qemu_domain.c 223.3 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_cgroup.h"
29
#include "qemu_command.h"
30
#include "qemu_process.h"
31
#include "qemu_parse_command.h"
32
#include "qemu_capabilities.h"
33
#include "qemu_migration.h"
34
#include "viralloc.h"
35
#include "virlog.h"
36
#include "virerror.h"
37
#include "c-ctype.h"
38
#include "cpu/cpu.h"
39
#include "viruuid.h"
E
Eric Blake 已提交
40
#include "virfile.h"
41
#include "domain_addr.h"
42
#include "domain_event.h"
43
#include "virtime.h"
44
#include "virstoragefile.h"
45
#include "virstring.h"
46
#include "virthreadjob.h"
47
#include "viratomic.h"
48
#include "virprocess.h"
49
#include "vircrypto.h"
50
#include "secret_util.h"
51
#include "logging/log_manager.h"
52
#include "locking/domain_lock.h"
53

54 55
#include "storage/storage_driver.h"

56
#include <sys/time.h>
57
#include <fcntl.h>
58 59 60
#if defined(HAVE_SYS_MOUNT_H)
# include <sys/mount.h>
#endif
61

62 63 64 65
#include <libxml/xpathInternals.h>

#define VIR_FROM_THIS VIR_FROM_QEMU

66 67
VIR_LOG_INIT("qemu.qemu_domain");

68 69
#define QEMU_NAMESPACE_HREF "http://libvirt.org/schemas/domain/qemu/1.0"

70 71 72 73 74 75
VIR_ENUM_IMPL(qemuDomainJob, QEMU_JOB_LAST,
              "none",
              "query",
              "destroy",
              "suspend",
              "modify",
76
              "abort",
77
              "migration operation",
78 79 80 81 82 83 84 85 86 87
              "none",   /* async job is never stored in job.active */
              "async nested",
);

VIR_ENUM_IMPL(qemuDomainAsyncJob, QEMU_ASYNC_JOB_LAST,
              "none",
              "migration out",
              "migration in",
              "save",
              "dump",
88
              "snapshot",
89
              "start",
90 91
);

92 93 94 95 96 97 98 99 100 101 102 103 104 105
VIR_ENUM_IMPL(qemuDomainNamespace, QEMU_DOMAIN_NS_LAST,
              "mount",
);


static struct {
    const char *path;
    const char *suffix;
} devPreserveMounts[] = {
    {"/dev/pts", "devpts"},
    {"/dev/shm", "devshm"},
    {"/dev/mqueue", "mqueue"},
};

106

107 108 109
struct _qemuDomainLogContext {
    int refs;
    int writefd;
110
    int readfd; /* Only used if manager == NULL */
111
    off_t pos;
112
    ino_t inode; /* Only used if manager != NULL */
113
    char *path;
114
    virLogManagerPtr manager;
115 116
};

J
Jiri Denemark 已提交
117
const char *
118
qemuDomainAsyncJobPhaseToString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
119 120 121 122 123
                                int phase ATTRIBUTE_UNUSED)
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
124 125
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
126 127
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
128
    case QEMU_ASYNC_JOB_SNAPSHOT:
129
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
130 131 132 133 134 135 136 137 138
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        ; /* fall through */
    }

    return "none";
}

int
139
qemuDomainAsyncJobPhaseFromString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
140 141 142 143 144 145 146 147
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
148 149
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
150 151
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
152
    case QEMU_ASYNC_JOB_SNAPSHOT:
153
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
154 155 156 157 158 159 160 161 162 163 164
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        ; /* fall through */
    }

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

165

166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
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;
}


static int
qemuDomainGetPreservedMounts(virQEMUDriverPtr driver,
                             virDomainObjPtr vm,
                             char ***devMountsPath,
                             size_t *ndevMountsPath)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    char **paths;
    size_t i;

    if (VIR_ALLOC_N(paths, ARRAY_CARDINALITY(devPreserveMounts)) < 0)
        goto error;

    for (i = 0; i < ARRAY_CARDINALITY(devPreserveMounts); i++) {
        if (virAsprintf(&paths[i], "%s/%s.%s",
                        cfg->stateDir, vm->def->name,
                        devPreserveMounts[i].suffix) < 0)
            goto error;
    }

    *devMountsPath = paths;
    *ndevMountsPath = ARRAY_CARDINALITY(devPreserveMounts);
    virObjectUnref(cfg);
    return 0;

 error:
    virStringListFreeCount(paths, ARRAY_CARDINALITY(devPreserveMounts));
    virObjectUnref(cfg);
    return -1;
}


230
void qemuDomainEventQueue(virQEMUDriverPtr driver,
231
                          virObjectEventPtr event)
232
{
233 234
    if (event)
        virObjectEventStateQueue(driver->domainEventState, event);
235 236 237
}


238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261
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);
    qemuDomainEventQueue(driver, event);
}


262 263 264 265 266 267 268 269
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

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

270
    if (virCondInit(&priv->job.asyncCond) < 0) {
271
        virCondDestroy(&priv->job.cond);
272 273 274
        return -1;
    }

275 276 277 278 279 280 281 282 283
    return 0;
}

static void
qemuDomainObjResetJob(qemuDomainObjPrivatePtr priv)
{
    struct qemuDomainJobObj *job = &priv->job;

    job->active = QEMU_JOB_NONE;
284
    job->owner = 0;
285
    job->ownerAPI = NULL;
J
Jiri Denemark 已提交
286
    job->started = 0;
287 288 289 290 291 292 293 294
}

static void
qemuDomainObjResetAsyncJob(qemuDomainObjPrivatePtr priv)
{
    struct qemuDomainJobObj *job = &priv->job;

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
295
    job->asyncOwner = 0;
296
    job->asyncOwnerAPI = NULL;
J
Jiri Denemark 已提交
297
    job->asyncStarted = 0;
J
Jiri Denemark 已提交
298
    job->phase = 0;
299
    job->mask = QEMU_JOB_DEFAULT_MASK;
300
    job->dump_memory_only = false;
301
    job->abortJob = false;
302
    job->spiceMigration = false;
303
    job->spiceMigrated = false;
304
    job->postcopyEnabled = false;
J
Jiri Denemark 已提交
305
    VIR_FREE(job->current);
306 307
}

308 309 310 311 312 313 314 315
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
                        struct qemuDomainJobObj *job)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
316
    job->owner = priv->job.owner;
317
    job->asyncJob = priv->job.asyncJob;
318
    job->asyncOwner = priv->job.asyncOwner;
J
Jiri Denemark 已提交
319
    job->phase = priv->job.phase;
320 321 322 323 324

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

325 326 327
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
J
Jiri Denemark 已提交
328
    VIR_FREE(priv->job.current);
329
    VIR_FREE(priv->job.completed);
330 331
    virCondDestroy(&priv->job.cond);
    virCondDestroy(&priv->job.asyncCond);
332 333
}

334
static bool
335
qemuDomainTrackJob(qemuDomainJob job)
336 337 338 339
{
    return (QEMU_DOMAIN_TRACK_JOBS & JOB_MASK(job)) != 0;
}

340

J
Jiri Denemark 已提交
341 342 343 344 345 346 347 348 349 350 351
int
qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->started)
        return 0;

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

352 353 354 355 356 357
    if (now < jobInfo->started) {
        VIR_WARN("Async job starts in the future");
        jobInfo->started = 0;
        return 0;
    }

J
Jiri Denemark 已提交
358 359 360 361
    jobInfo->timeElapsed = now - jobInfo->started;
    return 0;
}

362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378
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;
    }

379 380
    jobInfo->stats.downtime = now - jobInfo->stopped;
    jobInfo->stats.downtime_set = true;
381 382 383
    return 0;
}

J
Jiri Denemark 已提交
384 385 386 387
int
qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                        virDomainJobInfoPtr info)
{
388
    info->type = jobInfo->type;
J
Jiri Denemark 已提交
389 390 391
    info->timeElapsed = jobInfo->timeElapsed;
    info->timeRemaining = jobInfo->timeRemaining;

392 393 394
    info->memTotal = jobInfo->stats.ram_total;
    info->memRemaining = jobInfo->stats.ram_remaining;
    info->memProcessed = jobInfo->stats.ram_transferred;
J
Jiri Denemark 已提交
395

396 397 398
    info->fileTotal = jobInfo->stats.disk_total;
    info->fileRemaining = jobInfo->stats.disk_remaining;
    info->fileProcessed = jobInfo->stats.disk_transferred;
J
Jiri Denemark 已提交
399 400 401 402 403 404 405 406 407 408 409 410 411 412

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

    return 0;
}

int
qemuDomainJobInfoToParams(qemuDomainJobInfoPtr jobInfo,
                          int *type,
                          virTypedParameterPtr *params,
                          int *nparams)
{
413
    qemuMonitorMigrationStats *stats = &jobInfo->stats;
J
Jiri Denemark 已提交
414 415 416 417 418 419 420 421 422
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED,
                                jobInfo->timeElapsed) < 0)
        goto error;

423 424 425 426 427 428 429
    if (jobInfo->timeDeltaSet &&
        jobInfo->timeElapsed > jobInfo->timeDelta &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_ELAPSED_NET,
                                jobInfo->timeElapsed - jobInfo->timeDelta) < 0)
        goto error;

J
Jiri Denemark 已提交
430 431 432 433 434 435
    if (jobInfo->type == VIR_DOMAIN_JOB_BOUNDED &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_REMAINING,
                                jobInfo->timeRemaining) < 0)
        goto error;

436
    if (stats->downtime_set &&
J
Jiri Denemark 已提交
437 438
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME,
439
                                stats->downtime) < 0)
J
Jiri Denemark 已提交
440 441
        goto error;

442
    if (stats->downtime_set &&
443
        jobInfo->timeDeltaSet &&
444
        stats->downtime > jobInfo->timeDelta &&
445 446
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME_NET,
447
                                stats->downtime - jobInfo->timeDelta) < 0)
448 449
        goto error;

450
    if (stats->setup_time_set &&
451 452
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_SETUP_TIME,
453
                                stats->setup_time) < 0)
454 455
        goto error;

J
Jiri Denemark 已提交
456 457
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_TOTAL,
458 459
                                stats->ram_total +
                                stats->disk_total) < 0 ||
J
Jiri Denemark 已提交
460 461
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_PROCESSED,
462 463
                                stats->ram_transferred +
                                stats->disk_transferred) < 0 ||
J
Jiri Denemark 已提交
464 465
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_REMAINING,
466 467
                                stats->ram_remaining +
                                stats->disk_remaining) < 0)
J
Jiri Denemark 已提交
468 469 470 471
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
472
                                stats->ram_total) < 0 ||
J
Jiri Denemark 已提交
473 474
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
475
                                stats->ram_transferred) < 0 ||
J
Jiri Denemark 已提交
476 477
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
478
                                stats->ram_remaining) < 0)
J
Jiri Denemark 已提交
479 480
        goto error;

481
    if (stats->ram_bps &&
482 483
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_BPS,
484
                                stats->ram_bps) < 0)
485 486
        goto error;

487
    if (stats->ram_duplicate_set) {
J
Jiri Denemark 已提交
488 489
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_CONSTANT,
490
                                    stats->ram_duplicate) < 0 ||
J
Jiri Denemark 已提交
491 492
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL,
493
                                    stats->ram_normal) < 0 ||
J
Jiri Denemark 已提交
494 495
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL_BYTES,
496
                                    stats->ram_normal_bytes) < 0)
J
Jiri Denemark 已提交
497 498 499
            goto error;
    }

500 501 502 503 504 505 506 507
    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;

J
Jiri Denemark 已提交
508 509
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_TOTAL,
510
                                stats->disk_total) < 0 ||
J
Jiri Denemark 已提交
511 512
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_PROCESSED,
513
                                stats->disk_transferred) < 0 ||
J
Jiri Denemark 已提交
514 515
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_REMAINING,
516
                                stats->disk_remaining) < 0)
J
Jiri Denemark 已提交
517 518
        goto error;

519
    if (stats->disk_bps &&
520 521
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_BPS,
522
                                stats->disk_bps) < 0)
523 524
        goto error;

525
    if (stats->xbzrle_set) {
J
Jiri Denemark 已提交
526 527
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE,
528
                                    stats->xbzrle_cache_size) < 0 ||
J
Jiri Denemark 已提交
529 530
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_BYTES,
531
                                    stats->xbzrle_bytes) < 0 ||
J
Jiri Denemark 已提交
532 533
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_PAGES,
534
                                    stats->xbzrle_pages) < 0 ||
J
Jiri Denemark 已提交
535 536
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE_MISSES,
537
                                    stats->xbzrle_cache_miss) < 0 ||
J
Jiri Denemark 已提交
538 539
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_OVERFLOW,
540
                                    stats->xbzrle_overflow) < 0)
J
Jiri Denemark 已提交
541 542 543
            goto error;
    }

544 545 546 547 548 549
    if (stats->cpu_throttle_percentage &&
        virTypedParamsAddInt(&par, &npar, &maxpar,
                             VIR_DOMAIN_JOB_AUTO_CONVERGE_THROTTLE,
                             stats->cpu_throttle_percentage) < 0)
        goto error;

J
Jiri Denemark 已提交
550 551 552 553 554 555 556 557 558 559 560
    *type = jobInfo->type;
    *params = par;
    *nparams = npar;
    return 0;

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


J
John Ferlan 已提交
561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582
/* 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:
583 584
 * @driver: qemu driver data
 * @vm: Pointer to the vm object
J
John Ferlan 已提交
585 586 587 588 589
 *
 * 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
 */
590
int
M
Martin Kletzander 已提交
591 592
qemuDomainWriteMasterKeyFile(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
J
John Ferlan 已提交
593 594 595 596
{
    char *path;
    int fd = -1;
    int ret = -1;
M
Martin Kletzander 已提交
597
    qemuDomainObjPrivatePtr priv = vm->privateData;
J
John Ferlan 已提交
598

599 600 601 602
    /* Only gets filled in if we have the capability */
    if (!priv->masterKey)
        return 0;

J
John Ferlan 已提交
603 604 605 606 607 608 609 610 611 612 613 614 615 616 617
    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;
    }

618 619
    if (virSecurityManagerDomainSetPathLabel(driver->securityManager,
                                             vm->def, path) < 0)
M
Martin Kletzander 已提交
620 621
        goto cleanup;

J
John Ferlan 已提交
622 623 624 625 626 627 628 629 630 631
    ret = 0;

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

    return ret;
}


632 633 634 635 636 637 638 639 640
static void
qemuDomainMasterKeyFree(qemuDomainObjPrivatePtr priv)
{
    if (!priv->masterKey)
        return;

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

J
John Ferlan 已提交
641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738
/* 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 */
739
    qemuDomainMasterKeyFree(priv);
J
John Ferlan 已提交
740 741 742 743 744 745 746 747 748 749

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

    VIR_FREE(path);
}


/* qemuDomainMasterKeyCreate:
750
 * @vm: Pointer to the domain object
J
John Ferlan 已提交
751 752 753 754 755 756 757 758
 *
 * 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
759
qemuDomainMasterKeyCreate(virDomainObjPtr vm)
J
John Ferlan 已提交
760
{
M
Martin Kletzander 已提交
761 762
    qemuDomainObjPrivatePtr priv = vm->privateData;

J
John Ferlan 已提交
763 764 765 766 767
    /* If we don't have the capability, then do nothing. */
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

    if (!(priv->masterKey =
768
          virCryptoGenerateRandom(QEMU_DOMAIN_MASTER_KEY_LEN)))
769
        return -1;
J
John Ferlan 已提交
770 771 772 773 774 775 776

    priv->masterKeyLen = QEMU_DOMAIN_MASTER_KEY_LEN;

    return 0;
}


777
static void
778
qemuDomainSecretPlainClear(qemuDomainSecretPlain secret)
779 780
{
    VIR_FREE(secret.username);
781
    VIR_DISPOSE_N(secret.secret, secret.secretlen);
782 783 784
}


J
John Ferlan 已提交
785
static void
786
qemuDomainSecretAESClear(qemuDomainSecretAES secret)
J
John Ferlan 已提交
787 788 789 790 791 792 793 794
{
    VIR_FREE(secret.username);
    VIR_FREE(secret.alias);
    VIR_FREE(secret.iv);
    VIR_FREE(secret.ciphertext);
}


795 796 797 798 799 800
static void
qemuDomainSecretInfoFree(qemuDomainSecretInfoPtr *secinfo)
{
    if (!*secinfo)
        return;

J
John Ferlan 已提交
801 802
    switch ((qemuDomainSecretInfoType) (*secinfo)->type) {
    case VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN:
803
        qemuDomainSecretPlainClear((*secinfo)->s.plain);
J
John Ferlan 已提交
804 805
        break;

806
    case VIR_DOMAIN_SECRET_INFO_TYPE_AES:
807
        qemuDomainSecretAESClear((*secinfo)->s.aes);
J
John Ferlan 已提交
808 809 810 811 812 813
        break;

    case VIR_DOMAIN_SECRET_INFO_TYPE_LAST:
        break;
    }

814 815 816 817
    VIR_FREE(*secinfo);
}


818
static virClassPtr qemuDomainDiskPrivateClass;
819
static void qemuDomainDiskPrivateDispose(void *obj);
820 821 822 823 824 825 826

static int
qemuDomainDiskPrivateOnceInit(void)
{
    qemuDomainDiskPrivateClass = virClassNew(virClassForObject(),
                                             "qemuDomainDiskPrivate",
                                             sizeof(qemuDomainDiskPrivate),
827
                                             qemuDomainDiskPrivateDispose);
828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850
    if (!qemuDomainDiskPrivateClass)
        return -1;
    else
        return 0;
}

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;
}


851 852 853 854 855 856
static void
qemuDomainDiskPrivateDispose(void *obj)
{
    qemuDomainDiskPrivatePtr priv = obj;

    qemuDomainSecretInfoFree(&priv->secinfo);
J
John Ferlan 已提交
857
    qemuDomainSecretInfoFree(&priv->encinfo);
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
static virClassPtr qemuDomainHostdevPrivateClass;
static void qemuDomainHostdevPrivateDispose(void *obj);

static int
qemuDomainHostdevPrivateOnceInit(void)
{
    qemuDomainHostdevPrivateClass =
        virClassNew(virClassForObject(),
                    "qemuDomainHostdevPrivate",
                    sizeof(qemuDomainHostdevPrivate),
                    qemuDomainHostdevPrivateDispose);
    if (!qemuDomainHostdevPrivateClass)
        return -1;
    else
        return 0;
}

VIR_ONCE_GLOBAL_INIT(qemuDomainHostdevPrivate)

static virObjectPtr
qemuDomainHostdevPrivateNew(void)
{
    qemuDomainHostdevPrivatePtr priv;

    if (qemuDomainHostdevPrivateInitialize() < 0)
        return NULL;

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

    return (virObjectPtr) priv;
}


static void
qemuDomainHostdevPrivateDispose(void *obj)
{
    qemuDomainHostdevPrivatePtr priv = obj;

    qemuDomainSecretInfoFree(&priv->secinfo);
}


904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937
static virClassPtr qemuDomainVcpuPrivateClass;
static void qemuDomainVcpuPrivateDispose(void *obj);

static int
qemuDomainVcpuPrivateOnceInit(void)
{
    qemuDomainVcpuPrivateClass = virClassNew(virClassForObject(),
                                             "qemuDomainVcpuPrivate",
                                             sizeof(qemuDomainVcpuPrivate),
                                             qemuDomainVcpuPrivateDispose);
    if (!qemuDomainVcpuPrivateClass)
        return -1;
    else
        return 0;
}

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
938
qemuDomainVcpuPrivateDispose(void *obj)
939
{
940 941 942 943
    qemuDomainVcpuPrivatePtr priv = obj;

    VIR_FREE(priv->type);
    VIR_FREE(priv->alias);
944 945 946 947
    return;
}


948 949
static virClassPtr qemuDomainChrSourcePrivateClass;
static void qemuDomainChrSourcePrivateDispose(void *obj);
950 951

static int
952
qemuDomainChrSourcePrivateOnceInit(void)
953
{
954
    qemuDomainChrSourcePrivateClass =
955
        virClassNew(virClassForObject(),
956 957 958 959
                    "qemuDomainChrSourcePrivate",
                    sizeof(qemuDomainChrSourcePrivate),
                    qemuDomainChrSourcePrivateDispose);
    if (!qemuDomainChrSourcePrivateClass)
960 961 962 963 964
        return -1;
    else
        return 0;
}

965
VIR_ONCE_GLOBAL_INIT(qemuDomainChrSourcePrivate)
966 967

static virObjectPtr
968
qemuDomainChrSourcePrivateNew(void)
969
{
970
    qemuDomainChrSourcePrivatePtr priv;
971

972
    if (qemuDomainChrSourcePrivateInitialize() < 0)
973 974
        return NULL;

975
    if (!(priv = virObjectNew(qemuDomainChrSourcePrivateClass)))
976 977 978 979 980 981 982
        return NULL;

    return (virObjectPtr) priv;
}


static void
983
qemuDomainChrSourcePrivateDispose(void *obj)
984
{
985
    qemuDomainChrSourcePrivatePtr priv = obj;
986 987 988 989 990

    qemuDomainSecretInfoFree(&priv->secinfo);
}


991 992 993
/* qemuDomainSecretPlainSetup:
 * @conn: Pointer to connection
 * @secinfo: Pointer to secret info
994
 * @secretUsageType: The virSecretUsageType
995 996
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
997 998 999 1000 1001 1002 1003 1004
 *
 * Taking a secinfo, fill in the plaintext information
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
qemuDomainSecretPlainSetup(virConnectPtr conn,
                           qemuDomainSecretInfoPtr secinfo,
1005
                           virSecretUsageType secretUsageType,
1006 1007
                           const char *username,
                           virSecretLookupTypeDefPtr seclookupdef)
1008
{
1009
    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN;
1010
    if (VIR_STRDUP(secinfo->s.plain.username, username) < 0)
1011 1012
        return -1;

