qemu_domain.c 157.4 KB
Newer Older
1
/*
2
 * qemu_domain.c: QEMU domain private state
3
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include "qemu_domain.h"
J
John Ferlan 已提交
27
#include "qemu_alias.h"
28
#include "qemu_cgroup.h"
29
#include "qemu_command.h"
30
#include "qemu_parse_command.h"
31
#include "qemu_capabilities.h"
32
#include "qemu_migration.h"
33
#include "viralloc.h"
34
#include "virlog.h"
35
#include "virerror.h"
36
#include "c-ctype.h"
37
#include "cpu/cpu.h"
38
#include "viruuid.h"
E
Eric Blake 已提交
39
#include "virfile.h"
40
#include "domain_addr.h"
41
#include "domain_event.h"
42
#include "virtime.h"
43
#include "virstoragefile.h"
44
#include "virstring.h"
45
#include "virthreadjob.h"
46
#include "viratomic.h"
47
#include "virprocess.h"
J
John Ferlan 已提交
48
#include "virrandom.h"
49
#include "secret_util.h"
J
John Ferlan 已提交
50
#include "base64.h"
51 52 53 54 55
#ifdef WITH_GNUTLS
# include <gnutls/gnutls.h>
# if HAVE_GNUTLS_CRYPTO_H
#  include <gnutls/crypto.h>
# endif
J
John Ferlan 已提交
56
#endif
57
#include "logging/log_manager.h"
58
#include "locking/domain_lock.h"
59

60 61
#include "storage/storage_driver.h"

62
#include <sys/time.h>
63
#include <fcntl.h>
64

65 66 67 68
#include <libxml/xpathInternals.h>

#define VIR_FROM_THIS VIR_FROM_QEMU

69 70
VIR_LOG_INIT("qemu.qemu_domain");

71 72
#define QEMU_NAMESPACE_HREF "http://libvirt.org/schemas/domain/qemu/1.0"

73 74 75 76 77 78
VIR_ENUM_IMPL(qemuDomainJob, QEMU_JOB_LAST,
              "none",
              "query",
              "destroy",
              "suspend",
              "modify",
79
              "abort",
80
              "migration operation",
81 82 83 84 85 86 87 88 89 90
              "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",
91
              "snapshot",
92
              "start",
93 94
);

95

96 97 98
struct _qemuDomainLogContext {
    int refs;
    int writefd;
99
    int readfd; /* Only used if manager == NULL */
100
    off_t pos;
101
    ino_t inode; /* Only used if manager != NULL */
102
    char *path;
103
    virLogManagerPtr manager;
104 105
};

J
Jiri Denemark 已提交
106
const char *
107
qemuDomainAsyncJobPhaseToString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
108 109 110 111 112
                                int phase ATTRIBUTE_UNUSED)
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
113 114
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
115 116
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
117
    case QEMU_ASYNC_JOB_SNAPSHOT:
118
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
119 120 121 122 123 124 125 126 127
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        ; /* fall through */
    }

    return "none";
}

int
128
qemuDomainAsyncJobPhaseFromString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
129 130 131 132 133 134 135 136
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
137 138
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
139 140
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
141
    case QEMU_ASYNC_JOB_SNAPSHOT:
142
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
143 144 145 146 147 148 149 150 151 152 153
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        ; /* fall through */
    }

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

154

155
void qemuDomainEventQueue(virQEMUDriverPtr driver,
156
                          virObjectEventPtr event)
157
{
158 159
    if (event)
        virObjectEventStateQueue(driver->domainEventState, event);
160 161 162
}


163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186
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);
}


187 188 189 190 191 192 193 194
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

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

195
    if (virCondInit(&priv->job.asyncCond) < 0) {
196
        virCondDestroy(&priv->job.cond);
197 198 199
        return -1;
    }

200 201 202 203 204 205 206 207 208
    return 0;
}

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

    job->active = QEMU_JOB_NONE;
209
    job->owner = 0;
210
    job->ownerAPI = NULL;
J
Jiri Denemark 已提交
211
    job->started = 0;
212 213 214 215 216 217 218 219
}

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

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
220
    job->asyncOwner = 0;
221
    job->asyncOwnerAPI = NULL;
J
Jiri Denemark 已提交
222
    job->asyncStarted = 0;
J
Jiri Denemark 已提交
223
    job->phase = 0;
224
    job->mask = QEMU_JOB_DEFAULT_MASK;
225
    job->dump_memory_only = false;
226
    job->abortJob = false;
227
    job->spiceMigrated = false;
228
    job->postcopyEnabled = false;
J
Jiri Denemark 已提交
229
    VIR_FREE(job->current);
230 231
}

