        size_t i;
2391 2392
        virBufferAddLit(buf, "<qemuCaps>\n");
        virBufferAdjustIndent(buf, 2);
2393
        for (i = 0; i < QEMU_CAPS_LAST; i++) {
2394
            if (virQEMUCapsGet(priv->qemuCaps, i)) {
2395
                virBufferAsprintf(buf, "<flag name='%s'/>\n",
2396
                                  virQEMUCapsTypeToString(i));
2397 2398
            }
        }
2399 2400
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</qemuCaps>\n");
2401 2402
    }

2403
    if (priv->lockState)
2404
        virBufferAsprintf(buf, "<lockstate>%s</lockstate>\n", priv->lockState);
2405

2406 2407
    if (qemuDomainObjPrivateXMLFormatJob(buf, vm, priv) < 0)
        return -1;
2408

2409
    if (priv->fakeReboot)
2410
        virBufferAddLit(buf, "<fakereboot/>\n");
2411

2412 2413
    if (priv->qemuDevices && *priv->qemuDevices) {
        char **tmp = priv->qemuDevices;
2414 2415
        virBufferAddLit(buf, "<devices>\n");
        virBufferAdjustIndent(buf, 2);
2416
        while (*tmp) {
2417
            virBufferAsprintf(buf, "<device alias='%s'/>\n", *tmp);
2418 2419
            tmp++;
        }
2420 2421
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</devices>\n");
2422 2423
    }

2424
    if (qemuDomainObjPrivateXMLFormatAutomaticPlacement(buf, priv) < 0)
2425
        return -1;
2426

2427 2428 2429 2430 2431
    /* Various per-domain paths */
    virBufferEscapeString(buf, "<libDir path='%s'/>\n", priv->libDir);
    virBufferEscapeString(buf, "<channelTargetDir path='%s'/>\n",
                          priv->channelTargetDir);

2432
    virCPUDefFormatBufFull(buf, priv->origCPU, NULL);
2433

2434
    if (priv->chardevStdioLogd)
2435
        virBufferAddLit(buf, "<chardevStdioLogd/>\n");
2436

2437 2438
    qemuDomainObjPrivateXMLFormatAllowReboot(buf, priv->allowReboot);

2439 2440
    qemuDomainObjPrivateXMLFormatPR(buf, priv);

2441 2442 2443
    if (qemuDomainObjPrivateXMLFormatBlockjobs(buf, vm) < 0)
        return -1;

2444 2445 2446
    return 0;
}

2447 2448 2449 2450

static int
qemuDomainObjPrivateXMLParseVcpu(xmlNodePtr node,
                                 unsigned int idx,
2451
                                 virDomainDefPtr def)
2452
{
2453
    virDomainVcpuDefPtr vcpu;
2454
    char *idstr;
2455
    char *pidstr;
2456
    unsigned int tmp;
2457 2458
    int ret = -1;

2459 2460
    idstr = virXMLPropString(node, "id");

2461 2462
    if (idstr &&
        (virStrToLong_uip(idstr, NULL, 10, &idx) < 0)) {
2463
        virReportError(VIR_ERR_INTERNAL_ERROR,
2464 2465 2466 2467 2468 2469
                       _("cannot parse vcpu index '%s'"), idstr);
        goto cleanup;
    }
    if (!(vcpu = virDomainDefGetVcpu(def, idx))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("invalid vcpu index '%u'"), idx);
2470
        goto cleanup;
2471 2472
    }

2473 2474 2475
    if (!(pidstr = virXMLPropString(node, "pid")))
        goto cleanup;

2476
    if (virStrToLong_uip(pidstr, NULL, 10, &tmp) < 0)
2477 2478
        goto cleanup;

2479 2480
    QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid = tmp;

2481 2482 2483
    ret = 0;

 cleanup:
2484
    VIR_FREE(idstr);
2485 2486 2487 2488 2489
    VIR_FREE(pidstr);
    return ret;
}


2490 2491 2492 2493 2494 2495 2496
static int
qemuDomainObjPrivateXMLParseAutomaticPlacement(xmlXPathContextPtr ctxt,
                                               qemuDomainObjPrivatePtr priv,
                                               virQEMUDriverPtr driver)
{
    virCapsPtr caps = NULL;
    char *nodeset;
2497
    char *cpuset;
2498 2499
    int nodesetSize = 0;
    size_t i;
2500 2501 2502
    int ret = -1;

    nodeset = virXPathString("string(./numad/@nodeset)", ctxt);
2503
    cpuset = virXPathString("string(./numad/@cpuset)", ctxt);
2504

2505
    if (!nodeset && !cpuset)
2506 2507 2508 2509 2510
        return 0;

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

2511 2512 2513 2514 2515 2516
    /* 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);

2517
    if (nodeset &&
2518
        virBitmapParse(nodeset, &priv->autoNodeset, nodesetSize) < 0)
2519 2520
        goto cleanup;

2521 2522 2523 2524 2525 2526 2527 2528 2529 2530
    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;
    }
2531 2532 2533 2534 2535 2536

    ret = 0;

 cleanup:
    virObjectUnref(caps);
    VIR_FREE(nodeset);
2537
    VIR_FREE(cpuset);
2538 2539 2540 2541 2542

    return ret;
}


2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558
static int
qemuDomainObjPrivateXMLParseBlockjobs(qemuDomainObjPrivatePtr priv,
                                      xmlXPathContextPtr ctxt)
{
    char *active;
    int tmp;

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

    VIR_FREE(active);
    return 0;
}


2559
int
2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583
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;
}


2584 2585 2586 2587 2588 2589 2590 2591
static void
qemuDomainObjPrivateXMLParsePR(xmlXPathContextPtr ctxt,
                               bool *prDaemonRunning)
{
    *prDaemonRunning = virXPathBoolean("boolean(./prDaemon)", ctxt) > 0;
}


2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657
static int
qemuDomainObjPrivateXMLParseJobNBDSource(xmlNodePtr node,
                                         xmlXPathContextPtr ctxt,
                                         virDomainDiskDefPtr disk)
{
    xmlNodePtr savedNode = ctxt->node;
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
    virStorageSourcePtr migrSource = NULL;
    char *format = NULL;
    char *type = NULL;
    int ret = -1;

    ctxt->node = node;

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

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

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

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

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

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

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

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

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

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


2658 2659 2660 2661 2662 2663
static int
qemuDomainObjPrivateXMLParseJobNBD(virDomainObjPtr vm,
                                   qemuDomainObjPrivatePtr priv,
                                   xmlXPathContextPtr ctxt)
{
    xmlNodePtr *nodes = NULL;
2664
    char *dst = NULL;
2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680
    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;

2681 2682
            if ((dst = virXMLPropString(nodes[i], "dev")) &&
                (disk = virDomainDiskByName(vm->def, dst, false))) {
2683
                QEMU_DOMAIN_DISK_PRIVATE(disk)->migrating = true;
2684 2685 2686 2687 2688 2689

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

2690 2691 2692 2693 2694 2695 2696 2697
            VIR_FREE(dst);
        }
    }

    ret = 0;

 cleanup:
    VIR_FREE(nodes);
2698
    VIR_FREE(dst);
2699 2700 2701 2702
    return ret;
}


2703 2704 2705 2706 2707
static int
qemuDomainObjPrivateXMLParseJob(virDomainObjPtr vm,
                                qemuDomainObjPrivatePtr priv,
                                xmlXPathContextPtr ctxt)
{
2708
    xmlNodePtr savedNode = ctxt->node;
2709 2710 2711
    char *tmp = NULL;
    int ret = -1;

2712 2713 2714 2715 2716 2717
    if (!(ctxt->node = virXPathNode("./job[1]", ctxt))) {
        ret = 0;
        goto cleanup;
    }

    if ((tmp = virXPathString("string(@type)", ctxt))) {
2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728
        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;
    }

2729
    if ((tmp = virXPathString("string(@async)", ctxt))) {
2730 2731 2732 2733 2734 2735 2736 2737 2738 2739
        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;

2740
        if ((tmp = virXPathString("string(@phase)", ctxt))) {
2741 2742 2743 2744 2745 2746 2747 2748 2749 2750
            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);
        }
    }

2751 2752 2753 2754 2755
    if (virXPathULongHex("string(@flags)", ctxt, &priv->job.apiFlags) == -2) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("Invalid job flags"));
        goto cleanup;
    }

2756
    if (qemuDomainObjPrivateXMLParseJobNBD(vm, priv, ctxt) < 0)
2757 2758
        goto cleanup;

2759 2760 2761
    if (qemuMigrationParamsParse(ctxt, &priv->job.migParams) < 0)
        goto cleanup;

2762 2763 2764
    ret = 0;

 cleanup:
2765
    ctxt->node = savedNode;
2766 2767 2768 2769 2770
    VIR_FREE(tmp);
    return ret;
}


2771
static int
2772
qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt,
2773
                             virDomainObjPtr vm,
2774
                             virDomainDefParserConfigPtr config)
2775
{
2776
    qemuDomainObjPrivatePtr priv = vm->privateData;
2777
    virQEMUDriverPtr driver = config->priv;
2778
    char *monitorpath;
2779
    char *tmp = NULL;
2780 2781
    int n;
    size_t i;
2782
    xmlNodePtr *nodes = NULL;
2783
    xmlNodePtr node = NULL;
2784
    virQEMUCapsPtr qemuCaps = NULL;
2785

2786
    if (!(priv->monConfig = virDomainChrSourceDefNew(NULL)))
2787 2788 2789 2790
        goto error;

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
2791 2792
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("no monitor path"));
2793 2794 2795 2796 2797
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
2798
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
2799
    else
2800
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
2801 2802
    VIR_FREE(tmp);

E
Eric Blake 已提交
2803 2804
    priv->monJSON = virXPathBoolean("count(./monitor[@json = '1']) > 0",
                                    ctxt) > 0;
2805

2806
    switch (priv->monConfig->type) {
2807
    case VIR_DOMAIN_CHR_TYPE_PTY:
2808
        priv->monConfig->data.file.path = monitorpath;
2809 2810
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
2811
        priv->monConfig->data.nix.path = monitorpath;
2812 2813 2814
        break;
    default:
        VIR_FREE(monitorpath);
2815 2816 2817
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported monitor type '%s'"),
                       virDomainChrTypeToString(priv->monConfig->type));
2818 2819 2820
        goto error;
    }

2821 2822 2823 2824
    if ((node = virXPathNode("./namespaces", ctxt))) {
        xmlNodePtr next;

        for (next = node->children; next; next = next->next) {
2825
            int ns = qemuDomainNamespaceTypeFromString((const char *)next->name);
2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844

            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;
    }

2845
    if ((n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes)) < 0)
2846 2847
        goto error;

2848
    for (i = 0; i < n; i++) {
2849
        if (qemuDomainObjPrivateXMLParseVcpu(nodes[i], i, vm->def) < 0)
2850
            goto error;
2851
    }
2852
    VIR_FREE(nodes);
2853

2854
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
2855 2856
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse qemu capabilities flags"));
2857 2858 2859
        goto error;
    }
    if (n > 0) {
2860
        if (!(qemuCaps = virQEMUCapsNew()))
2861 2862
            goto error;

2863
        for (i = 0; i < n; i++) {
2864 2865
            char *str = virXMLPropString(nodes[i], "name");
            if (str) {
2866
                int flag = virQEMUCapsTypeFromString(str);
2867
                if (flag < 0) {
2868 2869
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Unknown qemu capabilities flag %s"), str);
2870
                    VIR_FREE(str);
2871 2872
                    goto error;
                }
2873
                VIR_FREE(str);
2874
                virQEMUCapsSet(qemuCaps, flag);
2875 2876 2877
            }
        }

2878
        priv->qemuCaps = qemuCaps;
2879
        qemuCaps = NULL;
2880 2881 2882
    }
    VIR_FREE(nodes);

2883
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
2884

2885
    if (qemuDomainObjPrivateXMLParseJob(vm, priv, ctxt) < 0)
2886 2887
        goto error;

2888 2889
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910
    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);

2911
    if (qemuDomainObjPrivateXMLParseAutomaticPlacement(ctxt, priv, driver) < 0)
2912 2913
        goto error;

2914 2915 2916 2917 2918 2919 2920 2921 2922
    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;

2923 2924 2925
    if (virCPUDefParseXML(ctxt, "./cpu", VIR_CPU_TYPE_GUEST, &priv->origCPU) < 0)
        goto error;

2926 2927 2928
    priv->chardevStdioLogd = virXPathBoolean("boolean(./chardevStdioLogd)",
                                             ctxt) == 1;

2929 2930
    qemuDomainObjPrivateXMLParseAllowReboot(ctxt, &priv->allowReboot);

2931 2932
    qemuDomainObjPrivateXMLParsePR(ctxt, &priv->prDaemonRunning);

2933 2934 2935
    if (qemuDomainObjPrivateXMLParseBlockjobs(priv, ctxt) < 0)
        goto error;

2936 2937
    return 0;

2938
 error:
2939
    VIR_FREE(nodes);
2940
    VIR_FREE(tmp);
2941 2942 2943 2944
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
    virDomainChrSourceDefFree(priv->monConfig);
    priv->monConfig = NULL;
2945
    virStringListFree(priv->qemuDevices);
2946
    priv->qemuDevices = NULL;
2947
    virObjectUnref(qemuCaps);
2948 2949 2950 2951
    return -1;
}


2952 2953 2954 2955 2956 2957 2958 2959 2960
static void *
qemuDomainObjPrivateXMLGetParseOpaque(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return priv->qemuCaps;
}


2961 2962 2963
virDomainXMLPrivateDataCallbacks virQEMUDriverPrivateDataCallbacks = {
    .alloc = qemuDomainObjPrivateAlloc,
    .free = qemuDomainObjPrivateFree,
2964
    .diskNew = qemuDomainDiskPrivateNew,
2965
    .vcpuNew = qemuDomainVcpuPrivateNew,
2966
    .chrSourceNew = qemuDomainChrSourcePrivateNew,
J
Ján Tomko 已提交
2967
    .vsockNew = qemuDomainVsockPrivateNew,
2968 2969
    .parse = qemuDomainObjPrivateXMLParse,
    .format = qemuDomainObjPrivateXMLFormat,
2970
    .getParseOpaque = qemuDomainObjPrivateXMLGetParseOpaque,
2971 2972
    .storageParse = qemuStorageSourcePrivateDataParse,
    .storageFormat = qemuStorageSourcePrivateDataFormat,
2973 2974 2975
};


2976 2977 2978 2979 2980
static void
qemuDomainDefNamespaceFree(void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;

2981
    qemuDomainCmdlineDefFree(cmd);
2982 2983 2984
}

static int
P
Philipp Hahn 已提交
2985 2986
qemuDomainDefNamespaceParse(xmlDocPtr xml ATTRIBUTE_UNUSED,
                            xmlNodePtr root ATTRIBUTE_UNUSED,
2987 2988 2989 2990
                            xmlXPathContextPtr ctxt,
                            void **data)
{
    qemuDomainCmdlineDefPtr cmd = NULL;
P
Philipp Hahn 已提交
2991
    bool uses_qemu_ns = false;
2992
    xmlNodePtr *nodes = NULL;
2993 2994
    int n;
    size_t i;
2995

P
Philipp Hahn 已提交
2996
    if (xmlXPathRegisterNs(ctxt, BAD_CAST "qemu", BAD_CAST QEMU_NAMESPACE_HREF) < 0) {
2997 2998 2999
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to register xml namespace '%s'"),
                       QEMU_NAMESPACE_HREF);
3000 3001 3002
        return -1;
    }

3003
    if (VIR_ALLOC(cmd) < 0)
3004 3005 3006 3007 3008 3009
        return -1;

    /* first handle the extra command-line arguments */
    n = virXPathNodeSet("./qemu:commandline/qemu:arg", ctxt, &nodes);
    if (n < 0)
        goto error;
P
Philipp Hahn 已提交
3010
    uses_qemu_ns |= n > 0;
3011 3012

    if (n && VIR_ALLOC_N(cmd->args, n) < 0)
3013
        goto error;
3014 3015 3016 3017

    for (i = 0; i < n; i++) {
        cmd->args[cmd->num_args] = virXMLPropString(nodes[i], "value");
        if (cmd->args[cmd->num_args] == NULL) {
3018 3019
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu command-line argument specified"));
3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030
            goto error;
        }
        cmd->num_args++;
    }

    VIR_FREE(nodes);

    /* now handle the extra environment variables */
    n = virXPathNodeSet("./qemu:commandline/qemu:env", ctxt, &nodes);
    if (n < 0)
        goto error;
P
Philipp Hahn 已提交
3031
    uses_qemu_ns |= n > 0;
3032 3033

    if (n && VIR_ALLOC_N(cmd->env_name, n) < 0)
3034
        goto error;
3035 3036

    if (n && VIR_ALLOC_N(cmd->env_value, n) < 0)
3037
        goto error;
3038 3039 3040 3041 3042 3043

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

        tmp = virXMLPropString(nodes[i], "name");
        if (tmp == NULL) {
3044 3045
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu environment name specified"));
3046 3047 3048
            goto error;
        }
        if (tmp[0] == '\0') {
3049 3050
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Empty qemu environment name specified"));
3051 3052 3053
            goto error;
        }
        if (!c_isalpha(tmp[0]) && tmp[0] != '_') {
3054 3055
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must begin with a letter or underscore"));
3056 3057 3058
            goto error;
        }
        if (strspn(tmp, "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789_") != strlen(tmp)) {
3059 3060
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must contain only alphanumerics and underscore"));
3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072
            goto error;
        }

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

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

    VIR_FREE(nodes);

P
Philipp Hahn 已提交
3073 3074 3075 3076
    if (uses_qemu_ns)
        *data = cmd;
    else
        VIR_FREE(cmd);
3077 3078 3079

    return 0;

3080
 error:
3081 3082 3083 3084 3085 3086 3087 3088 3089 3090
    VIR_FREE(nodes);
    qemuDomainDefNamespaceFree(cmd);
    return -1;
}

static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;
3091
    size_t i;
3092 3093 3094 3095

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

3096 3097 3098
    virBufferAddLit(buf, "<qemu:commandline>\n");
    virBufferAdjustIndent(buf, 2);

3099
    for (i = 0; i < cmd->num_args; i++)
3100
        virBufferEscapeString(buf, "<qemu:arg value='%s'/>\n",
3101 3102
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
3103
        virBufferAsprintf(buf, "<qemu:env name='%s'", cmd->env_name[i]);
3104 3105 3106 3107 3108
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }

3109 3110
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:commandline>\n");
3111 3112 3113 3114 3115 3116 3117 3118 3119 3120
    return 0;
}

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


3121 3122 3123 3124 3125 3126
virDomainXMLNamespace virQEMUDriverDomainXMLNamespace = {
    .parse = qemuDomainDefNamespaceParse,
    .free = qemuDomainDefNamespaceFree,
    .format = qemuDomainDefNamespaceFormatXML,
    .href = qemuDomainDefNamespaceHref,
};
3127

3128

P
Pavel Hrdina 已提交
3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147
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;
}


3148
static int
3149 3150
qemuDomainDefAddDefaultDevices(virDomainDefPtr def,
                               virQEMUCapsPtr qemuCaps)
3151
{
3152
    bool addDefaultUSB = true;
3153
    int usbModel = -1; /* "default for machinetype" */
3154
    int pciRoot;       /* index within def->controllers */
3155
    bool addImplicitSATA = false;
3156
    bool addPCIRoot = false;
L
Laine Stump 已提交
3157
    bool addPCIeRoot = false;
3158
    bool addDefaultMemballoon = true;
3159 3160
    bool addDefaultUSBKBD = false;
    bool addDefaultUSBMouse = false;
3161
    bool addPanicDevice = false;
3162
    int ret = -1;
3163

P
Pavel Hrdina 已提交
3164 3165 3166 3167
    /* add implicit input devices */
    if (qemuDomainDefAddImplicitInputDevice(def) < 0)
        goto cleanup;

3168 3169 3170 3171
    /* 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
3172
        if (STREQ(def->os.machine, "isapc")) {
3173
            addDefaultUSB = false;
3174
            break;
3175
        }
3176
        if (qemuDomainIsQ35(def)) {
3177 3178
            addPCIeRoot = true;
            addImplicitSATA = true;
3179

3180 3181 3182
            /* 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.
3183
             */
3184 3185 3186
            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))
3187 3188
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
            else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_USB_EHCI1))
3189 3190 3191
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_ICH9_EHCI1;
            else
                addDefaultUSB = false;
3192
            break;
L
Laine Stump 已提交
3193
        }
3194
        if (qemuDomainIsI440FX(def))
3195
            addPCIRoot = true;
3196 3197
        break;

3198
    case VIR_ARCH_ARMV7L:
3199
    case VIR_ARCH_AARCH64:
3200 3201
        addDefaultUSB = false;
        addDefaultMemballoon = false;
3202
        if (qemuDomainIsVirt(def))
3203 3204
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
        break;
3205

3206
    case VIR_ARCH_PPC64:
3207
    case VIR_ARCH_PPC64LE:
3208 3209 3210
        addPCIRoot = true;
        addDefaultUSBKBD = true;
        addDefaultUSBMouse = true;
3211 3212 3213
        /* For pSeries guests, the firmware provides the same
         * functionality as the pvpanic device, so automatically
         * add the definition if not already present */
3214
        if (qemuDomainIsPSeries(def))
3215
            addPanicDevice = true;
3216 3217
        break;

3218 3219 3220 3221 3222 3223 3224
    case VIR_ARCH_ALPHA:
    case VIR_ARCH_PPC:
    case VIR_ARCH_PPCEMB:
    case VIR_ARCH_SH4:
    case VIR_ARCH_SH4EB:
        addPCIRoot = true;
        break;
3225

3226 3227 3228
    case VIR_ARCH_S390:
    case VIR_ARCH_S390X:
        addDefaultUSB = false;
3229
        addPanicDevice = true;
3230
        break;
3231 3232 3233 3234 3235 3236

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

3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257
    case VIR_ARCH_ARMV6L:
    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:
3258 3259 3260 3261
    default:
        break;
    }

3262
    if (addDefaultUSB &&
3263 3264
        virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_USB, 0) < 0 &&
        virDomainDefAddUSBController(def, 0, usbModel) < 0)
3265
        goto cleanup;
3266

3267 3268 3269
    if (addImplicitSATA &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_SATA, 0, -1) < 0)
3270
        goto cleanup;
3271

3272 3273
    pciRoot = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

3274 3275 3276
    /* NB: any machine that sets addPCIRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
     */
3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291
    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;
        }
    }
3292

3293 3294 3295
    /* 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
3296 3297 3298
     *
     * NB: any machine that sets addPCIeRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
3299 3300
     */
    if (addPCIeRoot) {
3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313
        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;
        }
3314
    }
3315

3316
    if (addDefaultMemballoon && !def->memballoon) {
3317 3318
        virDomainMemballoonDefPtr memballoon;
        if (VIR_ALLOC(memballoon) < 0)
3319
            goto cleanup;
3320 3321 3322 3323 3324

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

3325 3326 3327 3328 3329
    if (addDefaultUSBKBD &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_KBD,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
3330
        goto cleanup;
3331 3332 3333 3334 3335 3336

    if (addDefaultUSBMouse &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
3337
        goto cleanup;
3338

D
Dmitry Andreev 已提交
3339 3340 3341 3342
    if (addPanicDevice) {
        size_t j;
        for (j = 0; j < def->npanics; j++) {
            if (def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT ||
3343 3344 3345 3346
                (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
3347 3348
                break;
        }
3349

D
Dmitry Andreev 已提交
3350 3351 3352 3353 3354 3355 3356 3357 3358
        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;
            }
        }
3359 3360
    }

3361 3362 3363 3364 3365 3366
    ret = 0;
 cleanup:
    return ret;
}


A
Andrea Bolognani 已提交
3367 3368 3369
/**
 * qemuDomainDefEnableDefaultFeatures:
 * @def: domain definition
3370
 * @qemuCaps: QEMU capabilities
A
Andrea Bolognani 已提交
3371 3372 3373 3374 3375
 *
 * Make sure that features that should be enabled by default are actually
 * enabled and configure default values related to those features.
 */
static void
3376 3377
qemuDomainDefEnableDefaultFeatures(virDomainDefPtr def,
                                   virQEMUCapsPtr qemuCaps)
A
Andrea Bolognani 已提交
3378
{
3379
    /* The virt machine type always uses GIC: if the relevant information
3380 3381
     * was not included in the domain XML, we need to choose a suitable
     * GIC version ourselves */
3382 3383 3384 3385 3386
    if ((def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ABSENT &&
         qemuDomainIsVirt(def)) ||
        (def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ON &&
         def->gic_version == VIR_GIC_VERSION_NONE)) {
        virGICVersion version;
3387 3388 3389 3390 3391

        VIR_DEBUG("Looking for usable GIC version in domain capabilities");
        for (version = VIR_GIC_VERSION_LAST - 1;
             version > VIR_GIC_VERSION_NONE;
             version--) {
3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406

            /* 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;
            }

3407 3408 3409 3410 3411 3412 3413 3414
            if (virQEMUCapsSupportsGICVersion(qemuCaps,
                                              def->virtType,
                                              version)) {
                VIR_DEBUG("Using GIC version %s",
                          virGICVersionTypeToString(version));
                def->gic_version = version;
                break;
            }
3415 3416
        }

3417 3418 3419 3420 3421 3422 3423
        /* 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;
        }

3424 3425 3426
        /* 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;
3427
    }
A
Andrea Bolognani 已提交
3428 3429 3430
}


3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450
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;
}


3451
static int
3452 3453 3454 3455 3456
qemuDomainRecheckInternalPaths(virDomainDefPtr def,
                               virQEMUDriverConfigPtr cfg,
                               unsigned int flags)
{
    size_t i = 0;
3457
    size_t j = 0;
3458 3459 3460 3461

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

3462 3463 3464 3465 3466 3467
        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
3468
             * the auto-generated socket based on config option from qemu.conf
3469 3470 3471 3472
             * 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 &&
3473
                !glisten->autoGenerated &&
3474 3475 3476 3477 3478 3479 3480
                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;
                }
3481
            }
3482 3483
        }
    }
3484 3485

    return 0;
3486 3487 3488
}


3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558
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;
}


3559 3560 3561 3562 3563 3564
static int
qemuDomainDefCPUPostParse(virDomainDefPtr def)
{
    if (!def->cpu)
        return 0;

3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618
    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;
        }
    }

3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651
    /* 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;
}


3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683
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;
}


3684 3685 3686 3687 3688 3689 3690 3691
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)))
3692
        return 1;
3693 3694 3695 3696 3697

    return 0;
}


3698 3699
static int
qemuDomainDefPostParse(virDomainDefPtr def,
3700
                       virCapsPtr caps ATTRIBUTE_UNUSED,
3701
                       unsigned int parseFlags,
3702
                       void *opaque,
3703
                       void *parseOpaque)
3704 3705
{
    virQEMUDriverPtr driver = opaque;
3706
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3707 3708 3709
    /* 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. */
3710
    virQEMUCapsPtr qemuCaps = parseOpaque;
3711 3712 3713 3714 3715
    int ret = -1;

    if (def->os.bootloader || def->os.bootloaderArgs) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bootloader is not supported by QEMU"));
3716
        goto cleanup;
3717 3718
    }

3719 3720 3721
    if (!def->os.machine) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing machine type"));
3722
        goto cleanup;
3723 3724
    }

3725 3726 3727 3728 3729 3730 3731 3732 3733
    if (def->os.loader &&
        def->os.loader->type == VIR_DOMAIN_LOADER_TYPE_PFLASH &&
        def->os.loader->readonly == VIR_TRISTATE_SWITCH_ON &&
        !def->os.loader->nvram) {
        if (virAsprintf(&def->os.loader->nvram, "%s/%s_VARS.fd",
                        cfg->nvramDir, def->name) < 0)
            goto cleanup;
    }

3734 3735 3736
    if (qemuDomainDefAddDefaultDevices(def, qemuCaps) < 0)
        goto cleanup;

3737 3738 3739
    if (qemuCanonicalizeMachine(def, qemuCaps) < 0)
        goto cleanup;