1013
    return virSecretGetSecretString(conn, seclookupdef, secretUsageType,
1014 1015
                                    &secinfo->s.plain.secret,
                                    &secinfo->s.plain.secretlen);
1016 1017 1018
}


1019 1020 1021 1022 1023
/* qemuDomainSecretAESSetup:
 * @conn: Pointer to connection
 * @priv: pointer to domain private object
 * @secinfo: Pointer to secret info
 * @srcalias: Alias of the disk/hostdev used to generate the secret alias
1024
 * @secretUsageType: The virSecretUsageType
1025 1026
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1027
 * @isLuks: True/False for is for luks (alias generation)
1028 1029 1030 1031 1032 1033 1034 1035 1036 1037
 *
 * Taking a secinfo, fill in the AES specific information using the
 *
 * Returns 0 on success, -1 on failure with error message
 */
static int
qemuDomainSecretAESSetup(virConnectPtr conn,
                         qemuDomainObjPrivatePtr priv,
                         qemuDomainSecretInfoPtr secinfo,
                         const char *srcalias,
1038
                         virSecretUsageType secretUsageType,
1039
                         const char *username,
1040 1041
                         virSecretLookupTypeDefPtr seclookupdef,
                         bool isLuks)
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051
{
    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;

    secinfo->type = VIR_DOMAIN_SECRET_INFO_TYPE_AES;
1052
    if (VIR_STRDUP(secinfo->s.aes.username, username) < 0)
1053 1054
        return -1;

1055
    if (!(secinfo->s.aes.alias = qemuDomainGetSecretAESAlias(srcalias, isLuks)))
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066
        return -1;

    /* Create a random initialization vector */
    if (!(raw_iv = virCryptoGenerateRandom(ivlen)))
        return -1;

    /* 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 */
1067
    if (virSecretGetSecretString(conn, seclookupdef, secretUsageType,
1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095
                                 &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);

    return ret;
}


1096 1097
/* qemuDomainSecretSetup:
 * @conn: Pointer to connection
1098
 * @priv: pointer to domain private object
1099
 * @secinfo: Pointer to secret info
1100
 * @srcalias: Alias of the disk/hostdev used to generate the secret alias
1101
 * @secretUsageType: The virSecretUsageType
1102 1103
 * @username: username to use for authentication (may be NULL)
 * @seclookupdef: Pointer to seclookupdef data
1104
 * @isLuks: True when is luks (generates different alias)
1105
 *
1106 1107 1108 1109
 * If we have the encryption API present and can support a secret object, then
 * build the AES secret; otherwise, build the Plain secret. This is the magic
 * decision point for utilizing the AES secrets for an RBD disk. For now iSCSI
 * disks and hostdevs will not be able to utilize this mechanism.
1110 1111 1112 1113 1114
 *
 * Returns 0 on success, -1 on failure
 */
static int
qemuDomainSecretSetup(virConnectPtr conn,
1115
                      qemuDomainObjPrivatePtr priv,
1116
                      qemuDomainSecretInfoPtr secinfo,
1117
                      const char *srcalias,
1118
                      virSecretUsageType secretUsageType,
1119
                      const char *username,
1120 1121
                      virSecretLookupTypeDefPtr seclookupdef,
                      bool isLuks)
1122
{
1123 1124
    if (virCryptoHaveCipher(VIR_CRYPTO_CIPHER_AES256CBC) &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET) &&
1125
        (secretUsageType == VIR_SECRET_USAGE_TYPE_CEPH ||
1126 1127
         secretUsageType == VIR_SECRET_USAGE_TYPE_VOLUME ||
         secretUsageType == VIR_SECRET_USAGE_TYPE_TLS)) {
1128
        if (qemuDomainSecretAESSetup(conn, priv, secinfo, srcalias,
1129
                                     secretUsageType, username,
1130
                                     seclookupdef, isLuks) < 0)
1131 1132
            return -1;
    } else {
1133
        if (qemuDomainSecretPlainSetup(conn, secinfo, secretUsageType,
1134
                                       username, seclookupdef) < 0)
1135 1136 1137
            return -1;
    }
    return 0;
1138 1139 1140
}


1141 1142 1143 1144 1145 1146 1147 1148 1149 1150
/* qemuDomainSecretDiskDestroy:
 * @disk: Pointer to a disk definition
 *
 * Clear and destroy memory associated with the secret
 */
void
qemuDomainSecretDiskDestroy(virDomainDiskDefPtr disk)
{
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

1151
    if (!diskPriv || !diskPriv->secinfo)
1152 1153 1154 1155 1156 1157
        return;

    qemuDomainSecretInfoFree(&diskPriv->secinfo);
}


1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171
bool
qemuDomainSecretDiskCapable(virStorageSourcePtr src)
{
    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;
}


1172 1173
/* qemuDomainSecretDiskPrepare:
 * @conn: Pointer to connection
J
John Ferlan 已提交
1174
 * @priv: pointer to domain private object
1175 1176 1177 1178 1179 1180 1181 1182
 * @disk: Pointer to a disk definition
 *
 * For the right disk, generate the qemuDomainSecretInfo structure.
 *
 * Returns 0 on success, -1 on failure
 */
int
qemuDomainSecretDiskPrepare(virConnectPtr conn,
1183
                            qemuDomainObjPrivatePtr priv,
1184 1185 1186
                            virDomainDiskDefPtr disk)
{
    virStorageSourcePtr src = disk->src;
1187
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
1188 1189
    qemuDomainSecretInfoPtr secinfo = NULL;

1190
    if (qemuDomainSecretDiskCapable(src)) {
1191
        virSecretUsageType secretUsageType = VIR_SECRET_USAGE_TYPE_ISCSI;
1192 1193 1194 1195

        if (VIR_ALLOC(secinfo) < 0)
            return -1;

1196 1197 1198
        if (src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD)
            secretUsageType = VIR_SECRET_USAGE_TYPE_CEPH;

1199
        if (qemuDomainSecretSetup(conn, priv, secinfo, disk->info.alias,
1200
                                  secretUsageType, src->auth->username,
1201
                                  &src->auth->seclookupdef, false) < 0)
1202 1203 1204 1205 1206
            goto error;

        diskPriv->secinfo = secinfo;
    }

1207
    if (!virStorageSourceIsEmpty(src) && src->encryption &&
1208
        src->encryption->format == VIR_STORAGE_ENCRYPTION_FORMAT_LUKS) {
1209 1210 1211 1212 1213 1214 1215 1216 1217 1218

        if (VIR_ALLOC(secinfo) < 0)
            return -1;

        if (qemuDomainSecretSetup(conn, priv, secinfo, disk->info.alias,
                                  VIR_SECRET_USAGE_TYPE_VOLUME, NULL,
                                  &src->encryption->secrets[0]->seclookupdef,
                                  true) < 0)
            goto error;

1219 1220 1221 1222 1223 1224 1225
        if (secinfo->type == VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("luks encryption requires encrypted secrets "
                             "to be supported"));
            goto error;
        }

1226 1227 1228
        diskPriv->encinfo = secinfo;
    }

1229 1230 1231 1232 1233 1234 1235 1236
    return 0;

 error:
    qemuDomainSecretInfoFree(&secinfo);
    return -1;
}


1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247
/* qemuDomainSecretHostdevDestroy:
 * @disk: Pointer to a hostdev definition
 *
 * Clear and destroy memory associated with the secret
 */
void
qemuDomainSecretHostdevDestroy(virDomainHostdevDefPtr hostdev)
{
    qemuDomainHostdevPrivatePtr hostdevPriv =
        QEMU_DOMAIN_HOSTDEV_PRIVATE(hostdev);

1248
    if (!hostdevPriv || !hostdevPriv->secinfo)
1249 1250 1251 1252 1253 1254 1255 1256
        return;

    qemuDomainSecretInfoFree(&hostdevPriv->secinfo);
}


/* qemuDomainSecretHostdevPrepare:
 * @conn: Pointer to connection
J
John Ferlan 已提交
1257
 * @priv: pointer to domain private object
1258 1259 1260 1261 1262 1263 1264 1265
 * @hostdev: Pointer to a hostdev definition
 *
 * For the right host device, generate the qemuDomainSecretInfo structure.
 *
 * Returns 0 on success, -1 on failure
 */
int
qemuDomainSecretHostdevPrepare(virConnectPtr conn,
1266
                               qemuDomainObjPrivatePtr priv,
1267 1268 1269 1270
                               virDomainHostdevDefPtr hostdev)
{
    qemuDomainSecretInfoPtr secinfo = NULL;

1271
    if (virHostdevIsSCSIDevice(hostdev)) {
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283
        virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;
        virDomainHostdevSubsysSCSIiSCSIPtr iscsisrc = &scsisrc->u.iscsi;

        if (scsisrc->protocol == VIR_DOMAIN_HOSTDEV_SCSI_PROTOCOL_TYPE_ISCSI &&
            iscsisrc->auth) {

            qemuDomainHostdevPrivatePtr hostdevPriv =
                QEMU_DOMAIN_HOSTDEV_PRIVATE(hostdev);

            if (VIR_ALLOC(secinfo) < 0)
                return -1;

1284
            if (qemuDomainSecretSetup(conn, priv, secinfo, hostdev->info->alias,
1285
                                      VIR_SECRET_USAGE_TYPE_ISCSI,
1286
                                      iscsisrc->auth->username,
1287
                                      &iscsisrc->auth->seclookupdef, false) < 0)
1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301
                goto error;

            hostdevPriv->secinfo = secinfo;
        }
    }

    return 0;

 error:
    qemuDomainSecretInfoFree(&secinfo);
    return -1;
}


1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388
/* 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:
 * @conn: Pointer to connection
 * @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
qemuDomainSecretChardevPrepare(virConnectPtr conn,
                               virQEMUDriverConfigPtr cfg,
                               qemuDomainObjPrivatePtr priv,
                               const char *chrAlias,
                               virDomainChrSourceDefPtr dev)
{
    virSecretLookupTypeDef seclookupdef = {0};
    qemuDomainSecretInfoPtr secinfo = NULL;
    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 (virUUIDParse(cfg->chardevTLSx509secretUUID,
                         seclookupdef.u.uuid) < 0) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("malformed chardev TLS secret uuid in qemu.conf"));
            goto error;
        }
        seclookupdef.type = VIR_SECRET_LOOKUP_TYPE_UUID;

        if (VIR_ALLOC(secinfo) < 0)
            goto error;

        if (!(charAlias = qemuAliasChardevFromDevAlias(chrAlias)))
            goto error;

        if (qemuDomainSecretSetup(conn, priv, secinfo, charAlias,
                                  VIR_SECRET_USAGE_TYPE_TLS, NULL,
                                  &seclookupdef, false) < 0)
            goto error;

        if (secinfo->type == VIR_DOMAIN_SECRET_INFO_TYPE_PLAIN) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("TLS X.509 requires encrypted secrets "
                             "to be supported"));
            goto error;
        }

        chrSourcePriv->secinfo = secinfo;
    }

    VIR_FREE(charAlias);
    return 0;

 error:
    qemuDomainSecretInfoFree(&secinfo);
    return -1;
}


1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401
/* qemuDomainSecretDestroy:
 * @vm: Domain object
 *
 * Once completed with the generation of the command line it is
 * expect to remove the secrets
 */
void
qemuDomainSecretDestroy(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++)
        qemuDomainSecretDiskDestroy(vm->def->disks[i]);
1402 1403 1404

    for (i = 0; i < vm->def->nhostdevs; i++)
        qemuDomainSecretHostdevDestroy(vm->def->hostdevs[i]);
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

    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);
1431 1432 1433 1434 1435
}


/* qemuDomainSecretPrepare:
 * @conn: Pointer to connection
1436
 * @driver: Pointer to driver object
1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448
 * @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
qemuDomainSecretPrepare(virConnectPtr conn,
1449
                        virQEMUDriverPtr driver,
1450 1451
                        virDomainObjPtr vm)
{
J
John Ferlan 已提交
1452
    qemuDomainObjPrivatePtr priv = vm->privateData;
1453
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1454
    size_t i;
1455
    int ret = -1;
1456 1457

    for (i = 0; i < vm->def->ndisks; i++) {
J
John Ferlan 已提交
1458
        if (qemuDomainSecretDiskPrepare(conn, priv, vm->def->disks[i]) < 0)
1459
            goto cleanup;
1460 1461
    }

1462
    for (i = 0; i < vm->def->nhostdevs; i++) {
J
John Ferlan 已提交
1463 1464
        if (qemuDomainSecretHostdevPrepare(conn, priv,
                                           vm->def->hostdevs[i]) < 0)
1465
            goto cleanup;
1466 1467
    }

1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523
    for (i = 0; i < vm->def->nserials; i++) {
        if (qemuDomainSecretChardevPrepare(conn, cfg, priv,
                                           vm->def->serials[i]->info.alias,
                                           vm->def->serials[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nparallels; i++) {
        if (qemuDomainSecretChardevPrepare(conn, cfg, priv,
                                           vm->def->parallels[i]->info.alias,
                                           vm->def->parallels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nchannels; i++) {
        if (qemuDomainSecretChardevPrepare(conn, cfg, priv,
                                           vm->def->channels[i]->info.alias,
                                           vm->def->channels[i]->source) < 0)
            goto cleanup;
    }

    for (i = 0; i < vm->def->nconsoles; i++) {
        if (qemuDomainSecretChardevPrepare(conn, cfg, priv,
                                           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 &&
            qemuDomainSecretChardevPrepare(conn, cfg, priv,
                                           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 &&
            qemuDomainSecretChardevPrepare(conn, cfg, priv,
                                           vm->def->rngs[i]->info.alias,
                                           vm->def->rngs[i]->source.chardev) < 0)
            goto cleanup;
    }

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

    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
1524 1525 1526
}


1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553
/* 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
1554 1555
qemuDomainSetPrivatePaths(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
1556
{
1557 1558
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainObjPrivatePtr priv = vm->privateData;
1559
    char *domname = virDomainObjGetShortName(vm->def);
1560
    int ret = -1;
1561

M
Martin Kletzander 已提交
1562 1563 1564
    if (!domname)
        goto cleanup;

1565
    if (!priv->libDir &&
M
Martin Kletzander 已提交
1566
        virAsprintf(&priv->libDir, "%s/domain-%s", cfg->libDir, domname) < 0)
1567
        goto cleanup;
1568

1569
    if (!priv->channelTargetDir &&
M
Martin Kletzander 已提交
1570 1571
        virAsprintf(&priv->channelTargetDir, "%s/domain-%s",
                    cfg->channelTargetDir, domname) < 0)
1572
        goto cleanup;
1573

1574 1575 1576
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
M
Martin Kletzander 已提交
1577
    VIR_FREE(domname);
1578
    return ret;
1579 1580 1581
}


1582 1583 1584 1585 1586 1587 1588 1589 1590 1591
void
qemuDomainClearPrivatePaths(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    VIR_FREE(priv->libDir);
    VIR_FREE(priv->channelTargetDir);
}


1592 1593
static void *
qemuDomainObjPrivateAlloc(void)
1594 1595 1596 1597 1598 1599
{
    qemuDomainObjPrivatePtr priv;

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

1600 1601 1602
    if (qemuDomainObjInitJob(priv) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to init qemu driver mutexes"));
1603
        goto error;
1604
    }
1605

1606
    if (!(priv->devs = virChrdevAlloc()))
1607 1608
        goto error;

1609
    priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
1610

1611
    return priv;
1612

1613
 error:
1614 1615
    VIR_FREE(priv);
    return NULL;
1616 1617
}

1618 1619
static void
qemuDomainObjPrivateFree(void *data)
1620 1621 1622
{
    qemuDomainObjPrivatePtr priv = data;

1623
    virObjectUnref(priv->qemuCaps);
1624

1625 1626
    virBitmapFree(priv->namespaces);

1627
    virCgroupFree(&priv->cgroup);
1628
    virDomainPCIAddressSetFree(priv->pciaddrs);
J
Ján Tomko 已提交
1629
    virDomainUSBAddressSetFree(priv->usbaddrs);
1630
    virDomainChrSourceDefFree(priv->monConfig);
1631
    qemuDomainObjFreeJob(priv);
1632
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
1633
    VIR_FREE(priv->origname);
1634

1635
    virStringListFree(priv->qemuDevices);
1636
    virChrdevFree(priv->devs);
1637

1638 1639
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
1640
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
1641 1642
        qemuMonitorClose(priv->mon);
    }
D
Daniel P. Berrange 已提交
1643 1644 1645 1646
    if (priv->agent) {
        VIR_ERROR(_("Unexpected QEMU agent still active during domain deletion"));
        qemuAgentClose(priv->agent);
    }
1647
    VIR_FREE(priv->cleanupCallbacks);
1648
    virBitmapFree(priv->autoNodeset);
1649
    virBitmapFree(priv->autoCpuset);
1650 1651 1652

    VIR_FREE(priv->libDir);
    VIR_FREE(priv->channelTargetDir);
1653
    qemuDomainMasterKeyFree(priv);
1654

1655 1656 1657 1658
    VIR_FREE(priv);
}


1659 1660
static void
qemuDomainObjPrivateXMLFormatVcpus(virBufferPtr buf,
1661
                                   virDomainDefPtr def)
1662 1663
{
    size_t i;
1664 1665 1666
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    pid_t tid;
1667 1668 1669 1670

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

1671 1672 1673 1674 1675 1676 1677 1678 1679
    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);
    }
1680 1681 1682 1683 1684 1685

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


1686
static int
1687 1688
qemuDomainObjPrivateXMLFormat(virBufferPtr buf,
                              virDomainObjPtr vm)
1689
{
1690
    qemuDomainObjPrivatePtr priv = vm->privateData;
1691
    const char *monitorpath;
1692
    qemuDomainJob job;
1693 1694 1695

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
1696
        switch (priv->monConfig->type) {
1697
        case VIR_DOMAIN_CHR_TYPE_UNIX:
1698
            monitorpath = priv->monConfig->data.nix.path;
1699 1700 1701
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
1702
            monitorpath = priv->monConfig->data.file.path;
1703 1704 1705
            break;
        }

1706
        virBufferEscapeString(buf, "<monitor path='%s'", monitorpath);
1707 1708
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
1709
        virBufferAsprintf(buf, " type='%s'/>\n",
1710
                          virDomainChrTypeToString(priv->monConfig->type));
1711 1712
    }

1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723
    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");
    }

1724
    qemuDomainObjPrivateXMLFormatVcpus(buf, vm->def);
1725

1726
    if (priv->qemuCaps) {
1727
        size_t i;
1728 1729
        virBufferAddLit(buf, "<qemuCaps>\n");
        virBufferAdjustIndent(buf, 2);
1730
        for (i = 0; i < QEMU_CAPS_LAST; i++) {
1731
            if (virQEMUCapsGet(priv->qemuCaps, i)) {
1732
                virBufferAsprintf(buf, "<flag name='%s'/>\n",
1733
                                  virQEMUCapsTypeToString(i));
1734 1735
            }
        }
1736 1737
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</qemuCaps>\n");
1738 1739
    }

1740
    if (priv->lockState)
1741
        virBufferAsprintf(buf, "<lockstate>%s</lockstate>\n", priv->lockState);
1742

1743 1744 1745 1746
    job = priv->job.active;
    if (!qemuDomainTrackJob(job))
        priv->job.active = QEMU_JOB_NONE;

1747
    if (priv->job.active || priv->job.asyncJob) {
1748
        virBufferAsprintf(buf, "<job type='%s' async='%s'",
1749 1750
                          qemuDomainJobTypeToString(priv->job.active),
                          qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
J
Jiri Denemark 已提交
1751 1752 1753 1754 1755
        if (priv->job.phase) {
            virBufferAsprintf(buf, " phase='%s'",
                              qemuDomainAsyncJobPhaseToString(
                                    priv->job.asyncJob, priv->job.phase));
        }
1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776
        if (priv->job.asyncJob != QEMU_ASYNC_JOB_MIGRATION_OUT) {
            virBufferAddLit(buf, "/>\n");
        } else {
            size_t i;
            virDomainDiskDefPtr disk;
            qemuDomainDiskPrivatePtr diskPriv;

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

            for (i = 0; i < vm->def->ndisks; i++) {
                disk = vm->def->disks[i];
                diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
                virBufferAsprintf(buf, "<disk dev='%s' migrating='%s'/>\n",
                                  disk->dst,
                                  diskPriv->migrating ? "yes" : "no");
            }

            virBufferAdjustIndent(buf, -2);
            virBufferAddLit(buf, "</job>\n");
        }
1777
    }
1778
    priv->job.active = job;
1779

1780
    if (priv->fakeReboot)
1781
        virBufferAddLit(buf, "<fakereboot/>\n");
1782

1783 1784
    if (priv->qemuDevices && *priv->qemuDevices) {
        char **tmp = priv->qemuDevices;
1785 1786
        virBufferAddLit(buf, "<devices>\n");
        virBufferAdjustIndent(buf, 2);
1787
        while (*tmp) {
1788
            virBufferAsprintf(buf, "<device alias='%s'/>\n", *tmp);
1789 1790
            tmp++;
        }
1791 1792
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</devices>\n");
1793 1794
    }