232 233 234 235 236 237 238 239
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
                        struct qemuDomainJobObj *job)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
240
    job->owner = priv->job.owner;
241
    job->asyncJob = priv->job.asyncJob;
242
    job->asyncOwner = priv->job.asyncOwner;
J
Jiri Denemark 已提交
243
    job->phase = priv->job.phase;
244 245 246 247 248

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

249 250 251
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
J
Jiri Denemark 已提交
252
    VIR_FREE(priv->job.current);
253
    VIR_FREE(priv->job.completed);
254 255
    virCondDestroy(&priv->job.cond);
    virCondDestroy(&priv->job.asyncCond);
256 257
}

258
static bool
259
qemuDomainTrackJob(qemuDomainJob job)
260 261 262 263
{
    return (QEMU_DOMAIN_TRACK_JOBS & JOB_MASK(job)) != 0;
}

264

J
Jiri Denemark 已提交
265 266 267 268 269 270 271 272 273 274 275
int
qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->started)
        return 0;

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

276 277 278 279 280 281
    if (now < jobInfo->started) {
        VIR_WARN("Async job starts in the future");
        jobInfo->started = 0;
        return 0;
    }

J
Jiri Denemark 已提交
282 283 284 285
    jobInfo->timeElapsed = now - jobInfo->started;
    return 0;
}

286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302
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;
    }

303 304
    jobInfo->stats.downtime = now - jobInfo->stopped;
    jobInfo->stats.downtime_set = true;
305 306 307
    return 0;
}

J
Jiri Denemark 已提交
308 309 310 311
int
qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                        virDomainJobInfoPtr info)
{
312
    info->type = jobInfo->type;
J
Jiri Denemark 已提交
313 314 315
    info->timeElapsed = jobInfo->timeElapsed;
    info->timeRemaining = jobInfo->timeRemaining;

316 317 318
    info->memTotal = jobInfo->stats.ram_total;
    info->memRemaining = jobInfo->stats.ram_remaining;
    info->memProcessed = jobInfo->stats.ram_transferred;
J
Jiri Denemark 已提交
319

320 321 322
    info->fileTotal = jobInfo->stats.disk_total;
    info->fileRemaining = jobInfo->stats.disk_remaining;
    info->fileProcessed = jobInfo->stats.disk_transferred;
J
Jiri Denemark 已提交
323 324 325 326 327 328 329 330 331 332 333 334 335 336

    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)
{
337
    qemuMonitorMigrationStats *stats = &jobInfo->stats;
J
Jiri Denemark 已提交
338 339 340 341 342 343 344 345 346
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;

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

347 348 349 350 351 352 353
    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
354 355 356 357 358 359
    if (jobInfo->type == VIR_DOMAIN_JOB_BOUNDED &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_REMAINING,
                                jobInfo->timeRemaining) < 0)
        goto error;

360
    if (stats->downtime_set &&
J
Jiri Denemark 已提交
361 362
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME,
363
                                stats->downtime) < 0)
J
Jiri Denemark 已提交
364 365
        goto error;

366
    if (stats->downtime_set &&
367
        jobInfo->timeDeltaSet &&
368
        stats->downtime > jobInfo->timeDelta &&
369 370
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME_NET,
371
                                stats->downtime - jobInfo->timeDelta) < 0)
372 373
        goto error;

374
    if (stats->setup_time_set &&
375 376
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_SETUP_TIME,
377
                                stats->setup_time) < 0)
378 379
        goto error;

J
Jiri Denemark 已提交
380 381
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_TOTAL,
382 383
                                stats->ram_total +
                                stats->disk_total) < 0 ||
J
Jiri Denemark 已提交
384 385
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_PROCESSED,
386 387
                                stats->ram_transferred +
                                stats->disk_transferred) < 0 ||
J
Jiri Denemark 已提交
388 389
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_REMAINING,
390 391
                                stats->ram_remaining +
                                stats->disk_remaining) < 0)
J
Jiri Denemark 已提交
392 393 394 395
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
396
                                stats->ram_total) < 0 ||
J
Jiri Denemark 已提交
397 398
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
399
                                stats->ram_transferred) < 0 ||
J
Jiri Denemark 已提交
400 401
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
402
                                stats->ram_remaining) < 0)
J
Jiri Denemark 已提交
403 404
        goto error;

405
    if (stats->ram_bps &&
406 407
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_BPS,
408
                                stats->ram_bps) < 0)
409 410
        goto error;

411
    if (stats->ram_duplicate_set) {
J
Jiri Denemark 已提交
412 413
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_CONSTANT,
414
                                    stats->ram_duplicate) < 0 ||
J
Jiri Denemark 已提交
415 416
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL,
417
                                    stats->ram_normal) < 0 ||
J
Jiri Denemark 已提交
418 419
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL_BYTES,
420
                                    stats->ram_normal_bytes) < 0)
J
Jiri Denemark 已提交
421 422 423
            goto error;
    }