3740
    qemuDomainDefEnableDefaultFeatures(def, qemuCaps);
A
Andrea Bolognani 已提交
3741

3742 3743
    if (qemuDomainRecheckInternalPaths(def, cfg, parseFlags) < 0)
        goto cleanup;
3744

3745
    if (qemuSecurityVerify(driver->securityManager, def) < 0)
3746 3747
        goto cleanup;

3748 3749 3750
    if (qemuDomainDefVcpusPostParse(def) < 0)
        goto cleanup;

3751 3752 3753
    if (qemuDomainDefCPUPostParse(def) < 0)
        goto cleanup;

3754 3755 3756
    if (qemuDomainDefTsegPostParse(def, qemuCaps) < 0)
        goto cleanup;

3757 3758
    ret = 0;
 cleanup:
3759
    virObjectUnref(cfg);
3760
    return ret;
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
/**
 * 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;
}


3794 3795 3796
#define QEMU_MAX_VCPUS_WITHOUT_EIM 255


3797
static int
3798 3799
qemuDomainDefValidateFeatures(const virDomainDef *def,
                              virQEMUCapsPtr qemuCaps)
3800
{
3801
    size_t i;
3802

3803 3804 3805 3806 3807
    for (i = 0; i < VIR_DOMAIN_FEATURE_LAST; i++) {
        const char *featureName = virDomainFeatureTypeToString(i);

        switch ((virDomainFeature) i) {
        case VIR_DOMAIN_FEATURE_IOAPIC:
3808
            if (def->features[i] != VIR_DOMAIN_IOAPIC_NONE &&
3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820
                !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:
3821
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832
                !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;

3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845
        case VIR_DOMAIN_FEATURE_GIC:
            if (def->features[i] == VIR_TRISTATE_SWITCH_ON &&
                !qemuDomainIsVirt(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;

3846 3847
        case VIR_DOMAIN_FEATURE_SMM:
            if (def->features[i] != VIR_TRISTATE_SWITCH_ABSENT &&
3848
                !virQEMUCapsGet(qemuCaps, QEMU_CAPS_MACHINE_SMM_OPT)) {
3849 3850 3851 3852 3853 3854
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("smm is not available with this QEMU binary"));
                return -1;
            }
            break;

3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870
        case VIR_DOMAIN_FEATURE_ACPI:
        case VIR_DOMAIN_FEATURE_APIC:
        case VIR_DOMAIN_FEATURE_PAE:
        case VIR_DOMAIN_FEATURE_HAP:
        case VIR_DOMAIN_FEATURE_VIRIDIAN:
        case VIR_DOMAIN_FEATURE_PRIVNET:
        case VIR_DOMAIN_FEATURE_HYPERV:
        case VIR_DOMAIN_FEATURE_KVM:
        case VIR_DOMAIN_FEATURE_PVSPINLOCK:
        case VIR_DOMAIN_FEATURE_CAPABILITIES:
        case VIR_DOMAIN_FEATURE_PMU:
        case VIR_DOMAIN_FEATURE_VMPORT:
        case VIR_DOMAIN_FEATURE_VMCOREINFO:
        case VIR_DOMAIN_FEATURE_LAST:
            break;
        }
3871 3872 3873 3874 3875 3876
    }

    return 0;
}


3877 3878
static int
qemuDomainDefValidate(const virDomainDef *def,
3879
                      virCapsPtr caps ATTRIBUTE_UNUSED,
3880
                      void *opaque)
3881
{
3882 3883 3884 3885
    virQEMUDriverPtr driver = opaque;
    virQEMUCapsPtr qemuCaps = NULL;
    int ret = -1;

3886
    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
3887 3888 3889
                                            def->emulator)))
        goto cleanup;

3890 3891 3892
    if (def->mem.min_guarantee) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Parameter 'min_guarantee' not supported by QEMU."));
3893
        goto cleanup;
3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913
    }

    /* 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;
3914 3915
    }

M
Michal Privoznik 已提交
3916 3917 3918 3919 3920
    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. */

3921
        if (!qemuDomainIsQ35(def)) {
M
Michal Privoznik 已提交
3922 3923
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot is supported with q35 machine types only"));
3924
            goto cleanup;
M
Michal Privoznik 已提交
3925 3926 3927 3928 3929 3930 3931 3932
        }

        /* 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"));
3933
            goto cleanup;
M
Michal Privoznik 已提交
3934 3935 3936 3937 3938
        }

        if (def->features[VIR_DOMAIN_FEATURE_SMM] != VIR_TRISTATE_SWITCH_ON) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot requires SMM feature enabled"));
3939
            goto cleanup;
M
Michal Privoznik 已提交
3940 3941 3942
        }
    }

3943 3944 3945 3946 3947
    /* 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;
3948
        unsigned int granularity;
3949 3950 3951 3952 3953 3954

        /* 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)) {
3955 3956 3957
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CPU topology doesn't match maximum vcpu count"));
            goto cleanup;
3958
        }
3959 3960 3961 3962 3963 3964 3965 3966 3967 3968

        /* 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;
        }
3969 3970
    }

3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988
    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;
        }
    }

3989 3990 3991 3992 3993 3994 3995
    if (def->ncachetunes &&
        def->virtType != VIR_DOMAIN_VIRT_KVM) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("cachetune is only supported for KVM domains"));
        goto cleanup;
    }

3996
    if (qemuDomainDefValidateFeatures(def, qemuCaps) < 0)
3997 3998
        goto cleanup;

3999 4000 4001 4002 4003
    ret = 0;

 cleanup:
    virObjectUnref(qemuCaps);
    return ret;
4004 4005 4006
}


4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030
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;
}


4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077
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;
}


4078 4079 4080
static int
qemuDomainChrSerialTargetTypeToAddressType(int targetType)
{
4081
    switch ((virDomainChrSerialTargetType)targetType) {
4082 4083 4084 4085 4086 4087
    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;
4088 4089
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
        return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_SPAPRVIO;
4090
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
4091
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
4092 4093 4094 4095 4096 4097 4098 4099 4100
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        break;
    }

    return VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE;
}


4101 4102 4103 4104 4105 4106 4107 4108 4109 4110
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;
4111 4112
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
4113 4114
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
4115 4116 4117
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
    case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
        return VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
4118 4119 4120 4121 4122 4123 4124 4125 4126
    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;
}


4127
static int
4128
qemuDomainChrTargetDefValidate(const virDomainChrDef *chr)
4129
{
4130 4131
    int expected;

4132
    switch ((virDomainChrDeviceType)chr->deviceType) {
4133 4134 4135
    case VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL:

        /* Validate target type */
4136
        switch ((virDomainChrSerialTargetType)chr->targetType) {
4137
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_ISA:
4138 4139
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_USB:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_PCI:
4140
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
4141

4142
            expected = qemuDomainChrSerialTargetTypeToAddressType(chr->targetType);
4143 4144

            if (chr->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
4145 4146 4147 4148 4149
                chr->info.type != expected) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Target type '%s' requires address type '%s'"),
                               virDomainChrSerialTargetTypeToString(chr->targetType),
                               virDomainDeviceAddressTypeToString(expected));
4150 4151 4152 4153
                return -1;
            }
            break;

4154
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
4155
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
4156 4157 4158 4159 4160 4161 4162 4163 4164
            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;

4165 4166 4167 4168
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            break;
        }
4169 4170 4171 4172 4173 4174

        /* 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:
4175
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY:
4176
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011:
4177 4178
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPLMCONSOLE:
4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194

            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;
        }
4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208
        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;
}


4209
static int
4210
qemuDomainChrDefValidate(const virDomainChrDef *dev,
4211
                         const virDomainDef *def)
4212
{
4213
    if (qemuDomainChrSourceDefValidate(dev->source) < 0)
4214 4215
        return -1;

4216
    if (qemuDomainChrTargetDefValidate(dev) < 0)
4217 4218
        return -1;

4219
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL &&
4220
        (ARCH_IS_S390(def->os.arch) || qemuDomainIsPSeries(def))) {
4221 4222 4223 4224 4225
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("parallel ports are not supported"));
            return -1;
    }

4226 4227 4228 4229 4230 4231 4232 4233 4234
    if (dev->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL) {
        bool isCompatible = true;

        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;
        }

4235 4236 4237 4238 4239 4240
        if (!qemuDomainIsVirt(def) &&
            (dev->targetType == VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM ||
             dev->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011)) {
            isCompatible = false;
        }

4241 4242 4243 4244 4245 4246 4247
        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;
        }

4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258
        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;
        }
    }

4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294
    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
qemuDomainRNGDefValidate(const virDomainRNGDef *def)
{
    if (def->backend == VIR_DOMAIN_RNG_BACKEND_EGD &&
        qemuDomainChrSourceDefValidate(def->source.chardev) < 0)
        return -1;

    return 0;
}


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

    return 0;
}


4295 4296 4297 4298 4299 4300 4301 4302 4303
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
4304
                           _("%s model of watchdog can go only on PCI bus"),
4305 4306 4307 4308 4309 4310 4311 4312 4313
                           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
4314
                           _("%s model of watchdog can go only on ISA bus"),
4315 4316 4317 4318 4319 4320 4321 4322
                           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
4323
                           _("%s model of watchdog is virtual and cannot go on any bus."),
4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342
                           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;
}


4343
static int
4344
qemuDomainDeviceDefValidateNetwork(const virDomainNetDef *net)
4345
{
4346 4347
    bool hasIPv4 = false;
    bool hasIPv6 = false;
4348
    size_t i;
4349

4350 4351 4352 4353 4354 4355 4356 4357 4358 4359
    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];
4360

4361
            if (VIR_SOCKET_ADDR_VALID(&net->guestIP.ips[i]->peer)) {
4362
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4363 4364
                               _("Invalid attempt to set peer IP for guest"));
                return -1;
4365
            }
4366

4367 4368
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ip->address, AF_INET)) {
                if (hasIPv4) {
4369
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4370 4371 4372
                                   _("Only one IPv4 address per "
                                     "interface is allowed"));
                    return -1;
4373
                }
4374
                hasIPv4 = true;
4375

4376
                if (ip->prefix < 4 || ip->prefix > 27) {
4377
                    virReportError(VIR_ERR_XML_ERROR, "%s",
4378
                                   _("invalid prefix, must be in range of 4-27"));
4379
                    return -1;
4380
                }
4381
            }
4382

4383 4384 4385 4386 4387 4388
            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;
4389
                }
4390
                hasIPv6 = true;
4391

4392 4393 4394 4395 4396
                if (ip->prefix > 120) {
                    virReportError(VIR_ERR_XML_ERROR, "%s",
                                   _("prefix too long"));
                    return -1;
                }
4397
            }
4398
        }
4399 4400 4401 4402 4403 4404 4405
    } 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;
    }
4406

4407 4408 4409 4410 4411 4412 4413 4414 4415 4416
    if (STREQ_NULLABLE(net->model, "virtio")) {
        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;
4417
        }
4418
    }
4419

4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438
    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;
}


4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455
static int
qemuDomainDeviceDefValidateHostdev(const virDomainHostdevDef *hostdev,
                                   const virDomainDef *def)
{
    /* 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;
    }

    return 0;
}


4456 4457 4458
static int
qemuDomainDeviceDefValidateVideo(const virDomainVideoDef *video)
{
4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503
    switch (video->type) {
    case VIR_DOMAIN_VIDEO_TYPE_XEN:
    case VIR_DOMAIN_VIDEO_TYPE_VBOX:
    case VIR_DOMAIN_VIDEO_TYPE_PARALLELS:
    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);
4504 4505
            return -1;
        }
4506 4507 4508 4509 4510 4511 4512
        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;
            }
4513

4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527
            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)"));
4528 4529 4530 4531 4532 4533 4534 4535
            return -1;
        }
    }

    return 0;
}


4536
int
4537 4538
qemuDomainValidateStorageSource(virStorageSourcePtr src,
                                virQEMUCapsPtr qemuCaps)
4539
{
4540 4541
    int actualType = virStorageSourceGetActualType(src);

4542 4543 4544 4545 4546 4547
    if (src->format == VIR_STORAGE_FILE_COW) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                      _("'cow' storage format is not supported"));
        return -1;
    }

4548 4549 4550 4551 4552 4553 4554
    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;
    }

4555 4556 4557 4558 4559 4560 4561
    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;
    }

4562 4563 4564 4565 4566 4567 4568 4569 4570 4571
    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;
    }

4572 4573 4574 4575 4576 4577 4578 4579 4580
    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",
                       _("LUKS encrypted QCOW2 images are not suppored by this QEMU"));
        return -1;
    }

4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603
    if (src->format == VIR_STORAGE_FILE_FAT &&
        actualType != VIR_STORAGE_TYPE_DIR) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("storage format 'fat' is supported only with 'dir' "
                         "storage type"));
        return -1;
    }

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

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

4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616
    if (src->pr) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_PR_MANAGER_HELPER)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("reservations not supported with this QEMU binary"));
            return -1;
        }

        if (virStoragePRDefIsManaged(src->pr) && src->pr->path) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'path' attribute should not be provided for "
                             "managed reservations"));
            return -1;
        }
4617 4618
    }

4619 4620 4621 4622
    return 0;
}


4623
int
4624 4625
qemuDomainDeviceDefValidateDisk(const virDomainDiskDef *disk,
                                virQEMUCapsPtr qemuCaps)
4626
{
4627
    const char *driverName = virDomainDiskGetDriver(disk);
4628 4629
    virStorageSourcePtr n;

4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645
    if (disk->src->shared && !disk->src->readonly) {
        if (disk->src->format <= VIR_STORAGE_FILE_AUTO) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("shared access for disk '%s' requires use of "
                             "explicitly specified disk format"), disk->dst);
            return -1;
        }

        if (!qemuBlockStorageSourceSupportsConcurrentAccess(disk->src)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("shared access for disk '%s' requires use of "
                             "supported storage format"), disk->dst);
            return -1;
        }
    }

4646 4647 4648 4649 4650 4651 4652
    if (driverName && STRNEQ(driverName, "qemu")) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("unsupported driver name '%s' for disk '%s'"),
                       driverName, disk->dst);
        return -1;
    }

4653
    for (n = disk->src; virStorageSourceIsBacking(n); n = n->backingStore) {
4654
        if (qemuDomainValidateStorageSource(n, qemuCaps) < 0)
4655 4656 4657
            return -1;
    }

4658 4659 4660 4661
    return 0;
}


4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686
static int
qemuDomainDeviceDefValidateControllerAttributes(const virDomainControllerDef *controller)
{
    if (!(controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
          controller->model == VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI)) {
        if (controller->queues) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'queues' is only supported by virtio-scsi controller"));
            return -1;
        }
        if (controller->cmd_per_lun) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'cmd_per_lun' is only supported by virtio-scsi controller"));
            return -1;
        }
        if (controller->max_sectors) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'max_sectors' is only supported by virtio-scsi controller"));
            return -1;
        }
        if (controller->ioeventfd) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'ioeventfd' is only supported by virtio-scsi controller"));
            return -1;
        }
4687 4688 4689 4690 4691
        if (controller->iothread) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("'iothread' is only supported for virtio-scsi controller"));
            return -1;
        }
4692 4693 4694 4695 4696 4697
    }

    return 0;
}


4698 4699 4700 4701 4702 4703
/**
 * @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
4704
 * Returns true if acceptable, false otherwise with error message set.
4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752
 */
static bool
qemuDomainCheckSCSIControllerModel(virQEMUCapsPtr qemuCaps,
                                   int model)
{
    switch ((virDomainControllerModelSCSI) model) {
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSILOGIC:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_LSI)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support "
                             "the LSI 53C895A SCSI controller"));
            return false;
        }
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_SCSI)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support "
                             "virtio scsi controller"));
            return false;
        }
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_IBMVSCSI:
        /*TODO: need checking work here if necessary */
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSISAS1068:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_MPTSAS1068)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support "
                             "the LSI SAS1068 (MPT Fusion) controller"));
            return false;
        }
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSISAS1078:
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_MEGASAS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("This QEMU doesn't support "
                             "the LSI SAS1078 (MegaRAID) controller"));
            return false;
        }
        break;
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_AUTO:
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_BUSLOGIC:
    case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VMPVSCSI:
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("Unsupported controller model: %s"),
                       virDomainControllerModelSCSITypeToString(model));
        return false;
4753 4754 4755 4756 4757 4758
    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;
4759 4760 4761 4762 4763 4764
    }

    return true;
}


4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790
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;
}


4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807
/* 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;

4808 4809
    if (controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI &&
4810 4811
        controller->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
       virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
4812 4813
                       _("virtio-scsi IOThreads only available for virtio "
                         "pci and virtio ccw controllers"));
4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845
       return false;
    }

    /* Can we find the controller iothread in the iothreadid list? */
    if (!virDomainIOThreadIDFind(def, controller->iothread)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("controller iothread '%u' not defined in iothreadid"),
                       controller->iothread);
        return false;
    }

    return true;
}


static int
qemuDomainDeviceDefValidateControllerSCSI(const virDomainControllerDef *controller,
                                          const virDomainDef *def)
{
    switch ((virDomainControllerModelSCSI) controller->model) {
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI:
            if (!qemuDomainCheckSCSIControllerIOThreads(controller, def))
                return -1;
            break;

        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_AUTO:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_BUSLOGIC:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSILOGIC:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSISAS1068:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VMPVSCSI:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_IBMVSCSI:
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSISAS1078:
4846
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_DEFAULT:
4847 4848 4849 4850 4851 4852 4853 4854
        case VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LAST:
            break;
    }

    return 0;
}


4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886
/**
 * 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;
4887 4888
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_PCIE_PCI_BRIDGE:
        return QEMU_CAPS_DEVICE_PCIE_PCI_BRIDGE;
4889 4890 4891 4892 4893
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_NONE:
        return 0;
    case VIR_DOMAIN_CONTROLLER_PCI_MODEL_NAME_LAST:
    default:
        return -1;
4894 4895
    }

4896
    return -1;
4897 4898 4899
}


4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916
#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));


4917 4918 4919 4920 4921 4922 4923 4924 4925
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);
4926
    int cap = virDomainControllerPCIModelNameToQEMUCaps(pciopts->modelName);
4927 4928 4929 4930 4931 4932 4933 4934 4935 4936

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

4937 4938 4939 4940 4941 4942 4943 4944 4945
    /* 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:
4946
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050
        /* 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;

5051 5052 5053 5054 5055 5056 5057
    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;

5058 5059 5060 5061 5062 5063 5064
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_DEFAULT:
    case VIR_DOMAIN_CONTROLLER_MODEL_PCI_LAST:
    default:
        virReportEnumRangeError(virDomainControllerModelPCI, cont->model);
        return -1;
    }

5065 5066 5067 5068 5069 5070 5071 5072 5073
    /* 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:
5074
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108
        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;
    }

5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135
    /* 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:
5136
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149
        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;
    }

5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169
    /* 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:
5170
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184
        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;
    }

5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201
    /* 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:
5202
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215
        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;
    }

5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246
    /* 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:
5247
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260
        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;
    }

5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277
    /* 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:
5278
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291
        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;
    }

5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312
    /* 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:
5313
    case VIR_DOMAIN_CONTROLLER_MODEL_PCIE_TO_PCI_BRIDGE:
5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329
        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);
    }

5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355
    /* 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;
    }

5356
    return 0;
5357 5358 5359
}


5360 5361 5362 5363 5364
#undef virReportControllerInvalidValue
#undef virReportControllerInvalidOption
#undef virReportControllerMissingOption


5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382
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;
}


5383
static int
5384 5385 5386
qemuDomainDeviceDefValidateController(const virDomainControllerDef *controller,
                                      const virDomainDef *def,
                                      virQEMUCapsPtr qemuCaps)
5387
{
5388 5389
    int ret = 0;

5390 5391 5392 5393
    if (!qemuDomainCheckCCWS390AddressSupport(def, controller->info, qemuCaps,
                                              "controller"))
        return -1;

5394 5395 5396 5397
    if (controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
        !qemuDomainCheckSCSIControllerModel(qemuCaps, controller->model))
        return -1;

5398 5399 5400
    if (qemuDomainDeviceDefValidateControllerAttributes(controller) < 0)
        return -1;

5401
    switch ((virDomainControllerType)controller->type) {
5402
    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
5403 5404 5405
        ret = qemuDomainDeviceDefValidateControllerIDE(controller, def);
        break;

5406
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
5407 5408 5409
        ret = qemuDomainDeviceDefValidateControllerSCSI(controller, def);
        break;

5410
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
5411 5412
        ret = qemuDomainDeviceDefValidateControllerPCI(controller, def,
                                                       qemuCaps);
5413 5414
        break;

5415
    case VIR_DOMAIN_CONTROLLER_TYPE_SATA:
5416 5417 5418 5419 5420
        ret = qemuDomainDeviceDefValidateControllerSATA(controller, def,
                                                        qemuCaps);
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
5421 5422 5423 5424 5425 5426 5427
    case VIR_DOMAIN_CONTROLLER_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_CONTROLLER_TYPE_CCID:
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
    }

5428
    return ret;
5429 5430 5431
}


5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455
static int
qemuDomainDeviceDefValidateMemory(const virDomainMemoryDef *memory ATTRIBUTE_UNUSED,
                                  const virDomainDef *def)
{
    const long system_page_size = virGetSystemPageSizeKB();

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

    return 0;
}


5456 5457 5458 5459 5460 5461 5462 5463 5464 5465
static int
qemuDomainDeviceDefValidateVsock(const virDomainVsockDef *vsock ATTRIBUTE_UNUSED,
                                 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;
    }
5466 5467
    return 0;
}
5468

5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493

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;
    }
5494 5495 5496 5497
    return 0;
}


5498 5499 5500
static int
qemuDomainDeviceDefValidate(const virDomainDeviceDef *dev,
                            const virDomainDef *def,
5501
                            void *opaque)
5502
{
5503
    int ret = 0;
5504 5505 5506 5507 5508 5509
    virQEMUDriverPtr driver = opaque;
    virQEMUCapsPtr qemuCaps = NULL;

    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
                                            def->emulator)))
        return -1;
5510

5511
    switch ((virDomainDeviceType)dev->type) {
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
    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:
        ret = qemuDomainRNGDefValidate(dev->data.rng);
        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:
        ret = qemuDomainDeviceDefValidateHostdev(dev->data.hostdev, def);
        break;

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

    case VIR_DOMAIN_DEVICE_DISK:
5545
        ret = qemuDomainDeviceDefValidateDisk(dev->data.disk, qemuCaps);
5546 5547
        break;

5548
    case VIR_DOMAIN_DEVICE_CONTROLLER:
5549 5550
        ret = qemuDomainDeviceDefValidateController(dev->data.controller, def,
                                                    qemuCaps);
5551 5552
        break;

5553 5554 5555 5556
    case VIR_DOMAIN_DEVICE_MEMORY:
        ret = qemuDomainDeviceDefValidateMemory(dev->data.memory, def);
        break;

5557 5558 5559 5560
    case VIR_DOMAIN_DEVICE_VSOCK:
        ret = qemuDomainDeviceDefValidateVsock(dev->data.vsock, qemuCaps);
        break;

5561 5562 5563 5564
    case VIR_DOMAIN_DEVICE_TPM:
        ret = qemuDomainDeviceDefValidateTPM(dev->data.tpm, def);
        break;

5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578
    case VIR_DOMAIN_DEVICE_LEASE:
    case VIR_DOMAIN_DEVICE_FS:
    case VIR_DOMAIN_DEVICE_INPUT:
    case VIR_DOMAIN_DEVICE_SOUND:
    case VIR_DOMAIN_DEVICE_GRAPHICS:
    case VIR_DOMAIN_DEVICE_HUB:
    case VIR_DOMAIN_DEVICE_MEMBALLOON:
    case VIR_DOMAIN_DEVICE_NVRAM:
    case VIR_DOMAIN_DEVICE_SHMEM:
    case VIR_DOMAIN_DEVICE_PANIC:
    case VIR_DOMAIN_DEVICE_IOMMU:
    case VIR_DOMAIN_DEVICE_NONE:
    case VIR_DOMAIN_DEVICE_LAST:
        break;
5579 5580
    }

5581
    virObjectUnref(qemuCaps);
5582 5583 5584 5585
    return ret;
}


5586 5587 5588 5589 5590 5591 5592 5593 5594
/**
 * 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.
 */
5595
static const char *
5596 5597
qemuDomainDefaultNetModel(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
5598
{
5599
    if (ARCH_IS_S390(def->os.arch))
5600 5601
        return "virtio";

5602 5603
    if (def->os.arch == VIR_ARCH_ARMV7L ||
        def->os.arch == VIR_ARCH_AARCH64) {
5604 5605 5606
        if (STREQ(def->os.machine, "versatilepb"))
            return "smc91c111";

5607
        if (qemuDomainIsVirt(def))
5608 5609
            return "virtio";

5610 5611 5612 5613 5614
        /* Incomplete. vexpress (and a few others) use this, but not all
         * arm boards */
        return "lan9118";
    }

5615 5616 5617 5618 5619
    /* In all other cases the model depends on the capabilities. If they were
     * not provided don't report any default. */
    if (!qemuCaps)
        return NULL;

5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631
    /* Try several network devices in turn; each of these devices is
     * less likely be supported out-of-the-box by the guest operating
     * system than the previous one */
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_RTL8139))
        return "rtl8139";
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_E1000))
        return "e1000";
    else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_NET))
        return "virtio";

    /* We've had no luck detecting support for any network device,
     * but we have to return something: might as well be rtl8139 */
5632 5633
    return "rtl8139";
}
5634

5635 5636

/*
5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652
 * 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.
5653
 */
5654
static int
5655 5656 5657
qemuDomainChrDefDropDefaultPath(virDomainChrDefPtr chr,
                                virQEMUDriverPtr driver)
{
5658 5659 5660 5661
    virQEMUDriverConfigPtr cfg;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *regexp = NULL;
    int ret = -1;
5662

5663 5664 5665 5666 5667
    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;
5668 5669
    }

5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686
    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);
5687
    virObjectUnref(cfg);
5688
    return ret;
5689 5690 5691
}


5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740 5741
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;
}


5742
#define QEMU_USB_XHCI_MAXPORTS 15
5743 5744


5745 5746 5747
static int
qemuDomainControllerDefPostParse(virDomainControllerDefPtr cont,
                                 const virDomainDef *def,
5748 5749
                                 virQEMUCapsPtr qemuCaps,
                                 unsigned int parseFlags)
5750
{
5751 5752
    switch ((virDomainControllerType)cont->type) {
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
5753 5754 5755
        /* Set the default SCSI controller model if not already set */
        if (qemuDomainSetSCSIControllerModel(def, cont, qemuCaps) < 0)
            return -1;
5756
        break;
5757

5758
    case VIR_DOMAIN_CONTROLLER_TYPE_USB:
5759
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT && qemuCaps) {
5760
            /* Pick a suitable default model for the USB controller if none
5761 5762
             * has been selected by the user and we have the qemuCaps for
             * figuring out which contollers are supported.
5763 5764 5765 5766 5767 5768 5769
             *
             * 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() */
5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780

            /* 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;
                }
5781
            } else if (ARCH_IS_PPC64(def->os.arch)) {
5782 5783
                /* To not break migration we need to set default USB controller
                 * for ppc64 to pci-ohci if we cannot change ABI of the VM.
5784 5785
                 * The nec-usb-xhci or qemu-xhci controller is used as default
                 * only for newly defined domains or devices. */
5786
                if ((parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE) &&
5787 5788 5789
                    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) &&
5790 5791 5792
                    virQEMUCapsGet(qemuCaps, QEMU_CAPS_NEC_USB_XHCI)) {
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
                } else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_OHCI)) {
5793
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_PCI_OHCI;
5794 5795 5796
                } else {
                    /* Explicitly fallback to legacy USB controller for PPC64. */
                    cont->model = -1;
5797
                }
5798
            } else if (def->os.arch == VIR_ARCH_AARCH64) {
5799 5800 5801
                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))
5802
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813
            }
        }
        /* 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;
        }
5814 5815 5816
        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) {
5817
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
5818 5819 5820
                           _("'%s' controller only supports up to '%u' ports"),
                           virDomainControllerModelUSBTypeToString(cont->model),
                           QEMU_USB_XHCI_MAXPORTS);
5821 5822 5823
            return -1;
        }
        break;
5824