1795 1796 1797 1798 1799 1800 1801 1802 1803 1804
    if (priv->autoNodeset) {
        char *nodeset = virBitmapFormat(priv->autoNodeset);

        if (!nodeset)
            return -1;

        virBufferAsprintf(buf, "<numad nodeset='%s'/>\n", nodeset);
        VIR_FREE(nodeset);
    }

1805 1806 1807 1808 1809
    /* Various per-domain paths */
    virBufferEscapeString(buf, "<libDir path='%s'/>\n", priv->libDir);
    virBufferEscapeString(buf, "<channelTargetDir path='%s'/>\n",
                          priv->channelTargetDir);

1810 1811 1812
    return 0;
}

1813 1814 1815 1816

static int
qemuDomainObjPrivateXMLParseVcpu(xmlNodePtr node,
                                 unsigned int idx,
1817
                                 virDomainDefPtr def)
1818
{
1819
    virDomainVcpuDefPtr vcpu;
1820
    char *idstr;
1821
    char *pidstr;
1822
    unsigned int tmp;
1823 1824
    int ret = -1;

1825 1826
    idstr = virXMLPropString(node, "id");

1827 1828
    if (idstr &&
        (virStrToLong_uip(idstr, NULL, 10, &idx) < 0)) {
1829
        virReportError(VIR_ERR_INTERNAL_ERROR,
1830 1831 1832 1833 1834 1835
                       _("cannot parse vcpu index '%s'"), idstr);
        goto cleanup;
    }
    if (!(vcpu = virDomainDefGetVcpu(def, idx))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("invalid vcpu index '%u'"), idx);
1836
        goto cleanup;
1837 1838
    }

1839 1840 1841
    if (!(pidstr = virXMLPropString(node, "pid")))
        goto cleanup;

1842
    if (virStrToLong_uip(pidstr, NULL, 10, &tmp) < 0)
1843 1844
        goto cleanup;

1845 1846
    QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid = tmp;

1847 1848 1849
    ret = 0;

 cleanup:
1850
    VIR_FREE(idstr);
1851 1852 1853 1854 1855
    VIR_FREE(pidstr);
    return ret;
}


1856
static int
1857
qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt,
1858
                             virDomainObjPtr vm,
1859
                             virDomainDefParserConfigPtr config)
1860
{
1861
    qemuDomainObjPrivatePtr priv = vm->privateData;
1862
    virQEMUDriverPtr driver = config->priv;
1863
    char *monitorpath;
1864
    char *tmp = NULL;
1865 1866
    int n;
    size_t i;
1867
    xmlNodePtr *nodes = NULL;
1868
    xmlNodePtr node = NULL;
1869
    virQEMUCapsPtr qemuCaps = NULL;
1870
    virCapsPtr caps = NULL;
1871

1872
    if (VIR_ALLOC(priv->monConfig) < 0)
1873 1874 1875 1876
        goto error;

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
1877 1878
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("no monitor path"));
1879 1880 1881 1882 1883
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
1884
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
1885
    else
1886
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
1887 1888
    VIR_FREE(tmp);

E
Eric Blake 已提交
1889 1890
    priv->monJSON = virXPathBoolean("count(./monitor[@json = '1']) > 0",
                                    ctxt) > 0;
1891

1892
    switch (priv->monConfig->type) {
1893
    case VIR_DOMAIN_CHR_TYPE_PTY:
1894
        priv->monConfig->data.file.path = monitorpath;
1895 1896
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
1897
        priv->monConfig->data.nix.path = monitorpath;
1898 1899 1900
        break;
    default:
        VIR_FREE(monitorpath);
1901 1902 1903
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported monitor type '%s'"),
                       virDomainChrTypeToString(priv->monConfig->type));
1904 1905 1906
        goto error;
    }

1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930
    if ((node = virXPathNode("./namespaces", ctxt))) {
        xmlNodePtr next;

        for (next = node->children; next; next = next->next) {
            int ns = qemuDomainNamespaceTypeFromString((const char *) next->name);

            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;
    }

1931
    if ((n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes)) < 0)
1932 1933
        goto error;

1934
    for (i = 0; i < n; i++) {
1935
        if (qemuDomainObjPrivateXMLParseVcpu(nodes[i], i, vm->def) < 0)
1936
            goto error;
1937
    }
1938
    VIR_FREE(nodes);
1939

1940
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
1941 1942
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse qemu capabilities flags"));
1943 1944 1945
        goto error;
    }
    if (n > 0) {
1946
        if (!(qemuCaps = virQEMUCapsNew()))
1947 1948
            goto error;

1949
        for (i = 0; i < n; i++) {
1950 1951
            char *str = virXMLPropString(nodes[i], "name");
            if (str) {
1952
                int flag = virQEMUCapsTypeFromString(str);
1953
                if (flag < 0) {
1954 1955
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Unknown qemu capabilities flag %s"), str);
1956
                    VIR_FREE(str);
1957 1958
                    goto error;
                }
1959
                VIR_FREE(str);
1960
                virQEMUCapsSet(qemuCaps, flag);
1961 1962 1963
            }
        }

1964
        priv->qemuCaps = qemuCaps;
1965
        qemuCaps = NULL;
1966 1967 1968
    }
    VIR_FREE(nodes);

1969
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
1970

1971 1972 1973 1974
    if ((tmp = virXPathString("string(./job[1]/@type)", ctxt))) {
        int type;

        if ((type = qemuDomainJobTypeFromString(tmp)) < 0) {
1975 1976
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown job type %s"), tmp);
1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987
            VIR_FREE(tmp);
            goto error;
        }
        VIR_FREE(tmp);
        priv->job.active = type;
    }

    if ((tmp = virXPathString("string(./job[1]/@async)", ctxt))) {
        int async;

        if ((async = qemuDomainAsyncJobTypeFromString(tmp)) < 0) {
1988 1989
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown async job type %s"), tmp);
1990 1991 1992 1993 1994
            VIR_FREE(tmp);
            goto error;
        }
        VIR_FREE(tmp);
        priv->job.asyncJob = async;
J
Jiri Denemark 已提交
1995 1996 1997 1998

        if ((tmp = virXPathString("string(./job[1]/@phase)", ctxt))) {
            priv->job.phase = qemuDomainAsyncJobPhaseFromString(async, tmp);
            if (priv->job.phase < 0) {
1999 2000
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Unknown job phase %s"), tmp);
J
Jiri Denemark 已提交
2001 2002 2003 2004 2005
                VIR_FREE(tmp);
                goto error;
            }
            VIR_FREE(tmp);
        }
2006 2007
    }

2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030
    if ((n = virXPathNodeSet("./job[1]/disk[@migrating='yes']",
                             ctxt, &nodes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("failed to parse list of disks marked for migration"));
        goto error;
    }
    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++) {
            char *dst = virXMLPropString(nodes[i], "dev");
            virDomainDiskDefPtr disk;

            if (dst && (disk = virDomainDiskByName(vm->def, dst, false)))
                QEMU_DOMAIN_DISK_PRIVATE(disk)->migrating = true;
            VIR_FREE(dst);
        }
    }
    VIR_FREE(nodes);

2031 2032
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053
    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);

2054 2055 2056 2057
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto error;

    if ((tmp = virXPathString("string(./numad/@nodeset)", ctxt))) {
2058
        if (virBitmapParse(tmp, &priv->autoNodeset,
2059 2060 2061 2062 2063 2064 2065 2066 2067 2068
                           caps->host.nnumaCell_max) < 0)
            goto error;

        if (!(priv->autoCpuset = virCapabilitiesGetCpusForNodemask(caps,
                                                                   priv->autoNodeset)))
            goto error;
    }
    virObjectUnref(caps);
    VIR_FREE(tmp);

2069 2070 2071 2072 2073 2074 2075 2076 2077
    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;

2078 2079
    return 0;

2080
 error:
2081
    VIR_FREE(nodes);
2082
    VIR_FREE(tmp);
2083 2084 2085 2086
    virBitmapFree(priv->namespaces);
    priv->namespaces = NULL;
    virDomainChrSourceDefFree(priv->monConfig);
    priv->monConfig = NULL;
2087
    virStringListFree(priv->qemuDevices);
2088
    priv->qemuDevices = NULL;
2089
    virObjectUnref(qemuCaps);
2090
    virObjectUnref(caps);
2091 2092 2093 2094
    return -1;
}


2095 2096 2097
virDomainXMLPrivateDataCallbacks virQEMUDriverPrivateDataCallbacks = {
    .alloc = qemuDomainObjPrivateAlloc,
    .free = qemuDomainObjPrivateFree,
2098
    .diskNew = qemuDomainDiskPrivateNew,
2099
    .vcpuNew = qemuDomainVcpuPrivateNew,
2100
    .hostdevNew = qemuDomainHostdevPrivateNew,
2101
    .chrSourceNew = qemuDomainChrSourcePrivateNew,
2102 2103 2104 2105 2106
    .parse = qemuDomainObjPrivateXMLParse,
    .format = qemuDomainObjPrivateXMLFormat,
};


2107 2108 2109 2110 2111
static void
qemuDomainDefNamespaceFree(void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;

2112
    qemuDomainCmdlineDefFree(cmd);
2113 2114 2115
}

static int
P
Philipp Hahn 已提交
2116 2117
qemuDomainDefNamespaceParse(xmlDocPtr xml ATTRIBUTE_UNUSED,
                            xmlNodePtr root ATTRIBUTE_UNUSED,
2118 2119 2120 2121
                            xmlXPathContextPtr ctxt,
                            void **data)
{
    qemuDomainCmdlineDefPtr cmd = NULL;
P
Philipp Hahn 已提交
2122
    bool uses_qemu_ns = false;
2123
    xmlNodePtr *nodes = NULL;
2124 2125
    int n;
    size_t i;
2126

P
Philipp Hahn 已提交
2127
    if (xmlXPathRegisterNs(ctxt, BAD_CAST "qemu", BAD_CAST QEMU_NAMESPACE_HREF) < 0) {
2128 2129 2130
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to register xml namespace '%s'"),
                       QEMU_NAMESPACE_HREF);
2131 2132 2133
        return -1;
    }

2134
    if (VIR_ALLOC(cmd) < 0)
2135 2136 2137 2138 2139 2140
        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
2141
    uses_qemu_ns |= n > 0;
2142 2143

    if (n && VIR_ALLOC_N(cmd->args, n) < 0)
2144
        goto error;
2145 2146 2147 2148

    for (i = 0; i < n; i++) {
        cmd->args[cmd->num_args] = virXMLPropString(nodes[i], "value");
        if (cmd->args[cmd->num_args] == NULL) {
2149 2150
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu command-line argument specified"));
2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161
            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
2162
    uses_qemu_ns |= n > 0;
2163 2164

    if (n && VIR_ALLOC_N(cmd->env_name, n) < 0)
2165
        goto error;
2166 2167

    if (n && VIR_ALLOC_N(cmd->env_value, n) < 0)
2168
        goto error;
2169 2170 2171 2172 2173 2174

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

        tmp = virXMLPropString(nodes[i], "name");
        if (tmp == NULL) {
2175 2176
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu environment name specified"));
2177 2178 2179
            goto error;
        }
        if (tmp[0] == '\0') {
2180 2181
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Empty qemu environment name specified"));
2182 2183 2184
            goto error;
        }
        if (!c_isalpha(tmp[0]) && tmp[0] != '_') {
2185 2186
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must begin with a letter or underscore"));
2187 2188 2189
            goto error;
        }
        if (strspn(tmp, "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789_") != strlen(tmp)) {
2190 2191
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must contain only alphanumerics and underscore"));
2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203
            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
2204 2205 2206 2207
    if (uses_qemu_ns)
        *data = cmd;
    else
        VIR_FREE(cmd);
2208 2209 2210

    return 0;

2211
 error:
2212 2213 2214 2215 2216 2217 2218 2219 2220 2221
    VIR_FREE(nodes);
    qemuDomainDefNamespaceFree(cmd);
    return -1;
}

static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;
2222
    size_t i;
2223 2224 2225 2226

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

2227 2228 2229
    virBufferAddLit(buf, "<qemu:commandline>\n");
    virBufferAdjustIndent(buf, 2);

2230
    for (i = 0; i < cmd->num_args; i++)
2231
        virBufferEscapeString(buf, "<qemu:arg value='%s'/>\n",
2232 2233
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
2234
        virBufferAsprintf(buf, "<qemu:env name='%s'", cmd->env_name[i]);
2235 2236 2237 2238 2239
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }

2240 2241
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:commandline>\n");
2242 2243 2244 2245 2246 2247 2248 2249 2250 2251
    return 0;
}

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


2252 2253 2254 2255 2256 2257
virDomainXMLNamespace virQEMUDriverDomainXMLNamespace = {
    .parse = qemuDomainDefNamespaceParse,
    .free = qemuDomainDefNamespaceFree,
    .format = qemuDomainDefNamespaceFormatXML,
    .href = qemuDomainDefNamespaceHref,
};
2258

2259

P
Pavel Hrdina 已提交
2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278
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;
}


2279
static int
2280 2281
qemuDomainDefAddDefaultDevices(virDomainDefPtr def,
                               virQEMUCapsPtr qemuCaps)
2282
{
2283
    bool addDefaultUSB = true;
2284
    int usbModel = -1; /* "default for machinetype" */
2285
    int pciRoot;       /* index within def->controllers */
2286
    bool addImplicitSATA = false;
2287
    bool addPCIRoot = false;
L
Laine Stump 已提交
2288
    bool addPCIeRoot = false;
2289
    bool addDefaultMemballoon = true;
2290 2291
    bool addDefaultUSBKBD = false;
    bool addDefaultUSBMouse = false;
2292
    bool addPanicDevice = false;
2293
    int ret = -1;
2294

P
Pavel Hrdina 已提交
2295 2296 2297 2298
    /* add implicit input devices */
    if (qemuDomainDefAddImplicitInputDevice(def) < 0)
        goto cleanup;

2299 2300 2301 2302
    /* 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
2303
        if (STREQ(def->os.machine, "isapc")) {
2304
            addDefaultUSB = false;
2305
            break;
2306
        }
2307
        if (qemuDomainMachineIsQ35(def)) {
2308 2309
            addPCIeRoot = true;
            addImplicitSATA = true;
2310

2311 2312 2313 2314
            /* Prefer adding USB3 controller if supported
             * (nec-usb-xhci). Failing that, add a USB2 controller set
             * if the ich9-usb-ehci1 device is supported. Otherwise
             * don't add anything.
2315
             */
2316 2317 2318
            if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_NEC_USB_XHCI))
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_NEC_XHCI;
            else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_USB_EHCI1))
2319 2320 2321
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_ICH9_EHCI1;
            else
                addDefaultUSB = false;
2322
            break;
L
Laine Stump 已提交
2323
        }
2324 2325
        if (qemuDomainMachineIsI440FX(def))
            addPCIRoot = true;
2326 2327
        break;

2328
    case VIR_ARCH_ARMV7L:
2329
    case VIR_ARCH_AARCH64:
2330 2331
        addDefaultUSB = false;
        addDefaultMemballoon = false;
J
Ján Tomko 已提交
2332
        if (qemuDomainMachineIsVirt(def))
2333 2334
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
        break;
2335

2336
    case VIR_ARCH_PPC64:
2337
    case VIR_ARCH_PPC64LE:
2338 2339 2340
        addPCIRoot = true;
        addDefaultUSBKBD = true;
        addDefaultUSBMouse = true;
2341 2342 2343
        /* For pSeries guests, the firmware provides the same
         * functionality as the pvpanic device, so automatically
         * add the definition if not already present */
2344
        if (qemuDomainMachineIsPSeries(def))
2345
            addPanicDevice = true;
2346 2347
        break;

2348 2349 2350 2351 2352 2353 2354
    case VIR_ARCH_ALPHA:
    case VIR_ARCH_PPC:
    case VIR_ARCH_PPCEMB:
    case VIR_ARCH_SH4:
    case VIR_ARCH_SH4EB:
        addPCIRoot = true;
        break;
2355

2356 2357 2358
    case VIR_ARCH_S390:
    case VIR_ARCH_S390X:
        addDefaultUSB = false;
2359
        addPanicDevice = true;
2360
        break;
2361 2362 2363 2364 2365 2366

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

2367 2368 2369 2370
    default:
        break;
    }

2371
    if (addDefaultUSB &&
2372 2373
        virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_USB, 0) < 0 &&
        virDomainDefAddUSBController(def, 0, usbModel) < 0)
2374
        goto cleanup;
2375

2376 2377 2378
    if (addImplicitSATA &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_SATA, 0, -1) < 0)
2379
        goto cleanup;
2380

2381 2382
    pciRoot = virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0);

2383 2384 2385
    /* NB: any machine that sets addPCIRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
     */
2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
    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;
        }
    }
2401

2402 2403 2404
    /* 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
2405 2406 2407
     *
     * NB: any machine that sets addPCIeRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
2408 2409
     */
    if (addPCIeRoot) {
2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422
        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;
        }
2423
    }
2424

2425
    if (addDefaultMemballoon && !def->memballoon) {
2426 2427
        virDomainMemballoonDefPtr memballoon;
        if (VIR_ALLOC(memballoon) < 0)
2428
            goto cleanup;
2429 2430 2431 2432 2433

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

2434 2435 2436 2437 2438
    if (addDefaultUSBKBD &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_KBD,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
2439
        goto cleanup;
2440 2441 2442 2443 2444 2445

    if (addDefaultUSBMouse &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
2446
        goto cleanup;
2447

D
Dmitry Andreev 已提交
2448 2449 2450 2451
    if (addPanicDevice) {
        size_t j;
        for (j = 0; j < def->npanics; j++) {
            if (def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT ||
2452 2453 2454 2455
                (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
2456 2457
                break;
        }
2458

D
Dmitry Andreev 已提交
2459 2460 2461 2462 2463 2464 2465 2466 2467
        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;
            }
        }
2468 2469
    }

2470 2471 2472 2473 2474 2475
    ret = 0;
 cleanup:
    return ret;
}


A
Andrea Bolognani 已提交
2476 2477 2478
/**
 * qemuDomainDefEnableDefaultFeatures:
 * @def: domain definition
2479
 * @qemuCaps: QEMU capabilities
A
Andrea Bolognani 已提交
2480 2481 2482 2483 2484
 *
 * Make sure that features that should be enabled by default are actually
 * enabled and configure default values related to those features.
 */
static void
2485 2486
qemuDomainDefEnableDefaultFeatures(virDomainDefPtr def,
                                   virQEMUCapsPtr qemuCaps)
A
Andrea Bolognani 已提交
2487
{
2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507
    virGICVersion version;

    /* The virt machine type always uses GIC: if the relevant element
     * was not included in the domain XML, we need to choose a suitable
     * GIC version ourselves */
    if (def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ABSENT &&
        qemuDomainMachineIsVirt(def)) {

        VIR_DEBUG("Looking for usable GIC version in domain capabilities");
        for (version = VIR_GIC_VERSION_LAST - 1;
             version > VIR_GIC_VERSION_NONE;
             version--) {
            if (virQEMUCapsSupportsGICVersion(qemuCaps,
                                              def->virtType,
                                              version)) {
                VIR_DEBUG("Using GIC version %s",
                          virGICVersionTypeToString(version));
                def->gic_version = version;
                break;
            }
2508 2509
        }

2510 2511 2512
        /* 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;
2513 2514
    }

2515
    /* Use the default GIC version if no version was specified */
A
Andrea Bolognani 已提交
2516 2517
    if (def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ON &&
        def->gic_version == VIR_GIC_VERSION_NONE)
2518
        def->gic_version = VIR_GIC_VERSION_DEFAULT;
A
Andrea Bolognani 已提交
2519 2520 2521
}


2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541
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;
}


2542
static int
2543 2544 2545 2546 2547
qemuDomainRecheckInternalPaths(virDomainDefPtr def,
                               virQEMUDriverConfigPtr cfg,
                               unsigned int flags)
{
    size_t i = 0;
2548
    size_t j = 0;
2549 2550 2551 2552

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

2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570
        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
             * the auto-generated socket base on config option from qemu.conf
             * 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 &&
                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;
                }
2571
            }
2572 2573
        }
    }
2574 2575

    return 0;
2576 2577 2578
}


2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 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
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;
}


2649 2650 2651
static int
qemuDomainDefPostParse(virDomainDefPtr def,
                       virCapsPtr caps,
2652
                       unsigned int parseFlags,
2653
                       void *opaque,
2654
                       void *parseOpaque)
2655 2656
{
    virQEMUDriverPtr driver = opaque;
2657
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2658
    virQEMUCapsPtr qemuCaps = parseOpaque;
2659 2660 2661 2662 2663
    int ret = -1;

    if (def->os.bootloader || def->os.bootloaderArgs) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bootloader is not supported by QEMU"));
2664
        goto cleanup;
2665 2666
    }

2667 2668 2669
    if (!def->os.machine) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing machine type"));
2670
        goto cleanup;
2671 2672
    }

2673 2674 2675 2676 2677 2678 2679 2680 2681
    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;
    }

2682 2683 2684
    /* check for emulator and create a default one if needed */
    if (!def->emulator &&
        !(def->emulator = virDomainDefGetDefaultEmulator(def, caps)))
2685
        goto cleanup;
2686

2687 2688 2689 2690 2691 2692 2693 2694
    if (qemuCaps) {
        virObjectRef(qemuCaps);
    } else {
        if (!(qemuCaps = virQEMUCapsCacheLookup(caps,
                                                driver->qemuCapsCache,
                                                def->emulator)))
            goto cleanup;
    }
2695 2696 2697 2698