424 425 426 427 428 429 430 431
    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
432 433
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_TOTAL,
434
                                stats->disk_total) < 0 ||
J
Jiri Denemark 已提交
435 436
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_PROCESSED,
437
                                stats->disk_transferred) < 0 ||
J
Jiri Denemark 已提交
438 439
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_REMAINING,
440
                                stats->disk_remaining) < 0)
J
Jiri Denemark 已提交
441 442
        goto error;

443
    if (stats->disk_bps &&
444 445
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_BPS,
446
                                stats->disk_bps) < 0)
447 448
        goto error;

449
    if (stats->xbzrle_set) {
J
Jiri Denemark 已提交
450 451
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE,
452
                                    stats->xbzrle_cache_size) < 0 ||
J
Jiri Denemark 已提交
453 454
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_BYTES,
455
                                    stats->xbzrle_bytes) < 0 ||
J
Jiri Denemark 已提交
456 457
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_PAGES,
458
                                    stats->xbzrle_pages) < 0 ||
J
Jiri Denemark 已提交
459 460
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE_MISSES,
461
                                    stats->xbzrle_cache_miss) < 0 ||
J
Jiri Denemark 已提交
462 463
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_OVERFLOW,
464
                                    stats->xbzrle_overflow) < 0)
J
Jiri Denemark 已提交
465 466 467 468 469 470 471 472 473 474 475 476 477 478
            goto error;
    }

    *type = jobInfo->type;
    *params = par;
    *nparams = npar;
    return 0;

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


J
John Ferlan 已提交
479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507
/* 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:
 * @priv: pointer to domain private object
 *
 * 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
 */
static int
M
Martin Kletzander 已提交
508 509
qemuDomainWriteMasterKeyFile(virQEMUDriverPtr driver,
                             virDomainObjPtr vm)
J
John Ferlan 已提交
510 511 512 513
{
    char *path;
    int fd = -1;
    int ret = -1;
M
Martin Kletzander 已提交
514
    qemuDomainObjPrivatePtr priv = vm->privateData;
J
John Ferlan 已提交
515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530

    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;
    }

531 532
    if (virSecurityManagerDomainSetPathLabel(driver->securityManager,
                                             vm->def, path) < 0)
M
Martin Kletzander 已提交
533 534
        goto cleanup;

J
John Ferlan 已提交
535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646
    ret = 0;

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

    return ret;
}


/* 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;
}


/* qemuDomainGenerateRandomKey
 * @nbytes: Size in bytes of random key to generate
 *
 * Generate a random key of nbytes length and return it.
 *
 * Since the gnutls_rnd could be missing, provide an alternate less
 * secure mechanism to at least have something.
 *
 * Returns pointer memory containing key on success, NULL on failure
 */
static uint8_t *
qemuDomainGenerateRandomKey(size_t nbytes)
{
    uint8_t *key;
    int ret;

    if (VIR_ALLOC_N(key, nbytes) < 0)
        return NULL;

647 648
#if HAVE_GNUTLS_RND
    /* Generate a master key using gnutls_rnd() if possible */
J
John Ferlan 已提交
649 650 651 652 653 654 655
    if ((ret = gnutls_rnd(GNUTLS_RND_RANDOM, key, nbytes)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("failed to generate master key, ret=%d"), ret);
        VIR_FREE(key);
        return NULL;
    }
#else
656
    /* If we don't have gnutls_rnd(), we will generate a less cryptographically
J
John Ferlan 已提交
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
     * strong master key from /dev/urandom.
     */
    if ((ret = virRandomBytes(key, nbytes)) < 0) {
        virReportSystemError(ret, "%s", _("failed to generate master key"));
        VIR_FREE(key);
        return NULL;
    }
#endif

    return key;
}


/* 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 */
    memset(priv->masterKey, 0, priv->masterKeyLen);
    VIR_FREE(priv->masterKey);
    priv->masterKeyLen = 0;

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

    VIR_FREE(path);
}


/* qemuDomainMasterKeyCreate:
 * @priv: Pointer to the domain private object
 *
 * 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
M
Martin Kletzander 已提交
707 708
qemuDomainMasterKeyCreate(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
J
John Ferlan 已提交
709
{
M
Martin Kletzander 已提交
710 711
    qemuDomainObjPrivatePtr priv = vm->privateData;

J
John Ferlan 已提交
712 713 714 715 716 717 718 719 720 721
    /* If we don't have the capability, then do nothing. */
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_SECRET))
        return 0;

    if (!(priv->masterKey =
          qemuDomainGenerateRandomKey(QEMU_DOMAIN_MASTER_KEY_LEN)))
        goto error;

    priv->masterKeyLen = QEMU_DOMAIN_MASTER_KEY_LEN;