5825
    case VIR_DOMAIN_CONTROLLER_TYPE_PCI:
5826 5827 5828 5829 5830 5831 5832

        /* 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) {
5833 5834 5835 5836 5837 5838
            virReportError(VIR_ERR_XML_ERROR, "%s",
                           _("pci-root and pcie-root controllers "
                             "should have index 0"));
            return -1;
        }

5839
        if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS &&
5840
            !qemuDomainIsI440FX(def)) {
5841 5842 5843 5844 5845 5846
            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 &&
5847
            !qemuDomainIsQ35(def)) {
5848 5849 5850 5851 5852 5853
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("pcie-expander-bus controllers are only supported "
                             "on q35-based machinetypes"));
            return -1;
        }

5854 5855 5856
        /* 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
5857 5858
         * from 0 .. size-1.
         */
5859 5860
        if (cont->opts.pciopts.numaNode >= 0 &&
            cont->opts.pciopts.numaNode >=
5861
            (int)virDomainNumaGetNodeCount(def->numa)) {
5862 5863 5864 5865 5866 5867 5868 5869 5870 5871
            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;
        }
5872 5873 5874 5875 5876 5877 5878 5879 5880
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_SATA:
    case VIR_DOMAIN_CONTROLLER_TYPE_VIRTIO_SERIAL:
    case VIR_DOMAIN_CONTROLLER_TYPE_CCID:
    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
    case VIR_DOMAIN_CONTROLLER_TYPE_LAST:
        break;
5881 5882 5883 5884 5885
    }

    return 0;
}

5886 5887 5888 5889 5890 5891
static int
qemuDomainChrDefPostParse(virDomainChrDefPtr chr,
                          const virDomainDef *def,
                          virQEMUDriverPtr driver,
                          unsigned int parseFlags)
{
5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905
    /* 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)) {
5906
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO;
5907 5908
        } else if (qemuDomainIsVirt(def)) {
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM;
5909 5910
        } else if (ARCH_IS_S390(def->os.arch)) {
            chr->targetType = VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP;
5911 5912 5913
        }
    }

5914 5915 5916
    /* Set the default target model */
    if (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL &&
        chr->targetModel == VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_NONE) {
5917
        switch ((virDomainChrSerialTargetType)chr->targetType) {
5918 5919 5920 5921 5922 5923 5924 5925 5926
        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;
5927 5928 5929
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SPAPR_VTY;
            break;
5930 5931 5932
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_PL011;
            break;
5933 5934 5935
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
            chr->targetModel = VIR_DOMAIN_CHR_SERIAL_TARGET_MODEL_SCLPCONSOLE;
            break;
5936 5937 5938 5939 5940 5941 5942
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
        case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
            /* Nothing to do */
            break;
        }
    }

5943 5944 5945 5946 5947 5948 5949 5950 5951 5952 5953 5954 5955 5956 5957 5958
    /* 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;
}
5959

5960

5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001 6002 6003 6004 6005 6006 6007 6008 6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027 6028 6029 6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040 6041
/**
 * 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;
}


6042 6043
static int
qemuDomainDeviceDiskDefPostParse(virDomainDiskDefPtr disk,
6044
                                 virQEMUCapsPtr qemuCaps,
6045
                                 unsigned int parseFlags)
6046 6047
{
    /* set default disk types and drivers */
6048 6049 6050
    if (!virDomainDiskGetDriver(disk) &&
        virDomainDiskSetDriver(disk, "qemu") < 0)
        return -1;
6051

6052 6053 6054 6055 6056
    /* default disk format for drives */
    if (virDomainDiskGetFormat(disk) == VIR_STORAGE_FILE_NONE &&
        (virDomainDiskGetType(disk) == VIR_STORAGE_TYPE_FILE ||
         virDomainDiskGetType(disk) == VIR_STORAGE_TYPE_BLOCK))
        virDomainDiskSetFormat(disk, VIR_STORAGE_FILE_RAW);
6057

6058 6059 6060 6061
    /* default disk format for mirrored drive */
    if (disk->mirror &&
        disk->mirror->format == VIR_STORAGE_FILE_NONE)
        disk->mirror->format = VIR_STORAGE_FILE_RAW;
6062

6063 6064 6065 6066
    if (qemuDomainDeviceDiskDefPostParseRestoreSecAlias(disk, qemuCaps,
                                                        parseFlags) < 0)
        return -1;

6067 6068 6069 6070 6071 6072 6073
    /* 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;

6074 6075 6076 6077
    return 0;
}


6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089 6090 6091 6092 6093
static int
qemuDomainDeviceNetDefPostParse(virDomainNetDefPtr net,
                                const virDomainDef *def,
                                virQEMUCapsPtr qemuCaps)
{
    if (net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
        !net->model) {
        if (VIR_STRDUP(net->model,
                       qemuDomainDefaultNetModel(def, qemuCaps)) < 0)
            return -1;
    }

    return 0;
}


6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112 6113 6114 6115
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;
        else if (qemuDomainIsVirt(def) || ARCH_IS_S390(def->os.arch))
            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;
}


6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132
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
6133 6134 6135 6136 6137 6138 6139 6140 6141 6142
static int
qemuDomainVsockDefPostParse(virDomainVsockDefPtr vsock)
{
    if (vsock->model == VIR_DOMAIN_VSOCK_MODEL_DEFAULT)
        vsock->model = VIR_DOMAIN_VSOCK_MODEL_VIRTIO;

    return 0;
}


6143 6144
static int
qemuDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
6145
                             const virDomainDef *def,
6146
                             virCapsPtr caps ATTRIBUTE_UNUSED,
6147
                             unsigned int parseFlags,
6148
                             void *opaque,
6149
                             void *parseOpaque)
6150
{
6151
    virQEMUDriverPtr driver = opaque;
6152 6153 6154
    /* 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. */
6155
    virQEMUCapsPtr qemuCaps = parseOpaque;
6156
    int ret = -1;
6157

6158 6159 6160 6161
    switch ((virDomainDeviceType) dev->type) {
    case VIR_DOMAIN_DEVICE_NET:
        ret = qemuDomainDeviceNetDefPostParse(dev->data.net, def, qemuCaps);
        break;
6162

6163
    case VIR_DOMAIN_DEVICE_DISK:
6164
        ret = qemuDomainDeviceDiskDefPostParse(dev->data.disk, qemuCaps,
6165
                                               parseFlags);
6166
        break;
6167

6168 6169 6170
    case VIR_DOMAIN_DEVICE_VIDEO:
        ret = qemuDomainDeviceVideoDefPostParse(dev->data.video, def);
        break;
6171

6172 6173 6174
    case VIR_DOMAIN_DEVICE_PANIC:
        ret = qemuDomainDevicePanicDefPostParse(dev->data.panic, def);
        break;
6175

6176 6177 6178 6179
    case VIR_DOMAIN_DEVICE_CONTROLLER:
        ret = qemuDomainControllerDefPostParse(dev->data.controller, def,
                                               qemuCaps, parseFlags);
        break;
6180

6181 6182 6183
    case VIR_DOMAIN_DEVICE_SHMEM:
        ret = qemuDomainShmemDefPostParse(dev->data.shmem);
        break;
6184

6185 6186 6187
    case VIR_DOMAIN_DEVICE_CHR:
        ret = qemuDomainChrDefPostParse(dev->data.chr, def, driver, parseFlags);
        break;
6188

J
Ján Tomko 已提交
6189 6190 6191 6192
    case VIR_DOMAIN_DEVICE_VSOCK:
        ret = qemuDomainVsockDefPostParse(dev->data.vsock);
        break;

6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221
    case VIR_DOMAIN_DEVICE_LEASE:
    case VIR_DOMAIN_DEVICE_FS:
    case VIR_DOMAIN_DEVICE_INPUT:
    case VIR_DOMAIN_DEVICE_SOUND:
    case VIR_DOMAIN_DEVICE_HOSTDEV:
    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;
    }
6222 6223

    return ret;
6224 6225 6226
}


6227 6228
static int
qemuDomainDefAssignAddresses(virDomainDef *def,
6229
                             virCapsPtr caps ATTRIBUTE_UNUSED,
6230
                             unsigned int parseFlags ATTRIBUTE_UNUSED,
6231
                             void *opaque,
6232
                             void *parseOpaque)
6233 6234
{
    virQEMUDriverPtr driver = opaque;
6235 6236 6237
    /* 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. */
6238
    virQEMUCapsPtr qemuCaps = parseOpaque;
6239
    bool newDomain = parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE;
6240

6241 6242 6243 6244 6245 6246 6247
    /* 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;

6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261
    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,
6262
                                                def->emulator)))
6263
        return 1;
6264

6265 6266 6267 6268 6269 6270 6271 6272
    return 0;
}


static void
qemuDomainPostParseDataFree(void *parseOpaque)
{
    virQEMUCapsPtr qemuCaps = parseOpaque;
6273 6274 6275 6276 6277

    virObjectUnref(qemuCaps);
}


6278
virDomainDefParserConfig virQEMUDriverDomainDefParserConfig = {
6279
    .domainPostParseBasicCallback = qemuDomainDefPostParseBasic,
6280 6281
    .domainPostParseDataAlloc = qemuDomainPostParseDataAlloc,
    .domainPostParseDataFree = qemuDomainPostParseDataFree,
6282
    .devicesPostParseCallback = qemuDomainDeviceDefPostParse,
6283
    .domainPostParseCallback = qemuDomainDefPostParse,
6284
    .assignAddressesCallback = qemuDomainDefAssignAddresses,
6285
    .domainValidateCallback = qemuDomainDefValidate,
6286 6287
    .deviceValidateCallback = qemuDomainDeviceDefValidate,

6288
    .features = VIR_DOMAIN_DEF_FEATURE_MEMORY_HOTPLUG |
6289
                VIR_DOMAIN_DEF_FEATURE_OFFLINE_VCPUPIN |
6290 6291
                VIR_DOMAIN_DEF_FEATURE_INDIVIDUAL_VCPUS |
                VIR_DOMAIN_DEF_FEATURE_USER_ALIAS,
6292 6293 6294
};


6295
static void
6296
qemuDomainObjSaveJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
6297
{
6298 6299 6300
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    if (virDomainObjIsActive(obj)) {
6301
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj, driver->caps) < 0)
6302
            VIR_WARN("Failed to save status on vm %s", obj->def->name);
6303
    }
6304

6305
    virObjectUnref(cfg);
6306 6307
}

J
Jiri Denemark 已提交
6308
void
6309
qemuDomainObjSetJobPhase(virQEMUDriverPtr driver,
J
Jiri Denemark 已提交
6310 6311 6312 6313
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
6314
    unsigned long long me = virThreadSelfID();
J
Jiri Denemark 已提交
6315 6316 6317 6318

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

6319 6320 6321 6322 6323
    VIR_DEBUG("Setting '%s' phase to '%s'",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob, phase));

    if (priv->job.asyncOwner && me != priv->job.asyncOwner) {
6324
        VIR_WARN("'%s' async job is owned by thread %llu",
6325 6326 6327 6328
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }

J
Jiri Denemark 已提交
6329
    priv->job.phase = phase;
6330
    priv->job.asyncOwner = me;
J
Jiri Denemark 已提交
6331 6332 6333
    qemuDomainObjSaveJob(driver, obj);
}

6334
void
6335 6336
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
6337 6338 6339
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

6340 6341 6342 6343 6344 6345 6346
    if (!priv->job.asyncJob)
        return;

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

void
6347
qemuDomainObjDiscardAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
6348 6349 6350 6351 6352 6353
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
6354
    qemuDomainObjSaveJob(driver, obj);
6355 6356
}

6357 6358 6359 6360 6361 6362 6363 6364 6365
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()) {
6366
        VIR_WARN("'%s' async job is owned by thread %llu",
6367 6368 6369 6370 6371 6372
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }
    priv->job.asyncOwner = 0;
}

6373
static bool
6374
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
6375 6376
{
    return !priv->job.asyncJob || (priv->job.mask & JOB_MASK(job)) != 0;
6377 6378
}

6379
bool
6380
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
6381 6382 6383 6384
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395
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));
}

6396 6397 6398
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

6399 6400 6401 6402 6403 6404
/**
 * qemuDomainObjBeginJobInternal:
 * @driver: qemu driver
 * @obj: domain object
 * @job: qemuDomainJob to start
 * @asyncJob: qemuDomainAsyncJob to start
6405
 * @nowait: don't wait trying to acquire @job
6406 6407 6408 6409
 *
 * 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
6410 6411 6412 6413 6414
 * 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.
6415 6416 6417 6418 6419
 *
 * Returns: 0 on success,
 *         -2 if unable to start job because of timeout or
 *            maxQueuedJobs limit,
 *         -1 otherwise.
6420
 */
6421
static int ATTRIBUTE_NONNULL(1)
6422
qemuDomainObjBeginJobInternal(virQEMUDriverPtr driver,
6423
                              virDomainObjPtr obj,
6424
                              qemuDomainJob job,
6425
                              qemuDomainAgentJob agentJob,
6426 6427
                              qemuDomainAsyncJob asyncJob,
                              bool nowait)
6428 6429
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
6430
    unsigned long long now;
6431
    unsigned long long then;
6432
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
J
Jiri Denemark 已提交
6433
    bool async = job == QEMU_JOB_ASYNC;
6434
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
6435
    const char *blocker = NULL;
6436
    const char *agentBlocker = NULL;
J
Jiri Denemark 已提交
6437
    int ret = -1;
J
Jiri Denemark 已提交
6438
    unsigned long long duration = 0;
6439
    unsigned long long agentDuration = 0;
J
Jiri Denemark 已提交
6440
    unsigned long long asyncDuration = 0;
6441

6442 6443 6444 6445 6446 6447
    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
6448
              qemuDomainJobTypeToString(priv->job.active),
6449
              qemuDomainAgentJobTypeToString(priv->job.agentActive),
J
Jiri Denemark 已提交
6450
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
6451

6452 6453
    if (virTimeMillisNow(&now) < 0) {
        virObjectUnref(cfg);
6454
        return -1;
6455 6456
    }

6457
    priv->jobs_queued++;
J
Jiri Denemark 已提交
6458
    then = now + QEMU_JOB_WAIT_TIME;
6459

6460
 retry:
6461 6462
    if ((!async && job != QEMU_JOB_DESTROY) &&
        cfg->maxQueuedJobs &&
6463
        priv->jobs_queued > cfg->maxQueuedJobs) {
6464 6465 6466
        goto error;
    }

6467
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
6468 6469 6470
        if (nowait)
            goto cleanup;

6471
        VIR_DEBUG("Waiting for async job (vm=%p name=%s)", obj, obj->def->name);
6472
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->parent.lock, then) < 0)
6473 6474 6475
            goto error;
    }

6476
    while (!qemuDomainObjCanSetJob(priv, job, agentJob)) {
6477 6478 6479
        if (nowait)
            goto cleanup;

6480
        VIR_DEBUG("Waiting for job (vm=%p name=%s)", obj, obj->def->name);
6481
        if (virCondWaitUntil(&priv->job.cond, &obj->parent.lock, then) < 0)
6482
            goto error;
6483
    }
6484 6485 6486

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
6487
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
6488 6489
        goto retry;

J
Jiri Denemark 已提交
6490 6491
    ignore_value(virTimeMillisNow(&now));

6492 6493 6494 6495 6496 6497 6498 6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517 6518 6519 6520 6521 6522 6523 6524 6525 6526 6527 6528 6529 6530 6531
    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;
6532
    }
6533

6534 6535
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
6536

6537
    virObjectUnref(cfg);
6538
    return 0;
6539

6540
 error:
J
Jiri Denemark 已提交
6541 6542 6543
    ignore_value(virTimeMillisNow(&now));
    if (priv->job.active && priv->job.started)
        duration = now - priv->job.started;
6544 6545
    if (priv->job.agentActive && priv->job.agentStarted)
        agentDuration = now - priv->job.agentStarted;
J
Jiri Denemark 已提交
6546 6547 6548
    if (priv->job.asyncJob && priv->job.asyncStarted)
        asyncDuration = now - priv->job.asyncStarted;

6549 6550 6551 6552
    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)",
6553
             qemuDomainJobTypeToString(job),
6554
             qemuDomainAgentJobTypeToString(agentJob),
6555 6556 6557
             qemuDomainAsyncJobTypeToString(asyncJob),
             obj->def->name,
             qemuDomainJobTypeToString(priv->job.active),
6558
             qemuDomainAgentJobTypeToString(priv->job.agentActive),
6559
             qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
6560
             priv->job.owner, NULLSTR(priv->job.ownerAPI),
6561
             priv->job.agentOwner, NULLSTR(priv->job.agentOwnerAPI),
J
Jiri Denemark 已提交
6562
             priv->job.asyncOwner, NULLSTR(priv->job.asyncOwnerAPI),
6563
             priv->job.apiFlags,
6564
             duration / 1000, agentDuration / 1000, asyncDuration / 1000);
6565

6566 6567 6568 6569 6570 6571 6572 6573 6574
    if (job) {
        if (nested || qemuDomainNestedJobAllowed(priv, job))
            blocker = priv->job.ownerAPI;
        else
            blocker = priv->job.asyncOwnerAPI;
    }

    if (agentJob)
        agentBlocker = priv->job.agentOwnerAPI;
6575

6576
    if (errno == ETIMEDOUT) {
6577 6578 6579 6580 6581 6582
        if (blocker && agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by monitor=%s agent=%s)"),
                           blocker, agentBlocker);
        } else if (blocker) {
6583
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
6584 6585
                           _("cannot acquire state change "
                             "lock (held by monitor=%s)"),
6586
                           blocker);
6587 6588 6589 6590 6591
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change "
                             "lock (held by agent=%s)"),
                           agentBlocker);
6592 6593 6594 6595
        } else {
            virReportError(VIR_ERR_OPERATION_TIMEOUT, "%s",
                           _("cannot acquire state change lock"));
        }
6596 6597 6598
        ret = -2;
    } else if (cfg->maxQueuedJobs &&
               priv->jobs_queued > cfg->maxQueuedJobs) {
6599 6600 6601 6602 6603 6604 6605
        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) {
6606
            virReportError(VIR_ERR_OPERATION_FAILED,
6607 6608
                           _("cannot acquire state change "
                             "lock (held by monitor=%s) "
6609 6610
                             "due to max_queued limit"),
                           blocker);
6611 6612 6613 6614 6615 6616
        } else if (agentBlocker) {
            virReportError(VIR_ERR_OPERATION_FAILED,
                           _("cannot acquire state change "
                             "lock (held by agent=%s) "
                             "due to max_queued limit"),
                           agentBlocker);
6617 6618 6619 6620 6621
        } else {
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("cannot acquire state change lock "
                             "due to max_queued limit"));
        }
6622 6623
        ret = -2;
    } else {
6624
        virReportSystemError(errno, "%s", _("cannot acquire job mutex"));
6625
    }
J
Jiri Denemark 已提交
6626 6627

 cleanup:
6628
    priv->jobs_queued--;
6629
    virObjectUnref(cfg);
6630
    return ret;
6631 6632 6633
}

/*
6634
 * obj must be locked before calling
6635 6636 6637 6638
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
6639
 * Successful calls must be followed by EndJob eventually
6640
 */
6641
int qemuDomainObjBeginJob(virQEMUDriverPtr driver,
6642
                          virDomainObjPtr obj,
6643
                          qemuDomainJob job)
6644
{
6645
    if (qemuDomainObjBeginJobInternal(driver, obj, job,
6646
                                      QEMU_AGENT_JOB_NONE,
6647
                                      QEMU_ASYNC_JOB_NONE, false) < 0)
6648 6649 6650
        return -1;
    else
        return 0;
6651 6652
}

6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676 6677 6678 6679 6680 6681 6682 6683 6684 6685 6686 6687 6688 6689
/**
 * 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);
}

6690
int qemuDomainObjBeginAsyncJob(virQEMUDriverPtr driver,
6691
                               virDomainObjPtr obj,
6692
                               qemuDomainAsyncJob asyncJob,
6693 6694
                               virDomainJobOperation operation,
                               unsigned long apiFlags)
6695
{
6696 6697
    qemuDomainObjPrivatePtr priv;

6698
    if (qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_ASYNC,
6699
                                      QEMU_AGENT_JOB_NONE,
6700
                                      asyncJob, false) < 0)
6701
        return -1;
6702 6703 6704

    priv = obj->privateData;
    priv->job.current->operation = operation;
6705
    priv->job.apiFlags = apiFlags;
6706
    return 0;
6707 6708
}

6709
int
6710 6711
qemuDomainObjBeginNestedJob(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
6712
                            qemuDomainAsyncJob asyncJob)
6713 6714 6715 6716 6717
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (asyncJob != priv->job.asyncJob) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
6718 6719
                       _("unexpected async job %d type expected %d"),
                       asyncJob, priv->job.asyncJob);
6720 6721 6722 6723
        return -1;
    }

    if (priv->job.asyncOwner != virThreadSelfID()) {
6724
        VIR_WARN("This thread doesn't seem to be the async job owner: %llu",
6725 6726 6727 6728 6729
                 priv->job.asyncOwner);
    }

    return qemuDomainObjBeginJobInternal(driver, obj,
                                         QEMU_JOB_ASYNC_NESTED,
6730
                                         QEMU_AGENT_JOB_NONE,
6731 6732
                                         QEMU_ASYNC_JOB_NONE,
                                         false);
6733 6734
}

6735 6736 6737 6738 6739 6740 6741 6742 6743 6744 6745 6746 6747 6748 6749 6750 6751 6752 6753
/**
 * 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,
6754
                                         QEMU_AGENT_JOB_NONE,
6755 6756
                                         QEMU_ASYNC_JOB_NONE, true);
}
6757

6758
/*
6759
 * obj must be locked and have a reference before calling
6760 6761 6762 6763
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 */
6764 6765
void
qemuDomainObjEndJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
6766 6767
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
6768
    qemuDomainJob job = priv->job.active;
6769

6770 6771
    priv->jobs_queued--;

6772
    VIR_DEBUG("Stopping job: %s (async=%s vm=%p name=%s)",
6773
              qemuDomainJobTypeToString(job),
6774 6775
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
6776

6777
    qemuDomainObjResetJob(priv);
6778 6779
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
6780 6781 6782 6783 6784 6785 6786 6787 6788 6789 6790 6791 6792 6793 6794 6795 6796 6797 6798 6799 6800 6801 6802 6803 6804 6805 6806 6807 6808 6809 6810 6811 6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824 6825 6826
    /* 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);
6827 6828
}

6829
void
6830
qemuDomainObjEndAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
6831 6832
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
6833

6834 6835
    priv->jobs_queued--;

6836 6837 6838
    VIR_DEBUG("Stopping async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
6839

6840
    qemuDomainObjResetAsyncJob(priv);
6841
    qemuDomainObjSaveJob(driver, obj);
6842 6843 6844
    virCondBroadcast(&priv->job.asyncCond);
}

6845 6846 6847 6848 6849
void
qemuDomainObjAbortAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

6850 6851 6852
    VIR_DEBUG("Requesting abort of async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
6853

6854 6855
    priv->job.abortJob = true;
    virDomainObjBroadcast(obj);
6856 6857
}

6858 6859 6860 6861
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU monitor API call
6862 6863 6864
 * Must have already either called qemuDomainObjBeginJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active; may not be used for nested async jobs.
6865 6866 6867
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
6868
static int
6869
qemuDomainObjEnterMonitorInternal(virQEMUDriverPtr driver,
6870
                                  virDomainObjPtr obj,
6871
                                  qemuDomainAsyncJob asyncJob)
6872 6873 6874
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

6875
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
6876 6877 6878
        int ret;
        if ((ret = qemuDomainObjBeginNestedJob(driver, obj, asyncJob)) < 0)
            return ret;
6879
        if (!virDomainObjIsActive(obj)) {
6880 6881
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
6882
            qemuDomainObjEndJob(driver, obj);
6883 6884
            return -1;
        }
6885 6886 6887
    } 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");
6888 6889
    }

6890 6891
    VIR_DEBUG("Entering monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
6892
    virObjectLock(priv->mon);
6893
    virObjectRef(priv->mon);
6894
    ignore_value(virTimeMillisNow(&priv->monStart));
6895
    virObjectUnlock(obj);
6896 6897

    return 0;
6898 6899
}

6900
static void ATTRIBUTE_NONNULL(1)
6901
qemuDomainObjExitMonitorInternal(virQEMUDriverPtr driver,
6902
                                 virDomainObjPtr obj)
6903 6904
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
6905
    bool hasRefs;
6906

6907
    hasRefs = virObjectUnref(priv->mon);
6908

6909
    if (hasRefs)
6910
        virObjectUnlock(priv->mon);
6911

6912
    virObjectLock(obj);
6913 6914
    VIR_DEBUG("Exited monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
6915

6916
    priv->monStart = 0;
6917
    if (!hasRefs)
6918
        priv->mon = NULL;
6919

J
Jiri Denemark 已提交
6920 6921
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjEndJob(driver, obj);
6922 6923
}

6924
void qemuDomainObjEnterMonitor(virQEMUDriverPtr driver,
6925
                               virDomainObjPtr obj)
6926
{
6927
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, obj,
6928
                                                   QEMU_ASYNC_JOB_NONE));
6929 6930
}

6931
/* obj must NOT be locked before calling
6932 6933
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
6934 6935 6936 6937 6938 6939
 *
 * 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.
6940
 */
6941 6942
int qemuDomainObjExitMonitor(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
6943
{
6944
    qemuDomainObjExitMonitorInternal(driver, obj);
6945
    if (!virDomainObjIsActive(obj)) {
6946
        if (virGetLastErrorCode() == VIR_ERR_OK)
6947 6948
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
6949 6950 6951
        return -1;
    }
    return 0;
6952
}
6953 6954

/*
6955
 * obj must be locked before calling
6956 6957
 *
 * To be called immediately before any QEMU monitor API call.
6958
 * Must have already either called qemuDomainObjBeginJob()
6959 6960 6961 6962 6963
 * 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
6964 6965 6966
 * 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).
6967 6968
 */
int
6969
qemuDomainObjEnterMonitorAsync(virQEMUDriverPtr driver,
6970
                               virDomainObjPtr obj,
6971
                               qemuDomainAsyncJob asyncJob)
6972
{
6973
    return qemuDomainObjEnterMonitorInternal(driver, obj, asyncJob);
6974 6975
}

D
Daniel P. Berrange 已提交
6976

6977 6978 6979 6980
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU agent API call.
6981 6982 6983
 * Must have already called qemuDomainObjBeginAgentJob() or
 * qemuDomainObjBeginJobWithAgent() and checked that the VM is
 * still active.
6984 6985 6986
 *
 * To be followed with qemuDomainObjExitAgent() once complete
 */
6987
qemuAgentPtr
6988
qemuDomainObjEnterAgent(virDomainObjPtr obj)
D
Daniel P. Berrange 已提交
6989 6990
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
6991
    qemuAgentPtr agent = priv->agent;
D
Daniel P. Berrange 已提交
6992

6993 6994
    VIR_DEBUG("Entering agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
6995 6996 6997

    virObjectLock(agent);
    virObjectRef(agent);
6998
    virObjectUnlock(obj);
6999 7000

    return agent;
D
Daniel P. Berrange 已提交
7001 7002
}

7003 7004 7005 7006 7007 7008

/* obj must NOT be locked before calling
 *
 * Should be paired with an earlier qemuDomainObjEnterAgent() call
 */
void
7009
qemuDomainObjExitAgent(virDomainObjPtr obj, qemuAgentPtr agent)
D
Daniel P. Berrange 已提交
7010
{
7011 7012
    virObjectUnlock(agent);
    virObjectUnref(agent);
7013
    virObjectLock(obj);
D
Daniel P. Berrange 已提交
7014

7015 7016
    VIR_DEBUG("Exited agent (agent=%p vm=%p name=%s)",
              agent, obj, obj->def->name);
D
Daniel P. Berrange 已提交
7017 7018
}

7019
void qemuDomainObjEnterRemote(virDomainObjPtr obj)
7020
{
7021 7022
    VIR_DEBUG("Entering remote (vm=%p name=%s)",
              obj, obj->def->name);
7023
    virObjectUnlock(obj);
7024 7025
}