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

2699 2700 2701
    if (qemuCanonicalizeMachine(def, qemuCaps) < 0)
        goto cleanup;

2702
    qemuDomainDefEnableDefaultFeatures(def, qemuCaps);
A
Andrea Bolognani 已提交
2703

2704 2705
    if (qemuDomainRecheckInternalPaths(def, cfg, parseFlags) < 0)
        goto cleanup;
2706

2707 2708 2709
    if (virSecurityManagerVerify(driver->securityManager, def) < 0)
        goto cleanup;

2710 2711 2712
    if (qemuDomainDefVcpusPostParse(def) < 0)
        goto cleanup;

2713 2714 2715
    ret = 0;
 cleanup:
    virObjectUnref(qemuCaps);
2716
    virObjectUnref(cfg);
2717
    return ret;
2718 2719
}

2720

2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747
static int
qemuDomainDefValidateVideo(const virDomainDef *def)
{
    size_t i;
    virDomainVideoDefPtr video;

    for (i = 0; i < def->nvideos; i++) {
        video = def->videos[i];

        switch (video->type) {
        case VIR_DOMAIN_VIDEO_TYPE_XEN:
        case VIR_DOMAIN_VIDEO_TYPE_VBOX:
        case VIR_DOMAIN_VIDEO_TYPE_PARALLELS:
            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 &&
2748 2749
            video->type != VIR_DOMAIN_VIDEO_TYPE_QXL &&
            video->type != VIR_DOMAIN_VIDEO_TYPE_VIRTIO) {
2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792
            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);
                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)"));
                return -1;
            }
        }
    }

    return 0;
}


2793 2794
static int
qemuDomainDefValidate(const virDomainDef *def,
2795
                      virCapsPtr caps,
2796
                      void *opaque)
2797
{
2798 2799
    virQEMUDriverPtr driver = opaque;
    virQEMUCapsPtr qemuCaps = NULL;
2800
    unsigned int topologycpus;
2801 2802
    int ret = -1;

2803 2804
    if (!(qemuCaps = virQEMUCapsCacheLookup(caps,
                                            driver->qemuCapsCache,
2805 2806 2807
                                            def->emulator)))
        goto cleanup;

2808 2809 2810
    if (def->mem.min_guarantee) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Parameter 'min_guarantee' not supported by QEMU."));
2811
        goto cleanup;
2812 2813
    }

M
Michal Privoznik 已提交
2814 2815 2816 2817 2818 2819 2820 2821
    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. */

        if (!qemuDomainMachineIsQ35(def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot is supported with q35 machine types only"));
2822
            goto cleanup;
M
Michal Privoznik 已提交
2823 2824 2825 2826 2827 2828 2829 2830
        }

        /* 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"));
2831
            goto cleanup;
M
Michal Privoznik 已提交
2832 2833 2834 2835 2836
        }

        if (def->features[VIR_DOMAIN_FEATURE_SMM] != VIR_TRISTATE_SWITCH_ON) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Secure boot requires SMM feature enabled"));
2837
            goto cleanup;
M
Michal Privoznik 已提交
2838 2839 2840
        }
    }

2841
    /* qemu as of 2.5.0 rejects SMP topologies that don't match the cpu count */
2842 2843 2844 2845 2846 2847 2848
    if (virDomainDefGetVcpusTopology(def, &topologycpus) == 0 &&
        topologycpus != virDomainDefGetVcpusMax(def)) {
        /* presence of query-hotpluggable-cpus should be a good enough witness */
        if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("CPU topology doesn't match maximum vcpu count"));
            goto cleanup;
2849 2850 2851
        }
    }

2852 2853 2854
    if (qemuDomainDefValidateVideo(def) < 0)
        goto cleanup;

2855 2856 2857 2858 2859
    ret = 0;

 cleanup:
    virObjectUnref(qemuCaps);
    return ret;
2860 2861 2862
}


2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881
static int
qemuDomainDeviceDefValidate(const virDomainDeviceDef *dev,
                            const virDomainDef *def ATTRIBUTE_UNUSED,
                            void *opaque)
{
    virQEMUDriverPtr driver = opaque;
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

    if (dev->type == VIR_DOMAIN_DEVICE_NET) {
        const virDomainNetDef *net = dev->data.net;

        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"));
            goto cleanup;
        }
2882 2883 2884 2885 2886 2887 2888

        if (STREQ_NULLABLE(net->model, "virtio") &&
            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"));
            goto cleanup;
        }
2889 2890 2891 2892 2893 2894 2895 2896 2897
    }

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


2898
static const char *
2899 2900
qemuDomainDefaultNetModel(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
2901
{
2902
    if (ARCH_IS_S390(def->os.arch))
2903 2904
        return "virtio";

2905 2906
    if (def->os.arch == VIR_ARCH_ARMV7L ||
        def->os.arch == VIR_ARCH_AARCH64) {
2907 2908 2909
        if (STREQ(def->os.machine, "versatilepb"))
            return "smc91c111";

J
Ján Tomko 已提交
2910
        if (qemuDomainMachineIsVirt(def))
2911 2912
            return "virtio";

2913 2914 2915 2916 2917
        /* Incomplete. vexpress (and a few others) use this, but not all
         * arm boards */
        return "lan9118";
    }

2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929
    /* 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 */
2930 2931
    return "rtl8139";
}
2932

2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945

/*
 * Clear auto generated unix socket path, i.e., the one which starts with our
 * channel directory.
 */
static void
qemuDomainChrDefDropDefaultPath(virDomainChrDefPtr chr,
                                virQEMUDriverPtr driver)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    if (chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_CHANNEL &&
        chr->targetType == VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO &&
2946 2947 2948 2949
        chr->source->type == VIR_DOMAIN_CHR_TYPE_UNIX &&
        chr->source->data.nix.path &&
        STRPREFIX(chr->source->data.nix.path, cfg->channelTargetDir)) {
        VIR_FREE(chr->source->data.nix.path);
2950 2951 2952 2953 2954 2955
    }

    virObjectUnref(cfg);
}


2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005
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;
}


3006 3007
static int
qemuDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
3008
                             const virDomainDef *def,
3009
                             virCapsPtr caps,
3010
                             unsigned int parseFlags,
3011
                             void *opaque,
3012
                             void *parseOpaque)
3013
{
3014
    virQEMUDriverPtr driver = opaque;
3015
    virQEMUCapsPtr qemuCaps = parseOpaque;
3016
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3017
    int ret = -1;
3018

3019 3020 3021 3022 3023 3024
    if (qemuCaps) {
        virObjectRef(qemuCaps);
    } else {
        qemuCaps = virQEMUCapsCacheLookup(caps, driver->qemuCapsCache,
                                          def->emulator);
    }
3025

3026
    if (dev->type == VIR_DOMAIN_DEVICE_NET &&
3027 3028
        dev->data.net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
        !dev->data.net->model) {
3029
        if (VIR_STRDUP(dev->data.net->model,
3030
                       qemuDomainDefaultNetModel(def, qemuCaps)) < 0)
3031
            goto cleanup;
3032
    }
3033

3034 3035 3036 3037
    /* set default disk types and drivers */
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
        virDomainDiskDefPtr disk = dev->data.disk;

3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065
        /* assign default storage format and driver according to config */
        if (cfg->allowDiskFormatProbing) {
            /* 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_AUTO);

            /* default disk format for mirrored drive */
            if (disk->mirror &&
                disk->mirror->format == VIR_STORAGE_FILE_NONE)
                disk->mirror->format = VIR_STORAGE_FILE_AUTO;
        } else {
            /* default driver if probing is forbidden */
            if (!virDomainDiskGetDriver(disk) &&
                virDomainDiskSetDriver(disk, "qemu") < 0)
                goto cleanup;

            /* 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);

            /* default disk format for mirrored drive */
            if (disk->mirror &&
                disk->mirror->format == VIR_STORAGE_FILE_NONE)
                disk->mirror->format = VIR_STORAGE_FILE_RAW;
3066 3067 3068
        }
    }

3069 3070 3071 3072
    /* set the default console type for S390 arches */
    if (dev->type == VIR_DOMAIN_DEVICE_CHR &&
        dev->data.chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_CONSOLE &&
        dev->data.chr->targetType == VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_NONE &&
3073
        ARCH_IS_S390(def->os.arch))
3074 3075
        dev->data.chr->targetType = VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_VIRTIO;

3076 3077 3078 3079 3080
    /* set the default USB model to none for s390 unless an address is found */
    if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER &&
        dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_USB &&
        dev->data.controller->model == -1 &&
        dev->data.controller->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
3081
        ARCH_IS_S390(def->os.arch))
3082 3083
        dev->data.controller->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NONE;

3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096
    /* forbid usb model 'qusb1' and 'qusb2' in this kind of hyperviosr */
    if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER &&
        dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_USB &&
        (dev->data.controller->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_QUSB1 ||
         dev->data.controller->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));
        goto cleanup;
    }


3097 3098 3099 3100 3101 3102 3103
    /* set the default SCSI controller model for S390 arches */
    if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER &&
        dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
        dev->data.controller->model == -1 &&
        ARCH_IS_S390(def->os.arch))
        dev->data.controller->model = VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI;

3104 3105
    /* clear auto generated unix socket path for inactive definitions */
    if ((parseFlags & VIR_DOMAIN_DEF_PARSE_INACTIVE) &&
3106 3107
        dev->type == VIR_DOMAIN_DEVICE_CHR)
        qemuDomainChrDefDropDefaultPath(dev->data.chr, driver);
3108

3109 3110 3111 3112 3113 3114 3115 3116 3117 3118
    /* forbid capabilities mode hostdev in this kind of hypervisor */
    if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV &&
        dev->data.hostdev->mode == VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("hostdev mode 'capabilities' is not "
                         "supported in %s"),
                       virDomainVirtTypeToString(def->virtType));
        goto cleanup;
    }

3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133
    if (dev->type == VIR_DOMAIN_DEVICE_VIDEO &&
        dev->data.video->type == VIR_DOMAIN_VIDEO_TYPE_QXL) {
        if (dev->data.video->vgamem) {
            if (dev->data.video->vgamem < 1024) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("value for 'vgamem' must be at least 1 MiB "
                                 "(1024 KiB)"));
                goto cleanup;
            }
            if (dev->data.video->vgamem != VIR_ROUND_UP_POWER_OF_TWO(dev->data.video->vgamem)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("value for 'vgamem' must be power of two"));
                goto cleanup;
            }
        } else {
3134
            dev->data.video->vgamem = QEMU_QXL_VGAMEM_DEFAULT;
3135 3136 3137
        }
    }

3138 3139
    if (dev->type == VIR_DOMAIN_DEVICE_PANIC &&
        dev->data.panic->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT) {
3140
        if (qemuDomainMachineIsPSeries(def))
3141
            dev->data.panic->model = VIR_DOMAIN_PANIC_MODEL_PSERIES;
3142 3143
        else if (ARCH_IS_S390(def->os.arch))
            dev->data.panic->model = VIR_DOMAIN_PANIC_MODEL_S390;
3144 3145 3146 3147
        else
            dev->data.panic->model = VIR_DOMAIN_PANIC_MODEL_ISA;
    }

3148 3149 3150 3151

    if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        virDomainControllerDefPtr cont = dev->data.controller;

3152 3153 3154
        if (cont->type == VIR_DOMAIN_CONTROLLER_TYPE_PCI) {
            if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS &&
                !qemuDomainMachineIsI440FX(def)) {
3155 3156 3157 3158 3159
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("pci-expander-bus controllers are only supported "
                                 "on 440fx-based machinetypes"));
                goto cleanup;
            }
3160 3161 3162 3163 3164 3165 3166
            if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS &&
                !qemuDomainMachineIsQ35(def)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("pcie-expander-bus controllers are only supported "
                                 "on q35-based machinetypes"));
                goto cleanup;
            }
3167 3168 3169 3170 3171 3172

            /* if a PCI expander bus 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
             * from 0 .. size-1.
             */
3173 3174 3175
            if ((cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS ||
                 cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS) &&
                (int) virDomainNumaGetNodeCount(def->numa)
3176 3177
                <= cont->opts.pciopts.numaNode) {
                virReportError(VIR_ERR_XML_ERROR,
3178
                               _("%s with index %d is "
3179 3180 3181
                                 "configured for a NUMA node (%d) "
                                 "not present in the domain's "
                                 "<cpu><numa> array (%zu)"),
3182
                               virDomainControllerModelPCITypeToString(cont->model),
3183 3184 3185 3186
                               cont->idx, cont->opts.pciopts.numaNode,
                               virDomainNumaGetNodeCount(def->numa));
                goto cleanup;
            }
3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206
        } else if (cont->type == VIR_DOMAIN_CONTROLLER_TYPE_USB &&
                   cont->model == -1) {
            /* Pick a suitable default model for the USB controller if none
             * has been selected by the user.
             *
             * 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() */
            if (ARCH_IS_PPC64(def->os.arch)) {
                /* Default USB controller for ppc64 is pci-ohci */
                if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_OHCI))
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_PCI_OHCI;
            } else {
                /* Default USB controller for anything else is piix3-uhci */
                if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PIIX3_USB_UHCI))
                    cont->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_PIIX3_UHCI;
            }
3207 3208 3209
        }
    }

3210 3211 3212 3213
    if (dev->type == VIR_DOMAIN_DEVICE_SHMEM &&
        qemuDomainShmemDefPostParse(dev->data.shmem) < 0)
        goto cleanup;

3214 3215
    ret = 0;

3216
 cleanup:
3217
    virObjectUnref(qemuCaps);
3218 3219
    virObjectUnref(cfg);
    return ret;
3220 3221 3222
}


3223 3224
static int
qemuDomainDefAssignAddresses(virDomainDef *def,
3225
                             virCapsPtr caps,
3226
                             unsigned int parseFlags ATTRIBUTE_UNUSED,
3227
                             void *opaque,
3228
                             void *parseOpaque)
3229 3230
{
    virQEMUDriverPtr driver = opaque;
3231
    virQEMUCapsPtr qemuCaps = parseOpaque;
3232
    int ret = -1;
3233
    bool newDomain = parseFlags & VIR_DOMAIN_DEF_PARSE_ABI_UPDATE;
3234

3235 3236 3237 3238 3239 3240 3241 3242
    if (qemuCaps) {
        virObjectRef(qemuCaps);
    } else {
        if (!(qemuCaps = virQEMUCapsCacheLookup(caps,
                                                driver->qemuCapsCache,
                                                def->emulator)))
            goto cleanup;
    }
3243

3244
    if (qemuDomainAssignAddresses(def, qemuCaps, driver, NULL, newDomain) < 0)
3245 3246 3247 3248 3249 3250 3251 3252 3253
        goto cleanup;

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


3254 3255
virDomainDefParserConfig virQEMUDriverDomainDefParserConfig = {
    .devicesPostParseCallback = qemuDomainDeviceDefPostParse,
3256
    .domainPostParseCallback = qemuDomainDefPostParse,
3257
    .assignAddressesCallback = qemuDomainDefAssignAddresses,
3258
    .domainValidateCallback = qemuDomainDefValidate,
3259 3260
    .deviceValidateCallback = qemuDomainDeviceDefValidate,

3261
    .features = VIR_DOMAIN_DEF_FEATURE_MEMORY_HOTPLUG |
3262 3263
                VIR_DOMAIN_DEF_FEATURE_OFFLINE_VCPUPIN |
                VIR_DOMAIN_DEF_FEATURE_INDIVIDUAL_VCPUS,
3264 3265 3266
};


3267
static void
3268
qemuDomainObjSaveJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
3269
{
3270 3271 3272
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    if (virDomainObjIsActive(obj)) {
3273
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj, driver->caps) < 0)
3274
            VIR_WARN("Failed to save status on vm %s", obj->def->name);
3275
    }
3276

3277
    virObjectUnref(cfg);
3278 3279
}

J
Jiri Denemark 已提交
3280
void
3281
qemuDomainObjSetJobPhase(virQEMUDriverPtr driver,
J
Jiri Denemark 已提交
3282 3283 3284 3285
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
3286
    unsigned long long me = virThreadSelfID();
J
Jiri Denemark 已提交
3287 3288 3289 3290

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

3291 3292 3293 3294 3295
    VIR_DEBUG("Setting '%s' phase to '%s'",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob, phase));

    if (priv->job.asyncOwner && me != priv->job.asyncOwner) {
3296
        VIR_WARN("'%s' async job is owned by thread %llu",
3297 3298 3299 3300
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }

J
Jiri Denemark 已提交
3301
    priv->job.phase = phase;
3302
    priv->job.asyncOwner = me;
J
Jiri Denemark 已提交
3303 3304 3305
    qemuDomainObjSaveJob(driver, obj);
}

3306
void
3307 3308
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
3309 3310 3311
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

3312 3313 3314 3315 3316 3317 3318
    if (!priv->job.asyncJob)
        return;

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

void
3319
qemuDomainObjDiscardAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
3320 3321 3322 3323 3324 3325
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
3326
    qemuDomainObjSaveJob(driver, obj);
3327 3328
}

3329 3330 3331 3332 3333 3334 3335 3336 3337
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()) {
3338
        VIR_WARN("'%s' async job is owned by thread %llu",
3339 3340 3341 3342 3343 3344
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }
    priv->job.asyncOwner = 0;
}

3345
static bool
3346
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
3347 3348
{
    return !priv->job.asyncJob || (priv->job.mask & JOB_MASK(job)) != 0;
3349 3350
}

3351
bool
3352
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
3353 3354 3355 3356
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

3357 3358 3359
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

3360
/*
3361
 * obj must be locked before calling
3362
 */
3363
static int ATTRIBUTE_NONNULL(1)
3364
qemuDomainObjBeginJobInternal(virQEMUDriverPtr driver,
3365
                              virDomainObjPtr obj,
3366 3367
                              qemuDomainJob job,
                              qemuDomainAsyncJob asyncJob)
3368 3369
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
3370
    unsigned long long now;
3371
    unsigned long long then;
3372
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
J
Jiri Denemark 已提交
3373
    bool async = job == QEMU_JOB_ASYNC;
3374
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3375
    const char *blocker = NULL;
J
Jiri Denemark 已提交
3376
    int ret = -1;
J
Jiri Denemark 已提交
3377 3378
    unsigned long long duration = 0;
    unsigned long long asyncDuration = 0;
J
Jiri Denemark 已提交
3379
    const char *jobStr;
3380

J
Jiri Denemark 已提交
3381 3382 3383 3384 3385 3386 3387 3388 3389
    if (async)
        jobStr = qemuDomainAsyncJobTypeToString(asyncJob);
    else
        jobStr = qemuDomainJobTypeToString(job);

    VIR_DEBUG("Starting %s: %s (vm=%p name=%s, current job=%s async=%s)",
              async ? "async job" : "job", jobStr, obj, obj->def->name,
              qemuDomainJobTypeToString(priv->job.active),
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
3390

3391 3392
    if (virTimeMillisNow(&now) < 0) {
        virObjectUnref(cfg);
3393
        return -1;
3394 3395
    }

3396
    priv->jobs_queued++;
J
Jiri Denemark 已提交
3397
    then = now + QEMU_JOB_WAIT_TIME;
3398

3399
 retry:
3400 3401
    if (cfg->maxQueuedJobs &&
        priv->jobs_queued > cfg->maxQueuedJobs) {
3402 3403 3404
        goto error;
    }

3405
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
3406
        VIR_DEBUG("Waiting for async job (vm=%p name=%s)", obj, obj->def->name);
3407
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->parent.lock, then) < 0)
3408 3409 3410
            goto error;
    }

3411
    while (priv->job.active) {
3412
        VIR_DEBUG("Waiting for job (vm=%p name=%s)", obj, obj->def->name);
3413
        if (virCondWaitUntil(&priv->job.cond, &obj->parent.lock, then) < 0)
3414
            goto error;
3415
    }
3416 3417 3418

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
3419
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
3420 3421
        goto retry;

3422
    qemuDomainObjResetJob(priv);
3423

J
Jiri Denemark 已提交
3424 3425
    ignore_value(virTimeMillisNow(&now));

3426
    if (job != QEMU_JOB_ASYNC) {
3427
        VIR_DEBUG("Started job: %s (async=%s vm=%p name=%s)",
3428
                   qemuDomainJobTypeToString(job),
3429 3430
                  qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                  obj, obj->def->name);
3431
        priv->job.active = job;
3432
        priv->job.owner = virThreadSelfID();
3433
        priv->job.ownerAPI = virThreadJobGet();
J
Jiri Denemark 已提交
3434
        priv->job.started = now;
3435
    } else {
3436 3437 3438
        VIR_DEBUG("Started async job: %s (vm=%p name=%s)",
                  qemuDomainAsyncJobTypeToString(asyncJob),
                  obj, obj->def->name);
3439
        qemuDomainObjResetAsyncJob(priv);
J
Jiri Denemark 已提交
3440 3441
        if (VIR_ALLOC(priv->job.current) < 0)
            goto cleanup;
3442
        priv->job.asyncJob = asyncJob;
3443
        priv->job.asyncOwner = virThreadSelfID();
3444
        priv->job.asyncOwnerAPI = virThreadJobGet();
J
Jiri Denemark 已提交
3445
        priv->job.asyncStarted = now;
J
Jiri Denemark 已提交
3446
        priv->job.current->started = now;
3447
    }
3448