M
Martin Kletzander 已提交
722
    if (qemuDomainWriteMasterKeyFile(driver, vm) < 0)
J
John Ferlan 已提交
723 724 725 726 727 728 729 730 731 732
        goto error;

    return 0;

 error:
    qemuDomainMasterKeyRemove(priv);
    return -1;
}


733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752
static void
qemuDomainSecretPlainFree(qemuDomainSecretPlain secret)
{
    VIR_FREE(secret.username);
    memset(secret.secret, 0, strlen(secret.secret));
    VIR_FREE(secret.secret);
}


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

    qemuDomainSecretPlainFree((*secinfo)->s.plain);
    VIR_FREE(*secinfo);
}


753
static virClassPtr qemuDomainDiskPrivateClass;
754
static void qemuDomainDiskPrivateDispose(void *obj);
755 756 757 758 759 760 761

static int
qemuDomainDiskPrivateOnceInit(void)
{
    qemuDomainDiskPrivateClass = virClassNew(virClassForObject(),
                                             "qemuDomainDiskPrivate",
                                             sizeof(qemuDomainDiskPrivate),
762
                                             qemuDomainDiskPrivateDispose);
763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
    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;
}


786 787 788 789 790 791 792 793 794
static void
qemuDomainDiskPrivateDispose(void *obj)
{
    qemuDomainDiskPrivatePtr priv = obj;

    qemuDomainSecretInfoFree(&priv->secinfo);
}


795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837
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);
}


838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977
/* qemuDomainSecretPlainSetup:
 * @conn: Pointer to connection
 * @secinfo: Pointer to secret info
 * @protocol: Protocol for secret
 * @authdef: Pointer to auth data
 *
 * 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,
                           virStorageNetProtocol protocol,
                           virStorageAuthDefPtr authdef)
{
    bool encode = false;
    int secretType = VIR_SECRET_USAGE_TYPE_ISCSI;
    const char *protocolstr = virStorageNetProtocolTypeToString(protocol);

    secinfo->type = VIR_DOMAIN_SECRET_INFO_PLAIN;
    if (VIR_STRDUP(secinfo->s.plain.username, authdef->username) < 0)
        return -1;

    if (protocol == VIR_STORAGE_NET_PROTOCOL_RBD) {
        /* qemu requires the secret to be encoded for RBD */
        encode = true;
        secretType = VIR_SECRET_USAGE_TYPE_CEPH;
    }

    if (!(secinfo->s.plain.secret =
          virSecretGetSecretString(conn, protocolstr, encode,
                                   authdef, secretType)))
        return -1;

    return 0;
}


/* 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);

    if (!diskPriv->secinfo)
        return;

    qemuDomainSecretInfoFree(&diskPriv->secinfo);
}


/* qemuDomainSecretDiskPrepare:
 * @conn: Pointer to connection
 * @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,
                            virDomainDiskDefPtr disk)
{
    virStorageSourcePtr src = disk->src;
    qemuDomainSecretInfoPtr secinfo = NULL;

    if (conn && !virStorageSourceIsEmpty(src) &&
        virStorageSourceGetActualType(src) == VIR_STORAGE_TYPE_NETWORK &&
        src->auth &&
        (src->protocol == VIR_STORAGE_NET_PROTOCOL_ISCSI ||
         src->protocol == VIR_STORAGE_NET_PROTOCOL_RBD)) {

        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

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

        if (qemuDomainSecretPlainSetup(conn, secinfo, src->protocol,
                                       src->auth) < 0)
            goto error;

        diskPriv->secinfo = secinfo;
    }

    return 0;

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


/* 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]);
}


/* qemuDomainSecretPrepare:
 * @conn: Pointer to connection
 * @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,
                        virDomainObjPtr vm)
{
    size_t i;

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

    return 0;
}


978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004
/* 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
1005 1006
qemuDomainSetPrivatePaths(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
1007
{
1008 1009
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainObjPrivatePtr priv = vm->privateData;
M
Martin Kletzander 已提交
1010
    char *domname = virDomainObjGetShortName(vm);
1011
    int ret = -1;
1012

M
Martin Kletzander 已提交
1013 1014 1015
    if (!domname)
        goto cleanup;

1016
    if (!priv->libDir &&
M
Martin Kletzander 已提交
1017
        virAsprintf(&priv->libDir, "%s/domain-%s", cfg->libDir, domname) < 0)
1018
        goto cleanup;
1019

1020
    if (!priv->channelTargetDir &&
M
Martin Kletzander 已提交
1021 1022
        virAsprintf(&priv->channelTargetDir, "%s/domain-%s",
                    cfg->channelTargetDir, domname) < 0)
1023
        goto cleanup;
1024