7026
void qemuDomainObjExitRemote(virDomainObjPtr obj)
7027
{
7028
    virObjectLock(obj);
7029 7030
    VIR_DEBUG("Exited remote (vm=%p name=%s)",
              obj, obj->def->name);
7031
}
7032 7033


7034 7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052
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;
}


7053 7054 7055 7056 7057 7058
virDomainDefPtr
qemuDomainDefCopy(virQEMUDriverPtr driver,
                  virDomainDefPtr src,
                  unsigned int flags)
{
    virDomainDefPtr ret = NULL;
7059
    char *xml;
7060

7061
    if (!(xml = qemuDomainDefFormatXML(driver, src, flags)))
7062
        return NULL;
7063

7064
    ret = qemuDomainDefFromXML(driver, xml);
7065 7066 7067 7068 7069

    VIR_FREE(xml);
    return ret;
}

7070 7071 7072 7073 7074 7075 7076

static int
qemuDomainDefFormatBufInternal(virQEMUDriverPtr driver,
                               virDomainDefPtr def,
                               virCPUDefPtr origCPU,
                               unsigned int flags,
                               virBuffer *buf)
7077
{
7078
    int ret = -1;
7079
    virDomainDefPtr copy = NULL;
7080
    virCapsPtr caps = NULL;
7081
    virQEMUCapsPtr qemuCaps = NULL;
7082 7083 7084

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;
7085

7086 7087 7088
    if (!(flags & (VIR_DOMAIN_XML_UPDATE_CPU | VIR_DOMAIN_XML_MIGRATABLE)))
        goto format;

7089
    if (!(copy = virDomainDefCopy(def, caps, driver->xmlopt, NULL,
7090 7091 7092 7093 7094
                                  flags & VIR_DOMAIN_XML_MIGRATABLE)))
        goto cleanup;

    def = copy;

7095
    /* Update guest CPU requirements according to host CPU */
7096
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) &&
7097
        def->cpu &&
7098 7099
        (def->cpu->mode != VIR_CPU_MODE_CUSTOM ||
         def->cpu->model)) {
7100 7101 7102 7103 7104 7105 7106 7107
        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)
7108 7109 7110
            goto cleanup;
    }

7111
    if ((flags & VIR_DOMAIN_XML_MIGRATABLE)) {
7112
        size_t i;
7113
        int toremove = 0;
7114
        virDomainControllerDefPtr usb = NULL, pci = NULL;
7115 7116 7117 7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129

        /* 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];
            }
        }
7130 7131 7132 7133 7134 7135 7136 7137 7138

        /* 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'/>.
7139
         */
7140
        if (ARCH_IS_X86(def->os.arch) && qemuDomainIsI440FX(def) &&
7141
            usb && usb->idx == 0 &&
7142
            (usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_DEFAULT ||
7143 7144
             usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_PIIX3_UHCI) &&
            !virDomainDeviceAliasIsUserAlias(usb->info.alias)) {
7145 7146
            VIR_DEBUG("Removing default USB controller from domain '%s'"
                      " for migration compatibility", def->name);
7147
            toremove++;
7148 7149 7150 7151 7152 7153 7154 7155 7156 7157 7158 7159 7160 7161 7162 7163 7164
        } 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 &&
7165 7166 7167
            pci->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT &&
            !virDomainDeviceAliasIsUserAlias(pci->info.alias) &&
            !pci->opts.pciopts.pcihole64) {
L
Laine Stump 已提交
7168
            VIR_DEBUG("Removing default pci-root from domain '%s'"
7169
                      " for migration compatibility", def->name);
7170
            toremove++;
7171 7172 7173 7174
        } else {
            pci = NULL;
        }

7175
        if (toremove) {
7176 7177 7178
            virDomainControllerDefPtr *controllers = def->controllers;
            int ncontrollers = def->ncontrollers;

7179
            if (VIR_ALLOC_N(def->controllers, ncontrollers - toremove) < 0) {
7180
                def->controllers = controllers;
7181 7182 7183 7184 7185
                goto cleanup;
            }

            def->ncontrollers = 0;
            for (i = 0; i < ncontrollers; i++) {
7186
                if (controllers[i] != usb && controllers[i] != pci)
7187 7188
                    def->controllers[def->ncontrollers++] = controllers[i];
            }
7189 7190 7191 7192

            VIR_FREE(controllers);
            virDomainControllerDefFree(pci);
            virDomainControllerDefFree(usb);
7193
        }
7194

7195 7196 7197 7198 7199 7200 7201 7202 7203
        /* 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;
            }
        }

7204 7205 7206 7207
        for (i = 0; i < def->nchannels; i++) {
            if (qemuDomainChrDefDropDefaultPath(def->channels[i], driver) < 0)
                goto cleanup;
        }
7208

7209 7210 7211 7212 7213 7214 7215 7216 7217
        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) {
7218
                switch ((virDomainChrSerialTargetType)serial->targetType) {
7219
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SPAPR_VIO:
7220
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SYSTEM:
7221 7222 7223 7224 7225 7226
                    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:
7227
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_SCLP:
7228 7229 7230 7231 7232 7233 7234 7235
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_NONE:
                case VIR_DOMAIN_CHR_SERIAL_TARGET_TYPE_LAST:
                    /* Nothing to do */
                    break;
                }
            }
        }

7236 7237 7238 7239 7240 7241 7242 7243 7244
        /* 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;
        }
7245 7246
    }

7247 7248
 format:
    ret = virDomainDefFormatInternal(def, caps,
7249
                                     virDomainDefFormatConvertXMLFlags(flags),
7250
                                     buf, driver->xmlopt);
7251

7252
 cleanup:
7253
    virDomainDefFree(copy);
7254
    virObjectUnref(caps);
7255
    virObjectUnref(qemuCaps);
7256 7257
    return ret;
}
7258

7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271 7272 7273 7274

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)
7275 7276 7277
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

7278
    if (qemuDomainDefFormatBufInternal(driver, def, origCPU, flags, &buf) < 0)
7279 7280 7281 7282 7283
        return NULL;

    return virBufferContentAndReset(&buf);
}

7284 7285 7286 7287 7288 7289 7290 7291 7292 7293

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


7294
char *qemuDomainFormatXML(virQEMUDriverPtr driver,
7295
                          virDomainObjPtr vm,
7296
                          unsigned int flags)
7297 7298
{
    virDomainDefPtr def;
7299 7300
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCPUDefPtr origCPU = NULL;
7301

7302
    if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef) {
7303
        def = vm->newDef;
7304
    } else {
7305
        def = vm->def;
7306
        origCPU = priv->origCPU;
7307
    }
7308

7309
    return qemuDomainDefFormatXMLInternal(driver, def, origCPU, flags);
7310 7311
}

7312
char *
7313
qemuDomainDefFormatLive(virQEMUDriverPtr driver,
7314
                        virDomainDefPtr def,
7315
                        virCPUDefPtr origCPU,
7316 7317
                        bool inactive,
                        bool compatible)
7318 7319 7320 7321 7322
{
    unsigned int flags = QEMU_DOMAIN_FORMAT_LIVE_FLAGS;

    if (inactive)
        flags |= VIR_DOMAIN_XML_INACTIVE;
7323 7324
    if (compatible)
        flags |= VIR_DOMAIN_XML_MIGRATABLE;
7325

7326
    return qemuDomainDefFormatXMLInternal(driver, def, origCPU, flags);
7327 7328
}

7329

7330
void qemuDomainObjTaint(virQEMUDriverPtr driver,
7331
                        virDomainObjPtr obj,
7332
                        virDomainTaintFlags taint,
7333
                        qemuDomainLogContextPtr logCtxt)
7334
{
7335
    virErrorPtr orig_err = NULL;
7336
    bool closeLog = false;
7337 7338
    char *timestamp = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
7339

7340 7341 7342 7343 7344 7345 7346 7347 7348 7349 7350 7351 7352 7353 7354
    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();
7355

7356 7357 7358 7359 7360 7361 7362 7363 7364
    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;
7365
        }
7366 7367 7368 7369 7370 7371 7372 7373 7374 7375 7376 7377 7378
        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)
7379
        virObjectUnref(logCtxt);
7380 7381 7382
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
7383 7384 7385 7386
    }
}


7387
void qemuDomainObjCheckTaint(virQEMUDriverPtr driver,
7388
                             virDomainObjPtr obj,
7389
                             qemuDomainLogContextPtr logCtxt)
7390
{
7391
    size_t i;
7392
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
7393
    qemuDomainObjPrivatePtr priv = obj->privateData;
7394

7395
    if (virQEMUDriverIsPrivileged(driver) &&
7396 7397 7398
        (!cfg->clearEmulatorCapabilities ||
         cfg->user == 0 ||
         cfg->group == 0))
7399
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
7400

7401
    if (priv->hookRun)
7402
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOOK, logCtxt);
7403

7404 7405 7406
    if (obj->def->namespaceData) {
        qemuDomainCmdlineDefPtr qemucmd = obj->def->namespaceData;
        if (qemucmd->num_args || qemucmd->num_env)
7407
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logCtxt);
7408 7409
    }

7410
    if (obj->def->cpu && obj->def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
7411
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOST_CPU, logCtxt);
7412

7413
    for (i = 0; i < obj->def->ndisks; i++)
7414
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logCtxt);
7415

7416 7417
    for (i = 0; i < obj->def->nhostdevs; i++)
        qemuDomainObjCheckHostdevTaint(driver, obj, obj->def->hostdevs[i],
7418
                                       logCtxt);
7419

7420
    for (i = 0; i < obj->def->nnets; i++)
7421
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logCtxt);
7422

7423
    if (obj->def->os.dtb)
7424
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_DTB, logCtxt);
7425

7426
    virObjectUnref(cfg);
7427 7428 7429
}


7430
void qemuDomainObjCheckDiskTaint(virQEMUDriverPtr driver,
7431
                                 virDomainObjPtr obj,
7432
                                 virDomainDiskDefPtr disk,
7433
                                 qemuDomainLogContextPtr logCtxt)
7434
{
7435
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
7436

7437 7438
    if (disk->rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
7439
                           logCtxt);
7440

7441 7442
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
        virStorageSourceGetActualType(disk->src) == VIR_STORAGE_TYPE_BLOCK &&
7443
        disk->src->path && virFileIsCDROM(disk->src->path) == 1)
7444
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CDROM_PASSTHROUGH,
7445
                           logCtxt);
7446

7447
    virObjectUnref(cfg);
7448 7449 7450
}


7451 7452 7453
void qemuDomainObjCheckHostdevTaint(virQEMUDriverPtr driver,
                                    virDomainObjPtr obj,
                                    virDomainHostdevDefPtr hostdev,
7454
                                    qemuDomainLogContextPtr logCtxt)
7455
{
7456
    if (!virHostdevIsSCSIDevice(hostdev))
7457
        return;
7458

7459 7460
    if (hostdev->source.subsys.u.scsi.rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
7461 7462 7463
}


7464
void qemuDomainObjCheckNetTaint(virQEMUDriverPtr driver,
7465
                                virDomainObjPtr obj,
7466
                                virDomainNetDefPtr net,
7467
                                qemuDomainLogContextPtr logCtxt)
7468
{
7469 7470 7471 7472 7473 7474
    /* 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)
7475
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logCtxt);
7476
}
7477 7478


7479 7480 7481 7482 7483 7484 7485
qemuDomainLogContextPtr qemuDomainLogContextNew(virQEMUDriverPtr driver,
                                                virDomainObjPtr vm,
                                                qemuDomainLogContextMode mode)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainLogContextPtr ctxt = NULL;

7486
    if (qemuDomainInitialize() < 0)
7487 7488 7489 7490
        goto cleanup;

    if (!(ctxt = virObjectNew(qemuDomainLogContextClass)))
        goto cleanup;
7491

7492
    VIR_DEBUG("Context new %p stdioLogD=%d", ctxt, cfg->stdioLogD);
7493 7494 7495
    ctxt->writefd = -1;
    ctxt->readfd = -1;

7496 7497 7498
    if (virAsprintf(&ctxt->path, "%s/%s.log", cfg->logDir, vm->def->name) < 0)
        goto error;

7499 7500 7501 7502
    if (cfg->stdioLogD) {
        ctxt->manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver));
        if (!ctxt->manager)
            goto error;
7503

7504 7505 7506 7507
        ctxt->writefd = virLogManagerDomainOpenLogFile(ctxt->manager,
                                                       "qemu",
                                                       vm->def->uuid,
                                                       vm->def->name,
7508
                                                       ctxt->path,
7509 7510 7511 7512 7513 7514
                                                       0,
                                                       &ctxt->inode,
                                                       &ctxt->pos);
        if (ctxt->writefd < 0)
            goto error;
    } else {
7515
        if ((ctxt->writefd = open(ctxt->path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
7516
            virReportSystemError(errno, _("failed to create logfile %s"),
7517
                                 ctxt->path);
7518 7519
            goto error;
        }
7520
        if (virSetCloseExec(ctxt->writefd) < 0) {
7521
            virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
7522
                                 ctxt->path);
7523 7524 7525
            goto error;
        }

7526 7527 7528 7529 7530 7531 7532
        /* 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"),
7533
                                 ctxt->path);
7534 7535 7536 7537
            goto error;
        }

        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START) {
7538
            if ((ctxt->readfd = open(ctxt->path, O_RDONLY, S_IRUSR | S_IWUSR)) < 0) {
7539
                virReportSystemError(errno, _("failed to open logfile %s"),
7540
                                     ctxt->path);
7541 7542 7543 7544
                goto error;
            }
            if (virSetCloseExec(ctxt->readfd) < 0) {
                virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
7545
                                     ctxt->path);
7546 7547 7548 7549
                goto error;
            }
        }

7550 7551
        if ((ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END)) < 0) {
            virReportSystemError(errno, _("failed to seek in log file %s"),
7552
                                 ctxt->path);
7553 7554 7555 7556
            goto error;
        }
    }

7557
 cleanup:
7558 7559 7560 7561
    virObjectUnref(cfg);
    return ctxt;

 error:
7562
    virObjectUnref(ctxt);
7563 7564
    ctxt = NULL;
    goto cleanup;
7565 7566 7567 7568 7569 7570 7571 7572 7573 7574 7575 7576 7577 7578
}


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;
7579 7580
    if (!ctxt->manager &&
        lseek(ctxt->writefd, 0, SEEK_END) < 0) {
7581
        virReportSystemError(errno, "%s",
7582
                             _("Unable to seek to end of domain logfile"));
7583 7584 7585 7586 7587 7588 7589 7590 7591 7592 7593 7594 7595 7596 7597 7598 7599 7600 7601 7602
        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)
{
7603 7604 7605 7606
    VIR_DEBUG("Context read %p manager=%p inode=%llu pos=%llu",
              ctxt, ctxt->manager,
              (unsigned long long)ctxt->inode,
              (unsigned long long)ctxt->pos);
7607
    char *buf;
7608 7609 7610
    size_t buflen;
    if (ctxt->manager) {
        buf = virLogManagerDomainReadLogFile(ctxt->manager,
7611
                                             ctxt->path,
7612 7613 7614 7615 7616 7617 7618 7619 7620
                                             ctxt->inode,
                                             ctxt->pos,
                                             1024 * 128,
                                             0);
        if (!buf)
            return -1;
        buflen = strlen(buf);
    } else {
        ssize_t got;
7621

7622
        buflen = 1024 * 128;
7623

7624 7625
        /* Best effort jump to start of messages */
        ignore_value(lseek(ctxt->readfd, ctxt->pos, SEEK_SET));
7626

7627 7628
        if (VIR_ALLOC_N(buf, buflen) < 0)
            return -1;
7629

7630 7631 7632 7633 7634 7635 7636 7637 7638 7639 7640 7641 7642
        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;
    }
7643 7644 7645

    *msg = buf;

7646
    return buflen;
7647 7648 7649
}


7650 7651 7652 7653 7654 7655 7656 7657 7658 7659 7660 7661 7662 7663 7664 7665 7666 7667 7668 7669 7670 7671 7672 7673 7674 7675 7676 7677 7678 7679 7680 7681 7682 7683 7684 7685 7686 7687 7688 7689 7690 7691 7692 7693 7694 7695 7696 7697 7698 7699 7700 7701 7702 7703 7704 7705 7706 7707 7708 7709 7710 7711
/**
 * 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;
}


7712 7713 7714 7715 7716 7717 7718 7719
int qemuDomainLogContextGetWriteFD(qemuDomainLogContextPtr ctxt)
{
    return ctxt->writefd;
}


void qemuDomainLogContextMarkPosition(qemuDomainLogContextPtr ctxt)
{
7720 7721
    if (ctxt->manager)
        virLogManagerDomainGetLogFilePosition(ctxt->manager,
7722
                                              ctxt->path,
7723 7724 7725 7726 7727
                                              0,
                                              &ctxt->inode,
                                              &ctxt->pos);
    else
        ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END);
7728 7729 7730
}


7731 7732 7733 7734 7735 7736
virLogManagerPtr qemuDomainLogContextGetManager(qemuDomainLogContextPtr ctxt)
{
    return ctxt->manager;
}


7737 7738
/* Locate an appropriate 'qemu-img' binary.  */
const char *
7739
qemuFindQemuImgBinary(virQEMUDriverPtr driver)
7740
{
7741 7742
    if (!driver->qemuImgBinary)
        virReportError(VIR_ERR_INTERNAL_ERROR,
7743
                       "%s", _("unable to find qemu-img"));
7744 7745 7746 7747 7748 7749 7750

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
                                virDomainSnapshotObjPtr snapshot,
7751
                                virCapsPtr caps,
7752
                                virDomainXMLOptionPtr xmlopt,
7753 7754 7755 7756 7757 7758 7759 7760 7761
                                char *snapshotDir)
{
    char *newxml = NULL;
    int ret = -1;
    char *snapDir = NULL;
    char *snapFile = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

    virUUIDFormat(vm->def->uuid, uuidstr);
7762
    newxml = virDomainSnapshotDefFormat(
7763
        uuidstr, snapshot->def, caps, xmlopt,
7764 7765
        virDomainDefFormatConvertXMLFlags(QEMU_DOMAIN_FORMAT_LIVE_FLAGS),
        1);
7766
    if (newxml == NULL)
7767 7768
        return -1;

7769
    if (virAsprintf(&snapDir, "%s/%s", snapshotDir, vm->def->name) < 0)
7770 7771 7772 7773 7774 7775 7776
        goto cleanup;
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
        goto cleanup;
    }

7777
    if (virAsprintf(&snapFile, "%s/%s.xml", snapDir, snapshot->def->name) < 0)
7778 7779
        goto cleanup;

J
Ján Tomko 已提交
7780
    ret = virXMLSaveFile(snapFile, NULL, "snapshot-edit", newxml);
7781

7782
 cleanup:
7783 7784 7785 7786 7787 7788 7789 7790
    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.  */
7791
static int
7792
qemuDomainSnapshotForEachQcow2Raw(virQEMUDriverPtr driver,
7793 7794 7795 7796 7797
                                  virDomainDefPtr def,
                                  const char *name,
                                  const char *op,
                                  bool try_all,
                                  int ndisks)
7798 7799
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
7800
    size_t i;
7801 7802 7803 7804 7805 7806 7807 7808 7809
    bool skipped = false;

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

    qemuimgarg[2] = op;
7810
    qemuimgarg[3] = name;
7811

7812
    for (i = 0; i < ndisks; i++) {
7813
        /* FIXME: we also need to handle LVM here */
7814
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
7815 7816 7817
            int format = virDomainDiskGetFormat(def->disks[i]);

            if (format > 0 && format != VIR_STORAGE_FILE_QCOW2) {
7818 7819 7820 7821 7822
                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",
7823
                             def->disks[i]->dst);
7824 7825
                    skipped = true;
                    continue;
7826 7827 7828 7829 7830
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
7831
                }
7832 7833 7834 7835
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("Disk device '%s' does not support"
                                 " snapshotting"),
                               def->disks[i]->dst);
7836 7837 7838
                return -1;
            }

7839
            qemuimgarg[4] = virDomainDiskGetSource(def->disks[i]);
7840 7841 7842 7843

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
7844
                             def->disks[i]->dst);
7845 7846
                    skipped = true;
                    continue;
7847 7848 7849 7850 7851
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
7852 7853 7854 7855 7856 7857 7858 7859 7860
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

7861 7862 7863
/* 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
7864
qemuDomainSnapshotForEachQcow2(virQEMUDriverPtr driver,
7865 7866 7867 7868 7869 7870 7871 7872 7873 7874 7875 7876 7877 7878 7879 7880
                               virDomainObjPtr vm,
                               virDomainSnapshotObjPtr snap,
                               const char *op,
                               bool try_all)
{
    /* Prefer action on the disks in use at the time the snapshot was
     * created; but fall back to current definition if dealing with a
     * snapshot created prior to libvirt 0.9.5.  */
    virDomainDefPtr def = snap->def->dom;

    if (!def)
        def = vm->def;
    return qemuDomainSnapshotForEachQcow2Raw(driver, def, snap->def->name,
                                             op, try_all, def->ndisks);
}

7881 7882
/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
7883
qemuDomainSnapshotDiscard(virQEMUDriverPtr driver,
7884 7885 7886 7887 7888 7889 7890 7891 7892
                          virDomainObjPtr vm,
                          virDomainSnapshotObjPtr snap,
                          bool update_current,
                          bool metadata_only)
{
    char *snapFile = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;
    virDomainSnapshotObjPtr parentsnap = NULL;
7893
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
7894 7895 7896 7897 7898 7899 7900 7901 7902

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
                goto cleanup;
        } else {
            priv = vm->privateData;
7903
            qemuDomainObjEnterMonitor(driver, vm);
7904 7905
            /* we continue on even in the face of error */
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
7906
            ignore_value(qemuDomainObjExitMonitor(driver, vm));
7907 7908 7909
        }
    }

7910
    if (virAsprintf(&snapFile, "%s/%s/%s.xml", cfg->snapshotDir,
7911
                    vm->def->name, snap->def->name) < 0)
7912 7913 7914 7915
        goto cleanup;

    if (snap == vm->current_snapshot) {
        if (update_current && snap->def->parent) {
7916
            parentsnap = virDomainSnapshotFindByName(vm->snapshots,
7917 7918 7919 7920 7921 7922
                                                     snap->def->parent);
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
                         snap->def->parent);
            } else {
                parentsnap->def->current = true;
7923
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap, driver->caps,
7924
                                                    driver->xmlopt,
7925
                                                    cfg->snapshotDir) < 0) {
7926 7927 7928 7929 7930 7931 7932 7933 7934 7935 7936 7937
                    VIR_WARN("failed to set parent snapshot '%s' as current",
                             snap->def->parent);
                    parentsnap->def->current = false;
                    parentsnap = NULL;
                }
            }
        }
        vm->current_snapshot = parentsnap;
    }

    if (unlink(snapFile) < 0)
        VIR_WARN("Failed to unlink %s", snapFile);
7938
    virDomainSnapshotObjListRemove(vm->snapshots, snap);
7939 7940 7941

    ret = 0;

7942
 cleanup:
7943
    VIR_FREE(snapFile);
7944
    virObjectUnref(cfg);
7945 7946 7947 7948
    return ret;
}

/* Hash iterator callback to discard multiple snapshots.  */
7949 7950 7951
int qemuDomainSnapshotDiscardAll(void *payload,
                                 const void *name ATTRIBUTE_UNUSED,
                                 void *data)
7952 7953
{
    virDomainSnapshotObjPtr snap = payload;
7954
    virQEMUSnapRemovePtr curr = data;
7955 7956 7957 7958 7959 7960 7961 7962
    int err;

    if (snap->def->current)
        curr->current = true;
    err = qemuDomainSnapshotDiscard(curr->driver, curr->vm, snap, false,
                                    curr->metadata_only);
    if (err && !curr->err)
        curr->err = err;
7963
    return 0;
7964 7965 7966
}

int
7967
qemuDomainSnapshotDiscardAllMetadata(virQEMUDriverPtr driver,
7968 7969
                                     virDomainObjPtr vm)
{
7970
    virQEMUSnapRemove rem;
7971 7972 7973 7974 7975

    rem.driver = driver;
    rem.vm = vm;
    rem.metadata_only = true;
    rem.err = 0;
7976 7977
    virDomainSnapshotForEach(vm->snapshots, qemuDomainSnapshotDiscardAll,
                             &rem);
7978 7979 7980 7981

    return rem.err;
}

7982 7983 7984 7985 7986

/**
 * qemuDomainRemoveInactive:
 *
 * The caller must hold a lock to the vm.
7987 7988
 */
void
7989
qemuDomainRemoveInactive(virQEMUDriverPtr driver,
7990 7991
                         virDomainObjPtr vm)
{
7992
    char *snapDir;
7993 7994 7995 7996 7997 7998 7999 8000
    virQEMUDriverConfigPtr cfg;

    if (vm->persistent) {
        /* Short-circuit, we don't want to remove a persistent domain */
        return;
    }

    cfg = virQEMUDriverGetConfig(driver);
8001

8002 8003 8004 8005 8006
    /* 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);
    }
8007
    else if (virAsprintf(&snapDir, "%s/%s", cfg->snapshotDir,
8008 8009
                         vm->def->name) < 0) {
        VIR_WARN("unable to remove snapshot directory %s/%s",
8010
                 cfg->snapshotDir, vm->def->name);
8011 8012 8013 8014 8015
    } else {
        if (rmdir(snapDir) < 0 && errno != ENOENT)
            VIR_WARN("unable to remove snapshot directory %s", snapDir);
        VIR_FREE(snapDir);
    }
8016
    qemuExtDevicesCleanupHost(driver, vm->def);
8017

8018
    virDomainObjListRemove(driver->domains, vm);
8019

8020
    virObjectUnref(cfg);
8021 8022 8023 8024 8025 8026 8027 8028 8029 8030 8031 8032 8033 8034 8035 8036 8037 8038 8039 8040
}


/**
 * 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);
8041 8042

    if (haveJob)
8043
        qemuDomainObjEndJob(driver, vm);
8044
}
8045

8046

8047
void
8048
qemuDomainSetFakeReboot(virQEMUDriverPtr driver,
8049 8050 8051 8052
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
8053
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8054 8055

    if (priv->fakeReboot == value)
8056
        goto cleanup;
8057 8058 8059

    priv->fakeReboot = value;

8060
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
8061
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
8062

8063
 cleanup:
8064
    virObjectUnref(cfg);
8065
}
M
Michal Privoznik 已提交
8066

8067
static void
8068 8069
qemuDomainCheckRemoveOptionalDisk(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
8070
                                  size_t diskIndex)
8071 8072
{
    char uuid[VIR_UUID_STRING_BUFLEN];
8073
    virObjectEventPtr event = NULL;
8074
    virDomainDiskDefPtr disk = vm->def->disks[diskIndex];
8075
    const char *src = virDomainDiskGetSource(disk);
8076 8077 8078 8079 8080

    virUUIDFormat(vm->def->uuid, uuid);

    VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
              "due to inaccessible source '%s'",
8081
              disk->dst, vm->def->name, uuid, src);
8082 8083 8084 8085

    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM ||
        disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY) {

8086
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
8087 8088
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
8089
        virDomainDiskEmptySource(disk);
8090 8091
        /* keeping the old startup policy would be invalid for new images */
        disk->startupPolicy = VIR_DOMAIN_STARTUP_POLICY_DEFAULT;
8092
    } else {
8093
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
8094 8095
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_DROP_MISSING_ON_START);
8096 8097
        virDomainDiskRemove(vm->def, diskIndex);
        virDomainDiskDefFree(disk);
8098 8099
    }

8100
    virObjectEventStateQueue(driver->domainEventState, event);
8101 8102
}

8103 8104 8105 8106 8107 8108 8109 8110 8111 8112 8113 8114 8115 8116 8117 8118 8119

/**
 * 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
8120 8121
qemuDomainCheckDiskStartupPolicy(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
8122
                                 size_t diskIndex,
8123 8124
                                 bool cold_boot)
{
8125
    int startupPolicy = vm->def->disks[diskIndex]->startupPolicy;
8126
    int device = vm->def->disks[diskIndex]->device;
8127