3449 3450
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
3451

3452
    virObjectUnref(cfg);
3453
    return 0;
3454

3455
 error:
J
Jiri Denemark 已提交
3456 3457 3458 3459 3460 3461
    ignore_value(virTimeMillisNow(&now));
    if (priv->job.active && priv->job.started)
        duration = now - priv->job.started;
    if (priv->job.asyncJob && priv->job.asyncStarted)
        asyncDuration = now - priv->job.asyncStarted;

3462
    VIR_WARN("Cannot start job (%s, %s) for domain %s; "
J
Jiri Denemark 已提交
3463 3464
             "current job is (%s, %s) owned by (%llu %s, %llu %s) "
             "for (%llus, %llus)",
3465 3466 3467 3468 3469
             qemuDomainJobTypeToString(job),
             qemuDomainAsyncJobTypeToString(asyncJob),
             obj->def->name,
             qemuDomainJobTypeToString(priv->job.active),
             qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
3470
             priv->job.owner, NULLSTR(priv->job.ownerAPI),
J
Jiri Denemark 已提交
3471 3472
             priv->job.asyncOwner, NULLSTR(priv->job.asyncOwnerAPI),
             duration / 1000, asyncDuration / 1000);
3473 3474 3475 3476 3477

    if (nested || qemuDomainNestedJobAllowed(priv, job))
        blocker = priv->job.ownerAPI;
    else
        blocker = priv->job.asyncOwnerAPI;
3478

3479 3480
    ret = -1;
    if (errno == ETIMEDOUT) {
3481 3482 3483 3484 3485 3486 3487 3488
        if (blocker) {
            virReportError(VIR_ERR_OPERATION_TIMEOUT,
                           _("cannot acquire state change lock (held by %s)"),
                           blocker);
        } else {
            virReportError(VIR_ERR_OPERATION_TIMEOUT, "%s",
                           _("cannot acquire state change lock"));
        }
3489 3490 3491
        ret = -2;
    } else if (cfg->maxQueuedJobs &&
               priv->jobs_queued > cfg->maxQueuedJobs) {
3492 3493 3494 3495 3496 3497 3498 3499 3500 3501
        if (blocker) {
            virReportError(VIR_ERR_OPERATION_FAILED,
                           _("cannot acquire state change lock (held by %s) "
                             "due to max_queued limit"),
                           blocker);
        } else {
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("cannot acquire state change lock "
                             "due to max_queued limit"));
        }
3502 3503
        ret = -2;
    } else {
3504
        virReportSystemError(errno, "%s", _("cannot acquire job mutex"));
3505
    }
J
Jiri Denemark 已提交
3506 3507

 cleanup:
3508
    priv->jobs_queued--;
3509
    virObjectUnref(cfg);
3510
    return ret;
3511 3512 3513
}

/*
3514
 * obj must be locked before calling
3515 3516 3517 3518
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
3519
 * Successful calls must be followed by EndJob eventually
3520
 */
3521
int qemuDomainObjBeginJob(virQEMUDriverPtr driver,
3522
                          virDomainObjPtr obj,
3523
                          qemuDomainJob job)
3524
{
3525 3526 3527 3528 3529
    if (qemuDomainObjBeginJobInternal(driver, obj, job,
                                      QEMU_ASYNC_JOB_NONE) < 0)
        return -1;
    else
        return 0;
3530 3531
}

3532
int qemuDomainObjBeginAsyncJob(virQEMUDriverPtr driver,
3533
                               virDomainObjPtr obj,
3534
                               qemuDomainAsyncJob asyncJob)
3535
{
3536 3537 3538 3539 3540
    if (qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_ASYNC,
                                      asyncJob) < 0)
        return -1;
    else
        return 0;
3541 3542
}

3543
int
3544 3545
qemuDomainObjBeginNestedJob(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
3546
                            qemuDomainAsyncJob asyncJob)
3547 3548 3549 3550 3551 3552 3553 3554 3555 3556
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (asyncJob != priv->job.asyncJob) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unexpected async job %d"), asyncJob);
        return -1;
    }

    if (priv->job.asyncOwner != virThreadSelfID()) {
3557
        VIR_WARN("This thread doesn't seem to be the async job owner: %llu",
3558 3559 3560 3561 3562 3563 3564 3565
                 priv->job.asyncOwner);
    }

    return qemuDomainObjBeginJobInternal(driver, obj,
                                         QEMU_JOB_ASYNC_NESTED,
                                         QEMU_ASYNC_JOB_NONE);
}

3566

3567
/*
3568
 * obj must be locked and have a reference before calling
3569 3570 3571 3572
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 */
3573 3574
void
qemuDomainObjEndJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
3575 3576
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
3577
    qemuDomainJob job = priv->job.active;
3578

3579 3580
    priv->jobs_queued--;

3581
    VIR_DEBUG("Stopping job: %s (async=%s vm=%p name=%s)",
3582
              qemuDomainJobTypeToString(job),
3583 3584
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
3585

3586
    qemuDomainObjResetJob(priv);
3587 3588
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
3589
    virCondSignal(&priv->job.cond);
3590 3591
}

3592
void
3593
qemuDomainObjEndAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
3594 3595
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
3596

3597 3598
    priv->jobs_queued--;

3599 3600 3601
    VIR_DEBUG("Stopping async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
3602

3603
    qemuDomainObjResetAsyncJob(priv);
3604
    qemuDomainObjSaveJob(driver, obj);
3605 3606 3607
    virCondBroadcast(&priv->job.asyncCond);
}

3608 3609 3610 3611 3612
void
qemuDomainObjAbortAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

3613 3614 3615
    VIR_DEBUG("Requesting abort of async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
3616

3617 3618
    priv->job.abortJob = true;
    virDomainObjBroadcast(obj);
3619 3620
}

3621 3622 3623 3624 3625 3626 3627 3628 3629
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU monitor API call
 * Must have already either called qemuDomainObjBeginJob() and checked
 * that the VM is still active; may not be used for nested async jobs.
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
3630
static int
3631
qemuDomainObjEnterMonitorInternal(virQEMUDriverPtr driver,
3632
                                  virDomainObjPtr obj,
3633
                                  qemuDomainAsyncJob asyncJob)
3634 3635 3636
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

3637
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
3638 3639 3640
        int ret;
        if ((ret = qemuDomainObjBeginNestedJob(driver, obj, asyncJob)) < 0)
            return ret;
3641
        if (!virDomainObjIsActive(obj)) {
3642 3643
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
3644
            qemuDomainObjEndJob(driver, obj);
3645 3646
            return -1;
        }
3647 3648 3649
    } 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");
3650 3651
    }

3652 3653
    VIR_DEBUG("Entering monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
3654
    virObjectLock(priv->mon);
3655
    virObjectRef(priv->mon);
3656
    ignore_value(virTimeMillisNow(&priv->monStart));
3657
    virObjectUnlock(obj);
3658 3659

    return 0;
3660 3661
}

3662
static void ATTRIBUTE_NONNULL(1)
3663
qemuDomainObjExitMonitorInternal(virQEMUDriverPtr driver,
3664
                                 virDomainObjPtr obj)
3665 3666
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
3667
    bool hasRefs;
3668

3669
    hasRefs = virObjectUnref(priv->mon);
3670

3671
    if (hasRefs)
3672
        virObjectUnlock(priv->mon);
3673

3674
    virObjectLock(obj);
3675 3676
    VIR_DEBUG("Exited monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
3677

3678
    priv->monStart = 0;
3679
    if (!hasRefs)
3680
        priv->mon = NULL;
3681

J
Jiri Denemark 已提交
3682 3683
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjEndJob(driver, obj);
3684 3685
}

3686
void qemuDomainObjEnterMonitor(virQEMUDriverPtr driver,
3687
                               virDomainObjPtr obj)
3688
{
3689
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, obj,
3690
                                                   QEMU_ASYNC_JOB_NONE));
3691 3692
}

3693
/* obj must NOT be locked before calling
3694 3695
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
3696 3697 3698 3699 3700 3701
 *
 * 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.
3702
 */
3703 3704
int qemuDomainObjExitMonitor(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
3705
{
3706
    qemuDomainObjExitMonitorInternal(driver, obj);
3707
    if (!virDomainObjIsActive(obj)) {
3708 3709 3710
        if (!virGetLastError())
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
3711 3712 3713
        return -1;
    }
    return 0;
3714
}
3715 3716

/*
3717
 * obj must be locked before calling
3718 3719
 *
 * To be called immediately before any QEMU monitor API call.
3720
 * Must have already either called qemuDomainObjBeginJob()
3721 3722 3723 3724 3725
 * 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
3726 3727 3728
 * 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).
3729 3730
 */
int
3731
qemuDomainObjEnterMonitorAsync(virQEMUDriverPtr driver,
3732
                               virDomainObjPtr obj,
3733
                               qemuDomainAsyncJob asyncJob)
3734
{
3735
    return qemuDomainObjEnterMonitorInternal(driver, obj, asyncJob);
3736 3737
}

D
Daniel P. Berrange 已提交
3738

3739 3740 3741 3742 3743 3744 3745 3746 3747
/*
 * obj must be locked before calling
 *
 * To be called immediately before any QEMU agent API call.
 * Must have already called qemuDomainObjBeginJob() and checked
 * that the VM is still active.
 *
 * To be followed with qemuDomainObjExitAgent() once complete
 */
3748
qemuAgentPtr
3749
qemuDomainObjEnterAgent(virDomainObjPtr obj)
D
Daniel P. Berrange 已提交
3750 3751
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
3752
    qemuAgentPtr agent = priv->agent;
D
Daniel P. Berrange 已提交
3753

3754 3755
    VIR_DEBUG("Entering agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
3756 3757 3758

    virObjectLock(agent);
    virObjectRef(agent);
3759
    virObjectUnlock(obj);
3760 3761

    return agent;
D
Daniel P. Berrange 已提交
3762 3763
}

3764 3765 3766 3767 3768 3769

/* obj must NOT be locked before calling
 *
 * Should be paired with an earlier qemuDomainObjEnterAgent() call
 */
void
3770
qemuDomainObjExitAgent(virDomainObjPtr obj, qemuAgentPtr agent)
D
Daniel P. Berrange 已提交
3771
{
3772 3773
    virObjectUnlock(agent);
    virObjectUnref(agent);
3774
    virObjectLock(obj);
D
Daniel P. Berrange 已提交
3775

3776 3777
    VIR_DEBUG("Exited agent (agent=%p vm=%p name=%s)",
              agent, obj, obj->def->name);
D
Daniel P. Berrange 已提交
3778 3779
}

3780
void qemuDomainObjEnterRemote(virDomainObjPtr obj)
3781
{
3782 3783
    VIR_DEBUG("Entering remote (vm=%p name=%s)",
              obj, obj->def->name);
3784
    virObjectUnlock(obj);
3785 3786
}

3787
void qemuDomainObjExitRemote(virDomainObjPtr obj)
3788
{
3789
    virObjectLock(obj);
3790 3791
    VIR_DEBUG("Exited remote (vm=%p name=%s)",
              obj, obj->def->name);
3792
}
3793 3794


3795 3796 3797 3798 3799 3800 3801 3802
virDomainDefPtr
qemuDomainDefCopy(virQEMUDriverPtr driver,
                  virDomainDefPtr src,
                  unsigned int flags)
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virDomainDefPtr ret = NULL;
    virCapsPtr caps = NULL;
3803
    char *xml = NULL;
3804 3805 3806 3807 3808 3809 3810 3811 3812

    if (qemuDomainDefFormatBuf(driver, src, flags, &buf) < 0)
        goto cleanup;

    xml = virBufferContentAndReset(&buf);

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

3813
    if (!(ret = virDomainDefParseString(xml, caps, driver->xmlopt, NULL,
3814
                                        VIR_DOMAIN_DEF_PARSE_INACTIVE |
3815
                                        VIR_DOMAIN_DEF_PARSE_SKIP_VALIDATE)))
3816 3817
        goto cleanup;

3818
 cleanup:
3819 3820 3821 3822 3823
    VIR_FREE(xml);
    virObjectUnref(caps);
    return ret;
}

3824
int
3825
qemuDomainDefFormatBuf(virQEMUDriverPtr driver,
3826 3827 3828
                       virDomainDefPtr def,
                       unsigned int flags,
                       virBuffer *buf)
3829
{
3830
    int ret = -1;
3831
    virDomainDefPtr copy = NULL;
3832 3833 3834 3835
    virCapsPtr caps = NULL;

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;
3836

3837 3838 3839
    if (!(flags & (VIR_DOMAIN_XML_UPDATE_CPU | VIR_DOMAIN_XML_MIGRATABLE)))
        goto format;

3840
    if (!(copy = virDomainDefCopy(def, caps, driver->xmlopt, NULL,
3841 3842 3843 3844 3845
                                  flags & VIR_DOMAIN_XML_MIGRATABLE)))
        goto cleanup;

    def = copy;

3846
    /* Update guest CPU requirements according to host CPU */
3847
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) &&
3848
        def->cpu &&
3849 3850
        (def->cpu->mode != VIR_CPU_MODE_CUSTOM ||
         def->cpu->model)) {
J
Jiri Denemark 已提交
3851
        if (virCPUUpdate(def->os.arch, def->cpu, caps->host.cpu) < 0)
3852 3853 3854
            goto cleanup;
    }

3855
    if ((flags & VIR_DOMAIN_XML_MIGRATABLE)) {
3856
        size_t i;
3857
        int toremove = 0;
3858
        virDomainControllerDefPtr usb = NULL, pci = NULL;
3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873

        /* 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];
            }
        }
3874 3875 3876 3877 3878 3879 3880 3881 3882

        /* 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'/>.
3883
         */
3884
        if (ARCH_IS_X86(def->os.arch) && qemuDomainMachineIsI440FX(def) &&
3885 3886 3887
            usb && usb->idx == 0 &&
            (usb->model == -1 ||
             usb->model == VIR_DOMAIN_CONTROLLER_MODEL_USB_PIIX3_UHCI)) {
3888 3889
            VIR_DEBUG("Removing default USB controller from domain '%s'"
                      " for migration compatibility", def->name);
3890
            toremove++;
3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908
        } 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 &&
            pci->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT) {
L
Laine Stump 已提交
3909
            VIR_DEBUG("Removing default pci-root from domain '%s'"
3910
                      " for migration compatibility", def->name);
3911
            toremove++;
3912 3913 3914 3915
        } else {
            pci = NULL;
        }

3916
        if (toremove) {
3917 3918 3919
            virDomainControllerDefPtr *controllers = def->controllers;
            int ncontrollers = def->ncontrollers;

3920
            if (VIR_ALLOC_N(def->controllers, ncontrollers - toremove) < 0) {
3921
                def->controllers = controllers;
3922 3923 3924 3925 3926
                goto cleanup;
            }

            def->ncontrollers = 0;
            for (i = 0; i < ncontrollers; i++) {
3927
                if (controllers[i] != usb && controllers[i] != pci)
3928 3929
                    def->controllers[def->ncontrollers++] = controllers[i];
            }
3930 3931 3932 3933

            VIR_FREE(controllers);
            virDomainControllerDefFree(pci);
            virDomainControllerDefFree(usb);
3934
        }
3935

3936 3937 3938 3939 3940 3941 3942 3943 3944
        /* 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;
            }
        }

3945 3946
        for (i = 0; i < def->nchannels; i++)
            qemuDomainChrDefDropDefaultPath(def->channels[i], driver);
3947 3948
    }

3949 3950
 format:
    ret = virDomainDefFormatInternal(def, caps,
3951 3952
                                     virDomainDefFormatConvertXMLFlags(flags),
                                     buf);
3953

3954
 cleanup:
3955
    virDomainDefFree(copy);
3956
    virObjectUnref(caps);
3957 3958
    return ret;
}
3959

3960
char *qemuDomainDefFormatXML(virQEMUDriverPtr driver,
3961
                             virDomainDefPtr def,
3962
                             unsigned int flags)
3963 3964 3965
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

3966
    if (qemuDomainDefFormatBuf(driver, def, flags, &buf) < 0) {
3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979
        virBufferFreeAndReset(&buf);
        return NULL;
    }

    if (virBufferError(&buf)) {
        virReportOOMError();
        virBufferFreeAndReset(&buf);
        return NULL;
    }

    return virBufferContentAndReset(&buf);
}

3980
char *qemuDomainFormatXML(virQEMUDriverPtr driver,
3981
                          virDomainObjPtr vm,
3982
                          unsigned int flags)
3983 3984 3985
{
    virDomainDefPtr def;

3986
    if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef) {
3987
        def = vm->newDef;
3988
    } else {
3989
        def = vm->def;
3990 3991 3992
        if (virDomainObjIsActive(vm))
            flags &= ~VIR_DOMAIN_XML_UPDATE_CPU;
    }
3993

3994
    return qemuDomainDefFormatXML(driver, def, flags);
3995 3996
}

3997
char *
3998
qemuDomainDefFormatLive(virQEMUDriverPtr driver,
3999
                        virDomainDefPtr def,
4000 4001
                        bool inactive,
                        bool compatible)
4002 4003 4004 4005 4006
{
    unsigned int flags = QEMU_DOMAIN_FORMAT_LIVE_FLAGS;

    if (inactive)
        flags |= VIR_DOMAIN_XML_INACTIVE;
4007 4008
    if (compatible)
        flags |= VIR_DOMAIN_XML_MIGRATABLE;
4009

4010
    return qemuDomainDefFormatXML(driver, def, flags);
4011 4012
}

4013

4014
void qemuDomainObjTaint(virQEMUDriverPtr driver,
4015
                        virDomainObjPtr obj,
4016
                        virDomainTaintFlags taint,
4017
                        qemuDomainLogContextPtr logCtxt)
4018
{
4019
    virErrorPtr orig_err = NULL;
4020
    bool closeLog = false;
4021

4022 4023 4024 4025 4026 4027 4028 4029 4030
    if (virDomainObjTaint(obj, taint)) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        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));
4031 4032 4033 4034 4035

        /* We don't care about errors logging taint info, so
         * preserve original error, and clear any error that
         * is raised */
        orig_err = virSaveLastError();
4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053
        if (logCtxt == NULL) {
            logCtxt = qemuDomainLogContextNew(driver, obj,
                                              QEMU_DOMAIN_LOG_CONTEXT_MODE_ATTACH);
            if (!logCtxt) {
                if (orig_err) {
                    virSetError(orig_err);
                    virFreeError(orig_err);
                }
                VIR_WARN("Unable to open domainlog");
                return;
            }
            closeLog = true;
        }

        if (qemuDomainLogContextWrite(logCtxt,
                                      "Domain id=%d is tainted: %s\n",
                                      obj->def->id,
                                      virDomainTaintTypeToString(taint)) < 0)
4054
            virResetLastError();
4055 4056
        if (closeLog)
            qemuDomainLogContextFree(logCtxt);
4057 4058 4059 4060
        if (orig_err) {
            virSetError(orig_err);
            virFreeError(orig_err);
        }
4061 4062 4063 4064
    }
}


4065
void qemuDomainObjCheckTaint(virQEMUDriverPtr driver,
4066
                             virDomainObjPtr obj,
4067
                             qemuDomainLogContextPtr logCtxt)
4068
{
4069
    size_t i;
4070
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4071
    qemuDomainObjPrivatePtr priv = obj->privateData;
4072

4073
    if (virQEMUDriverIsPrivileged(driver) &&
4074 4075 4076
        (!cfg->clearEmulatorCapabilities ||
         cfg->user == 0 ||
         cfg->group == 0))
4077
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
4078

4079
    if (priv->hookRun)
4080
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOOK, logCtxt);
4081

4082 4083 4084
    if (obj->def->namespaceData) {
        qemuDomainCmdlineDefPtr qemucmd = obj->def->namespaceData;
        if (qemucmd->num_args || qemucmd->num_env)
4085
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logCtxt);
4086 4087
    }

4088
    if (obj->def->cpu && obj->def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
4089
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOST_CPU, logCtxt);
4090

4091
    for (i = 0; i < obj->def->ndisks; i++)
4092
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logCtxt);
4093

4094 4095
    for (i = 0; i < obj->def->nhostdevs; i++)
        qemuDomainObjCheckHostdevTaint(driver, obj, obj->def->hostdevs[i],
4096
                                       logCtxt);
4097

4098
    for (i = 0; i < obj->def->nnets; i++)
4099
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logCtxt);
4100

4101
    if (obj->def->os.dtb)
4102
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_DTB, logCtxt);
4103

4104
    virObjectUnref(cfg);
4105 4106 4107
}


4108
void qemuDomainObjCheckDiskTaint(virQEMUDriverPtr driver,
4109
                                 virDomainObjPtr obj,
4110
                                 virDomainDiskDefPtr disk,
4111
                                 qemuDomainLogContextPtr logCtxt)
4112
{
4113
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4114
    int format = virDomainDiskGetFormat(disk);
4115

4116
    if ((!format || format == VIR_STORAGE_FILE_AUTO) &&
4117
        cfg->allowDiskFormatProbing)
4118
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_DISK_PROBING, logCtxt);
4119

4120 4121
    if (disk->rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
4122
                           logCtxt);
4123

4124 4125 4126 4127
    if (disk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
        virStorageSourceGetActualType(disk->src) == VIR_STORAGE_TYPE_BLOCK &&
        disk->src->path)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CDROM_PASSTHROUGH,
4128
                           logCtxt);
4129

4130
    virObjectUnref(cfg);
4131 4132 4133
}


4134 4135 4136
void qemuDomainObjCheckHostdevTaint(virQEMUDriverPtr driver,
                                    virDomainObjPtr obj,
                                    virDomainHostdevDefPtr hostdev,
4137
                                    qemuDomainLogContextPtr logCtxt)
4138
{
4139
    if (!virHostdevIsSCSIDevice(hostdev))
4140
        return;
4141