1025 1026 1027
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
M
Martin Kletzander 已提交
1028
    VIR_FREE(domname);
1029
    return ret;
1030 1031 1032
}


1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
void
qemuDomainClearPrivatePaths(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

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


1043 1044
static void *
qemuDomainObjPrivateAlloc(void)
1045 1046 1047 1048 1049 1050
{
    qemuDomainObjPrivatePtr priv;

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

1051 1052 1053
    if (qemuDomainObjInitJob(priv) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to init qemu driver mutexes"));
1054
        goto error;
1055
    }
1056

1057
    if (!(priv->devs = virChrdevAlloc()))
1058 1059
        goto error;

1060
    priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
1061

1062
    return priv;
1063

1064
 error:
1065 1066
    VIR_FREE(priv);
    return NULL;
1067 1068
}

1069 1070
static void
qemuDomainObjPrivateFree(void *data)
1071 1072 1073
{
    qemuDomainObjPrivatePtr priv = data;

1074
    virObjectUnref(priv->qemuCaps);
1075

1076
    virCgroupFree(&priv->cgroup);
1077
    virDomainPCIAddressSetFree(priv->pciaddrs);
J
Ján Tomko 已提交
1078
    virDomainCCWAddressSetFree(priv->ccwaddrs);
1079
    virDomainVirtioSerialAddrSetFree(priv->vioserialaddrs);
1080
    virDomainChrSourceDefFree(priv->monConfig);
1081
    qemuDomainObjFreeJob(priv);
1082
    VIR_FREE(priv->vcpupids);
1083
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
1084
    VIR_FREE(priv->origname);
1085

1086
    virStringFreeList(priv->qemuDevices);
1087
    virChrdevFree(priv->devs);
1088

1089 1090
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
1091
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
1092 1093
        qemuMonitorClose(priv->mon);
    }
D
Daniel P. Berrange 已提交
1094 1095 1096 1097
    if (priv->agent) {
        VIR_ERROR(_("Unexpected QEMU agent still active during domain deletion"));
        qemuAgentClose(priv->agent);
    }
1098
    VIR_FREE(priv->cleanupCallbacks);
1099
    virBitmapFree(priv->autoNodeset);
1100
    virBitmapFree(priv->autoCpuset);
1101 1102 1103 1104

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

1105 1106 1107 1108
    VIR_FREE(priv);
}


1109
static int
1110 1111
qemuDomainObjPrivateXMLFormat(virBufferPtr buf,
                              virDomainObjPtr vm)
1112
{
1113
    qemuDomainObjPrivatePtr priv = vm->privateData;
1114
    const char *monitorpath;
1115
    qemuDomainJob job;
1116 1117 1118

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
1119
        switch (priv->monConfig->type) {
1120
        case VIR_DOMAIN_CHR_TYPE_UNIX:
1121
            monitorpath = priv->monConfig->data.nix.path;
1122 1123 1124
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
1125
            monitorpath = priv->monConfig->data.file.path;
1126 1127 1128
            break;
        }

1129
        virBufferEscapeString(buf, "<monitor path='%s'", monitorpath);
1130 1131
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
1132
        virBufferAsprintf(buf, " type='%s'/>\n",
1133
                          virDomainChrTypeToString(priv->monConfig->type));
1134 1135 1136 1137
    }


    if (priv->nvcpupids) {
1138
        size_t i;
1139 1140
        virBufferAddLit(buf, "<vcpus>\n");
        virBufferAdjustIndent(buf, 2);
1141
        for (i = 0; i < priv->nvcpupids; i++)
1142 1143 1144
            virBufferAsprintf(buf, "<vcpu pid='%d'/>\n", priv->vcpupids[i]);
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</vcpus>\n");
1145 1146
    }

1147
    if (priv->qemuCaps) {
1148
        size_t i;
1149 1150
        virBufferAddLit(buf, "<qemuCaps>\n");
        virBufferAdjustIndent(buf, 2);
1151
        for (i = 0; i < QEMU_CAPS_LAST; i++) {
1152
            if (virQEMUCapsGet(priv->qemuCaps, i)) {
1153
                virBufferAsprintf(buf, "<flag name='%s'/>\n",
1154
                                  virQEMUCapsTypeToString(i));
1155 1156
            }
        }
1157 1158
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</qemuCaps>\n");
1159 1160
    }

1161
    if (priv->lockState)
1162
        virBufferAsprintf(buf, "<lockstate>%s</lockstate>\n", priv->lockState);
1163

1164 1165 1166 1167
    job = priv->job.active;
    if (!qemuDomainTrackJob(job))
        priv->job.active = QEMU_JOB_NONE;