8128
    switch ((virDomainStartupPolicy) startupPolicy) {
8129
        case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
8130 8131 8132 8133 8134 8135
            /* 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)
8136
                return -1;
8137 8138
            break;

8139
        case VIR_DOMAIN_STARTUP_POLICY_DEFAULT:
8140
        case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
8141
            return -1;
8142 8143

        case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
8144
            if (cold_boot)
8145
                return -1;
8146 8147 8148 8149 8150 8151 8152
            break;

        case VIR_DOMAIN_STARTUP_POLICY_LAST:
            /* this should never happen */
            break;
    }

8153
    qemuDomainCheckRemoveOptionalDisk(driver, vm, diskIndex);
8154
    virResetLastError();
8155 8156 8157
    return 0;
}

8158

8159 8160 8161 8162 8163 8164 8165 8166 8167 8168

/*
 * The vm must be locked when any of the following cleanup functions is
 * called.
 */
int
qemuDomainCleanupAdd(virDomainObjPtr vm,
                     qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
8169
    size_t i;
8170 8171 8172 8173 8174 8175 8176 8177 8178 8179

    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,
8180
                     priv->ncleanupCallbacks, 1) < 0)
8181 8182 8183 8184 8185 8186 8187 8188 8189 8190 8191
        return -1;

    priv->cleanupCallbacks[priv->ncleanupCallbacks++] = cb;
    return 0;
}

void
qemuDomainCleanupRemove(virDomainObjPtr vm,
                        qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
8192
    size_t i;
8193 8194 8195 8196

    VIR_DEBUG("vm=%s, cb=%p", vm->def->name, cb);

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
8197 8198 8199
        if (priv->cleanupCallbacks[i] == cb)
            VIR_DELETE_ELEMENT_INPLACE(priv->cleanupCallbacks,
                                       i, priv->ncleanupCallbacks);
8200 8201 8202 8203 8204 8205 8206 8207
    }

    VIR_SHRINK_N(priv->cleanupCallbacks,
                 priv->ncleanupCallbacks_max,
                 priv->ncleanupCallbacks_max - priv->ncleanupCallbacks);
}

void
8208
qemuDomainCleanupRun(virQEMUDriverPtr driver,
8209 8210 8211
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
8212
    size_t i;
8213 8214 8215 8216

    VIR_DEBUG("driver=%p, vm=%s", driver, vm->def->name);

    /* run cleanup callbacks in reverse order */
8217 8218
    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[priv->ncleanupCallbacks - (i + 1)])
8219 8220 8221 8222 8223 8224 8225
            priv->cleanupCallbacks[i](driver, vm);
    }

    VIR_FREE(priv->cleanupCallbacks);
    priv->ncleanupCallbacks = 0;
    priv->ncleanupCallbacks_max = 0;
}
8226

8227 8228 8229
static void
qemuDomainGetImageIds(virQEMUDriverConfigPtr cfg,
                      virDomainObjPtr vm,
8230
                      virStorageSourcePtr src,
8231
                      virStorageSourcePtr parentSrc,
8232 8233 8234 8235 8236 8237 8238 8239 8240 8241 8242 8243 8244 8245 8246 8247 8248 8249
                      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;
    }

8250 8251
    if (vm && (vmlabel = virDomainDefGetSecurityLabelDef(vm->def, "dac")) &&
        vmlabel->label)
8252 8253
        virParseOwnershipIds(vmlabel->label, uid, gid);

8254 8255 8256 8257 8258
    if (parentSrc &&
        (disklabel = virStorageSourceGetSecurityLabelDef(parentSrc, "dac")) &&
        disklabel->label)
        virParseOwnershipIds(disklabel->label, uid, gid);

8259
    if ((disklabel = virStorageSourceGetSecurityLabelDef(src, "dac")) &&
8260
        disklabel->label)
8261 8262 8263 8264
        virParseOwnershipIds(disklabel->label, uid, gid);
}


8265 8266 8267
int
qemuDomainStorageFileInit(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
8268 8269
                          virStorageSourcePtr src,
                          virStorageSourcePtr parent)
8270 8271 8272 8273 8274 8275
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    uid_t uid;
    gid_t gid;
    int ret = -1;

8276
    qemuDomainGetImageIds(cfg, vm, src, parent, &uid, &gid);
8277 8278 8279 8280 8281 8282 8283 8284 8285 8286 8287 8288

    if (virStorageFileInitAs(src, uid, gid) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
}


8289 8290 8291 8292 8293
char *
qemuDomainStorageAlias(const char *device, int depth)
{
    char *alias;

8294
    device = qemuAliasDiskDriveSkipPrefix(device);
8295 8296 8297 8298 8299 8300 8301 8302 8303

    if (!depth)
        ignore_value(VIR_STRDUP(alias, device));
    else
        ignore_value(virAsprintf(&alias, "%s.%d", device, depth));
    return alias;
}


8304
int
8305
qemuDomainDetermineDiskChain(virQEMUDriverPtr driver,
8306
                             virDomainObjPtr vm,
8307
                             virDomainDiskDefPtr disk,
8308 8309
                             bool force_probe,
                             bool report_broken)
8310
{
8311
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
8312
    virStorageSourcePtr src = disk->src;
8313
    virStorageSourcePtr n;
8314
    qemuDomainObjPrivatePtr priv = vm->privateData;
8315
    int ret = -1;
8316 8317
    uid_t uid;
    gid_t gid;
8318

8319 8320
    if (virStorageSourceIsEmpty(src)) {
        ret = 0;
8321
        goto cleanup;
8322
    }
8323

8324 8325
    if (force_probe)
        virStorageSourceBackingStoreClear(src);
8326

8327 8328 8329 8330 8331 8332 8333 8334 8335 8336 8337 8338 8339
    /* There is no need to check the backing chain for disks without backing
     * support */
    if (virStorageSourceIsLocalStorage(src) &&
        src->format > VIR_STORAGE_FILE_NONE &&
        src->format < VIR_STORAGE_FILE_BACKING) {

        if (!virFileExists(src->path)) {
            if (report_broken)
                virStorageFileReportBrokenChain(errno, src, disk->src);

            goto cleanup;
        }

8340 8341 8342 8343 8344
        /* terminate the chain for such images as the code below would do */
        if (!src->backingStore &&
            VIR_ALLOC(src->backingStore) < 0)
            goto cleanup;

8345 8346 8347 8348 8349 8350 8351 8352
        /* 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 &&
            src->format == VIR_STORAGE_FILE_RAW &&
            virStorageSourceIsBlockLocal(src) &&
            virFileIsCDROM(src->path) == 1)
            src->hostcdrom = true;

8353 8354 8355 8356
        ret = 0;
        goto cleanup;
    }

8357 8358
    /* skip to the end of the chain if there is any */
    while (virStorageSourceHasBacking(src)) {
8359 8360
        if (report_broken) {
            int rv = virStorageFileSupportsAccess(src);
8361

8362
            if (rv < 0)
8363
                goto cleanup;
8364

8365 8366 8367 8368 8369 8370 8371 8372 8373 8374
            if (rv > 0) {
                if (qemuDomainStorageFileInit(driver, vm, src, disk->src) < 0)
                    goto cleanup;

                if (virStorageFileAccess(src, F_OK) < 0) {
                    virStorageFileReportBrokenChain(errno, src, disk->src);
                    virStorageFileDeinit(src);
                    goto cleanup;
                }

8375
                virStorageFileDeinit(src);
8376 8377
            }
        }
8378
        src = src->backingStore;
8379 8380 8381 8382 8383 8384 8385
    }

    /* 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;
8386
    }
8387

8388
    qemuDomainGetImageIds(cfg, vm, src, disk->src, &uid, &gid);
8389

8390
    if (virStorageFileGetMetadata(src, uid, gid, report_broken) < 0)
8391 8392
        goto cleanup;

8393
    for (n = src; virStorageSourceIsBacking(n); n = n->backingStore) {
8394 8395 8396
        if (qemuDomainValidateStorageSource(n, priv->qemuCaps) < 0)
            goto cleanup;

8397 8398 8399
        if (qemuDomainPrepareDiskSourceData(disk, n, cfg, priv->qemuCaps) < 0)
            goto cleanup;
    }
8400

8401
    ret = 0;
8402

8403
 cleanup:
8404 8405
    virObjectUnref(cfg);
    return ret;
8406
}
8407

8408

8409 8410 8411 8412 8413 8414 8415 8416 8417 8418 8419
/**
 * qemuDomainDiskChainElementRevoke:
 *
 * Revoke access to a single backing chain element. This restores the labels,
 * removes cgroup ACLs for devices and removes locks.
 */
void
qemuDomainDiskChainElementRevoke(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
                                 virStorageSourcePtr elem)
{
8420 8421 8422 8423
    if (qemuTeardownImageCgroup(vm, elem) < 0)
        VIR_WARN("Failed to teardown cgroup for disk path %s",
                 NULLSTR(elem->path));

8424
    if (qemuSecurityRestoreImageLabel(driver, vm, elem) < 0)
8425 8426
        VIR_WARN("Unable to restore security label on %s", NULLSTR(elem->path));

8427
    if (qemuDomainNamespaceTeardownDisk(vm, elem) < 0)
8428
        VIR_WARN("Unable to remove /dev entry for %s", NULLSTR(elem->path));
8429 8430 8431 8432 8433 8434 8435 8436

    if (virDomainLockImageDetach(driver->lockManager, vm, elem) < 0)
        VIR_WARN("Unable to release lock on %s", NULLSTR(elem->path));
}


/**
 * qemuDomainDiskChainElementPrepare:
8437 8438 8439 8440 8441
 * @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
8442 8443 8444
 *
 * 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
8445 8446 8447 8448 8449
 * 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.
 */
8450 8451 8452 8453
int
qemuDomainDiskChainElementPrepare(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
                                  virStorageSourcePtr elem,
8454 8455
                                  bool readonly,
                                  bool newSource)
8456 8457 8458 8459 8460 8461 8462 8463 8464 8465 8466 8467
{
    bool was_readonly = elem->readonly;
    virQEMUDriverConfigPtr cfg = NULL;
    int ret = -1;

    cfg = virQEMUDriverGetConfig(driver);

    elem->readonly = readonly;

    if (virDomainLockImageAttach(driver->lockManager, cfg->uri, vm, elem) < 0)
        goto cleanup;

8468
    if (newSource &&
8469
        qemuDomainNamespaceSetupDisk(vm, elem) < 0)
8470 8471
        goto cleanup;

8472 8473 8474
    if (qemuSetupImageCgroup(vm, elem) < 0)
        goto cleanup;

8475
    if (qemuSecuritySetImageLabel(driver, vm, elem) < 0)
8476 8477 8478 8479 8480 8481 8482 8483 8484 8485 8486
        goto cleanup;

    ret = 0;

 cleanup:
    elem->readonly = was_readonly;
    virObjectUnref(cfg);
    return ret;
}


8487
bool
8488
qemuDomainDiskSourceDiffers(virDomainDiskDefPtr disk,
8489 8490 8491 8492 8493 8494 8495 8496 8497 8498 8499 8500 8501 8502 8503
                            virDomainDiskDefPtr origDisk)
{
    char *diskSrc = NULL, *origDiskSrc = NULL;
    bool diskEmpty, origDiskEmpty;
    bool ret = true;

    diskEmpty = virStorageSourceIsEmpty(disk->src);
    origDiskEmpty = virStorageSourceIsEmpty(origDisk->src);

    if (diskEmpty && origDiskEmpty)
        return false;

    if (diskEmpty ^ origDiskEmpty)
        return true;

8504 8505 8506 8507
    /* This won't be a network storage, so no need to get the diskPriv
     * in order to fetch the secret, thus NULL for param2 */
    if (qemuGetDriveSourceString(disk->src, NULL, &diskSrc) < 0 ||
        qemuGetDriveSourceString(origDisk->src, NULL, &origDiskSrc) < 0)
8508 8509 8510 8511 8512 8513 8514 8515 8516 8517 8518 8519 8520
        goto cleanup;

    /* So far in qemu disk sources are considered different
     * if either path to disk or its format changes. */
    ret = virDomainDiskGetFormat(disk) != virDomainDiskGetFormat(origDisk) ||
          STRNEQ_NULLABLE(diskSrc, origDiskSrc);
 cleanup:
    VIR_FREE(diskSrc);
    VIR_FREE(origDiskSrc);
    return ret;
}


8521 8522 8523 8524 8525 8526 8527 8528 8529 8530
/*
 * 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)
{
8531 8532 8533 8534 8535 8536 8537 8538 8539 8540
#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; \
        } \
8541 8542 8543
    } while (0)

    CHECK_EQ(device, "device", false);
8544 8545 8546 8547
    CHECK_EQ(bus, "bus", false);
    if (STRNEQ(disk->dst, orig_disk->dst)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
8548
                       "target");
8549 8550 8551 8552
        return false;
    }
    CHECK_EQ(tray_status, "tray", true);
    CHECK_EQ(removable, "removable", true);
8553 8554 8555 8556 8557 8558 8559 8560 8561 8562 8563 8564 8565 8566 8567 8568 8569 8570 8571 8572 8573 8574 8575 8576 8577 8578 8579 8580 8581 8582 8583 8584 8585 8586 8587 8588 8589 8590 8591 8592 8593 8594 8595 8596 8597 8598 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

    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);

    if (disk->serial && STRNEQ_NULLABLE(disk->serial, orig_disk->serial)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "serial");
        return false;
    }

    if (disk->wwn && STRNEQ_NULLABLE(disk->wwn, orig_disk->wwn)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "wwn");
        return false;
    }

    if (disk->vendor && STRNEQ_NULLABLE(disk->vendor, orig_disk->vendor)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "vendor");
        return false;
    }

    if (disk->product && STRNEQ_NULLABLE(disk->product, orig_disk->product)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "product");
        return false;
    }

8636 8637 8638 8639 8640 8641 8642
    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);
8643
    /* "snapshot" is a libvirt internal field and thus can be changed */
8644
    /* startupPolicy is allowed to be updated. Therefore not checked here. */
8645
    CHECK_EQ(transient, "transient", true);
8646 8647 8648 8649 8650 8651 8652 8653 8654 8655 8656 8657 8658

    /* 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;
    }

8659
    /* device alias is checked already in virDomainDefCompatibleDevice */
8660

8661
    CHECK_EQ(info.bootIndex, "boot order", true);
8662 8663 8664 8665 8666 8667 8668 8669 8670 8671 8672 8673
    CHECK_EQ(rawio, "rawio", true);
    CHECK_EQ(sgio, "sgio", true);
    CHECK_EQ(discard, "discard", true);
    CHECK_EQ(iothread, "iothread", true);

    if (disk->domain_name &&
        STRNEQ_NULLABLE(disk->domain_name, orig_disk->domain_name)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "backenddomain");
        return false;
    }
8674

8675 8676 8677 8678 8679 8680
    /* 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);

8681 8682 8683 8684 8685 8686 8687 8688
    if (!virStoragePRDefIsEqual(disk->src->pr,
                                orig_disk->src->pr)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "reservations");
        return false;
    }

8689 8690 8691 8692 8693
#undef CHECK_EQ

    return true;
}

8694 8695 8696
bool
qemuDomainDiskBlockJobIsActive(virDomainDiskDefPtr disk)
{
8697 8698
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

8699 8700 8701 8702 8703 8704 8705 8706
    if (disk->mirror) {
        virReportError(VIR_ERR_BLOCK_COPY_ACTIVE,
                       _("disk '%s' already in active block job"),
                       disk->dst);

        return true;
    }

8707
    if (diskPriv->blockjob) {
8708 8709 8710 8711 8712 8713 8714 8715 8716 8717
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("disk '%s' already in active block job"),
                       disk->dst);
        return true;
    }

    return false;
}


8718 8719 8720 8721 8722 8723 8724 8725 8726 8727 8728 8729 8730 8731 8732
/**
 * 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++) {
8733 8734 8735 8736
        virDomainDiskDefPtr disk = vm->def->disks[i];
        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

        if (!copy_only && diskPriv->blockjob)
8737 8738
            return true;

8739
        if (disk->mirror && disk->mirrorJob == VIR_DOMAIN_BLOCK_JOB_TYPE_COPY)
8740 8741 8742 8743 8744 8745 8746
            return true;
    }

    return false;
}


8747 8748
int
qemuDomainUpdateDeviceList(virQEMUDriverPtr driver,
8749 8750
                           virDomainObjPtr vm,
                           int asyncJob)
8751 8752 8753
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char **aliases;
8754
    int rc;
8755 8756 8757 8758

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

8759 8760
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
8761 8762 8763 8764
    rc = qemuMonitorGetDeviceAliases(priv->mon, &aliases);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
    if (rc < 0)
8765 8766
        return -1;

8767
    virStringListFree(priv->qemuDevices);
8768 8769 8770
    priv->qemuDevices = aliases;
    return 0;
}
8771

8772 8773 8774 8775 8776 8777 8778 8779 8780 8781 8782 8783 8784 8785 8786 8787 8788 8789 8790

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);

8791 8792
    if (qemuDomainObjExitMonitor(driver, vm) < 0) {
        virHashFree(meminfo);
8793
        return -1;
8794
    }
8795 8796 8797 8798 8799 8800 8801 8802 8803 8804 8805 8806 8807 8808 8809 8810 8811 8812 8813 8814 8815 8816 8817 8818 8819 8820 8821 8822

    /* 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;
}


8823 8824 8825 8826
static bool
qemuDomainABIStabilityCheck(const virDomainDef *src,
                            const virDomainDef *dst)
{
8827 8828
    size_t i;

8829 8830 8831 8832 8833 8834 8835 8836 8837
    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;
    }

8838 8839 8840 8841 8842 8843 8844 8845 8846 8847 8848 8849 8850
    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;
        }
    }

8851 8852 8853 8854 8855 8856 8857 8858 8859
    return true;
}


virDomainABIStability virQEMUDriverDomainABIStability = {
    .domain = qemuDomainABIStabilityCheck,
};


8860 8861 8862 8863 8864 8865 8866 8867 8868 8869 8870 8871 8872 8873 8874 8875 8876 8877 8878 8879
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;
}


8880 8881 8882
#define COPY_FLAGS (VIR_DOMAIN_XML_SECURE | \
                    VIR_DOMAIN_XML_MIGRATABLE)

8883 8884 8885 8886 8887 8888 8889 8890 8891
bool
qemuDomainDefCheckABIStability(virQEMUDriverPtr driver,
                               virDomainDefPtr src,
                               virDomainDefPtr dst)
{
    virDomainDefPtr migratableDefSrc = NULL;
    virDomainDefPtr migratableDefDst = NULL;
    bool ret = false;

8892 8893
    if (!(migratableDefSrc = qemuDomainDefCopy(driver, src, COPY_FLAGS)) ||
        !(migratableDefDst = qemuDomainDefCopy(driver, dst, COPY_FLAGS)))
8894 8895
        goto cleanup;

8896 8897 8898
    ret = qemuDomainMigratableDefCheckABIStability(driver,
                                                   src, migratableDefSrc,
                                                   dst, migratableDefDst);
8899

8900
 cleanup:
8901 8902 8903 8904
    virDomainDefFree(migratableDefSrc);
    virDomainDefFree(migratableDefDst);
    return ret;
}
8905

8906 8907 8908 8909 8910 8911 8912 8913 8914 8915 8916 8917 8918 8919 8920 8921 8922 8923 8924 8925 8926 8927 8928 8929 8930 8931 8932 8933 8934 8935

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


8936
bool
8937
qemuDomainAgentAvailable(virDomainObjPtr vm,
8938 8939
                         bool reportError)
{
8940 8941
    qemuDomainObjPrivatePtr priv = vm->privateData;

8942 8943 8944 8945 8946 8947 8948
    if (virDomainObjGetState(vm, NULL) != VIR_DOMAIN_RUNNING) {
        if (reportError) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
        }
        return false;
    }
8949 8950 8951 8952 8953 8954 8955 8956 8957
    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) {
8958 8959 8960 8961 8962 8963 8964 8965 8966 8967 8968 8969
        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;
8970 8971 8972 8973
        }
    }
    return true;
}
8974

8975

8976
static unsigned long long
8977
qemuDomainGetMemorySizeAlignment(virDomainDefPtr def)
8978
{
8979 8980 8981 8982 8983
    /* 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;

8984 8985 8986 8987 8988 8989 8990
    /* Align memory size. QEMU requires rounding to next 4KiB block.
     * We'll take the "traditional" path and round it to 1MiB*/

    return 1024;
}


8991 8992 8993 8994 8995 8996 8997 8998 8999 9000 9001 9002 9003 9004 9005
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;
}


9006 9007 9008
int
qemuDomainAlignMemorySizes(virDomainDefPtr def)
{
9009 9010
    unsigned long long maxmemkb = virMemoryMaxValue(false) >> 10;
    unsigned long long maxmemcapped = virMemoryMaxValue(true) >> 10;
9011
    unsigned long long initialmem = 0;
9012
    unsigned long long hotplugmem = 0;
9013
    unsigned long long mem;
9014
    unsigned long long align = qemuDomainGetMemorySizeAlignment(def);
9015 9016 9017 9018 9019
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    /* align NUMA cell sizes if relevant */
    for (i = 0; i < ncells; i++) {
9020 9021
        mem = VIR_ROUND_UP(virDomainNumaGetNodeMemorySize(def->numa, i), align);
        initialmem += mem;
9022 9023 9024 9025 9026 9027 9028

        if (mem > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("memory size of NUMA node '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
9029
        virDomainNumaSetNodeMemorySize(def->numa, i, mem);
9030 9031
    }

9032 9033 9034 9035 9036
    /* 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);

9037 9038 9039 9040 9041 9042
    if (initialmem > maxmemcapped) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("initial memory size overflowed after alignment"));
        return -1;
    }

9043
    def->mem.max_memory = VIR_ROUND_UP(def->mem.max_memory, align);
9044 9045 9046 9047 9048
    if (def->mem.max_memory > maxmemkb) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("maximum memory size overflowed after alignment"));
        return -1;
    }
9049

9050
    /* Align memory module sizes */
9051 9052
    for (i = 0; i < def->nmems; i++) {
        align = qemuDomainGetMemoryModuleSizeAlignment(def, def->mems[i]);
9053
        def->mems[i]->size = VIR_ROUND_UP(def->mems[i]->size, align);
9054
        hotplugmem += def->mems[i]->size;
9055 9056 9057 9058 9059 9060 9061

        if (def->mems[i]->size > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("size of memory module '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
9062
    }
9063

9064 9065
    virDomainDefSetMemoryTotal(def, initialmem + hotplugmem);

9066 9067
    return 0;
}
9068 9069 9070 9071 9072 9073 9074 9075 9076 9077 9078


/**
 * 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
9079 9080
qemuDomainMemoryDeviceAlignSize(virDomainDefPtr def,
                                virDomainMemoryDefPtr mem)
9081
{
9082
    mem->size = VIR_ROUND_UP(mem->size, qemuDomainGetMemorySizeAlignment(def));
9083
}
9084 9085 9086 9087 9088 9089 9090 9091 9092 9093 9094 9095 9096


/**
 * qemuDomainGetMonitor:
 * @vm: domain object
 *
 * Returns the monitor pointer corresponding to the domain object @vm.
 */
qemuMonitorPtr
qemuDomainGetMonitor(virDomainObjPtr vm)
{
    return ((qemuDomainObjPrivatePtr) vm->privateData)->mon;
}
9097 9098 9099 9100 9101 9102 9103


/**
 * qemuDomainSupportsBlockJobs:
 * @vm: domain object
 *
 * Returns -1 in case when qemu does not support block jobs at all. Otherwise
9104
 * returns 0.
9105 9106
 */
int
9107
qemuDomainSupportsBlockJobs(virDomainObjPtr vm)
9108 9109
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
9110
    bool asynchronous = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKJOB_ASYNC);
9111

9112
    if (!asynchronous) {
9113 9114 9115 9116 9117 9118 9119
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("block jobs not supported with this QEMU binary"));
        return -1;
    }

    return 0;
}
9120 9121 9122 9123 9124 9125 9126 9127 9128


/**
 * 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.
 */
9129
virDomainChrDefPtr
9130 9131 9132 9133 9134 9135 9136 9137 9138 9139
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;

9140 9141
        if (STREQ_NULLABLE(channel->target.name, "org.qemu.guest_agent.0"))
            return channel;
9142 9143
    }

9144
    return NULL;
9145
}
9146 9147 9148


bool
9149
qemuDomainIsQ35(const virDomainDef *def)
9150
{
9151
    return qemuDomainMachineIsQ35(def->os.machine);
9152 9153 9154 9155
}


bool
9156
qemuDomainMachineIsQ35(const char *machine)
9157
{
9158 9159
    return (STRPREFIX(machine, "pc-q35") ||
            STREQ(machine, "q35"));
9160
}
9161 9162


9163
bool
9164 9165 9166 9167 9168 9169 9170 9171 9172 9173 9174 9175 9176 9177 9178 9179 9180 9181 9182
qemuDomainIsI440FX(const virDomainDef *def)
{
    return qemuDomainMachineIsI440FX(def->os.machine);
}


bool
qemuDomainMachineIsI440FX(const char *machine)
{
    return (STREQ(machine, "pc") ||
            STRPREFIX(machine, "pc-0.") ||
            STRPREFIX(machine, "pc-1.") ||
            STRPREFIX(machine, "pc-i440") ||
            STRPREFIX(machine, "rhel"));
}


bool
qemuDomainHasPCIRoot(const virDomainDef *def)
9183 9184 9185 9186 9187 9188 9189 9190 9191 9192 9193 9194 9195 9196
{
    int root = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

    if (root < 0)
        return false;

    if (def->controllers[root]->model != VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT)
        return false;

    return true;
}


bool
9197
qemuDomainHasPCIeRoot(const virDomainDef *def)
9198 9199 9200 9201 9202 9203 9204 9205 9206 9207 9208 9209 9210
{
    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;
}


9211
bool
9212
qemuDomainNeedsFDC(const virDomainDef *def)
9213
{
9214 9215 9216 9217 9218 9219 9220 9221
    return qemuDomainMachineNeedsFDC(def->os.machine);
}


bool
qemuDomainMachineNeedsFDC(const char *machine)
{
    const char *p = STRSKIP(machine, "pc-q35-");
9222 9223 9224 9225 9226 9227 9228 9229 9230 9231 9232 9233 9234 9235

    if (p) {
        if (STRPREFIX(p, "1.") ||
            STRPREFIX(p, "2.0") ||
            STRPREFIX(p, "2.1") ||
            STRPREFIX(p, "2.2") ||
            STRPREFIX(p, "2.3"))
            return false;
        return true;
    }
    return false;
}


9236
bool
9237
qemuDomainIsS390CCW(const virDomainDef *def)
9238
{
9239
    return qemuDomainMachineIsS390CCW(def->os.machine);
9240 9241
}

9242

J
Ján Tomko 已提交
9243
bool
9244
qemuDomainMachineIsS390CCW(const char *machine)
J
Ján Tomko 已提交
9245
{
9246 9247 9248 9249 9250 9251 9252 9253 9254 9255 9256 9257 9258 9259 9260 9261 9262
    return STRPREFIX(machine, "s390-ccw");
}


bool
qemuDomainIsVirt(const virDomainDef *def)
{
    return qemuDomainMachineIsVirt(def->os.machine, def->os.arch);
}


bool
qemuDomainMachineIsVirt(const char *machine,
                        const virArch arch)
{
    if (arch != VIR_ARCH_ARMV7L &&
        arch != VIR_ARCH_AARCH64)
9263 9264
        return false;

9265 9266
    if (STRNEQ(machine, "virt") &&
        !STRPREFIX(machine, "virt-"))
9267 9268 9269
        return false;

    return true;
J
Ján Tomko 已提交
9270 9271 9272
}


9273
bool
9274
qemuDomainIsPSeries(const virDomainDef *def)
9275
{
9276 9277 9278 9279 9280 9281 9282 9283 9284
    return qemuDomainMachineIsPSeries(def->os.machine, def->os.arch);
}


bool
qemuDomainMachineIsPSeries(const char *machine,
                           const virArch arch)
{
    if (!ARCH_IS_PPC64(arch))
9285 9286
        return false;