4142 4143
    if (hostdev->source.subsys.u.scsi.rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
4144 4145 4146
}


4147
void qemuDomainObjCheckNetTaint(virQEMUDriverPtr driver,
4148
                                virDomainObjPtr obj,
4149
                                virDomainNetDefPtr net,
4150
                                qemuDomainLogContextPtr logCtxt)
4151
{
4152 4153 4154 4155 4156 4157
    /* 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)
4158
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logCtxt);
4159
}
4160 4161


4162 4163 4164 4165 4166 4167 4168 4169 4170 4171
qemuDomainLogContextPtr qemuDomainLogContextNew(virQEMUDriverPtr driver,
                                                virDomainObjPtr vm,
                                                qemuDomainLogContextMode mode)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainLogContextPtr ctxt = NULL;

    if (VIR_ALLOC(ctxt) < 0)
        goto error;

4172
    VIR_DEBUG("Context new %p stdioLogD=%d", ctxt, cfg->stdioLogD);
4173 4174 4175 4176
    ctxt->writefd = -1;
    ctxt->readfd = -1;
    virAtomicIntSet(&ctxt->refs, 1);

4177 4178 4179
    if (virAsprintf(&ctxt->path, "%s/%s.log", cfg->logDir, vm->def->name) < 0)
        goto error;

4180 4181 4182 4183
    if (cfg->stdioLogD) {
        ctxt->manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver));
        if (!ctxt->manager)
            goto error;
4184

4185 4186 4187 4188
        ctxt->writefd = virLogManagerDomainOpenLogFile(ctxt->manager,
                                                       "qemu",
                                                       vm->def->uuid,
                                                       vm->def->name,
4189
                                                       ctxt->path,
4190 4191 4192 4193 4194 4195
                                                       0,
                                                       &ctxt->inode,
                                                       &ctxt->pos);
        if (ctxt->writefd < 0)
            goto error;
    } else {
4196
        if ((ctxt->writefd = open(ctxt->path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
4197
            virReportSystemError(errno, _("failed to create logfile %s"),
4198
                                 ctxt->path);
4199 4200
            goto error;
        }
4201
        if (virSetCloseExec(ctxt->writefd) < 0) {
4202
            virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
4203
                                 ctxt->path);
4204 4205 4206
            goto error;
        }

4207 4208 4209 4210 4211 4212 4213
        /* 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"),
4214
                                 ctxt->path);
4215 4216 4217 4218
            goto error;
        }

        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START) {
4219
            if ((ctxt->readfd = open(ctxt->path, O_RDONLY, S_IRUSR | S_IWUSR)) < 0) {
4220
                virReportSystemError(errno, _("failed to open logfile %s"),
4221
                                     ctxt->path);
4222 4223 4224 4225
                goto error;
            }
            if (virSetCloseExec(ctxt->readfd) < 0) {
                virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
4226
                                     ctxt->path);
4227 4228 4229 4230
                goto error;
            }
        }

4231 4232
        if ((ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END)) < 0) {
            virReportSystemError(errno, _("failed to seek in log file %s"),
4233
                                 ctxt->path);
4234 4235 4236 4237
            goto error;
        }
    }

4238
 cleanup:
4239 4240 4241 4242 4243
    virObjectUnref(cfg);
    return ctxt;

 error:
    qemuDomainLogContextFree(ctxt);
4244 4245
    ctxt = NULL;
    goto cleanup;
4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259
}


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;
4260 4261
    if (!ctxt->manager &&
        lseek(ctxt->writefd, 0, SEEK_END) < 0) {
4262
        virReportSystemError(errno, "%s",
4263
                             _("Unable to seek to end of domain logfile"));
4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283
        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)
{
4284 4285 4286 4287
    VIR_DEBUG("Context read %p manager=%p inode=%llu pos=%llu",
              ctxt, ctxt->manager,
              (unsigned long long)ctxt->inode,
              (unsigned long long)ctxt->pos);
4288
    char *buf;
4289 4290 4291
    size_t buflen;
    if (ctxt->manager) {
        buf = virLogManagerDomainReadLogFile(ctxt->manager,
4292
                                             ctxt->path,
4293 4294 4295 4296 4297 4298 4299 4300 4301
                                             ctxt->inode,
                                             ctxt->pos,
                                             1024 * 128,
                                             0);
        if (!buf)
            return -1;
        buflen = strlen(buf);
    } else {
        ssize_t got;
4302

4303
        buflen = 1024 * 128;
4304

4305 4306
        /* Best effort jump to start of messages */
        ignore_value(lseek(ctxt->readfd, ctxt->pos, SEEK_SET));
4307

4308 4309
        if (VIR_ALLOC_N(buf, buflen) < 0)
            return -1;
4310

4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323
        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;
    }
4324 4325 4326

    *msg = buf;

4327
    return buflen;
4328 4329 4330
}


4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392
/**
 * 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;
}


4393 4394 4395 4396 4397 4398 4399 4400
int qemuDomainLogContextGetWriteFD(qemuDomainLogContextPtr ctxt)
{
    return ctxt->writefd;
}


void qemuDomainLogContextMarkPosition(qemuDomainLogContextPtr ctxt)
{
4401 4402
    if (ctxt->manager)
        virLogManagerDomainGetLogFilePosition(ctxt->manager,
4403
                                              ctxt->path,
4404 4405 4406 4407 4408
                                              0,
                                              &ctxt->inode,
                                              &ctxt->pos);
    else
        ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END);
4409 4410 4411 4412 4413
}


void qemuDomainLogContextRef(qemuDomainLogContextPtr ctxt)
{
4414
    VIR_DEBUG("Context ref %p", ctxt);
4415 4416 4417 4418
    virAtomicIntInc(&ctxt->refs);
}


4419 4420 4421 4422 4423 4424
virLogManagerPtr qemuDomainLogContextGetManager(qemuDomainLogContextPtr ctxt)
{
    return ctxt->manager;
}


4425 4426 4427 4428 4429 4430 4431 4432
void qemuDomainLogContextFree(qemuDomainLogContextPtr ctxt)
{
    bool lastRef;

    if (!ctxt)
        return;

    lastRef = virAtomicIntDecAndTest(&ctxt->refs);
4433
    VIR_DEBUG("Context free %p lastref=%d", ctxt, lastRef);
4434 4435 4436
    if (!lastRef)
        return;

4437
    virLogManagerFree(ctxt->manager);
4438
    VIR_FREE(ctxt->path);
4439 4440 4441 4442 4443 4444
    VIR_FORCE_CLOSE(ctxt->writefd);
    VIR_FORCE_CLOSE(ctxt->readfd);
    VIR_FREE(ctxt);
}


4445 4446
/* Locate an appropriate 'qemu-img' binary.  */
const char *
4447
qemuFindQemuImgBinary(virQEMUDriverPtr driver)
4448
{
4449 4450
    if (!driver->qemuImgBinary)
        virReportError(VIR_ERR_INTERNAL_ERROR,
4451
                       "%s", _("unable to find qemu-img"));
4452 4453 4454 4455 4456 4457 4458

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
                                virDomainSnapshotObjPtr snapshot,
4459
                                virCapsPtr caps,
4460 4461 4462 4463 4464 4465 4466 4467 4468
                                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);
4469
    newxml = virDomainSnapshotDefFormat(
4470
        uuidstr, snapshot->def, caps,
4471 4472
        virDomainDefFormatConvertXMLFlags(QEMU_DOMAIN_FORMAT_LIVE_FLAGS),
        1);
4473
    if (newxml == NULL)
4474 4475
        return -1;

4476
    if (virAsprintf(&snapDir, "%s/%s", snapshotDir, vm->def->name) < 0)
4477 4478 4479 4480 4481 4482 4483
        goto cleanup;
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
        goto cleanup;
    }

4484
    if (virAsprintf(&snapFile, "%s/%s.xml", snapDir, snapshot->def->name) < 0)
4485 4486
        goto cleanup;

J
Ján Tomko 已提交
4487
    ret = virXMLSaveFile(snapFile, NULL, "snapshot-edit", newxml);
4488

4489
 cleanup:
4490 4491 4492 4493 4494 4495 4496 4497
    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.  */
4498
static int
4499
qemuDomainSnapshotForEachQcow2Raw(virQEMUDriverPtr driver,
4500 4501 4502 4503 4504
                                  virDomainDefPtr def,
                                  const char *name,
                                  const char *op,
                                  bool try_all,
                                  int ndisks)
4505 4506
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
4507
    size_t i;
4508 4509 4510 4511 4512 4513 4514 4515 4516
    bool skipped = false;

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

    qemuimgarg[2] = op;
4517
    qemuimgarg[3] = name;
4518

4519
    for (i = 0; i < ndisks; i++) {
4520
        /* FIXME: we also need to handle LVM here */
4521
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
4522 4523 4524
            int format = virDomainDiskGetFormat(def->disks[i]);

            if (format > 0 && format != VIR_STORAGE_FILE_QCOW2) {
4525 4526 4527 4528 4529
                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",
4530
                             def->disks[i]->dst);
4531 4532
                    skipped = true;
                    continue;
4533 4534 4535 4536 4537
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
4538
                }
4539 4540 4541 4542
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("Disk device '%s' does not support"
                                 " snapshotting"),
                               def->disks[i]->dst);
4543 4544 4545
                return -1;
            }

4546
            qemuimgarg[4] = virDomainDiskGetSource(def->disks[i]);
4547 4548 4549 4550

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
4551
                             def->disks[i]->dst);
4552 4553
                    skipped = true;
                    continue;
4554 4555 4556 4557 4558
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
4559 4560 4561 4562 4563 4564 4565 4566 4567
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

4568 4569 4570
/* 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
4571
qemuDomainSnapshotForEachQcow2(virQEMUDriverPtr driver,
4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587
                               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);
}

4588 4589
/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
4590
qemuDomainSnapshotDiscard(virQEMUDriverPtr driver,
4591 4592 4593 4594 4595 4596 4597 4598 4599
                          virDomainObjPtr vm,
                          virDomainSnapshotObjPtr snap,
                          bool update_current,
                          bool metadata_only)
{
    char *snapFile = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;
    virDomainSnapshotObjPtr parentsnap = NULL;
4600
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4601 4602 4603 4604 4605 4606 4607 4608 4609

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
                goto cleanup;
        } else {
            priv = vm->privateData;
4610
            qemuDomainObjEnterMonitor(driver, vm);
4611 4612
            /* we continue on even in the face of error */
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
4613
            ignore_value(qemuDomainObjExitMonitor(driver, vm));
4614 4615 4616
        }
    }

4617
    if (virAsprintf(&snapFile, "%s/%s/%s.xml", cfg->snapshotDir,
4618
                    vm->def->name, snap->def->name) < 0)
4619 4620 4621 4622
        goto cleanup;

    if (snap == vm->current_snapshot) {
        if (update_current && snap->def->parent) {
4623
            parentsnap = virDomainSnapshotFindByName(vm->snapshots,
4624 4625 4626 4627 4628 4629
                                                     snap->def->parent);
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
                         snap->def->parent);
            } else {
                parentsnap->def->current = true;
4630
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap, driver->caps,
4631
                                                    cfg->snapshotDir) < 0) {
4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643
                    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);
4644
    virDomainSnapshotObjListRemove(vm->snapshots, snap);
4645 4646 4647

    ret = 0;

4648
 cleanup:
4649
    VIR_FREE(snapFile);
4650
    virObjectUnref(cfg);
4651 4652 4653 4654
    return ret;
}

/* Hash iterator callback to discard multiple snapshots.  */
4655 4656 4657
int qemuDomainSnapshotDiscardAll(void *payload,
                                 const void *name ATTRIBUTE_UNUSED,
                                 void *data)
4658 4659
{
    virDomainSnapshotObjPtr snap = payload;
4660
    virQEMUSnapRemovePtr curr = data;
4661 4662 4663 4664 4665 4666 4667 4668
    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;
4669
    return 0;
4670 4671 4672
}

int
4673
qemuDomainSnapshotDiscardAllMetadata(virQEMUDriverPtr driver,
4674 4675
                                     virDomainObjPtr vm)
{
4676
    virQEMUSnapRemove rem;
4677 4678 4679 4680 4681

    rem.driver = driver;
    rem.vm = vm;
    rem.metadata_only = true;
    rem.err = 0;
4682 4683
    virDomainSnapshotForEach(vm->snapshots, qemuDomainSnapshotDiscardAll,
                             &rem);
4684 4685 4686 4687 4688

    return rem.err;
}

/*
4689
 * The caller must hold a lock the vm.
4690 4691
 */
void
4692
qemuDomainRemoveInactive(virQEMUDriverPtr driver,
4693 4694
                         virDomainObjPtr vm)
{
4695
    bool haveJob = true;
4696
    char *snapDir;
4697 4698 4699 4700 4701 4702 4703 4704
    virQEMUDriverConfigPtr cfg;

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

    cfg = virQEMUDriverGetConfig(driver);
4705

4706 4707 4708
    if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0)
        haveJob = false;

4709 4710 4711 4712 4713
    /* 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);
    }
4714
    else if (virAsprintf(&snapDir, "%s/%s", cfg->snapshotDir,
4715 4716
                         vm->def->name) < 0) {
        VIR_WARN("unable to remove snapshot directory %s/%s",
4717
                 cfg->snapshotDir, vm->def->name);
4718 4719 4720 4721 4722
    } else {
        if (rmdir(snapDir) < 0 && errno != ENOENT)
            VIR_WARN("unable to remove snapshot directory %s", snapDir);
        VIR_FREE(snapDir);
    }
4723 4724 4725

    virObjectRef(vm);

4726
    virDomainObjListRemove(driver->domains, vm);
4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739
    /*
     * virDomainObjListRemove() leaves the domain unlocked so it can
     * be unref'd for other drivers that depend on that, but we still
     * need to reset a job and we have a reference from the API that
     * called this function.  So we need to lock it back.  This is
     * just a workaround for the qemu driver.
     *
     * XXX: Ideally, the global handling of domain objects and object
     *      lists would be refactored so we don't need hacks like
     *      this, but since that requires refactor of all drivers,
     *      it's a work for another day.
     */
    virObjectLock(vm);
4740
    virObjectUnref(cfg);
4741 4742

    if (haveJob)
4743
        qemuDomainObjEndJob(driver, vm);
4744 4745

    virObjectUnref(vm);
4746
}
4747 4748

void
4749
qemuDomainSetFakeReboot(virQEMUDriverPtr driver,
4750 4751 4752 4753
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
4754
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4755 4756

    if (priv->fakeReboot == value)
4757
        goto cleanup;
4758 4759 4760

    priv->fakeReboot = value;

4761
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
4762
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
4763

4764
 cleanup:
4765
    virObjectUnref(cfg);
4766
}
M
Michal Privoznik 已提交
4767

4768
static void
4769 4770
qemuDomainCheckRemoveOptionalDisk(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
4771
                                  size_t diskIndex)
4772 4773
{
    char uuid[VIR_UUID_STRING_BUFLEN];
4774
    virObjectEventPtr event = NULL;
4775
    virDomainDiskDefPtr disk = vm->def->disks[diskIndex];
4776
    const char *src = virDomainDiskGetSource(disk);
4777 4778 4779 4780 4781

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

    VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
              "due to inaccessible source '%s'",
4782
              disk->dst, vm->def->name, uuid, src);
4783 4784 4785 4786

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

4787
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
4788 4789
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
4790
        ignore_value(virDomainDiskSetSource(disk, NULL));
4791 4792
        /* keeping the old startup policy would be invalid for new images */
        disk->startupPolicy = VIR_DOMAIN_STARTUP_POLICY_DEFAULT;
4793
    } else {
4794
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
4795 4796
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_DROP_MISSING_ON_START);
4797 4798
        virDomainDiskRemove(vm->def, diskIndex);
        virDomainDiskDefFree(disk);
4799 4800
    }

4801
    qemuDomainEventQueue(driver, event);
4802 4803
}

4804 4805 4806
static int
qemuDomainCheckDiskStartupPolicy(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
4807
                                 size_t diskIndex,
4808 4809
                                 bool cold_boot)
{
4810
    int startupPolicy = vm->def->disks[diskIndex]->startupPolicy;
4811
    int device = vm->def->disks[diskIndex]->device;
4812

4813
    switch ((virDomainStartupPolicy) startupPolicy) {
4814
        case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
4815 4816 4817 4818 4819 4820
            /* 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)
4821
                return -1;
4822 4823
            break;

4824
        case VIR_DOMAIN_STARTUP_POLICY_DEFAULT:
4825
        case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
4826
            return -1;
4827 4828

        case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
4829
            if (cold_boot)
4830
                return -1;
4831 4832 4833 4834 4835 4836 4837
            break;

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

4838
    qemuDomainCheckRemoveOptionalDisk(driver, vm, diskIndex);
4839
    virResetLastError();
4840 4841 4842
    return 0;
}

4843

M
Michal Privoznik 已提交
4844
int
4845 4846
qemuDomainCheckDiskPresence(virConnectPtr conn,
                            virQEMUDriverPtr driver,
M
Michal Privoznik 已提交
4847
                            virDomainObjPtr vm,
4848
                            unsigned int flags)
M
Michal Privoznik 已提交
4849
{
4850
    size_t i;
4851 4852
    bool pretend = flags & VIR_QEMU_PROCESS_START_PRETEND;
    bool cold_boot = flags & VIR_QEMU_PROCESS_START_COLD;
M
Michal Privoznik 已提交
4853

4854
    VIR_DEBUG("Checking for disk presence");
4855
    for (i = vm->def->ndisks; i > 0; i--) {
4856 4857
        size_t idx = i - 1;
        virDomainDiskDefPtr disk = vm->def->disks[idx];
4858
        virStorageFileFormat format = virDomainDiskGetFormat(disk);
M
Michal Privoznik 已提交
4859

4860 4861 4862 4863 4864 4865 4866 4867 4868 4869
        if (virStorageTranslateDiskSourcePool(conn, vm->def->disks[idx]) < 0) {
            if (pretend ||
                qemuDomainCheckDiskStartupPolicy(driver, vm, idx, cold_boot) < 0)
                return -1;
            continue;
        }

        if (pretend)
            continue;

4870
        if (virStorageSourceIsEmpty(disk->src))
4871 4872 4873 4874 4875
            continue;

        /* There is no need to check the backing chain for disks
         * without backing support, the fact that the file exists is
         * more than enough */
4876
        if (virStorageSourceIsLocalStorage(disk->src) &&
4877
            format > VIR_STORAGE_FILE_NONE &&
4878
            format < VIR_STORAGE_FILE_BACKING &&
4879
            virFileExists(virDomainDiskGetSource(disk)))
M
Michal Privoznik 已提交
4880 4881
            continue;

4882
        if (qemuDomainDetermineDiskChain(driver, vm, disk, true, true) >= 0)
M
Michal Privoznik 已提交
4883
            continue;
4884

4885
        if (qemuDomainCheckDiskStartupPolicy(driver, vm, idx, cold_boot) >= 0)
4886
            continue;
M
Michal Privoznik 已提交
4887

4888
        return -1;
M
Michal Privoznik 已提交
4889 4890
    }

4891
    return 0;
M
Michal Privoznik 已提交
4892
}
4893 4894 4895 4896 4897 4898 4899 4900 4901 4902

/*
 * The vm must be locked when any of the following cleanup functions is
 * called.
 */
int
qemuDomainCleanupAdd(virDomainObjPtr vm,
                     qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
4903
    size_t i;
4904 4905 4906 4907 4908 4909 4910 4911 4912 4913

    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,
4914
                     priv->ncleanupCallbacks, 1) < 0)
4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925
        return -1;

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

void
qemuDomainCleanupRemove(virDomainObjPtr vm,
                        qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
4926
    size_t i;
4927 4928 4929 4930

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

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
4931 4932 4933
        if (priv->cleanupCallbacks[i] == cb)
            VIR_DELETE_ELEMENT_INPLACE(priv->cleanupCallbacks,
                                       i, priv->ncleanupCallbacks);
4934 4935 4936 4937 4938 4939 4940 4941
    }

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

void
4942
qemuDomainCleanupRun(virQEMUDriverPtr driver,
4943 4944 4945
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
4946
    size_t i;
4947 4948 4949 4950

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

    /* run cleanup callbacks in reverse order */
4951 4952
    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[priv->ncleanupCallbacks - (i + 1)])
4953 4954 4955 4956 4957 4958 4959
            priv->cleanupCallbacks[i](driver, vm);
    }

    VIR_FREE(priv->cleanupCallbacks);
    priv->ncleanupCallbacks = 0;
    priv->ncleanupCallbacks_max = 0;
}
4960

4961 4962 4963
static void
qemuDomainGetImageIds(virQEMUDriverConfigPtr cfg,
                      virDomainObjPtr vm,
4964
                      virStorageSourcePtr src,
4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982
                      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;
    }

4983 4984
    if (vm && (vmlabel = virDomainDefGetSecurityLabelDef(vm->def, "dac")) &&
        vmlabel->label)
4985 4986
        virParseOwnershipIds(vmlabel->label, uid, gid);

4987
    if ((disklabel = virStorageSourceGetSecurityLabelDef(src, "dac")) &&
4988
        disklabel->label)
4989 4990 4991 4992
        virParseOwnershipIds(disklabel->label, uid, gid);
}


4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015
int
qemuDomainStorageFileInit(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
                          virStorageSourcePtr src)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    uid_t uid;
    gid_t gid;
    int ret = -1;

    qemuDomainGetImageIds(cfg, vm, src, &uid, &gid);