1168
    if (priv->job.active || priv->job.asyncJob) {
1169
        virBufferAsprintf(buf, "<job type='%s' async='%s'",
1170 1171
                          qemuDomainJobTypeToString(priv->job.active),
                          qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
J
Jiri Denemark 已提交
1172 1173 1174 1175 1176
        if (priv->job.phase) {
            virBufferAsprintf(buf, " phase='%s'",
                              qemuDomainAsyncJobPhaseToString(
                                    priv->job.asyncJob, priv->job.phase));
        }
1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197
        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");
        }
1198
    }
1199
    priv->job.active = job;
1200

1201
    if (priv->fakeReboot)
1202
        virBufferAddLit(buf, "<fakereboot/>\n");
1203

1204 1205
    if (priv->qemuDevices && *priv->qemuDevices) {
        char **tmp = priv->qemuDevices;
1206 1207
        virBufferAddLit(buf, "<devices>\n");
        virBufferAdjustIndent(buf, 2);
1208
        while (*tmp) {
1209
            virBufferAsprintf(buf, "<device alias='%s'/>\n", *tmp);
1210 1211
            tmp++;
        }
1212 1213
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</devices>\n");
1214 1215
    }

1216 1217 1218 1219 1220 1221 1222 1223 1224 1225
    if (priv->autoNodeset) {
        char *nodeset = virBitmapFormat(priv->autoNodeset);

        if (!nodeset)
            return -1;

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

1226 1227 1228 1229 1230
    /* Various per-domain paths */
    virBufferEscapeString(buf, "<libDir path='%s'/>\n", priv->libDir);
    virBufferEscapeString(buf, "<channelTargetDir path='%s'/>\n",
                          priv->channelTargetDir);

1231 1232 1233
    return 0;
}

1234
static int
1235
qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt,
1236
                             virDomainObjPtr vm,
1237
                             virDomainDefParserConfigPtr config)
1238
{
1239
    qemuDomainObjPrivatePtr priv = vm->privateData;
1240
    virQEMUDriverPtr driver = config->priv;
1241
    char *monitorpath;
1242
    char *tmp = NULL;
1243 1244
    int n;
    size_t i;
1245
    xmlNodePtr *nodes = NULL;
1246
    virQEMUCapsPtr qemuCaps = NULL;
1247
    virCapsPtr caps = NULL;
1248

1249
    if (VIR_ALLOC(priv->monConfig) < 0)
1250 1251 1252 1253
        goto error;

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
1254 1255
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("no monitor path"));
1256 1257 1258 1259 1260
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
1261
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
1262
    else
1263
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
1264 1265
    VIR_FREE(tmp);

E
Eric Blake 已提交
1266 1267
    priv->monJSON = virXPathBoolean("count(./monitor[@json = '1']) > 0",
                                    ctxt) > 0;
1268

1269
    switch (priv->monConfig->type) {
1270
    case VIR_DOMAIN_CHR_TYPE_PTY:
1271
        priv->monConfig->data.file.path = monitorpath;
1272 1273
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
1274
        priv->monConfig->data.nix.path = monitorpath;
1275 1276 1277
        break;
    default:
        VIR_FREE(monitorpath);
1278 1279 1280
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported monitor type '%s'"),
                       virDomainChrTypeToString(priv->monConfig->type));
1281 1282 1283 1284 1285 1286 1287 1288
        goto error;
    }

    n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes);
    if (n < 0)
        goto error;
    if (n) {
        priv->nvcpupids = n;
1289
        if (VIR_REALLOC_N(priv->vcpupids, priv->nvcpupids) < 0)
1290 1291
            goto error;

1292
        for (i = 0; i < n; i++) {
1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304
            char *pidstr = virXMLPropString(nodes[i], "pid");
            if (!pidstr)
                goto error;

            if (virStrToLong_i(pidstr, NULL, 10, &(priv->vcpupids[i])) < 0) {
                VIR_FREE(pidstr);
                goto error;
            }
            VIR_FREE(pidstr);
        }
        VIR_FREE(nodes);
    }
1305

1306
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
1307 1308
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse qemu capabilities flags"));
1309 1310 1311
        goto error;
    }
    if (n > 0) {
1312
        if (!(qemuCaps = virQEMUCapsNew()))
1313 1314
            goto error;

1315
        for (i = 0; i < n; i++) {
1316 1317
            char *str = virXMLPropString(nodes[i], "name");
            if (str) {
1318
                int flag = virQEMUCapsTypeFromString(str);
1319
                if (flag < 0) {
1320 1321
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Unknown qemu capabilities flag %s"), str);
1322
                    VIR_FREE(str);
1323 1324
                    goto error;
                }
1325
                VIR_FREE(str);
1326
                virQEMUCapsSet(qemuCaps, flag);
1327 1328 1329
            }
        }