9287 9288
    if (STRNEQ(machine, "pseries") &&
        !STRPREFIX(machine, "pseries-"))
9289 9290 9291 9292 9293 9294
        return false;

    return true;
}


9295 9296 9297 9298 9299 9300 9301 9302 9303 9304 9305 9306 9307 9308 9309 9310 9311 9312 9313 9314 9315 9316 9317 9318 9319 9320 9321 9322 9323 9324 9325 9326 9327 9328 9329 9330 9331 9332 9333
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
9334
    case VIR_DOMAIN_MEMORY_MODEL_NVDIMM:
9335 9336 9337 9338 9339 9340 9341 9342
        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;
        }

9343 9344 9345
        if (virDomainNumaGetNodeCount(def->numa) != 0) {
            if (mem->targetNode == -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
Y
Yuri Chornoivan 已提交
9346
                               _("target NUMA node needs to be specified for "
9347 9348 9349
                                 "memory device"));
                return -1;
            }
9350 9351
        }

9352 9353 9354 9355 9356 9357 9358 9359 9360 9361 9362 9363 9364 9365 9366 9367 9368 9369 9370 9371 9372 9373 9374 9375
        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;
}


9376 9377 9378 9379 9380 9381 9382 9383 9384 9385 9386 9387 9388 9389 9390 9391 9392 9393 9394 9395 9396
/**
 * 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
9397 9398
    bool needPCDimmCap = false;
    bool needNvdimmCap = false;
9399 9400 9401 9402 9403 9404 9405
    size_t i;

    hotplugSpace = def->mem.max_memory - virDomainDefGetMemoryInitial(def);

    if (mem) {
        nmems++;
        hotplugMemory = mem->size;
9406 9407 9408

        if (qemuDomainDefValidateMemoryHotplugDevice(mem, def) < 0)
            return -1;
9409 9410 9411 9412 9413 9414 9415 9416 9417 9418 9419 9420 9421
    }

    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;
    }

9422 9423 9424 9425 9426 9427 9428 9429 9430 9431 9432
    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;
        }
9433 9434 9435 9436 9437 9438 9439 9440 9441
    }

    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;
    }

9442
    for (i = 0; i < def->nmems; i++) {
9443 9444
        hotplugMemory += def->mems[i]->size;

M
Michal Privoznik 已提交
9445 9446 9447 9448 9449 9450 9451 9452 9453 9454 9455 9456 9457 9458
        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;
        }

9459 9460 9461 9462 9463 9464
        /* 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
9465 9466 9467 9468 9469 9470 9471 9472 9473 9474 9475 9476 9477 9478
    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;
    }

9479 9480 9481 9482 9483 9484 9485 9486 9487 9488
    if (hotplugMemory > hotplugSpace) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory device total size exceeds hotplug space"));
        return -1;
    }

    return 0;
}


9489
bool
9490 9491 9492 9493 9494 9495 9496 9497
qemuDomainHasBuiltinIDE(const virDomainDef *def)
{
    return qemuDomainMachineHasBuiltinIDE(def->os.machine);
}


bool
qemuDomainMachineHasBuiltinIDE(const char *machine)
9498
{
9499 9500 9501 9502
    return qemuDomainMachineIsI440FX(machine) ||
        STREQ(machine, "malta") ||
        STREQ(machine, "sun4u") ||
        STREQ(machine, "g3beige");
9503 9504 9505
}


9506 9507 9508 9509 9510 9511 9512 9513 9514 9515 9516 9517 9518 9519 9520 9521 9522 9523 9524 9525 9526 9527 9528 9529
/**
 * qemuDomainUpdateCurrentMemorySize:
 *
 * Updates the current balloon size from the monitor if necessary. In case when
 * the balloon is not present for the domain, the function recalculates the
 * maximum size to reflect possible changes.
 *
 * Returns 0 on success and updates vm->def->mem.cur_balloon if necessary, -1 on
 * error and reports libvirt error.
 */
int
qemuDomainUpdateCurrentMemorySize(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    unsigned long long balloon;
    int ret = -1;

    /* inactive domain doesn't need size update */
    if (!virDomainObjIsActive(vm))
        return 0;

    /* if no balloning is available, the current size equals to the current
     * full memory size */
9530
    if (!virDomainDefHasMemballoon(vm->def)) {
9531
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
9532 9533 9534 9535 9536 9537 9538 9539 9540 9541 9542 9543 9544 9545 9546
        return 0;
    }

    /* current size is always automagically updated via the event */
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BALLOON_EVENT))
        return 0;

    /* here we need to ask the monitor */

    /* Don't delay if someone's using the monitor, just use existing most
     * recent data instead */
    if (qemuDomainJobAllowed(priv, QEMU_JOB_QUERY)) {
        if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_QUERY) < 0)
            return -1;

9547
        if (virDomainObjCheckActive(vm) < 0)
9548 9549 9550 9551 9552 9553 9554 9555 9556 9557 9558 9559 9560
            goto endjob;

        qemuDomainObjEnterMonitor(driver, vm);
        ret = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
        if (qemuDomainObjExitMonitor(driver, vm) < 0)
            ret = -1;

 endjob:
        qemuDomainObjEndJob(driver, vm);

        if (ret < 0)
            return -1;

9561 9562
        vm->def->mem.cur_balloon = balloon;
    }
9563 9564 9565

    return 0;
}
9566 9567 9568


/**
9569
 * qemuDomainGetMemLockLimitBytes:
9570 9571
 * @def: domain definition
 *
9572 9573 9574 9575 9576 9577
 * 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
9578 9579
 */
unsigned long long
9580
qemuDomainGetMemLockLimitBytes(virDomainDefPtr def)
9581
{
9582 9583
    unsigned long long memKB = 0;
    size_t i;
9584

9585 9586 9587 9588 9589 9590
    /* prefer the hard limit */
    if (virMemoryLimitIsSet(def->mem.hard_limit)) {
        memKB = def->mem.hard_limit;
        goto done;
    }

9591 9592 9593 9594 9595 9596 9597
    /* 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;
9598 9599

    if (ARCH_IS_PPC64(def->os.arch) && def->virtType == VIR_DOMAIN_VIRT_KVM) {
9600 9601 9602 9603
        unsigned long long maxMemory;
        unsigned long long memory;
        unsigned long long baseLimit;
        unsigned long long passthroughLimit;
9604
        size_t nPCIHostBridges = 0;
9605 9606
        bool usesVFIO = false;

9607 9608 9609
        for (i = 0; i < def->ncontrollers; i++) {
            virDomainControllerDefPtr cont = def->controllers[i];

9610
            if (!virDomainControllerIsPSeriesPHB(cont))
9611 9612 9613 9614
                continue;

            nPCIHostBridges++;
        }
9615 9616 9617 9618 9619 9620 9621 9622 9623 9624 9625 9626

        for (i = 0; i < def->nhostdevs; i++) {
            virDomainHostdevDefPtr dev = def->hostdevs[i];

            if (dev->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS &&
                dev->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI &&
                dev->source.subsys.u.pci.backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO) {
                usesVFIO = true;
                break;
            }
        }

9627
        memory = virDomainDefGetMemoryTotal(def);
9628 9629 9630 9631 9632 9633 9634 9635 9636 9637 9638 9639 9640 9641 9642 9643 9644 9645 9646 9647 9648 9649 9650 9651 9652 9653 9654 9655 9656 9657 9658 9659 9660 9661 9662 9663 9664 9665 9666 9667 9668 9669 9670 9671 9672 9673 9674 9675 9676 9677 9678 9679 9680

        if (def->mem.max_memory)
            maxMemory = def->mem.max_memory;
        else
            maxMemory = memory;

        /* baseLimit := maxMemory / 128                                  (a)
         *              + 4 MiB * #PHBs + 8 MiB                          (b)
         *
         * (a) is the hash table
         *
         * (b) is accounting for the 32-bit DMA window - it could be either the
         * KVM accelerated TCE tables for emulated devices, or the VFIO
         * userspace view. The 4 MiB per-PHB (including the default one) covers
         * a 2GiB DMA window: default is 1GiB, but it's possible it'll be
         * increased to help performance. The 8 MiB extra should be plenty for
         * the TCE table index for any reasonable number of PHBs and several
         * spapr-vlan or spapr-vscsi devices (512kB + a tiny bit each) */
        baseLimit = maxMemory / 128 +
                    4096 * nPCIHostBridges +
                    8192;

        /* passthroughLimit := max( 2 GiB * #PHBs,                       (c)
         *                          memory                               (d)
         *                          + memory * 1/512 * #PHBs + 8 MiB )   (e)
         *
         * (c) is the pre-DDW VFIO DMA window accounting. We're allowing 2 GiB
         * rather than 1 GiB
         *
         * (d) is the with-DDW (and memory pre-registration and related
         * features) DMA window accounting - assuming that we only account RAM
         * once, even if mapped to multiple PHBs
         *
         * (e) is the with-DDW userspace view and overhead for the 64-bit DMA
         * window. This is based a bit on expected guest behaviour, but there
         * really isn't a way to completely avoid that. We assume the guest
         * requests a 64-bit DMA window (per PHB) just big enough to map all
         * its RAM. 4 kiB page size gives the 1/512; it will be less with 64
         * kiB pages, less still if the guest is mapped with hugepages (unlike
         * the default 32-bit DMA window, DDW windows can use large IOMMU
         * pages). 8 MiB is for second and further level overheads, like (b) */
        passthroughLimit = MAX(2 * 1024 * 1024 * nPCIHostBridges,
                               memory +
                               memory / 512 * nPCIHostBridges + 8192);

        if (usesVFIO)
            memKB = baseLimit + passthroughLimit;
        else
            memKB = baseLimit;

        goto done;
    }

9681 9682 9683 9684 9685 9686 9687 9688 9689 9690 9691 9692 9693 9694 9695 9696 9697 9698
    /* 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.
     */
9699
    for (i = 0; i < def->nhostdevs; i++) {
9700
        virDomainHostdevSubsysPtr subsys = &def->hostdevs[i]->source.subsys;
9701

9702 9703 9704
        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 &&
9705 9706 9707 9708
              subsys->u.pci.backend == VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO))) {
            memKB = virDomainDefGetMemoryTotal(def) + 1024 * 1024;
            goto done;
        }
9709 9710
    }

9711 9712
 done:
    return memKB << 10;
9713
}
9714

9715

9716 9717 9718 9719 9720 9721 9722
/**
 * qemuDomainAdjustMaxMemLock:
 * @vm: domain
 *
 * Adjust the memory locking limit for the QEMU process associated to @vm, in
 * order to comply with VFIO or architecture requirements.
 *
9723 9724 9725 9726
 * 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.
9727 9728 9729 9730 9731 9732 9733 9734 9735
 *
 * Returns: 0 on success, <0 on failure
 */
int
qemuDomainAdjustMaxMemLock(virDomainObjPtr vm)
{
    unsigned long long bytes = 0;
    int ret = -1;

9736 9737 9738
    bytes = qemuDomainGetMemLockLimitBytes(vm->def);

    if (bytes) {
9739 9740 9741 9742 9743 9744 9745 9746 9747 9748 9749 9750 9751 9752
        /* 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;
    }
9753 9754 9755 9756 9757 9758 9759 9760 9761 9762

    /* 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;
}
9763 9764 9765 9766 9767 9768 9769 9770 9771 9772

/**
 * qemuDomainHasVcpuPids:
 * @vm: Domain object
 *
 * Returns true if we were able to successfully detect vCPU pids for the VM.
 */
bool
qemuDomainHasVcpuPids(virDomainObjPtr vm)
{
9773 9774 9775 9776 9777 9778
    size_t i;
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    virDomainVcpuDefPtr vcpu;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
9779

9780 9781 9782 9783 9784
        if (QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid > 0)
            return true;
    }

    return false;
9785
}
9786 9787 9788 9789 9790 9791 9792 9793 9794 9795 9796


/**
 * 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,
9797
                     unsigned int vcpuid)
9798
{
9799 9800
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
    return QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid;
9801
}
9802 9803


9804 9805 9806 9807 9808 9809 9810 9811 9812 9813 9814 9815 9816 9817 9818 9819 9820 9821 9822 9823 9824 9825 9826 9827 9828 9829 9830 9831 9832 9833 9834 9835 9836 9837 9838 9839 9840 9841 9842 9843 9844 9845
/**
 * 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;
}


9846 9847 9848 9849 9850 9851 9852 9853 9854
bool
qemuDomainSupportsNewVcpuHotplug(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
}


9855
/**
9856
 * qemuDomainRefreshVcpuInfo:
9857 9858 9859
 * @driver: qemu driver data
 * @vm: domain object
 * @asyncJob: current asynchronous job type
9860
 * @state: refresh vcpu state
9861
 *
9862 9863
 * Updates vCPU information private data of @vm. Due to historical reasons this
 * function returns success even if some data were not reported by qemu.
9864
 *
9865 9866
 * If @state is true, the vcpu state is refreshed as reported by the monitor.
 *
9867
 * Returns 0 on success and -1 on fatal error.
9868 9869
 */
int
9870 9871
qemuDomainRefreshVcpuInfo(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
9872 9873
                          int asyncJob,
                          bool state)
9874
{
9875
    virDomainVcpuDefPtr vcpu;
9876 9877
    qemuDomainVcpuPrivatePtr vcpupriv;
    qemuMonitorCPUInfoPtr info = NULL;
9878 9879
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    size_t i;
9880
    bool hotplug;
9881
    bool fast;
9882
    int rc;
9883
    int ret = -1;
9884

9885
    hotplug = qemuDomainSupportsNewVcpuHotplug(vm);
9886 9887
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
9888 9889 9890

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
9891

9892 9893
    rc = qemuMonitorGetCPUInfo(qemuDomainGetMonitor(vm), &info, maxvcpus,
                               hotplug, fast);
9894

9895
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
9896 9897
        goto cleanup;

9898
    if (rc < 0)
9899 9900 9901 9902
        goto cleanup;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
9903
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
9904

9905 9906 9907 9908 9909 9910 9911
        /*
         * Current QEMU *can* report info about host threads mapped
         * to vCPUs, but it is not in a manner we can correctly
         * deal with. The TCG CPU emulation does have a separate vCPU
         * thread, but it runs every vCPU in that same thread. So it
         * is impossible to setup different affinity per thread.
         *
9912
         * What's more the 'query-cpus[-fast]' command returns bizarre
9913 9914 9915 9916 9917 9918 9919 9920 9921 9922 9923 9924 9925 9926 9927 9928 9929 9930 9931 9932 9933 9934 9935 9936 9937 9938
         * data for the threads. It gives the TCG thread for the
         * vCPU 0, but for vCPUs 1-> N, it actually replies with
         * the main process thread ID.
         *
         * The result is that when we try to set affinity for
         * vCPU 1, it will actually change the affinity of the
         * emulator thread :-( When you try to set affinity for
         * vCPUs 2, 3.... it will fail if the affinity was
         * different from vCPU 1.
         *
         * We *could* allow vcpu pinning with TCG, if we made the
         * restriction that all vCPUs had the same mask. This would
         * at least let us separate emulator from vCPUs threads, as
         * we do for KVM. It would need some changes to our cgroups
         * CPU layout though, and error reporting for the config
         * restrictions.
         *
         * Just disable CPU pinning with TCG until someone wants
         * to try to do this hard work.
         */
        if (vm->def->virtType != VIR_DOMAIN_VIRT_QEMU)
            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;
9939
        vcpupriv->node_id = info[i].node_id;
9940 9941 9942 9943 9944 9945
        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;
9946
        vcpupriv->qemu_id = info[i].qemu_id;
9947

9948
        if (hotplug && state) {
9949 9950 9951 9952 9953
            vcpu->online = info[i].online;
            if (info[i].hotpluggable)
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_YES;
            else
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_NO;
9954
        }
9955 9956
    }

9957
    ret = 0;
9958 9959

 cleanup:
9960
    qemuMonitorCPUInfoFree(info, maxvcpus);
9961
    return ret;
9962
}
9963

9964 9965 9966 9967 9968 9969 9970 9971 9972 9973 9974 9975 9976 9977 9978 9979 9980 9981 9982 9983 9984 9985 9986 9987 9988 9989 9990 9991 9992 9993 9994
/**
 * 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;
9995
    qemuDomainVcpuPrivatePtr vcpupriv;
9996
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
9997
    virBitmapPtr haltedmap = NULL;
9998 9999
    size_t i;
    int ret = -1;
10000
    bool fast;
10001 10002 10003 10004 10005

    /* Not supported currently for TCG, see qemuDomainRefreshVcpuInfo */
    if (vm->def->virtType == VIR_DOMAIN_VIRT_QEMU)
        return 0;

10006
    /* The halted state is interresting only on s390(x). On other platforms
10007 10008 10009 10010 10011 10012 10013
     * 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))
10014 10015
        return 0;

10016 10017 10018
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;

10019 10020 10021 10022
    fast = virQEMUCapsGet(QEMU_DOMAIN_PRIVATE(vm)->qemuCaps,
                          QEMU_CAPS_QUERY_CPUS_FAST);
    haltedmap = qemuMonitorGetCpuHalted(qemuDomainGetMonitor(vm), maxvcpus,
                                        fast);
10023
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || !haltedmap)
10024 10025 10026 10027
        goto cleanup;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
10028
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
10029 10030
        vcpupriv->halted = virTristateBoolFromBool(virBitmapIsBitSet(haltedmap,
                                                                     vcpupriv->qemu_id));
10031 10032 10033 10034 10035
    }

    ret = 0;

 cleanup:
10036
    virBitmapFree(haltedmap);
10037 10038
    return ret;
}
10039 10040 10041 10042 10043 10044 10045 10046 10047 10048 10049 10050 10051 10052 10053

bool
qemuDomainSupportsNicdev(virDomainDefPtr def,
                         virDomainNetDefPtr net)
{
    /* non-virtio ARM nics require legacy -net nic */
    if (((def->os.arch == VIR_ARCH_ARMV7L) ||
        (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;
}

10054 10055 10056 10057
bool
qemuDomainNetSupportsMTU(virDomainNetType type)
{
    switch (type) {
10058 10059
    case VIR_DOMAIN_NET_TYPE_NETWORK:
    case VIR_DOMAIN_NET_TYPE_BRIDGE:
10060 10061
    case VIR_DOMAIN_NET_TYPE_ETHERNET:
    case VIR_DOMAIN_NET_TYPE_VHOSTUSER:
10062 10063
        return true;
    case VIR_DOMAIN_NET_TYPE_USER:
10064 10065 10066 10067 10068 10069 10070 10071 10072 10073 10074 10075
    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
10076

P
Peter Krempa 已提交
10077 10078 10079 10080 10081 10082 10083 10084 10085 10086 10087 10088 10089 10090 10091

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;
}
10092 10093 10094 10095 10096 10097 10098 10099 10100 10101 10102 10103 10104 10105 10106 10107 10108 10109 10110 10111 10112 10113 10114 10115 10116 10117 10118 10119 10120 10121 10122


/**
 * 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;
}
10123 10124 10125 10126 10127 10128


int
qemuDomainPrepareChannel(virDomainChrDefPtr channel,
                         const char *domainChannelTargetDir)
{
S
Scott Garfinkle 已提交
10129 10130 10131 10132 10133 10134
    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) {
10135
        if (virAsprintf(&channel->source->data.nix.path,
10136
                        "%s/%s", domainChannelTargetDir,
S
Scott Garfinkle 已提交
10137 10138
                        channel->target.name) < 0)
            return -1;
10139 10140
    } else {
        /* Generate a unique name */
S
Scott Garfinkle 已提交
10141 10142 10143 10144 10145 10146
        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)
10147 10148 10149 10150 10151
            return -1;
    }

    return 0;
}
10152 10153


10154
/* qemuDomainPrepareChardevSourceTLS:
10155 10156 10157 10158 10159 10160 10161 10162 10163 10164 10165 10166 10167 10168 10169 10170 10171
 * @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;
10172
            source->data.tcp.tlsFromConfig = true;
10173 10174 10175 10176 10177
        }
    }
}


10178
/* qemuDomainPrepareChardevSource:
10179
 * @def: live domain definition
10180
 * @cfg: driver configuration
10181 10182 10183 10184 10185 10186
 *
 * Iterate through all devices that use virDomainChrSourceDefPtr as host
 * interface part.
 */
void
qemuDomainPrepareChardevSource(virDomainDefPtr def,
10187
                               virQEMUDriverConfigPtr cfg)
10188 10189 10190 10191 10192 10193 10194 10195 10196 10197 10198 10199 10200 10201 10202 10203 10204 10205 10206 10207 10208 10209 10210 10211 10212 10213 10214 10215 10216
{
    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);
}


10217 10218 10219 10220 10221 10222 10223 10224 10225 10226 10227 10228 10229 10230 10231 10232 10233 10234 10235 10236 10237 10238 10239 10240
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;
}


10241 10242 10243 10244 10245 10246 10247 10248 10249 10250 10251 10252 10253 10254 10255 10256 10257 10258 10259 10260 10261 10262 10263 10264 10265 10266 10267 10268
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;
}


10269
/* qemuProcessPrepareStorageSourceTLS:
10270
 * @source: source for a disk
10271
 * @cfg: driver configuration
10272
 * @parentAlias: alias of the parent device
10273 10274 10275 10276 10277 10278 10279
 *
 * 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
 */
10280
static int
10281
qemuDomainPrepareStorageSourceTLS(virStorageSourcePtr src,
10282
                                  virQEMUDriverConfigPtr cfg,
10283 10284
                                  const char *parentAlias,
                                  virQEMUCapsPtr qemuCaps)
10285
{
10286 10287 10288 10289 10290 10291 10292 10293
    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;
10294

10295
    case VIR_STORAGE_NET_PROTOCOL_NBD:
10296 10297 10298 10299
        if (qemuProcessPrepareStorageSourceTLSNBD(src, cfg, qemuCaps) < 0)
            return -1;
        break;

10300 10301 10302 10303 10304 10305 10306 10307 10308 10309
    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:
10310 10311 10312 10313 10314 10315
        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;
        }
10316 10317 10318 10319 10320 10321
        break;

    case VIR_STORAGE_NET_PROTOCOL_NONE:
    case VIR_STORAGE_NET_PROTOCOL_LAST:
    default:
        virReportEnumRangeError(virStorageNetProtocol, src->protocol);
10322
        return -1;
10323
    }
10324

10325 10326 10327 10328
    if (src->haveTLS == VIR_TRISTATE_BOOL_YES &&
        !(src->tlsAlias = qemuAliasTLSObjFromSrcAlias(parentAlias)))
        return -1;

10329 10330 10331 10332
    return 0;
}


10333 10334 10335 10336 10337 10338 10339 10340 10341 10342 10343 10344 10345
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);
}


10346 10347 10348 10349 10350 10351 10352 10353 10354 10355 10356 10357 10358 10359 10360 10361 10362 10363 10364 10365 10366 10367 10368 10369 10370 10371 10372 10373 10374 10375 10376 10377 10378
/**
 * 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);
}
10379 10380 10381 10382 10383 10384 10385 10386 10387 10388 10389 10390 10391 10392 10393 10394 10395 10396 10397 10398 10399 10400 10401 10402 10403 10404 10405 10406 10407 10408 10409 10410 10411 10412


/**
 * 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;
        }
    }
}
10413 10414 10415 10416 10417 10418 10419 10420 10421 10422 10423 10424 10425 10426 10427 10428 10429 10430 10431 10432


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;
}
10433 10434 10435 10436 10437 10438 10439 10440 10441 10442 10443 10444


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;
}
10445 10446


10447 10448
/**
 * qemuDomainGetHostdevPath:
10449
 * @def: domain definition
10450
 * @dev: host device definition
10451
 * @teardown: true if device will be removed
10452
 * @npaths: number of items in @path and @perms arrays
10453 10454 10455
 * @path: resulting path to @dev
 * @perms: Optional pointer to VIR_CGROUP_DEVICE_* perms
 *
10456 10457 10458 10459 10460 10461
 * 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.
10462 10463 10464 10465
 *
 * Returns 0 on success, -1 otherwise.
 */
int
10466 10467 10468
qemuDomainGetHostdevPath(virDomainDefPtr def,
                         virDomainHostdevDefPtr dev,
                         bool teardown,
10469 10470 10471
                         size_t *npaths,
                         char ***path,
                         int **perms)
10472 10473 10474 10475 10476 10477
{
    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;
10478
    virDomainHostdevSubsysMediatedDevPtr mdevsrc = &dev->source.subsys.u.mdev;
10479 10480 10481 10482 10483 10484
    virPCIDevicePtr pci = NULL;
    virUSBDevicePtr usb = NULL;
    virSCSIDevicePtr scsi = NULL;
    virSCSIVHostDevicePtr host = NULL;
    char *tmpPath = NULL;
    bool freeTmpPath = false;
10485 10486 10487 10488 10489
    bool includeVFIO = false;
    char **tmpPaths = NULL;
    int *tmpPerms = NULL;
    size_t i, tmpNpaths = 0;
    int perm = 0;
10490

10491
    *npaths = 0;
10492 10493 10494

    switch ((virDomainHostdevMode) dev->mode) {
    case VIR_DOMAIN_HOSTDEV_MODE_SUBSYS:
10495
        switch ((virDomainHostdevSubsysType)dev->source.subsys.type) {
10496 10497 10498 10499 10500 10501 10502 10503 10504 10505 10506 10507
        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;
10508 10509

                perm = VIR_CGROUP_DEVICE_RW;
10510 10511 10512 10513 10514 10515 10516 10517 10518 10519 10520 10521 10522 10523 10524
                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;
                }
10525 10526 10527 10528 10529 10530 10531 10532 10533 10534 10535 10536
            }
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
            if (dev->missing)
                break;
            usb = virUSBDeviceNew(usbsrc->bus,
                                  usbsrc->device,
                                  NULL);
            if (!usb)
                goto cleanup;

10537
            if (!(tmpPath = (char *)virUSBDeviceGetPath(usb)))
10538
                goto cleanup;
10539
            perm = VIR_CGROUP_DEVICE_RW;
10540 10541 10542 10543 10544 10545 10546 10547
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI:
            if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI) {
                virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;
                /* Follow qemuSetupDiskCgroup() and qemuSetImageCgroupInternal()
                 * which does nothing for non local storage
                 */
10548
                VIR_DEBUG("Not updating /dev for hostdev iSCSI path '%s'", iscsisrc->src->path);
10549 10550 10551 10552 10553 10554 10555 10556 10557 10558 10559 10560 10561
            } 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;

10562
                if (!(tmpPath = (char *)virSCSIDeviceGetPath(scsi)))
10563
                    goto cleanup;
10564 10565
                perm = virSCSIDeviceGetReadonly(scsi) ?
                    VIR_CGROUP_DEVICE_READ : VIR_CGROUP_DEVICE_RW;
10566 10567 10568 10569 10570 10571 10572 10573 10574
            }
            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;

10575
                if (!(tmpPath = (char *)virSCSIVHostDeviceGetPath(host)))
10576
                    goto cleanup;
10577
                perm = VIR_CGROUP_DEVICE_RW;
10578 10579 10580 10581
            }
            break;
        }

10582
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV:
10583
            if (!(tmpPath = virMediatedDeviceGetIOMMUGroupDev(mdevsrc->uuidstr)))
10584 10585 10586 10587 10588 10589
                goto cleanup;

            freeTmpPath = true;
            includeVFIO = true;
            perm = VIR_CGROUP_DEVICE_RW;
            break;
10590 10591 10592 10593 10594 10595 10596 10597 10598 10599 10600
        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
            break;
        }
        break;

    case VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES:
    case VIR_DOMAIN_HOSTDEV_MODE_LAST:
        /* nada */
        break;
    }