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

    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
}


5016 5017 5018 5019 5020
char *
qemuDomainStorageAlias(const char *device, int depth)
{
    char *alias;

5021
    device = qemuAliasDiskDriveSkipPrefix(device);
5022 5023 5024 5025 5026 5027 5028 5029 5030

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


5031
int
5032
qemuDomainDetermineDiskChain(virQEMUDriverPtr driver,
5033
                             virDomainObjPtr vm,
5034
                             virDomainDiskDefPtr disk,
5035 5036
                             bool force_probe,
                             bool report_broken)
5037
{
5038 5039
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = 0;
5040 5041
    uid_t uid;
    gid_t gid;
5042

5043
    if (virStorageSourceIsEmpty(disk->src))
5044
        goto cleanup;
5045

5046
    if (disk->src->backingStore) {
5047
        if (force_probe)
5048
            virStorageSourceBackingStoreClear(disk->src);
5049
        else
5050
            goto cleanup;
5051
    }
5052

5053
    qemuDomainGetImageIds(cfg, vm, disk->src, &uid, &gid);
5054

5055
    if (virStorageFileGetMetadata(disk->src,
5056
                                  uid, gid,
5057
                                  cfg->allowDiskFormatProbing,
5058
                                  report_broken) < 0)
5059 5060
        ret = -1;

5061
 cleanup:
5062 5063
    virObjectUnref(cfg);
    return ret;
5064
}
5065

5066

5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129
/**
 * 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)
{
    if (virSecurityManagerRestoreImageLabel(driver->securityManager,
                                            vm->def, elem) < 0)
        VIR_WARN("Unable to restore security label on %s", NULLSTR(elem->path));

    if (qemuTeardownImageCgroup(vm, elem) < 0)
        VIR_WARN("Failed to teardown cgroup for disk path %s",
                 NULLSTR(elem->path));

    if (virDomainLockImageDetach(driver->lockManager, vm, elem) < 0)
        VIR_WARN("Unable to release lock on %s", NULLSTR(elem->path));
}


/**
 * qemuDomainDiskChainElementPrepare:
 *
 * 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
 * labelling are all aware of each new file before it is added to a chain */
int
qemuDomainDiskChainElementPrepare(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
                                  virStorageSourcePtr elem,
                                  bool readonly)
{
    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;

    if (qemuSetupImageCgroup(vm, elem) < 0)
        goto cleanup;

    if (virSecurityManagerSetImageLabel(driver->securityManager, vm->def,
                                        elem) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    elem->readonly = was_readonly;
    virObjectUnref(cfg);
    return ret;
}


5130
bool
5131
qemuDomainDiskSourceDiffers(virDomainDiskDefPtr disk,
5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146
                            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;

5147 5148 5149 5150
    /* 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)
5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163
        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;
}


5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186
/*
 * 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)
{
#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;                                               \
        }                                                               \
    } while (0)

    CHECK_EQ(device, "device", false);
5187 5188 5189 5190
    CHECK_EQ(bus, "bus", false);
    if (STRNEQ(disk->dst, orig_disk->dst)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
5191
                       "target");
5192 5193 5194 5195
        return false;
    }
    CHECK_EQ(tray_status, "tray", true);
    CHECK_EQ(removable, "removable", true);
5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 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 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278

    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;
    }

5279 5280 5281 5282 5283 5284 5285
    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);
5286
    /* "snapshot" is a libvirt internal field and thus can be changed */
5287
    /* startupPolicy is allowed to be updated. Therefore not checked here. */
5288
    CHECK_EQ(transient, "transient", true);
5289
    CHECK_EQ(info.bootIndex, "boot order", true);
5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301
    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;
    }
5302

5303 5304 5305 5306 5307 5308
    /* 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);

5309 5310 5311 5312 5313
#undef CHECK_EQ

    return true;
}

5314 5315 5316
bool
qemuDomainDiskBlockJobIsActive(virDomainDiskDefPtr disk)
{
5317 5318
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

5319 5320 5321 5322 5323 5324 5325 5326
    if (disk->mirror) {
        virReportError(VIR_ERR_BLOCK_COPY_ACTIVE,
                       _("disk '%s' already in active block job"),
                       disk->dst);

        return true;
    }

5327
    if (diskPriv->blockjob) {
5328 5329 5330 5331 5332 5333 5334 5335 5336 5337
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("disk '%s' already in active block job"),
                       disk->dst);
        return true;
    }

    return false;
}


5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352
/**
 * 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++) {
5353 5354 5355 5356
        virDomainDiskDefPtr disk = vm->def->disks[i];
        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

        if (!copy_only && diskPriv->blockjob)
5357 5358
            return true;

5359
        if (disk->mirror && disk->mirrorJob == VIR_DOMAIN_BLOCK_JOB_TYPE_COPY)
5360 5361 5362 5363 5364 5365 5366
            return true;
    }

    return false;
}


5367 5368
int
qemuDomainUpdateDeviceList(virQEMUDriverPtr driver,
5369 5370
                           virDomainObjPtr vm,
                           int asyncJob)
5371 5372 5373
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char **aliases;
5374
    int rc;
5375 5376 5377 5378

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

5379 5380
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
5381 5382 5383 5384
    rc = qemuMonitorGetDeviceAliases(priv->mon, &aliases);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
    if (rc < 0)
5385 5386
        return -1;

5387
    virStringListFree(priv->qemuDevices);
5388 5389 5390
    priv->qemuDevices = aliases;
    return 0;
}
5391

5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440

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);

    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;

    /* 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;
}


5441 5442 5443 5444 5445 5446 5447
bool
qemuDomainDefCheckABIStability(virQEMUDriverPtr driver,
                               virDomainDefPtr src,
                               virDomainDefPtr dst)
{
    virDomainDefPtr migratableDefSrc = NULL;
    virDomainDefPtr migratableDefDst = NULL;
5448 5449 5450 5451
    const unsigned int flags = VIR_DOMAIN_XML_SECURE |
                               VIR_DOMAIN_XML_UPDATE_CPU |
                               VIR_DOMAIN_XML_MIGRATABLE;
    const unsigned int check_flags = VIR_DOMAIN_DEF_ABI_CHECK_SKIP_VOLATILE;
5452 5453 5454 5455 5456 5457
    bool ret = false;

    if (!(migratableDefSrc = qemuDomainDefCopy(driver, src, flags)) ||
        !(migratableDefDst = qemuDomainDefCopy(driver, dst, flags)))
        goto cleanup;

5458 5459 5460 5461 5462 5463 5464 5465 5466
    if (!virDomainDefCheckABIStabilityFlags(migratableDefSrc,
                                            migratableDefDst,
                                            check_flags))
        goto cleanup;

    /* Force update any skipped values from the volatile flag */
    dst->mem.cur_balloon = src->mem.cur_balloon;

    ret = true;
5467

5468
 cleanup:
5469 5470 5471 5472
    virDomainDefFree(migratableDefSrc);
    virDomainDefFree(migratableDefDst);
    return ret;
}
5473 5474

bool
5475
qemuDomainAgentAvailable(virDomainObjPtr vm,
5476 5477
                         bool reportError)
{
5478 5479
    qemuDomainObjPrivatePtr priv = vm->privateData;

5480 5481 5482 5483 5484 5485 5486
    if (virDomainObjGetState(vm, NULL) != VIR_DOMAIN_RUNNING) {
        if (reportError) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
        }
        return false;
    }
5487 5488 5489 5490 5491 5492 5493 5494 5495
    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) {
5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507
        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;
5508 5509 5510 5511
        }
    }
    return true;
}
5512

5513

5514
static unsigned long long
5515
qemuDomainGetMemorySizeAlignment(virDomainDefPtr def)
5516
{
5517 5518 5519 5520 5521
    /* 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;

5522 5523 5524 5525 5526 5527 5528
    /* Align memory size. QEMU requires rounding to next 4KiB block.
     * We'll take the "traditional" path and round it to 1MiB*/

    return 1024;
}


5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543
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;
}


5544 5545 5546
int
qemuDomainAlignMemorySizes(virDomainDefPtr def)
{
5547 5548
    unsigned long long maxmemkb = virMemoryMaxValue(false) >> 10;
    unsigned long long maxmemcapped = virMemoryMaxValue(true) >> 10;
5549
    unsigned long long initialmem = 0;
5550
    unsigned long long hotplugmem = 0;
5551
    unsigned long long mem;
5552
    unsigned long long align = qemuDomainGetMemorySizeAlignment(def);
5553 5554 5555 5556 5557
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    /* align NUMA cell sizes if relevant */
    for (i = 0; i < ncells; i++) {
5558 5559
        mem = VIR_ROUND_UP(virDomainNumaGetNodeMemorySize(def->numa, i), align);
        initialmem += mem;
5560 5561 5562 5563 5564 5565 5566

        if (mem > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("memory size of NUMA node '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
5567
        virDomainNumaSetNodeMemorySize(def->numa, i, mem);
5568 5569
    }

5570 5571 5572 5573 5574
    /* 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);

5575 5576 5577 5578 5579 5580
    if (initialmem > maxmemcapped) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("initial memory size overflowed after alignment"));
        return -1;
    }

5581
    def->mem.max_memory = VIR_ROUND_UP(def->mem.max_memory, align);
5582 5583 5584 5585 5586
    if (def->mem.max_memory > maxmemkb) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("maximum memory size overflowed after alignment"));
        return -1;
    }
5587

5588
    /* Align memory module sizes */
5589 5590
    for (i = 0; i < def->nmems; i++) {
        align = qemuDomainGetMemoryModuleSizeAlignment(def, def->mems[i]);
5591
        def->mems[i]->size = VIR_ROUND_UP(def->mems[i]->size, align);
5592
        hotplugmem += def->mems[i]->size;
5593 5594 5595 5596 5597 5598 5599

        if (def->mems[i]->size > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("size of memory module '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
5600
    }
5601

5602 5603
    virDomainDefSetMemoryTotal(def, initialmem + hotplugmem);

5604 5605
    return 0;
}
5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616


/**
 * 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
5617 5618
qemuDomainMemoryDeviceAlignSize(virDomainDefPtr def,
                                virDomainMemoryDefPtr mem)
5619
{
5620
    mem->size = VIR_ROUND_UP(mem->size, qemuDomainGetMemorySizeAlignment(def));
5621
}
5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634


/**
 * qemuDomainGetMonitor:
 * @vm: domain object
 *
 * Returns the monitor pointer corresponding to the domain object @vm.
 */
qemuMonitorPtr
qemuDomainGetMonitor(virDomainObjPtr vm)
{
    return ((qemuDomainObjPrivatePtr) vm->privateData)->mon;
}
5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650


/**
 * qemuDomainSupportsBlockJobs:
 * @vm: domain object
 * @modern: pointer to bool that returns whether modern block jobs are supported
 *
 * Returns -1 in case when qemu does not support block jobs at all. Otherwise
 * returns 0 and optionally fills @modern to denote that modern (async) block
 * jobs are supported.
 */
int
qemuDomainSupportsBlockJobs(virDomainObjPtr vm,
                            bool *modern)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
5651 5652
    bool asynchronous = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKJOB_ASYNC);
    bool synchronous = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKJOB_SYNC);
5653

5654
    if (!synchronous && !asynchronous) {
5655 5656 5657 5658 5659 5660
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("block jobs not supported with this QEMU binary"));
        return -1;
    }

    if (modern)
5661
        *modern = asynchronous;
5662 5663 5664

    return 0;
}
5665 5666 5667 5668 5669 5670 5671 5672 5673


/**
 * 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.
 */
5674
virDomainChrDefPtr
5675 5676 5677 5678 5679 5680 5681 5682 5683 5684
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;

5685 5686
        if (STREQ_NULLABLE(channel->target.name, "org.qemu.guest_agent.0"))
            return channel;
5687 5688
    }

5689
    return NULL;
5690
}
5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709


bool
qemuDomainMachineIsQ35(const virDomainDef *def)
{
    return (STRPREFIX(def->os.machine, "pc-q35") ||
            STREQ(def->os.machine, "q35"));
}


bool
qemuDomainMachineIsI440FX(const virDomainDef *def)
{
    return (STREQ(def->os.machine, "pc") ||
            STRPREFIX(def->os.machine, "pc-0.") ||
            STRPREFIX(def->os.machine, "pc-1.") ||
            STRPREFIX(def->os.machine, "pc-i440") ||
            STRPREFIX(def->os.machine, "rhel"));
}
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
bool
qemuDomainMachineHasPCIRoot(const virDomainDef *def)
{
    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
qemuDomainMachineHasPCIeRoot(const virDomainDef *def)
{
    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;
}


5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759
bool
qemuDomainMachineNeedsFDC(const virDomainDef *def)
{
    char *p = STRSKIP(def->os.machine, "pc-q35-");

    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;
}


5760 5761 5762 5763 5764 5765
bool
qemuDomainMachineIsS390CCW(const virDomainDef *def)
{
    return STRPREFIX(def->os.machine, "s390-ccw");
}

5766

J
Ján Tomko 已提交
5767 5768 5769
bool
qemuDomainMachineIsVirt(const virDomainDef *def)
{
5770 5771 5772 5773 5774 5775 5776 5777 5778
    if (def->os.arch != VIR_ARCH_ARMV7L &&
        def->os.arch != VIR_ARCH_AARCH64)
        return false;

    if (STRNEQ(def->os.machine, "virt") &&
        !STRPREFIX(def->os.machine, "virt-"))
        return false;

    return true;
J
Ján Tomko 已提交
5779 5780 5781
}


5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795
bool
qemuDomainMachineIsPSeries(const virDomainDef *def)
{
    if (!ARCH_IS_PPC64(def->os.arch))
        return false;

    if (STRNEQ(def->os.machine, "pseries") &&
        !STRPREFIX(def->os.machine, "pseries-"))
        return false;

    return true;
}


5796 5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831 5832 5833 5834 5835 5836 5837 5838 5839 5840 5841 5842
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:
        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;
        }

5843 5844 5845
        if (virDomainNumaGetNodeCount(def->numa) != 0) {
            if (mem->targetNode == -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
Y
Yuri Chornoivan 已提交
5846
                               _("target NUMA node needs to be specified for "
5847 5848 5849
                                 "memory device"));
                return -1;
            }
5850 5851
        }

5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875
        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;
}


5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903
/**
 * 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;
    size_t i;

    hotplugSpace = def->mem.max_memory - virDomainDefGetMemoryInitial(def);

    if (mem) {
        nmems++;
        hotplugMemory = mem->size;
5904 5905 5906

        if (qemuDomainDefValidateMemoryHotplugDevice(mem, def) < 0)
            return -1;
5907 5908 5909 5910 5911 5912 5913 5914 5915 5916 5917 5918 5919 5920 5921 5922 5923 5924 5925
    }

    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;
    }

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_PC_DIMM)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory hotplug isn't supported by this QEMU binary"));
        return -1;
    }

5926 5927 5928 5929 5930 5931 5932 5933 5934 5935 5936
    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;
        }
5937 5938 5939 5940 5941 5942 5943 5944 5945
    }

    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;
    }

5946
    for (i = 0; i < def->nmems; i++) {
5947 5948
        hotplugMemory += def->mems[i]->size;

5949 5950 5951 5952 5953 5954
        /* already existing devices don't need to be checked on hotplug */
        if (!mem &&
            qemuDomainDefValidateMemoryHotplugDevice(def->mems[i], def) < 0)
            return -1;
    }

5955 5956 5957 5958 5959 5960 5961 5962 5963 5964
    if (hotplugMemory > hotplugSpace) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory device total size exceeds hotplug space"));
        return -1;
    }

    return 0;
}


5965 5966 5967 5968 5969 5970 5971 5972 5973 5974
bool
qemuDomainMachineHasBuiltinIDE(const virDomainDef *def)
{
    return qemuDomainMachineIsI440FX(def) ||
        STREQ(def->os.machine, "malta") ||
        STREQ(def->os.machine, "sun4u") ||
        STREQ(def->os.machine, "g3beige");
}


5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998
/**
 * 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 */
5999
    if (!virDomainDefHasMemballoon(vm->def)) {
6000
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
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
        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;

        if (!virDomainObjIsActive(vm)) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
            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;

6033 6034
        vm->def->mem.cur_balloon = balloon;
    }
6035 6036 6037

    return 0;
}
6038 6039 6040


/**
6041
 * qemuDomainGetMemLockLimitBytes:
6042 6043 6044 6045
 *
 * @def: domain definition
 *
 * Returns the size of the memory in bytes that needs to be set as
6046
 * RLIMIT_MEMLOCK for the QEMU process.
6047 6048 6049 6050
 * If a mem.hard_limit is set, then that value is preferred; otherwise, the
 * value returned may depend upon the architecture or devices present.
 */
unsigned long long
6051
qemuDomainGetMemLockLimitBytes(virDomainDefPtr def)
6052 6053 6054
{
    unsigned long long memKB;

6055 6056 6057 6058 6059 6060
    /* prefer the hard limit */
    if (virMemoryLimitIsSet(def->mem.hard_limit)) {
        memKB = def->mem.hard_limit;
        goto done;
    }

6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083 6084
    if (ARCH_IS_PPC64(def->os.arch)) {
        unsigned long long maxMemory;
        unsigned long long memory;
        unsigned long long baseLimit;
        unsigned long long passthroughLimit;
        size_t nPCIHostBridges;
        size_t i;
        bool usesVFIO = false;

        /* TODO: Detect at runtime once we start using more than just
         *       the default PCI Host Bridge */
        nPCIHostBridges = 1;

        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;
            }
        }

6085
        memory = virDomainDefGetMemoryTotal(def);
6086 6087 6088 6089 6090 6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112 6113 6114 6115 6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133 6134 6135 6136 6137 6138

        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;
    }

6139 6140 6141 6142 6143 6144 6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156
    /* 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.
     */
6157
    memKB = virDomainDefGetMemoryTotal(def) + 1024 * 1024;
6158

6159
 done:
6160 6161
    return memKB << 10;
}
6162 6163 6164 6165 6166


/**
 * @def: domain definition
 *
6167 6168 6169
 * Returns true if the locked memory limit needs to be set or updated because
 * of domain configuration, VFIO passthrough devices or architecture-specific
 * requirements.
6170 6171
 * */
bool
6172
qemuDomainRequiresMemLock(virDomainDefPtr def)
6173 6174 6175 6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187
{
    size_t i;

    if (def->mem.locked)
        return true;

    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)
            return true;
    }

6188 6189 6190 6191
    /* ppc64 KVM domains need to lock some memory even when VFIO is not used */
    if (ARCH_IS_PPC64(def->os.arch) && def->virtType == VIR_DOMAIN_VIRT_KVM)
        return true;

6192 6193
    return false;
}
6194

6195 6196 6197 6198 6199 6200 6201
/**
 * qemuDomainAdjustMaxMemLock:
 * @vm: domain
 *
 * Adjust the memory locking limit for the QEMU process associated to @vm, in
 * order to comply with VFIO or architecture requirements.
 *
6202 6203 6204 6205
 * 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.
6206 6207 6208 6209 6210 6211 6212 6213 6214
 *
 * Returns: 0 on success, <0 on failure
 */
int
qemuDomainAdjustMaxMemLock(virDomainObjPtr vm)
{
    unsigned long long bytes = 0;
    int ret = -1;

6215
    if (qemuDomainRequiresMemLock(vm->def)) {
6216 6217 6218 6219 6220 6221 6222 6223
        /* 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;
        }
6224
        bytes = qemuDomainGetMemLockLimitBytes(vm->def);
6225 6226 6227 6228 6229 6230
    } 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;
    }
6231 6232 6233 6234 6235 6236 6237 6238 6239 6240

    /* 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;
}
6241 6242 6243 6244 6245 6246 6247 6248 6249 6250

/**
 * qemuDomainHasVcpuPids:
 * @vm: Domain object
 *
 * Returns true if we were able to successfully detect vCPU pids for the VM.
 */
bool
qemuDomainHasVcpuPids(virDomainObjPtr vm)
{
6251 6252 6253 6254 6255 6256
    size_t i;
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    virDomainVcpuDefPtr vcpu;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
6257

6258 6259 6260 6261 6262
        if (QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid > 0)
            return true;
    }

    return false;
6263
}
6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274


/**
 * 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,
6275
                     unsigned int vcpuid)
6276
{
6277 6278
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
    return QEMU_DOMAIN_VCPU_PRIVATE(vcpu)->tid;
6279
}
6280 6281


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


6324 6325 6326 6327 6328 6329 6330 6331 6332
bool
qemuDomainSupportsNewVcpuHotplug(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QUERY_HOTPLUGGABLE_CPUS);
}


6333
/**
6334
 * qemuDomainRefreshVcpuInfo:
6335 6336 6337
 * @driver: qemu driver data
 * @vm: domain object
 * @asyncJob: current asynchronous job type
6338
 * @state: refresh vcpu state
6339
 *
6340 6341
 * Updates vCPU information private data of @vm. Due to historical reasons this
 * function returns success even if some data were not reported by qemu.
6342
 *
6343 6344
 * If @state is true, the vcpu state is refreshed as reported by the monitor.
 *
6345
 * Returns 0 on success and -1 on fatal error.
6346 6347
 */
int
6348 6349
qemuDomainRefreshVcpuInfo(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
6350 6351
                          int asyncJob,
                          bool state)
6352
{
6353
    virDomainVcpuDefPtr vcpu;
6354 6355
    qemuDomainVcpuPrivatePtr vcpupriv;
    qemuMonitorCPUInfoPtr info = NULL;
6356 6357
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
    size_t i;
6358
    bool hotplug;
6359
    int rc;
6360
    int ret = -1;
6361