1330
        priv->qemuCaps = qemuCaps;
1331
        qemuCaps = NULL;
1332 1333 1334
    }
    VIR_FREE(nodes);

1335
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
1336

1337 1338 1339 1340
    if ((tmp = virXPathString("string(./job[1]/@type)", ctxt))) {
        int type;

        if ((type = qemuDomainJobTypeFromString(tmp)) < 0) {
1341 1342
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown job type %s"), tmp);
1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353
            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) {
1354 1355
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown async job type %s"), tmp);
1356 1357 1358 1359 1360
            VIR_FREE(tmp);
            goto error;
        }
        VIR_FREE(tmp);
        priv->job.asyncJob = async;
J
Jiri Denemark 已提交
1361 1362 1363 1364

        if ((tmp = virXPathString("string(./job[1]/@phase)", ctxt))) {
            priv->job.phase = qemuDomainAsyncJobPhaseFromString(async, tmp);
            if (priv->job.phase < 0) {
1365 1366
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Unknown job phase %s"), tmp);
J
Jiri Denemark 已提交
1367 1368 1369 1370 1371
                VIR_FREE(tmp);
                goto error;
            }
            VIR_FREE(tmp);
        }
1372 1373
    }

1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396
    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);

1397 1398
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419
    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);

1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto error;

    if ((tmp = virXPathString("string(./numad/@nodeset)", ctxt))) {
        if (virBitmapParse(tmp, 0, &priv->autoNodeset,
                           caps->host.nnumaCell_max) < 0)
            goto error;

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

1435 1436 1437 1438 1439 1440 1441 1442 1443
    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;

1444 1445
    return 0;

1446
 error:
1447
    virDomainChrSourceDefFree(priv->monConfig);
1448 1449
    priv->monConfig = NULL;
    VIR_FREE(nodes);
1450
    VIR_FREE(tmp);
1451 1452
    virStringFreeList(priv->qemuDevices);
    priv->qemuDevices = NULL;
1453
    virObjectUnref(qemuCaps);
1454
    virObjectUnref(caps);
1455 1456 1457 1458
    return -1;
}


1459 1460 1461
virDomainXMLPrivateDataCallbacks virQEMUDriverPrivateDataCallbacks = {
    .alloc = qemuDomainObjPrivateAlloc,
    .free = qemuDomainObjPrivateFree,
1462
    .diskNew = qemuDomainDiskPrivateNew,
1463
    .hostdevNew = qemuDomainHostdevPrivateNew,
1464 1465 1466 1467 1468
    .parse = qemuDomainObjPrivateXMLParse,
    .format = qemuDomainObjPrivateXMLFormat,
};


1469 1470 1471 1472 1473
static void
qemuDomainDefNamespaceFree(void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;

1474
    qemuDomainCmdlineDefFree(cmd);
1475 1476 1477
}

static int
P
Philipp Hahn 已提交
1478 1479
qemuDomainDefNamespaceParse(xmlDocPtr xml ATTRIBUTE_UNUSED,
                            xmlNodePtr root ATTRIBUTE_UNUSED,
1480 1481 1482 1483
                            xmlXPathContextPtr ctxt,
                            void **data)
{
    qemuDomainCmdlineDefPtr cmd = NULL;
P
Philipp Hahn 已提交
1484
    bool uses_qemu_ns = false;
1485
    xmlNodePtr *nodes = NULL;
1486 1487
    int n;
    size_t i;
1488

P
Philipp Hahn 已提交
1489
    if (xmlXPathRegisterNs(ctxt, BAD_CAST "qemu", BAD_CAST QEMU_NAMESPACE_HREF) < 0) {
1490 1491 1492
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to register xml namespace '%s'"),
                       QEMU_NAMESPACE_HREF);
1493 1494 1495
        return -1;
    }

1496
    if (VIR_ALLOC(cmd) < 0)
1497 1498 1499 1500 1501 1502
        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
1503
    uses_qemu_ns |= n > 0;
1504 1505

    if (n && VIR_ALLOC_N(cmd->args, n) < 0)
1506
        goto error;
1507 1508 1509 1510

    for (i = 0; i < n; i++) {
        cmd->args[cmd->num_args] = virXMLPropString(nodes[i], "value");
        if (cmd->args[cmd->num_args] == NULL) {
1511 1512
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu command-line argument specified"));
1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523
            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
1524
    uses_qemu_ns |= n > 0;
1525 1526

    if (n && VIR_ALLOC_N(cmd->env_name, n) < 0)
1527
        goto error;
1528 1529

    if (n && VIR_ALLOC_N(cmd->env_value, n) < 0)
1530
        goto error;
1531 1532 1533 1534 1535 1536