10601 10602 10603 10604 10605
    if (tmpPath) {
        size_t toAlloc = 1;

        if (includeVFIO)
            toAlloc = 2;
10606

10607 10608 10609 10610 10611 10612 10613 10614 10615 10616 10617 10618 10619 10620 10621 10622 10623 10624 10625 10626 10627 10628
        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;
    }
10629 10630
    ret = 0;
 cleanup:
10631 10632 10633 10634
    for (i = 0; i < tmpNpaths; i++)
        VIR_FREE(tmpPaths[i]);
    VIR_FREE(tmpPaths);
    VIR_FREE(tmpPerms);
10635 10636 10637 10638 10639 10640 10641 10642 10643 10644
    virPCIDeviceFree(pci);
    virUSBDeviceFree(usb);
    virSCSIDeviceFree(scsi);
    virSCSIVHostDeviceFree(host);
    if (freeTmpPath)
        VIR_FREE(tmpPath);
    return ret;
}


10645 10646 10647 10648
/**
 * qemuDomainGetPreservedMountPath:
 * @cfg: driver configuration data
 * @vm: domain object
10649
 * @mountpoint: mount point path to convert
10650
 *
10651
 * For given @mountpoint return new path where the mount point
10652 10653 10654 10655 10656 10657 10658 10659
 * 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,
10660
                                const char *mountpoint)
10661 10662 10663
{
    char *path = NULL;
    char *tmp;
10664
    const char *suffix = mountpoint + strlen(DEVPREFIX);
10665
    char *domname = virDomainDefGetShortName(vm->def);
10666 10667
    size_t off;

10668 10669 10670
    if (!domname)
        return NULL;

10671
    if (STREQ(mountpoint, "/dev"))
10672 10673 10674
        suffix = "dev";

    if (virAsprintf(&path, "%s/%s.%s",
10675 10676
                    cfg->stateDir, domname, suffix) < 0)
        goto cleanup;
10677

10678
    /* Now consider that @mountpoint is "/dev/blah/blah2".
10679 10680 10681 10682 10683 10684 10685 10686 10687 10688 10689 10690 10691
     * @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++;
    }

10692 10693
 cleanup:
    VIR_FREE(domname);
10694 10695 10696 10697
    return path;
}


10698 10699 10700 10701 10702 10703 10704 10705 10706 10707 10708 10709
/**
 * 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
10710
qemuDomainGetPreservedMounts(virQEMUDriverConfigPtr cfg,
10711 10712 10713 10714 10715 10716
                             virDomainObjPtr vm,
                             char ***devPath,
                             char ***devSavePath,
                             size_t *ndevPath)
{
    char **paths = NULL, **mounts = NULL;
10717
    size_t i, j, nmounts;
10718 10719 10720 10721 10722 10723 10724 10725 10726 10727 10728

    if (virFileGetMountSubtree(PROC_MOUNTS, "/dev",
                               &mounts, &nmounts) < 0)
        goto error;

    if (!nmounts) {
        if (ndevPath)
            *ndevPath = 0;
        return 0;
    }

10729 10730 10731 10732 10733 10734 10735 10736 10737 10738 10739 10740
    /* 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) {
10741 10742
            char *c = STRSKIP(mounts[j], mounts[i]);

10743
            if (c && (*c == '/' || *c == '\0')) {
10744 10745 10746 10747 10748 10749 10750 10751
                VIR_DEBUG("Dropping path %s because of %s", mounts[j], mounts[i]);
                VIR_DELETE_ELEMENT(mounts, j, nmounts);
            } else {
                j++;
            }
        }
    }

10752 10753 10754 10755
    if (VIR_ALLOC_N(paths, nmounts) < 0)
        goto error;

    for (i = 0; i < nmounts; i++) {
10756
        if (!(paths[i] = qemuDomainGetPreservedMountPath(cfg, vm, mounts[i])))
10757 10758 10759 10760 10761 10762 10763 10764 10765 10766 10767 10768 10769 10770 10771 10772 10773 10774 10775 10776 10777 10778 10779 10780 10781
            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;
}


10782 10783
struct qemuDomainCreateDeviceData {
    const char *path;     /* Path to temp new /dev location */
10784 10785
    char * const *devMountsPath;
    size_t ndevMountsPath;
10786 10787 10788
};


10789
static int
10790
qemuDomainCreateDeviceRecursive(const char *device,
10791
                                const struct qemuDomainCreateDeviceData *data,
10792 10793
                                bool allow_noent,
                                unsigned int ttl)
10794 10795
{
    char *devicePath = NULL;
10796
    char *target = NULL;
10797 10798
    struct stat sb;
    int ret = -1;
10799
    bool isLink = false;
10800
    bool isDev = false;
10801
    bool isReg = false;
10802
    bool isDir = false;
10803
    bool create = false;
10804 10805 10806
#ifdef WITH_SELINUX
    char *tcon = NULL;
#endif
10807

10808 10809 10810 10811 10812 10813 10814
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             device);
        return ret;
    }

10815
    if (lstat(device, &sb) < 0) {
10816 10817
        if (errno == ENOENT && allow_noent) {
            /* Ignore non-existent device. */
10818
            return 0;
10819
        }
10820 10821
        virReportSystemError(errno, _("Unable to stat %s"), device);
        return ret;
10822 10823
    }

10824
    isLink = S_ISLNK(sb.st_mode);
10825
    isDev = S_ISCHR(sb.st_mode) || S_ISBLK(sb.st_mode);
10826
    isReg = S_ISREG(sb.st_mode) || S_ISFIFO(sb.st_mode) || S_ISSOCK(sb.st_mode);
10827
    isDir = S_ISDIR(sb.st_mode);
10828 10829 10830 10831 10832 10833 10834 10835 10836 10837 10838 10839 10840 10841 10842 10843 10844

    /* 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)) {
10845
        size_t i;
10846

10847 10848 10849 10850 10851 10852 10853 10854 10855 10856 10857 10858 10859 10860 10861 10862 10863 10864 10865 10866 10867 10868 10869 10870 10871
        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]);
10872
        }
10873 10874
    }

10875 10876 10877 10878 10879 10880 10881 10882 10883
    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;
        }
10884

10885 10886 10887 10888 10889 10890 10891 10892 10893
        if (create &&
            symlink(target, devicePath) < 0) {
            if (errno == EEXIST) {
                ret = 0;
            } else {
                virReportSystemError(errno,
                                     _("unable to create symlink %s"),
                                     devicePath);
            }
10894 10895 10896
            goto cleanup;
        }

10897 10898 10899 10900 10901 10902 10903 10904 10905 10906
        /* 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;
10907

10908 10909
            if (VIR_STRDUP(devTmp, device) < 0)
                goto cleanup;
10910

10911 10912 10913 10914 10915 10916 10917 10918 10919 10920 10921 10922 10923
            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);
            target = tmp;
            tmp = NULL;
        }

10924
        if (qemuDomainCreateDeviceRecursive(target, data,
10925
                                            allow_noent, ttl - 1) < 0)
10926
            goto cleanup;
10927
    } else if (isDev) {
10928 10929 10930 10931 10932 10933 10934 10935 10936 10937
        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;
10938
        }
10939
    } else if (isReg) {
10940
        if (create &&
10941
            virFileTouch(devicePath, sb.st_mode) < 0)
10942
            goto cleanup;
10943 10944
        /* Just create the file here so that code below sets
         * proper owner and mode. Bind mount only after that. */
10945 10946 10947 10948
    } else if (isDir) {
        if (create &&
            virFileMakePathWithMode(devicePath, sb.st_mode) < 0)
            goto cleanup;
10949 10950 10951 10952 10953
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       device, sb.st_mode);
        goto cleanup;
10954 10955 10956 10957
    }

    if (!create) {
        ret = 0;
10958 10959 10960
        goto cleanup;
    }

10961
    if (lchown(devicePath, sb.st_uid, sb.st_gid) < 0) {
10962 10963 10964 10965 10966 10967
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             devicePath);
        goto cleanup;
    }

10968 10969 10970 10971 10972 10973 10974 10975 10976
    /* 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;
    }

10977 10978 10979
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileCopyACLs(device, devicePath) < 0 &&
10980 10981 10982 10983 10984 10985 10986
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Failed to copy ACLs on device %s"),
                             devicePath);
        goto cleanup;
    }

10987
#ifdef WITH_SELINUX
10988
    if (lgetfilecon_raw(device, &tcon) < 0 &&
10989 10990 10991
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"),
10992
                             device);
10993 10994 10995 10996
        goto cleanup;
    }

    if (tcon &&
10997
        lsetfilecon_raw(devicePath, (VIR_SELINUX_CTX_CONST char *)tcon) < 0) {
10998 10999 11000 11001 11002 11003 11004 11005 11006 11007 11008
        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

11009
    /* Finish mount process started earlier. */
11010
    if ((isReg || isDir) &&
11011 11012 11013
        virFileBindMountDevice(device, devicePath) < 0)
        goto cleanup;

11014 11015
    ret = 0;
 cleanup:
11016
    VIR_FREE(target);
11017
    VIR_FREE(devicePath);
11018 11019 11020
#ifdef WITH_SELINUX
    freecon(tcon);
#endif
11021 11022 11023 11024
    return ret;
}


11025 11026
static int
qemuDomainCreateDevice(const char *device,
11027
                       const struct qemuDomainCreateDeviceData *data,
11028 11029 11030 11031
                       bool allow_noent)
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

11032
    return qemuDomainCreateDeviceRecursive(device, data,
11033 11034 11035
                                           allow_noent, symloop_max);
}

11036 11037

static int
11038
qemuDomainPopulateDevices(virQEMUDriverConfigPtr cfg,
11039
                          virDomainObjPtr vm ATTRIBUTE_UNUSED,
11040
                          const struct qemuDomainCreateDeviceData *data)
11041 11042 11043 11044 11045 11046 11047 11048 11049
{
    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++) {
11050
        if (qemuDomainCreateDevice(devices[i], data, true) < 0)
11051 11052 11053 11054 11055 11056 11057 11058 11059 11060
            goto cleanup;
    }

    ret = 0;
 cleanup:
    return ret;
}


static int
11061 11062
qemuDomainSetupDev(virQEMUDriverConfigPtr cfg,
                   virSecurityManagerPtr mgr,
11063
                   virDomainObjPtr vm,
11064
                   const struct qemuDomainCreateDeviceData *data)
11065 11066 11067 11068 11069 11070 11071
{
    char *mount_options = NULL;
    char *opts = NULL;
    int ret = -1;

    VIR_DEBUG("Setting up /dev/ for domain %s", vm->def->name);

11072
    mount_options = qemuSecurityGetMountOptions(mgr, vm->def);
11073 11074 11075 11076 11077 11078 11079 11080 11081 11082 11083 11084 11085

    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;

11086
    if (virFileSetupDev(data->path, opts) < 0)
11087 11088
        goto cleanup;

11089
    if (qemuDomainPopulateDevices(cfg, vm, data) < 0)
11090 11091 11092 11093 11094 11095 11096 11097 11098 11099
        goto cleanup;

    ret = 0;
 cleanup:
    VIR_FREE(opts);
    VIR_FREE(mount_options);
    return ret;
}


11100
static int
11101
qemuDomainSetupDisk(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
11102
                    virDomainDiskDefPtr disk,
11103
                    const struct qemuDomainCreateDeviceData *data)
11104 11105 11106 11107 11108
{
    virStorageSourcePtr next;
    char *dst = NULL;
    int ret = -1;

11109
    for (next = disk->src; virStorageSourceIsBacking(next); next = next->backingStore) {
11110
        if (!next->path || !virStorageSourceIsLocalStorage(next)) {
11111 11112 11113 11114
            /* Not creating device. Just continue. */
            continue;
        }

11115
        if (qemuDomainCreateDevice(next->path, data, false) < 0)
11116 11117 11118
            goto cleanup;
    }

11119
    /* qemu-pr-helper might require access to /dev/mapper/control. */
11120
    if (disk->src->pr &&
11121 11122 11123
        qemuDomainCreateDevice(DEVICE_MAPPER_CONTROL_PATH, data, true) < 0)
        goto cleanup;

11124 11125 11126 11127 11128 11129 11130 11131
    ret = 0;
 cleanup:
    VIR_FREE(dst);
    return ret;
}


static int
11132
qemuDomainSetupAllDisks(virQEMUDriverConfigPtr cfg,
11133
                        virDomainObjPtr vm,
11134
                        const struct qemuDomainCreateDeviceData *data)
11135 11136 11137 11138 11139
{
    size_t i;
    VIR_DEBUG("Setting up disks");

    for (i = 0; i < vm->def->ndisks; i++) {
11140
        if (qemuDomainSetupDisk(cfg,
11141
                                vm->def->disks[i],
11142
                                data) < 0)
11143 11144 11145 11146 11147 11148 11149 11150
            return -1;
    }

    VIR_DEBUG("Setup all disks");
    return 0;
}


11151
static int
11152
qemuDomainSetupHostdev(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
11153
                       virDomainHostdevDefPtr dev,
11154
                       const struct qemuDomainCreateDeviceData *data)
11155 11156
{
    int ret = -1;
11157 11158
    char **path = NULL;
    size_t i, npaths = 0;
11159

11160
    if (qemuDomainGetHostdevPath(NULL, dev, false, &npaths, &path, NULL) < 0)
11161 11162
        goto cleanup;

11163
    for (i = 0; i < npaths; i++) {
11164
        if (qemuDomainCreateDevice(path[i], data, false) < 0)
11165
            goto cleanup;
11166 11167 11168 11169
    }

    ret = 0;
 cleanup:
11170 11171
    for (i = 0; i < npaths; i++)
        VIR_FREE(path[i]);
11172 11173 11174 11175 11176 11177
    VIR_FREE(path);
    return ret;
}


static int
11178
qemuDomainSetupAllHostdevs(virQEMUDriverConfigPtr cfg,
11179
                           virDomainObjPtr vm,
11180
                           const struct qemuDomainCreateDeviceData *data)
11181 11182 11183 11184 11185
{
    size_t i;

    VIR_DEBUG("Setting up hostdevs");
    for (i = 0; i < vm->def->nhostdevs; i++) {
11186
        if (qemuDomainSetupHostdev(cfg,
11187
                                   vm->def->hostdevs[i],
11188
                                   data) < 0)
11189 11190 11191 11192 11193 11194 11195
            return -1;
    }
    VIR_DEBUG("Setup all hostdevs");
    return 0;
}


M
Michal Privoznik 已提交
11196 11197 11198
static int
qemuDomainSetupMemory(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
                      virDomainMemoryDefPtr mem,
11199
                      const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
11200 11201 11202 11203
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

11204
    return qemuDomainCreateDevice(mem->nvdimmPath, data, false);
M
Michal Privoznik 已提交
11205 11206 11207 11208 11209 11210
}


static int
qemuDomainSetupAllMemories(virQEMUDriverConfigPtr cfg,
                           virDomainObjPtr vm,
11211
                           const struct qemuDomainCreateDeviceData *data)
M
Michal Privoznik 已提交
11212 11213 11214 11215 11216 11217 11218
{
    size_t i;

    VIR_DEBUG("Setting up memories");
    for (i = 0; i < vm->def->nmems; i++) {
        if (qemuDomainSetupMemory(cfg,
                                  vm->def->mems[i],
11219
                                  data) < 0)
M
Michal Privoznik 已提交
11220 11221 11222 11223 11224 11225 11226
            return -1;
    }
    VIR_DEBUG("Setup all memories");
    return 0;
}


11227 11228 11229 11230 11231
static int
qemuDomainSetupChardev(virDomainDefPtr def ATTRIBUTE_UNUSED,
                       virDomainChrDefPtr dev,
                       void *opaque)
{
11232
    const struct qemuDomainCreateDeviceData *data = opaque;
11233
    const char *path = NULL;
11234

11235
    if (!(path = virDomainChrSourceDefGetPath(dev->source)))
11236 11237
        return 0;

11238 11239 11240 11241 11242 11243
    /* 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);
11244 11245 11246 11247
}


static int
11248
qemuDomainSetupAllChardevs(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
11249
                           virDomainObjPtr vm,
11250
                           const struct qemuDomainCreateDeviceData *data)
11251 11252 11253 11254 11255 11256
{
    VIR_DEBUG("Setting up chardevs");

    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuDomainSetupChardev,
11257
                               (void *)data) < 0)
11258 11259 11260 11261 11262 11263 11264
        return -1;

    VIR_DEBUG("Setup all chardevs");
    return 0;
}


11265
static int
11266
qemuDomainSetupTPM(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
11267
                   virDomainObjPtr vm,
11268
                   const struct qemuDomainCreateDeviceData *data)
11269 11270 11271 11272 11273 11274 11275 11276 11277 11278 11279
{
    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,
11280
                                   data, false) < 0)
11281 11282 11283
            return -1;
        break;

11284
    case VIR_DOMAIN_TPM_TYPE_EMULATOR:
11285 11286 11287 11288 11289 11290 11291 11292 11293 11294
    case VIR_DOMAIN_TPM_TYPE_LAST:
        /* nada */
        break;
    }

    VIR_DEBUG("Setup TPM");
    return 0;
}


11295
static int
11296
qemuDomainSetupGraphics(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
11297
                        virDomainGraphicsDefPtr gfx,
11298
                        const struct qemuDomainCreateDeviceData *data)
11299 11300 11301 11302 11303 11304 11305 11306
{
    const char *rendernode = gfx->data.spice.rendernode;

    if (gfx->type != VIR_DOMAIN_GRAPHICS_TYPE_SPICE ||
        gfx->data.spice.gl != VIR_TRISTATE_BOOL_YES ||
        !rendernode)
        return 0;

11307
    return qemuDomainCreateDevice(rendernode, data, false);
11308 11309 11310 11311
}


static int
11312
qemuDomainSetupAllGraphics(virQEMUDriverConfigPtr cfg,
11313
                           virDomainObjPtr vm,
11314
                           const struct qemuDomainCreateDeviceData *data)
11315 11316 11317 11318 11319
{
    size_t i;

    VIR_DEBUG("Setting up graphics");
    for (i = 0; i < vm->def->ngraphics; i++) {
11320
        if (qemuDomainSetupGraphics(cfg,
11321
                                    vm->def->graphics[i],
11322
                                    data) < 0)
11323 11324 11325 11326 11327 11328 11329 11330
            return -1;
    }

    VIR_DEBUG("Setup all graphics");
    return 0;
}


11331
static int
11332
qemuDomainSetupInput(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
11333
                     virDomainInputDefPtr input,
11334
                     const struct qemuDomainCreateDeviceData *data)
11335
{
J
Ján Tomko 已提交
11336
    const char *path = virDomainInputDefGetPath(input);
11337

J
Ján Tomko 已提交
11338 11339
    if (path && qemuDomainCreateDevice(path, data, false) < 0)
        return -1;
11340

J
Ján Tomko 已提交
11341
    return 0;
11342 11343 11344 11345
}


static int
11346
qemuDomainSetupAllInputs(virQEMUDriverConfigPtr cfg,
11347
                         virDomainObjPtr vm,
11348
                         const struct qemuDomainCreateDeviceData *data)
11349 11350 11351
{
    size_t i;

11352
    VIR_DEBUG("Setting up inputs");
11353
    for (i = 0; i < vm->def->ninputs; i++) {
11354
        if (qemuDomainSetupInput(cfg,
11355
                                 vm->def->inputs[i],
11356
                                 data) < 0)
11357 11358
            return -1;
    }
11359
    VIR_DEBUG("Setup all inputs");
11360 11361 11362 11363
    return 0;
}


11364
static int
11365
qemuDomainSetupRNG(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
11366
                   virDomainRNGDefPtr rng,
11367
                   const struct qemuDomainCreateDeviceData *data)
11368 11369 11370
{
    switch ((virDomainRNGBackend) rng->backend) {
    case VIR_DOMAIN_RNG_BACKEND_RANDOM:
11371
        if (qemuDomainCreateDevice(rng->source.file, data, false) < 0)
11372 11373 11374 11375 11376 11377 11378 11379 11380 11381 11382 11383 11384
            return -1;

    case VIR_DOMAIN_RNG_BACKEND_EGD:
    case VIR_DOMAIN_RNG_BACKEND_LAST:
        /* nada */
        break;
    }

    return 0;
}


static int
11385
qemuDomainSetupAllRNGs(virQEMUDriverConfigPtr cfg,
11386
                       virDomainObjPtr vm,
11387
                       const struct qemuDomainCreateDeviceData *data)
11388 11389 11390 11391 11392
{
    size_t i;

    VIR_DEBUG("Setting up RNGs");
    for (i = 0; i < vm->def->nrngs; i++) {
11393
        if (qemuDomainSetupRNG(cfg,
11394
                               vm->def->rngs[i],
11395
                               data) < 0)
11396 11397 11398 11399 11400 11401 11402 11403
            return -1;
    }

    VIR_DEBUG("Setup all RNGs");
    return 0;
}


11404 11405 11406 11407 11408 11409 11410 11411 11412 11413 11414 11415 11416 11417 11418 11419 11420 11421 11422 11423 11424 11425 11426 11427 11428 11429 11430 11431 11432 11433 11434 11435 11436 11437 11438 11439 11440 11441
static int
qemuDomainSetupLoader(virQEMUDriverConfigPtr cfg ATTRIBUTE_UNUSED,
                      virDomainObjPtr vm,
                      const struct qemuDomainCreateDeviceData *data)
{
    virDomainLoaderDefPtr loader = vm->def->os.loader;
    int ret = -1;

    VIR_DEBUG("Setting up loader");

    if (loader) {
        switch ((virDomainLoader) loader->type) {
        case VIR_DOMAIN_LOADER_TYPE_ROM:
            if (qemuDomainCreateDevice(loader->path, data, false) < 0)
                goto cleanup;
            break;

        case VIR_DOMAIN_LOADER_TYPE_PFLASH:
            if (qemuDomainCreateDevice(loader->path, data, false) < 0)
                goto cleanup;

            if (loader->nvram &&
                qemuDomainCreateDevice(loader->nvram, data, false) < 0)
                goto cleanup;
            break;

        case VIR_DOMAIN_LOADER_TYPE_LAST:
            break;
        }
    }

    VIR_DEBUG("Setup loader");
    ret = 0;
 cleanup:
    return ret;
}


11442
int
11443 11444
qemuDomainBuildNamespace(virQEMUDriverConfigPtr cfg,
                         virSecurityManagerPtr mgr,
11445 11446
                         virDomainObjPtr vm)
{
11447
    struct qemuDomainCreateDeviceData data;
11448
    char *devPath = NULL;
11449
    char **devMountsPath = NULL, **devMountsSavePath = NULL;
11450 11451 11452 11453 11454 11455 11456 11457
    size_t ndevMountsPath = 0, i;
    int ret = -1;

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT)) {
        ret = 0;
        goto cleanup;
    }

11458
    if (qemuDomainGetPreservedMounts(cfg, vm,
11459 11460
                                     &devMountsPath, &devMountsSavePath,
                                     &ndevMountsPath) < 0)
11461 11462
        goto cleanup;

11463 11464 11465 11466 11467 11468 11469 11470 11471 11472 11473 11474 11475
    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;
    }

11476
    data.path = devPath;
11477 11478
    data.devMountsPath = devMountsPath;
    data.ndevMountsPath = ndevMountsPath;
11479

11480 11481 11482
    if (virProcessSetupPrivateMountNS() < 0)
        goto cleanup;

11483
    if (qemuDomainSetupDev(cfg, mgr, vm, &data) < 0)
11484 11485
        goto cleanup;

11486
    if (qemuDomainSetupAllDisks(cfg, vm, &data) < 0)
11487 11488
        goto cleanup;

11489
    if (qemuDomainSetupAllHostdevs(cfg, vm, &data) < 0)
11490 11491
        goto cleanup;

11492
    if (qemuDomainSetupAllMemories(cfg, vm, &data) < 0)
11493 11494
        goto cleanup;

11495
    if (qemuDomainSetupAllChardevs(cfg, vm, &data) < 0)
11496 11497
        goto cleanup;

11498
    if (qemuDomainSetupTPM(cfg, vm, &data) < 0)
11499 11500
        goto cleanup;

11501
    if (qemuDomainSetupAllGraphics(cfg, vm, &data) < 0)
11502 11503
        goto cleanup;

11504
    if (qemuDomainSetupAllInputs(cfg, vm, &data) < 0)
11505 11506
        goto cleanup;

11507
    if (qemuDomainSetupAllRNGs(cfg, vm, &data) < 0)
11508 11509
        goto cleanup;

11510 11511 11512
    if (qemuDomainSetupLoader(cfg, vm, &data) < 0)
        goto cleanup;

11513 11514
    /* Save some mount points because we want to share them with the host */
    for (i = 0; i < ndevMountsPath; i++) {
11515 11516
        struct stat sb;

11517 11518 11519
        if (devMountsSavePath[i] == devPath)
            continue;

11520 11521 11522 11523 11524 11525 11526
        if (stat(devMountsPath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsPath[i]);
            goto cleanup;
        }

11527 11528 11529
        /* At this point, devMountsPath is either:
         * a file (regular or special), or
         * a directory. */
11530
        if ((S_ISDIR(sb.st_mode) && virFileMakePath(devMountsSavePath[i]) < 0) ||
11531
            (!S_ISDIR(sb.st_mode) && virFileTouch(devMountsSavePath[i], sb.st_mode) < 0)) {
11532 11533 11534 11535 11536 11537
            virReportSystemError(errno,
                                 _("Failed to create %s"),
                                 devMountsSavePath[i]);
            goto cleanup;
        }

11538
        if (virFileMoveMount(devMountsPath[i], devMountsSavePath[i]) < 0)
11539 11540 11541
            goto cleanup;
    }

11542
    if (virFileMoveMount(devPath, "/dev") < 0)
11543 11544 11545
        goto cleanup;

    for (i = 0; i < ndevMountsPath; i++) {
11546 11547
        struct stat sb;

11548 11549 11550
        if (devMountsSavePath[i] == devPath)
            continue;

11551 11552 11553 11554
        if (stat(devMountsSavePath[i], &sb) < 0) {
            virReportSystemError(errno,
                                 _("Unable to stat: %s"),
                                 devMountsSavePath[i]);
11555 11556 11557
            goto cleanup;
        }

11558 11559 11560 11561 11562 11563 11564 11565 11566 11567 11568 11569 11570 11571 11572
        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;
            }
        }

11573
        if (virFileMoveMount(devMountsSavePath[i], devMountsPath[i]) < 0)
11574 11575 11576 11577 11578
            goto cleanup;
    }

    ret = 0;
 cleanup:
11579 11580 11581 11582 11583 11584 11585
    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]);
    }
11586
    virStringListFreeCount(devMountsPath, ndevMountsPath);
11587
    virStringListFreeCount(devMountsSavePath, ndevMountsPath);
11588 11589 11590 11591 11592 11593 11594 11595 11596 11597 11598
    return ret;
}


int
qemuDomainCreateNamespace(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

11599 11600
    if (virBitmapIsBitSet(cfg->namespaces, QEMU_DOMAIN_NS_MOUNT) &&
        qemuDomainEnableNamespace(vm, QEMU_DOMAIN_NS_MOUNT) < 0)
M
Michal Privoznik 已提交
11601
        goto cleanup;
11602 11603 11604 11605 11606 11607 11608 11609

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


11610 11611 11612 11613 11614 11615 11616 11617 11618
void
qemuDomainDestroyNamespace(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                           virDomainObjPtr vm)
{
    if (qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        qemuDomainDisableNamespace(vm, QEMU_DOMAIN_NS_MOUNT);
}


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
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__) */
}


11648 11649 11650 11651
struct qemuDomainAttachDeviceMknodData {
    virQEMUDriverPtr driver;
    virDomainObjPtr vm;
    const char *file;
11652
    const char *target;
11653 11654
    struct stat sb;
    void *acl;
11655 11656 11657
#ifdef WITH_SELINUX
    char *tcon;
#endif
11658 11659 11660
};


11661 11662
/* Our way of creating devices is highly linux specific */
#if defined(__linux__)
11663 11664 11665 11666 11667 11668
static int
qemuDomainAttachDeviceMknodHelper(pid_t pid ATTRIBUTE_UNUSED,
                                  void *opaque)
{
    struct qemuDomainAttachDeviceMknodData *data = opaque;
    int ret = -1;
11669
    bool delDevice = false;
11670
    bool isLink = S_ISLNK(data->sb.st_mode);
11671
    bool isDev = S_ISCHR(data->sb.st_mode) || S_ISBLK(data->sb.st_mode);
11672
    bool isReg = S_ISREG(data->sb.st_mode) || S_ISFIFO(data->sb.st_mode) || S_ISSOCK(data->sb.st_mode);
11673
    bool isDir = S_ISDIR(data->sb.st_mode);
11674