6362
    hotplug = qemuDomainSupportsNewVcpuHotplug(vm);
6363 6364 6365

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
6366

6367
    rc = qemuMonitorGetCPUInfo(qemuDomainGetMonitor(vm), &info, maxvcpus, hotplug);
6368

6369
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
6370 6371
        goto cleanup;

6372
    if (rc < 0)
6373 6374 6375 6376
        goto cleanup;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
6377
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
6378

6379 6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397 6398 6399 6400 6401 6402 6403 6404 6405 6406 6407 6408 6409 6410 6411 6412 6413 6414 6415 6416 6417 6418
        /*
         * 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.
         *
         * What's more the 'query-cpus' command returns bizarre
         * 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;
        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;
6419
        vcpupriv->qemu_id = info[i].qemu_id;
6420

6421
        if (hotplug && state) {
6422 6423 6424 6425 6426
            vcpu->online = info[i].online;
            if (info[i].hotpluggable)
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_YES;
            else
                vcpu->hotpluggable = VIR_TRISTATE_BOOL_NO;
6427
        }
6428 6429
    }

6430
    ret = 0;
6431 6432

 cleanup:
6433
    qemuMonitorCPUInfoFree(info, maxvcpus);
6434
    return ret;
6435
}
6436

6437 6438 6439 6440 6441 6442 6443 6444 6445 6446 6447 6448 6449 6450 6451 6452 6453 6454 6455 6456 6457 6458 6459 6460 6461 6462 6463 6464 6465 6466 6467
/**
 * 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;
6468
    qemuDomainVcpuPrivatePtr vcpupriv;
6469
    size_t maxvcpus = virDomainDefGetVcpusMax(vm->def);
6470
    virBitmapPtr haltedmap = NULL;
6471 6472 6473 6474 6475 6476 6477 6478 6479 6480
    size_t i;
    int ret = -1;

    /* Not supported currently for TCG, see qemuDomainRefreshVcpuInfo */
    if (vm->def->virtType == VIR_DOMAIN_VIRT_QEMU)
        return 0;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;

6481
    haltedmap = qemuMonitorGetCpuHalted(qemuDomainGetMonitor(vm), maxvcpus);
6482

6483
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || !haltedmap)
6484 6485 6486 6487
        goto cleanup;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(vm->def, i);
6488 6489
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);
        vcpupriv->halted = virBitmapIsBitSet(haltedmap, vcpupriv->qemu_id);
6490 6491 6492 6493 6494
    }

    ret = 0;

 cleanup:
6495
    virBitmapFree(haltedmap);
6496 6497
    return ret;
}
6498 6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517

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;
}

bool
qemuDomainSupportsNetdev(virDomainDefPtr def,
                         virQEMUCapsPtr qemuCaps,
                         virDomainNetDefPtr net)
{
6518
    if (!qemuDomainSupportsNicdev(def, net))
6519 6520 6521
        return false;
    return virQEMUCapsGet(qemuCaps, QEMU_CAPS_NETDEV);
}
J
John Ferlan 已提交
6522 6523 6524 6525 6526 6527 6528


int
qemuDomainNetVLAN(virDomainNetDefPtr def)
{
    return qemuDomainDeviceAliasIndex(&def->info, "net");
}
P
Peter Krempa 已提交
6529 6530 6531 6532 6533 6534 6535 6536 6537 6538 6539 6540 6541 6542 6543 6544


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;
}
6545 6546 6547 6548 6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562 6563 6564 6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575


/**
 * 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;
}
6576 6577 6578 6579 6580 6581 6582


int
qemuDomainPrepareChannel(virDomainChrDefPtr channel,
                         const char *domainChannelTargetDir)
{
    if (channel->targetType == VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO &&
6583 6584 6585
        channel->source->type == VIR_DOMAIN_CHR_TYPE_UNIX &&
        !channel->source->data.nix.path) {
        if (virAsprintf(&channel->source->data.nix.path,
6586 6587 6588 6589 6590
                        "%s/%s", domainChannelTargetDir,
                        channel->target.name ? channel->target.name
                        : "unknown.sock") < 0)
            return -1;

6591
        channel->source->data.nix.listen = true;
6592 6593 6594 6595
    }

    return 0;
}
6596 6597


6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608 6609 6610 6611 6612 6613 6614 6615
/* qemuProcessPrepareDomainChardevSourceTLS:
 * @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;
6616
            source->data.tcp.tlsFromConfig = true;
6617 6618 6619 6620 6621 6622 6623 6624 6625 6626 6627 6628 6629 6630 6631 6632 6633 6634 6635 6636 6637 6638 6639 6640 6641 6642 6643 6644 6645 6646 6647 6648 6649 6650 6651 6652 6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664
        }
    }
}


/* qemuProcessPrepareDomainChardevSource:
 * @def: live domain definition
 * @driver: qemu driver
 *
 * Iterate through all devices that use virDomainChrSourceDefPtr as host
 * interface part.
 */
void
qemuDomainPrepareChardevSource(virDomainDefPtr def,
                               virQEMUDriverPtr driver)
{
    size_t i;
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    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);

    virObjectUnref(cfg);
}



6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676 6677
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);
}


6678 6679 6680 6681 6682 6683 6684 6685 6686 6687 6688 6689 6690 6691 6692 6693 6694 6695 6696 6697 6698 6699 6700 6701 6702 6703 6704 6705 6706 6707 6708 6709 6710
/**
 * 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);
}
6711 6712 6713 6714 6715 6716 6717 6718 6719 6720 6721 6722 6723 6724 6725 6726 6727 6728 6729 6730 6731 6732 6733 6734 6735 6736 6737 6738 6739 6740 6741 6742 6743 6744


/**
 * 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;
        }
    }
}
6745 6746 6747 6748 6749 6750 6751 6752 6753 6754 6755 6756 6757 6758 6759 6760 6761 6762 6763 6764


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;
}
6765 6766 6767 6768 6769 6770 6771 6772 6773 6774 6775 6776


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;
}
6777 6778 6779 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 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836 6837 6838 6839 6840 6841 6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857 6858 6859 6860 6861 6862 6863 6864 6865 6866 6867 6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882 6883 6884 6885 6886 6887 6888 6889 6890 6891 6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906 6907 6908 6909 6910 6911


static int
qemuDomainCreateDevice(const char *device,
                       const char *path,
                       bool allow_noent)
{
    char *devicePath = NULL;
    struct stat sb;
    int ret = -1;

    if (!STRPREFIX(device, "/dev")) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("invalid device: %s"),
                       device);
        goto cleanup;
    }

    if (virAsprintf(&devicePath, "%s/%s",
                    path, device + 4) < 0)
        goto cleanup;

    if (stat(device, &sb) < 0) {
        if (errno == ENOENT && allow_noent) {
            /* Ignore non-existent device. */
            ret = 0;
            goto cleanup;
        }

        virReportSystemError(errno, _("Unable to stat %s"), device);
        goto cleanup;
    }

    if (virFileMakeParentPath(devicePath) < 0) {
        virReportSystemError(errno,
                             _("Unable to create %s"),
                             devicePath);
        goto cleanup;
    }

    if (mknod(devicePath, sb.st_mode, sb.st_rdev) < 0) {
        virReportSystemError(errno,
                             _("Failed to make device %s"),
                             devicePath);
        goto cleanup;
    }

    if (chown(devicePath, sb.st_uid, sb.st_gid) < 0) {
        virReportSystemError(errno,
                             _("Failed to chown device %s"),
                             devicePath);
        goto cleanup;
    }

    if (virFileCopyACLs(device, devicePath) < 0 &&
        errno != ENOTSUP) {
        virReportSystemError(errno,
                             _("Failed to copy ACLs on device %s"),
                             devicePath);
        goto cleanup;
    }

    ret = 0;
 cleanup:
    VIR_FREE(devicePath);
    return ret;
}



static int
qemuDomainPopulateDevices(virQEMUDriverPtr driver,
                          virDomainObjPtr vm ATTRIBUTE_UNUSED,
                          const char *path)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    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++) {
        if (qemuDomainCreateDevice(devices[i], path, true) < 0)
            goto cleanup;
    }

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


static int
qemuDomainSetupDev(virQEMUDriverPtr driver,
                   virDomainObjPtr vm,
                   const char *path)
{
    char *mount_options = NULL;
    char *opts = NULL;
    int ret = -1;

    VIR_DEBUG("Setting up /dev/ for domain %s", vm->def->name);

    mount_options = virSecurityManagerGetMountOptions(driver->securityManager,
                                                      vm->def);

    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;

    if (virFileSetupDev(path, opts) < 0)
        goto cleanup;

    if (qemuDomainPopulateDevices(driver, vm, path) < 0)
        goto cleanup;

    ret = 0;
 cleanup:
    VIR_FREE(opts);
    VIR_FREE(mount_options);
    return ret;
}


6912 6913 6914 6915 6916 6917 6918 6919 6920 6921 6922 6923 6924 6925 6926 6927 6928 6929 6930 6931 6932 6933 6934 6935 6936 6937 6938 6939 6940 6941 6942 6943 6944 6945 6946 6947 6948 6949 6950 6951 6952 6953 6954 6955 6956 6957 6958
static int
qemuDomainSetupDisk(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                    virDomainDiskDefPtr disk,
                    const char *devPath)
{
    virStorageSourcePtr next;
    char *dst = NULL;
    int ret = -1;

    for (next = disk->src; next; next = next->backingStore) {
        if (!next->path || !virStorageSourceIsLocalStorage(next) ||
            !STRPREFIX(next->path, "/dev")) {
            /* Not creating device. Just continue. */
            continue;
        }

        if (qemuDomainCreateDevice(next->path, devPath, false) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    VIR_FREE(dst);
    return ret;
}


static int
qemuDomainSetupAllDisks(virQEMUDriverPtr driver,
                        virDomainObjPtr vm,
                        const char *devPath)
{
    size_t i;
    VIR_DEBUG("Setting up disks");

    for (i = 0; i < vm->def->ndisks; i++) {
        if (qemuDomainSetupDisk(driver,
                                vm->def->disks[i],
                                devPath) < 0)
            return -1;
    }

    VIR_DEBUG("Setup all disks");
    return 0;
}


6959 6960 6961 6962 6963 6964 6965 6966 6967 6968 6969 6970 6971 6972 6973 6974 6975 6976 6977 6978 6979 6980 6981 6982 6983 6984 6985 6986 6987 6988 6989 6990 6991 6992 6993 6994 6995 6996 6997 6998 6999 7000 7001 7002 7003 7004 7005 7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020 7021 7022 7023 7024 7025 7026 7027 7028 7029 7030 7031 7032 7033 7034 7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052 7053 7054 7055 7056 7057 7058 7059 7060 7061 7062 7063 7064 7065 7066 7067 7068 7069 7070 7071 7072 7073 7074 7075 7076 7077 7078 7079 7080 7081 7082 7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093 7094 7095 7096 7097 7098 7099 7100 7101 7102 7103 7104 7105 7106 7107 7108 7109 7110 7111 7112 7113 7114 7115 7116
static int
qemuDomainGetHostdevPath(virDomainHostdevDefPtr dev,
                         char **path)
{
    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;
    virPCIDevicePtr pci = NULL;
    virUSBDevicePtr usb = NULL;
    virSCSIDevicePtr scsi = NULL;
    virSCSIVHostDevicePtr host = NULL;
    char *tmpPath = NULL;
    bool freeTmpPath = false;

    *path = NULL;

    switch ((virDomainHostdevMode) dev->mode) {
    case VIR_DOMAIN_HOSTDEV_MODE_SUBSYS:
        switch ((virDomainHostdevSubsysType) dev->source.subsys.type) {
        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;
            }
            break;

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
            if (dev->missing)
                break;
            usb = virUSBDeviceNew(usbsrc->bus,
                                  usbsrc->device,
                                  NULL);
            if (!usb)
                goto cleanup;

            if (!(tmpPath = (char *) virUSBDeviceGetPath(usb)))
                goto cleanup;
            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
                 */
                VIR_DEBUG("Not updating /dev for hostdev iSCSI path '%s'", iscsisrc->path);
            } 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;

                if (!(tmpPath = (char *) virSCSIDeviceGetPath(scsi)))
                    goto cleanup;
            }
            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;

                if (!(tmpPath = (char *) virSCSIVHostDeviceGetPath(host)))
                    goto cleanup;
            }
            break;
        }

        case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST:
            break;
        }
        break;

    case VIR_DOMAIN_HOSTDEV_MODE_CAPABILITIES:
    case VIR_DOMAIN_HOSTDEV_MODE_LAST:
        /* nada */
        break;
    }

    if (VIR_STRDUP(*path, tmpPath) < 0)
        goto cleanup;

    ret = 0;
 cleanup:
    virPCIDeviceFree(pci);
    virUSBDeviceFree(usb);
    virSCSIDeviceFree(scsi);
    virSCSIVHostDeviceFree(host);
    if (freeTmpPath)
        VIR_FREE(tmpPath);
    return ret;
}


static int
qemuDomainSetupHostdev(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                       virDomainHostdevDefPtr dev,
                       const char *devPath)
{
    int ret = -1;
    char *path = NULL;

    if (qemuDomainGetHostdevPath(dev, &path) < 0)
        goto cleanup;

    if (!path) {
        /* There's no /dev device that we need to create. Claim success. */
        ret = 0;
        goto cleanup;
    }

    if (qemuDomainCreateDevice(path, devPath, false) < 0)
        goto cleanup;

    ret = 0;
 cleanup:
    VIR_FREE(path);
    return ret;
}


static int
qemuDomainSetupAllHostdevs(virQEMUDriverPtr driver,
                           virDomainObjPtr vm,
                           const char *devPath)
{
    size_t i;

    VIR_DEBUG("Setting up hostdevs");
    for (i = 0; i < vm->def->nhostdevs; i++) {
        if (qemuDomainSetupHostdev(driver,
                                   vm->def->hostdevs[i],
                                   devPath) < 0)
            return -1;
    }
    VIR_DEBUG("Setup all hostdevs");
    return 0;
}


7117 7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131 7132 7133 7134 7135 7136 7137 7138 7139 7140 7141 7142 7143 7144 7145 7146 7147 7148
static int
qemuDomainSetupChardev(virDomainDefPtr def ATTRIBUTE_UNUSED,
                       virDomainChrDefPtr dev,
                       void *opaque)
{
    const char *devPath = opaque;

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

    return qemuDomainCreateDevice(dev->source->data.file.path, devPath, false);
}


static int
qemuDomainSetupAllChardevs(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                           virDomainObjPtr vm,
                           const char *devPath)
{
    VIR_DEBUG("Setting up chardevs");

    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuDomainSetupChardev,
                               (void *) devPath) < 0)
        return -1;

    VIR_DEBUG("Setup all chardevs");
    return 0;
}


7149 7150 7151 7152 7153 7154 7155 7156 7157 7158 7159 7160 7161 7162 7163 7164 7165 7166 7167 7168 7169 7170 7171 7172 7173 7174 7175 7176 7177
static int
qemuDomainSetupTPM(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                   virDomainObjPtr vm,
                   const char *devPath)
{
    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,
                                   devPath, false) < 0)
            return -1;
        break;

    case VIR_DOMAIN_TPM_TYPE_LAST:
        /* nada */
        break;
    }

    VIR_DEBUG("Setup TPM");
    return 0;
}


7178 7179 7180 7181 7182 7183 7184 7185 7186 7187 7188 7189 7190 7191 7192 7193 7194 7195 7196 7197 7198 7199 7200 7201 7202 7203 7204 7205 7206 7207 7208 7209 7210 7211 7212 7213 7214 7215
int
qemuDomainBuildNamespace(virQEMUDriverPtr driver,
                         virDomainObjPtr vm)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    const unsigned long mount_flags = MS_MOVE;
    char *devPath = NULL;
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0, i;
    int ret = -1;

    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT)) {
        ret = 0;
        goto cleanup;
    }

    if (virAsprintf(&devPath, "%s/%s.dev",
                    cfg->stateDir, vm->def->name) < 0)
        goto cleanup;

    if (qemuDomainGetPreservedMounts(driver, vm,
                                     &devMountsPath, &ndevMountsPath) < 0)
        goto cleanup;

    if (qemuDomainSetupDev(driver, vm, devPath) < 0)
        goto cleanup;

    /* Save some mount points because we want to share them with the host */
    for (i = 0; i < ndevMountsPath; i++) {
        if (mount(devPreserveMounts[i].path, devMountsPath[i],
                  NULL, mount_flags, NULL) < 0) {
            virReportSystemError(errno,
                                 _("Unable to move %s mount"),
                                 devPreserveMounts[i].path);
            goto cleanup;
        }
    }

7216 7217
    if (qemuDomainSetupAllDisks(driver, vm, devPath) < 0)
        goto cleanup;
7218 7219 7220

    if (qemuDomainSetupAllHostdevs(driver, vm, devPath) < 0)
        goto cleanup;
7221 7222 7223

    if (qemuDomainSetupAllChardevs(driver, vm, devPath) < 0)
        goto cleanup;
7224 7225 7226

    if (qemuDomainSetupTPM(driver, vm, devPath) < 0)
        goto cleanup;
7227

7228 7229 7230 7231 7232 7233 7234 7235 7236 7237 7238 7239 7240 7241 7242 7243 7244 7245 7246 7247 7248 7249 7250 7251 7252 7253 7254 7255 7256 7257 7258 7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271 7272 7273 7274 7275 7276 7277 7278 7279 7280 7281 7282 7283 7284 7285 7286 7287 7288 7289 7290 7291 7292 7293 7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320 7321 7322 7323 7324 7325 7326 7327 7328 7329 7330 7331 7332 7333 7334 7335 7336 7337 7338 7339 7340 7341 7342 7343 7344 7345 7346 7347 7348 7349 7350 7351 7352 7353 7354 7355 7356 7357 7358 7359 7360 7361 7362 7363 7364 7365 7366 7367 7368 7369 7370
    if (mount(devPath, "/dev", NULL, mount_flags, NULL) < 0) {
        virReportSystemError(errno,
                             _("Failed to mount %s on /dev"),
                             devPath);
        goto cleanup;
    }

    for (i = 0; i < ndevMountsPath; i++) {
        if (virFileMakePath(devPreserveMounts[i].path) < 0) {
            virReportSystemError(errno, _("Cannot create %s"),
                                 devPreserveMounts[i].path);
            goto cleanup;
        }

        if (mount(devMountsPath[i], devPreserveMounts[i].path,
                  NULL, mount_flags, NULL) < 0) {
            virReportSystemError(errno,
                                 _("Failed to mount %s on %s"),
                                 devMountsPath[i],
                                 devPreserveMounts[i].path);
            goto cleanup;
        }
    }

    ret = 0;
 cleanup:
    virObjectUnref(cfg);
    VIR_FREE(devPath);
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    return ret;
}


#if defined(__linux__)
int
qemuDomainCreateNamespace(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;
    char *devPath = NULL;
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0, i;

    if (!virQEMUDriverIsPrivileged(driver)) {
        ret = 0;
        goto cleanup;
    }

    if (virAsprintf(&devPath, "%s/%s.dev",
                    cfg->stateDir, vm->def->name) < 0)
        goto cleanup;

    if (qemuDomainGetPreservedMounts(driver, vm,
                                     &devMountsPath, &ndevMountsPath) < 0)
        goto cleanup;

    if (virFileMakePath(devPath) < 0) {
        virReportSystemError(errno,
                             _("Failed to create %s"),
                             devPath);
        goto cleanup;
    }

    for (i = 0; i < ndevMountsPath; i++) {
        if (virFileMakePath(devMountsPath[i]) < 0) {
            virReportSystemError(errno,
                                 _("Failed to create %s"),
                                 devMountsPath[i]);
            goto cleanup;
        }
    }

    /* Enabling of the mount namespace goes here. */

    ret = 0;
 cleanup:
    if (ret < 0) {
        if (devPath)
            unlink(devPath);
        for (i = 0; i < ndevMountsPath; i++)
            unlink(devMountsPath[i]);
    }
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    VIR_FREE(devPath);
    virObjectUnref(cfg);
    return ret;
}

#else /* !defined(__linux__) */

int
qemuDomainCreateNamespace(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm ATTRIBUTE_UNUSED)
{
    /* Namespaces are Linux specific. On other platforms just
     * carry on with the old behaviour. */
    return 0;
}
#endif /* !defined(__linux__) */


void
qemuDomainDeleteNamespace(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    char *devPath = NULL;
    char **devMountsPath = NULL;
    size_t ndevMountsPath = 0, i;


    if (!qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT))
        return;

    if (virAsprintf(&devPath, "%s/%s.dev",
                    cfg->stateDir, vm->def->name) < 0)
        goto cleanup;

    if (qemuDomainGetPreservedMounts(driver, vm,
                                     &devMountsPath, &ndevMountsPath) < 0)
        goto cleanup;

    if (rmdir(devPath) < 0) {
        virReportSystemError(errno,
                             _("Unable to remove %s"),
                             devPath);
        /* Bet effort. Fall through. */
    }

    for (i = 0; i < ndevMountsPath; i++) {
        if (rmdir(devMountsPath[i]) < 0) {
            virReportSystemError(errno,
                                 _("Unable to remove %s"),
                                 devMountsPath[i]);
            /* Bet effort. Fall through. */
        }
    }
 cleanup:
    virObjectUnref(cfg);
    virStringListFreeCount(devMountsPath, ndevMountsPath);
    VIR_FREE(devPath);
}