11675
    qemuSecurityPostFork(data->driver->securityManager);
11676 11677 11678 11679 11680 11681 11682

    if (virFileMakeParentPath(data->file) < 0) {
        virReportSystemError(errno,
                             _("Unable to create %s"), data->file);
        goto cleanup;
    }

11683 11684
    if (isLink) {
        VIR_DEBUG("Creating symlink %s -> %s", data->file, data->target);
11685 11686 11687 11688 11689 11690 11691 11692 11693 11694 11695 11696

        /* 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;
        }

11697
        if (symlink(data->target, data->file) < 0) {
11698 11699 11700 11701
            virReportSystemError(errno,
                                 _("Unable to create symlink %s (pointing to %s)"),
                                 data->file, data->target);
            goto cleanup;
11702 11703
        } else {
            delDevice = true;
11704
        }
11705
    } else if (isDev) {
11706 11707 11708 11709 11710 11711 11712 11713 11714 11715 11716 11717 11718 11719 11720 11721
        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;
        }
11722
    } else if (isReg || isDir) {
11723 11724 11725 11726 11727
        /* 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 &&
11728
            errno != ENOENT && errno != EINVAL) {
11729 11730 11731 11732 11733
            virReportSystemError(errno,
                                 _("Unable to umount %s"),
                                 data->file);
            goto cleanup;
        }
11734 11735
        if ((isReg && virFileTouch(data->file, data->sb.st_mode) < 0) ||
            (isDir && virFileMakePathWithMode(data->file, data->sb.st_mode) < 0))
11736 11737 11738 11739
            goto cleanup;
        delDevice = true;
        /* Just create the file here so that code below sets
         * proper owner and mode. Move the mount only after that. */
11740 11741 11742 11743 11744
    } else {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("unsupported device type %s 0%o"),
                       data->file, data->sb.st_mode);
        goto cleanup;
11745 11746
    }

11747 11748 11749 11750 11751 11752 11753
    if (lchown(data->file, data->sb.st_uid, data->sb.st_gid) < 0) {
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             data->file);
        goto cleanup;
    }

11754 11755 11756 11757 11758 11759 11760 11761 11762
    /* 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;
    }

11763 11764 11765
    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileSetACLs(data->file, data->acl) < 0 &&
11766 11767 11768 11769 11770 11771
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to set ACLs on %s"), data->file);
        goto cleanup;
    }

11772
# ifdef WITH_SELINUX
11773
    if (data->tcon &&
11774
        lsetfilecon_raw(data->file, (VIR_SELINUX_CTX_CONST char *)data->tcon) < 0) {
11775 11776 11777 11778 11779 11780 11781 11782 11783
        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;
        }
    }
11784
# endif
11785

11786
    /* Finish mount process started earlier. */
11787
    if ((isReg || isDir) &&
11788 11789 11790
        virFileMoveMount(data->target, data->file) < 0)
        goto cleanup;

11791 11792
    ret = 0;
 cleanup:
11793 11794 11795 11796 11797 11798
    if (ret < 0 && delDevice) {
        if (isDir)
            virFileDeleteTree(data->file);
        else
            unlink(data->file);
    }
11799
# ifdef WITH_SELINUX
11800
    freecon(data->tcon);
11801
# endif
11802 11803 11804 11805 11806 11807
    virFileFreeACLs(&data->acl);
    return ret;
}


static int
11808 11809 11810
qemuDomainAttachDeviceMknodRecursive(virQEMUDriverPtr driver,
                                     virDomainObjPtr vm,
                                     const char *file,
11811 11812
                                     char * const *devMountsPath,
                                     size_t ndevMountsPath,
11813
                                     unsigned int ttl)
11814
{
11815
    virQEMUDriverConfigPtr cfg = NULL;
11816 11817
    struct qemuDomainAttachDeviceMknodData data;
    int ret = -1;
11818 11819
    char *target = NULL;
    bool isLink;
11820
    bool isReg;
11821
    bool isDir;
11822

11823 11824 11825 11826 11827 11828 11829
    if (!ttl) {
        virReportSystemError(ELOOP,
                             _("Too many levels of symbolic links: %s"),
                             file);
        return ret;
    }

11830 11831 11832 11833 11834 11835
    memset(&data, 0, sizeof(data));

    data.driver = driver;
    data.vm = vm;
    data.file = file;

11836
    if (lstat(file, &data.sb) < 0) {
11837 11838 11839 11840 11841
        virReportSystemError(errno,
                             _("Unable to access %s"), file);
        return ret;
    }

11842
    isLink = S_ISLNK(data.sb.st_mode);
11843
    isReg = S_ISREG(data.sb.st_mode) || S_ISFIFO(data.sb.st_mode) || S_ISSOCK(data.sb.st_mode);
11844
    isDir = S_ISDIR(data.sb.st_mode);
11845

11846
    if ((isReg || isDir) && STRPREFIX(file, DEVPREFIX)) {
11847 11848 11849 11850 11851 11852 11853 11854 11855
        cfg = virQEMUDriverGetConfig(driver);
        if (!(target = qemuDomainGetPreservedMountPath(cfg, vm, file)))
            goto cleanup;

        if (virFileBindMountDevice(file, target) < 0)
            goto cleanup;

        data.target = target;
    } else if (isLink) {
11856 11857 11858 11859 11860 11861 11862 11863 11864 11865 11866 11867 11868 11869 11870 11871 11872 11873 11874 11875 11876 11877 11878 11879 11880 11881 11882 11883 11884 11885 11886 11887
        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);
            target = tmp;
            tmp = NULL;
        }

        data.target = target;
    }

    /* Symlinks don't have ACLs. */
    if (!isLink &&
        virFileGetACLs(file, &data.acl) < 0 &&
11888 11889 11890
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Unable to get ACLs on %s"), file);
11891
        goto cleanup;
11892 11893
    }

11894
# ifdef WITH_SELINUX
11895
    if (lgetfilecon_raw(file, &data.tcon) < 0 &&
11896 11897 11898 11899 11900
        (errno != ENOTSUP && errno != ENODATA)) {
        virReportSystemError(errno,
                             _("Unable to get SELinux label from %s"), file);
        goto cleanup;
    }
11901
# endif
11902

11903
    if (STRPREFIX(file, DEVPREFIX)) {
11904 11905 11906 11907 11908 11909 11910 11911 11912 11913 11914 11915
        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;
11916

11917 11918 11919 11920 11921 11922
            if (virProcessRunInMountNamespace(vm->pid,
                                              qemuDomainAttachDeviceMknodHelper,
                                              &data) < 0) {
                qemuSecurityPostFork(driver->securityManager);
                goto cleanup;
            }
11923
            qemuSecurityPostFork(driver->securityManager);
11924 11925 11926
        } else {
            VIR_DEBUG("Skipping dev %s because of %s mount point",
                      file, devMountsPath[i]);
11927
        }
11928 11929
    }

11930
    if (isLink &&
11931 11932 11933
        qemuDomainAttachDeviceMknodRecursive(driver, vm, target,
                                             devMountsPath, ndevMountsPath,
                                             ttl -1) < 0)
11934
        goto cleanup;
11935 11936 11937

    ret = 0;
 cleanup:
11938
# ifdef WITH_SELINUX
11939
    freecon(data.tcon);
11940
# endif
11941
    virFileFreeACLs(&data.acl);
11942 11943
    if (isReg && target)
        umount(target);
11944
    VIR_FREE(target);
11945
    virObjectUnref(cfg);
11946
    return ret;
11947 11948 11949
}


11950 11951 11952 11953 11954 11955 11956 11957 11958 11959 11960 11961 11962 11963 11964 11965 11966 11967 11968 11969
#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__) */


11970 11971 11972
static int
qemuDomainAttachDeviceMknod(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
11973 11974 11975
                            const char *file,
                            char * const *devMountsPath,
                            size_t ndevMountsPath)
11976 11977 11978
{
    long symloop_max = sysconf(_SC_SYMLOOP_MAX);

11979 11980 11981
    return qemuDomainAttachDeviceMknodRecursive(driver, vm, file,
                                                devMountsPath, ndevMountsPath,
                                                symloop_max);
11982 11983 11984
}


11985 11986 11987 11988 11989 11990 11991 11992 11993 11994 11995 11996 11997 11998 11999 12000 12001 12002
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
12003
qemuDomainDetachDeviceUnlink(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
12004
                             virDomainObjPtr vm,
12005 12006 12007
                             const char *file,
                             char * const *devMountsPath,
                             size_t ndevMountsPath)
12008
{
12009 12010
    int ret = -1;
    size_t i;
12011

12012 12013 12014 12015 12016 12017 12018 12019 12020 12021 12022 12023 12024 12025 12026 12027 12028 12029 12030
    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;
12031 12032 12033
}


12034 12035 12036 12037
static int
qemuDomainNamespaceMknodPaths(virDomainObjPtr vm,
                              const char **paths,
                              size_t npaths)
12038
{
12039 12040 12041
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    virQEMUDriverConfigPtr cfg;
12042 12043
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0;
12044
    int ret = -1;
12045
    size_t i;
12046 12047 12048 12049

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        return 0;

12050 12051 12052 12053 12054 12055
    cfg = virQEMUDriverGetConfig(driver);
    if (qemuDomainGetPreservedMounts(cfg, vm,
                                     &devMountsPath, NULL,
                                     &ndevMountsPath) < 0)
        goto cleanup;

12056 12057 12058 12059 12060 12061 12062 12063 12064 12065 12066 12067 12068 12069 12070 12071
    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;
}


12072 12073 12074 12075 12076 12077 12078 12079 12080 12081
static int
qemuDomainNamespaceMknodPath(virDomainObjPtr vm,
                             const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceMknodPaths(vm, paths, 1);
}


12082 12083 12084 12085 12086 12087 12088 12089 12090 12091 12092 12093 12094 12095 12096 12097 12098 12099 12100 12101 12102 12103 12104 12105 12106 12107 12108 12109 12110
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;

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        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;
    }

12111
    ret = 0;
12112 12113 12114 12115 12116 12117 12118
 cleanup:
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    virObjectUnref(cfg);
    return ret;
}


12119 12120 12121 12122 12123 12124 12125 12126 12127 12128
static int
qemuDomainNamespaceUnlinkPath(virDomainObjPtr vm,
                              const char *path)
{
    const char *paths[] = { path };

    return qemuDomainNamespaceUnlinkPaths(vm, paths, 1);
}


12129
int
12130
qemuDomainNamespaceSetupDisk(virDomainObjPtr vm,
12131 12132 12133
                             virStorageSourcePtr src)
{
    virStorageSourcePtr next;
12134
    const char **paths = NULL;
12135
    size_t npaths = 0;
12136
    char *dmPath = NULL;
12137 12138 12139 12140 12141
    int ret = -1;

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        return 0;

12142
    for (next = src; virStorageSourceIsBacking(next); next = next->backingStore) {
12143 12144
        if (virStorageSourceIsEmpty(next) ||
            !virStorageSourceIsLocalStorage(next)) {
12145 12146 12147 12148
            /* Not creating device. Just continue. */
            continue;
        }

12149
        if (VIR_APPEND_ELEMENT_COPY(paths, npaths, next->path) < 0)
12150 12151 12152
            goto cleanup;
    }

12153
    /* qemu-pr-helper might require access to /dev/mapper/control. */
12154
    if (src->pr &&
12155 12156 12157 12158
        (VIR_STRDUP(dmPath, DEVICE_MAPPER_CONTROL_PATH) < 0 ||
         VIR_APPEND_ELEMENT_COPY(paths, npaths, dmPath) < 0))
        goto cleanup;

12159
    if (qemuDomainNamespaceMknodPaths(vm, paths, npaths) < 0)
12160
        goto cleanup;
12161

12162 12163
    ret = 0;
 cleanup:
12164
    VIR_FREE(dmPath);
12165
    VIR_FREE(paths);
12166 12167 12168 12169 12170
    return ret;
}


int
12171
qemuDomainNamespaceTeardownDisk(virDomainObjPtr vm ATTRIBUTE_UNUSED,
12172
                                virStorageSourcePtr src ATTRIBUTE_UNUSED)
12173 12174 12175 12176 12177 12178 12179 12180 12181
{
    /* 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;
}
12182 12183 12184


int
12185
qemuDomainNamespaceSetupHostdev(virDomainObjPtr vm,
12186 12187 12188
                                virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
12189
    char **paths = NULL;
12190
    size_t i, npaths = 0;
12191 12192 12193 12194

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        return 0;

12195
    if (qemuDomainGetHostdevPath(NULL, hostdev, false, &npaths, &paths, NULL) < 0)
12196 12197
        goto cleanup;

12198
    if (qemuDomainNamespaceMknodPaths(vm, (const char **)paths, npaths) < 0)
12199 12200
        goto cleanup;

12201 12202
    ret = 0;
 cleanup:
12203
    for (i = 0; i < npaths; i++)
12204 12205
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
12206 12207 12208 12209 12210
    return ret;
}


int
12211
qemuDomainNamespaceTeardownHostdev(virDomainObjPtr vm,
12212 12213 12214
                                   virDomainHostdevDefPtr hostdev)
{
    int ret = -1;
12215
    char **paths = NULL;
12216
    size_t i, npaths = 0;
12217 12218 12219 12220

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        return 0;

12221
    if (qemuDomainGetHostdevPath(vm->def, hostdev, true,
12222
                                 &npaths, &paths, NULL) < 0)
12223 12224
        goto cleanup;

12225 12226
    if (npaths != 0 &&
        qemuDomainNamespaceUnlinkPaths(vm, (const char **)paths, npaths) < 0)
12227 12228
        goto cleanup;

12229 12230
    ret = 0;
 cleanup:
12231
    for (i = 0; i < npaths; i++)
12232 12233
        VIR_FREE(paths[i]);
    VIR_FREE(paths);
12234 12235
    return ret;
}
12236 12237


M
Michal Privoznik 已提交
12238
int
12239
qemuDomainNamespaceSetupMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
12240 12241 12242 12243 12244
                               virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

12245 12246
    if (qemuDomainNamespaceMknodPath(vm, mem->nvdimmPath) < 0)
        return -1;
12247

12248
    return 0;
M
Michal Privoznik 已提交
12249 12250 12251 12252
}


int
12253
qemuDomainNamespaceTeardownMemory(virDomainObjPtr vm,
M
Michal Privoznik 已提交
12254 12255 12256 12257 12258
                                  virDomainMemoryDefPtr mem)
{
    if (mem->model != VIR_DOMAIN_MEMORY_MODEL_NVDIMM)
        return 0;

12259 12260
    if (qemuDomainNamespaceUnlinkPath(vm, mem->nvdimmPath) < 0)
        return -1;
12261

12262
    return 0;
M
Michal Privoznik 已提交
12263 12264 12265
}


12266
int
12267
qemuDomainNamespaceSetupChardev(virDomainObjPtr vm,
12268 12269 12270 12271
                                virDomainChrDefPtr chr)
{
    const char *path;

12272
    if (!(path = virDomainChrSourceDefGetPath(chr->source)))
12273 12274
        return 0;

12275 12276 12277 12278
    /* 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;
12279

12280 12281
    if (qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
12282

12283
    return 0;
12284 12285 12286 12287
}


int
12288
qemuDomainNamespaceTeardownChardev(virDomainObjPtr vm,
12289 12290 12291 12292 12293 12294 12295 12296 12297
                                   virDomainChrDefPtr chr)
{
    const char *path = NULL;

    if (chr->source->type != VIR_DOMAIN_CHR_TYPE_DEV)
        return 0;

    path = chr->source->data.file.path;

12298 12299
    if (qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
12300

12301
    return 0;
12302
}
12303 12304 12305


int
12306
qemuDomainNamespaceSetupRNG(virDomainObjPtr vm,
12307 12308 12309 12310 12311 12312 12313 12314 12315 12316 12317
                            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:
12318
        break;
12319 12320
    }

12321 12322
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
12323

12324
    return 0;
12325 12326 12327 12328
}


int
12329
qemuDomainNamespaceTeardownRNG(virDomainObjPtr vm,
12330 12331 12332 12333 12334 12335 12336 12337 12338 12339 12340
                               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:
12341
        break;
12342 12343
    }

12344 12345
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
12346

12347
    return 0;
12348 12349 12350 12351 12352 12353 12354 12355 12356 12357 12358 12359
}


int
qemuDomainNamespaceSetupInput(virDomainObjPtr vm,
                              virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

12360 12361 12362
    if (path && qemuDomainNamespaceMknodPath(vm, path) < 0)
        return -1;
    return 0;
12363 12364 12365 12366 12367 12368 12369 12370 12371 12372 12373 12374
}


int
qemuDomainNamespaceTeardownInput(virDomainObjPtr vm,
                                 virDomainInputDefPtr input)
{
    const char *path = NULL;

    if (!(path = virDomainInputDefGetPath(input)))
        return 0;

12375 12376
    if (path && qemuDomainNamespaceUnlinkPath(vm, path) < 0)
        return -1;
12377

12378
    return 0;
12379
}
12380 12381 12382 12383 12384 12385 12386 12387 12388 12389 12390 12391 12392 12393 12394 12395 12396 12397 12398 12399 12400 12401 12402 12403 12404 12405 12406 12407 12408 12409 12410 12411 12412 12413 12414 12415 12416 12417 12418 12419 12420 12421 12422


/**
 * 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;
}
12423 12424 12425 12426 12427 12428 12429 12430 12431 12432 12433 12434 12435 12436 12437 12438 12439 12440 12441 12442 12443 12444


/**
 * 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;
}
12445 12446 12447 12448 12449 12450 12451 12452 12453 12454 12455 12456 12457 12458 12459 12460 12461 12462 12463 12464 12465 12466 12467 12468 12469 12470 12471


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
12472
                       _("failed to find disk '%s'"), target);
12473 12474 12475
        goto cleanup;
    }

12476 12477 12478 12479
    if (idx == 0)
        src = disk->src;
    else
        src = virStorageFileChainLookup(disk->src, NULL, NULL, idx, NULL);
12480 12481 12482 12483 12484

 cleanup:
    VIR_FREE(target);
    return src;
}
12485 12486 12487 12488 12489 12490 12491 12492


static void
qemuDomainSaveCookieDispose(void *obj)
{
    qemuDomainSaveCookiePtr cookie = obj;

    VIR_DEBUG("cookie=%p", cookie);
12493 12494

    virCPUDefFree(cookie->cpu);
12495 12496 12497 12498 12499 12500
}


qemuDomainSaveCookiePtr
qemuDomainSaveCookieNew(virDomainObjPtr vm ATTRIBUTE_UNUSED)
{
12501
    qemuDomainObjPrivatePtr priv = vm->privateData;
12502 12503 12504 12505 12506 12507 12508 12509
    qemuDomainSaveCookiePtr cookie = NULL;

    if (qemuDomainInitialize() < 0)
        goto error;

    if (!(cookie = virObjectNew(qemuDomainSaveCookieClass)))
        goto error;

12510 12511 12512 12513
    if (priv->origCPU && !(cookie->cpu = virCPUDefCopy(vm->def->cpu)))
        goto error;

    VIR_DEBUG("Save cookie %p, cpu=%p", cookie, cookie->cpu);
12514 12515 12516 12517 12518 12519 12520 12521 12522 12523 12524 12525 12526 12527 12528 12529 12530 12531 12532 12533 12534

    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;

12535 12536 12537 12538
    if (virCPUDefParseXML(ctxt, "./cpu[1]", VIR_CPU_TYPE_GUEST,
                          &cookie->cpu) < 0)
        goto error;

12539 12540 12541 12542 12543 12544 12545 12546 12547 12548
    *obj = (virObjectPtr) cookie;
    return 0;

 error:
    virObjectUnref(cookie);
    return -1;
}


static int
12549 12550
qemuDomainSaveCookieFormat(virBufferPtr buf,
                           virObjectPtr obj)
12551
{
12552 12553 12554
    qemuDomainSaveCookiePtr cookie = (qemuDomainSaveCookiePtr) obj;

    if (cookie->cpu &&
12555
        virCPUDefFormatBufFull(buf, cookie->cpu, NULL) < 0)
12556 12557
        return -1;

12558 12559 12560 12561 12562 12563 12564 12565
    return 0;
}


virSaveCookieCallbacks virQEMUDriverDomainSaveCookie = {
    .parse = qemuDomainSaveCookieParse,
    .format = qemuDomainSaveCookieFormat,
};
12566 12567 12568 12569 12570 12571 12572 12573 12574 12575 12576 12577 12578 12579 12580 12581 12582 12583 12584 12585 12586 12587 12588 12589 12590 12591 12592 12593 12594 12595 12596 12597 12598 12599 12600 12601 12602 12603 12604 12605


/**
 * 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;
}
12606

12607 12608 12609 12610 12611 12612 12613 12614 12615 12616 12617 12618 12619 12620 12621 12622 12623 12624 12625 12626 12627 12628 12629 12630 12631 12632 12633 12634

/**
 * 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;
12635
    int ret = -1;
12636 12637 12638 12639 12640 12641 12642 12643 12644 12645 12646 12647 12648 12649 12650 12651 12652 12653 12654 12655 12656 12657 12658 12659 12660 12661 12662 12663 12664 12665 12666 12667 12668 12669 12670 12671 12672 12673 12674 12675 12676 12677 12678 12679 12680 12681 12682

    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;
}


12683 12684 12685 12686 12687 12688 12689
char *
qemuDomainGetMachineName(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    char *ret = NULL;

12690
    if (vm->pid > 0) {
12691 12692 12693 12694 12695 12696 12697 12698 12699 12700 12701
        ret = virSystemdGetMachineNameByPID(vm->pid);
        if (!ret)
            virResetLastError();
    }

    if (!ret)
        ret = virDomainGenerateMachineName("qemu", vm->def->id, vm->def->name,
                                           virQEMUDriverIsPrivileged(driver));

    return ret;
}
12702 12703 12704 12705 12706 12707 12708 12709 12710 12711 12712 12713 12714 12715 12716 12717 12718 12719 12720 12721 12722 12723 12724 12725


/* Check whether the device address is using either 'ccw' or default s390
 * address format and whether that's "legal" for the current qemu and/or
 * guest os.machine type. This is the corollary to the code which doesn't
 * find the address type set using an emulator that supports either 'ccw'
 * or s390 and sets the address type based on the capabilities.
 *
 * If the address is using 'ccw' or s390 and it's not supported, generate
 * an error and return false; otherwise, return true.
 */
bool
qemuDomainCheckCCWS390AddressSupport(const virDomainDef *def,
                                     virDomainDeviceInfo info,
                                     virQEMUCapsPtr qemuCaps,
                                     const char *devicename)
{
    if (info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) {
        if (!qemuDomainIsS390CCW(def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("cannot use CCW address type for device "
                             "'%s' using machine type '%s'"),
                       devicename, def->os.machine);
            return false;
12726
        } else if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_CCW)) {
12727 12728 12729 12730 12731 12732 12733 12734 12735 12736 12737 12738 12739 12740 12741
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CCW address type is not supported by "
                             "this QEMU"));
            return false;
        }
    } else if (info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_VIRTIO_S390) {
        if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_S390)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("virtio S390 address type is not supported by "
                             "this QEMU"));
            return false;
        }
    }
    return true;
}
12742 12743


12744
/**
12745
 * qemuDomainPrepareDiskSourceData:
12746 12747 12748 12749 12750
 *
 * @disk: Disk config object
 * @src: source to start from
 * @cfg: qemu driver config object
 *
12751 12752 12753
 * Prepares various aspects of a storage source belonging to a disk backing
 * chain. This function should be also called for detected backing chain
 * members.
12754 12755
 */
int
12756 12757 12758 12759
qemuDomainPrepareDiskSourceData(virDomainDiskDefPtr disk,
                                virStorageSourcePtr src,
                                virQEMUDriverConfigPtr cfg,
                                virQEMUCapsPtr qemuCaps)
12760
{
12761
    /* transfer properties valid only for the top level image */
12762 12763
    if (src == disk->src)
        src->detect_zeroes = disk->detect_zeroes;
12764

12765 12766 12767 12768 12769 12770 12771
    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;
    }
12772

12773 12774 12775 12776
    /* transfer properties valid for the full chain */
    src->iomode = disk->iomode;
    src->cachemode = disk->cachemode;
    src->discard = disk->discard;
12777

12778 12779
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY)
        src->floppyimg = true;
12780 12781 12782 12783 12784

    return 0;
}


12785 12786 12787 12788 12789 12790 12791 12792 12793
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;
}


12794 12795
static int
qemuDomainPrepareStorageSourcePR(virStorageSourcePtr src,
12796 12797
                                 qemuDomainObjPrivatePtr priv,
                                 const char *parentalias)
12798 12799 12800 12801 12802 12803 12804
{
    if (!src->pr)
        return 0;

    if (virStoragePRDefIsManaged(src->pr)) {
        if (!(src->pr->path = qemuDomainGetManagedPRSocketPath(priv)))
            return -1;
12805 12806 12807 12808 12809
        if (VIR_STRDUP(src->pr->mgralias, qemuDomainGetManagedPRAlias()) < 0)
            return -1;
    } else {
        if (!(src->pr->mgralias = qemuDomainGetUnmanagedPRAlias(parentalias)))
            return -1;
12810 12811 12812 12813 12814 12815
    }

    return 0;
}


12816 12817 12818 12819 12820 12821 12822 12823 12824 12825 12826 12827
/**
 * 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)
12828
{
12829 12830 12831
    if (qemuDomainValidateStorageSource(disk->src, priv->qemuCaps) < 0)
        return -1;

12832
    if (qemuDomainPrepareDiskSourceData(disk, disk->src, cfg, priv->qemuCaps) < 0)
12833 12834
        return -1;

12835 12836 12837
    if (qemuDomainSecretStorageSourcePrepare(priv, disk->src,
                                             disk->info.alias,
                                             disk->info.alias) < 0)
12838 12839
        return -1;

12840
    if (qemuDomainPrepareStorageSourcePR(disk->src, priv, disk->info.alias) < 0)
12841
        return -1;
12842

12843 12844
    if (qemuDomainPrepareStorageSourceTLS(disk->src, cfg, disk->info.alias,
                                          priv->qemuCaps) < 0)
12845 12846
        return -1;

12847 12848 12849 12850 12851 12852 12853 12854 12855 12856 12857 12858 12859 12860
    return 0;
}


int
qemuDomainPrepareDiskSource(virDomainDiskDefPtr disk,
                            qemuDomainObjPrivatePtr priv,
                            virQEMUDriverConfigPtr cfg)
{
    qemuDomainPrepareDiskCachemode(disk);

    if (qemuDomainPrepareDiskSourceLegacy(disk, priv, cfg) < 0)
        return -1;

12861 12862
    return 0;
}
12863 12864


12865 12866 12867 12868 12869 12870 12871 12872 12873 12874 12875 12876 12877 12878 12879 12880 12881 12882 12883 12884 12885 12886 12887 12888 12889 12890 12891 12892 12893 12894 12895 12896 12897 12898 12899 12900 12901 12902 12903 12904 12905 12906 12907 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 12937 12938 12939
/**
 * 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;
}


12940 12941 12942 12943 12944 12945 12946 12947 12948 12949 12950 12951 12952 12953 12954 12955 12956 12957 12958 12959 12960 12961 12962
void
qemuProcessEventFree(struct qemuProcessEvent *event)
{
    if (!event)
        return;

    switch (event->eventType) {
    case QEMU_PROCESS_EVENT_GUESTPANIC:
        qemuMonitorEventPanicInfoFree(event->data);
        break;
    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;
    case QEMU_PROCESS_EVENT_LAST:
        break;
    }
    VIR_FREE(event);
}
12963 12964 12965


char *
12966
qemuDomainGetManagedPRSocketPath(qemuDomainObjPrivatePtr priv)
12967 12968 12969
{
    char *ret = NULL;

12970 12971
    ignore_value(virAsprintf(&ret, "%s/%s.sock", priv->libDir,
                             qemuDomainGetManagedPRAlias()));
12972 12973 12974

    return ret;
}