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

        tmp = virXMLPropString(nodes[i], "name");
        if (tmp == NULL) {
1537 1538
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu environment name specified"));
1539 1540 1541
            goto error;
        }
        if (tmp[0] == '\0') {
1542 1543
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Empty qemu environment name specified"));
1544 1545 1546
            goto error;
        }
        if (!c_isalpha(tmp[0]) && tmp[0] != '_') {
1547 1548
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must begin with a letter or underscore"));
1549 1550 1551
            goto error;
        }
        if (strspn(tmp, "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789_") != strlen(tmp)) {
1552 1553
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must contain only alphanumerics and underscore"));
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565
            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
1566 1567 1568 1569
    if (uses_qemu_ns)
        *data = cmd;
    else
        VIR_FREE(cmd);
1570 1571 1572

    return 0;

1573
 error:
1574 1575 1576 1577 1578 1579 1580 1581 1582 1583
    VIR_FREE(nodes);
    qemuDomainDefNamespaceFree(cmd);
    return -1;
}

static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;
1584
    size_t i;
1585 1586 1587 1588

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

1589 1590 1591
    virBufferAddLit(buf, "<qemu:commandline>\n");
    virBufferAdjustIndent(buf, 2);

1592
    for (i = 0; i < cmd->num_args; i++)
1593
        virBufferEscapeString(buf, "<qemu:arg value='%s'/>\n",
1594 1595
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
1596
        virBufferAsprintf(buf, "<qemu:env name='%s'", cmd->env_name[i]);
1597 1598 1599 1600 1601
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }

1602 1603
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:commandline>\n");
1604 1605 1606 1607 1608 1609 1610 1611 1612 1613
    return 0;
}

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


1614 1615 1616 1617 1618 1619
virDomainXMLNamespace virQEMUDriverDomainXMLNamespace = {
    .parse = qemuDomainDefNamespaceParse,
    .free = qemuDomainDefNamespaceFree,
    .format = qemuDomainDefNamespaceFormatXML,
    .href = qemuDomainDefNamespaceHref,
};
1620

1621

P
Pavel Hrdina 已提交
1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640
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;
}


1641
static int
1642 1643
qemuDomainDefAddDefaultDevices(virDomainDefPtr def,
                               virQEMUCapsPtr qemuCaps)
1644
{
1645
    bool addDefaultUSB = true;
1646
    int usbModel = -1; /* "default for machinetype" */
1647
    bool addImplicitSATA = false;
1648
    bool addPCIRoot = false;
L
Laine Stump 已提交
1649
    bool addPCIeRoot = false;
1650
    bool addDefaultMemballoon = true;
1651 1652
    bool addDefaultUSBKBD = false;
    bool addDefaultUSBMouse = false;
1653
    bool addPanicDevice = false;
1654
    int ret = -1;
1655

P
Pavel Hrdina 已提交
1656 1657 1658 1659
    /* add implicit input devices */
    if (qemuDomainDefAddImplicitInputDevice(def) < 0)
        goto cleanup;

1660 1661 1662 1663 1664 1665
    /* Add implicit PCI root controller if the machine has one */
    switch (def->os.arch) {
    case VIR_ARCH_I686:
    case VIR_ARCH_X86_64:
        if (!def->os.machine)
            break;
L
Laine Stump 已提交
1666
        if (STREQ(def->os.machine, "isapc")) {
1667
            addDefaultUSB = false;
1668
            break;
1669
        }
L
Laine Stump 已提交
1670 1671
        if (STRPREFIX(def->os.machine, "pc-q35") ||
            STREQ(def->os.machine, "q35")) {
1672 1673
            addPCIeRoot = true;
            addImplicitSATA = true;
1674 1675 1676 1677 1678 1679 1680 1681

            /* add a USB2 controller set, but only if the
             * ich9-usb-ehci1 device is supported
             */
            if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_ICH9_USB_EHCI1))
                usbModel = VIR_DOMAIN_CONTROLLER_MODEL_USB_ICH9_EHCI1;
            else
                addDefaultUSB = false;
1682
            break;
L
Laine Stump 已提交
1683
        }
1684 1685
        if (!STRPREFIX(def->os.machine, "pc-0.") &&
            !STRPREFIX(def->os.machine, "pc-1.") &&
1686
            !STRPREFIX(def->os.machine, "pc-i440") &&
1687
            STRNEQ(def->os.machine, "pc") &&
1688 1689 1690 1691 1692
            !STRPREFIX(def->os.machine, "rhel"))
            break;
        addPCIRoot = true;
        break;

1693
    case VIR_ARCH_ARMV7L:
1694
    case VIR_ARCH_AARCH64:
1695 1696 1697 1698 1699 1700 1701
        addDefaultUSB = false;
        addDefaultMemballoon = false;
        if (STREQ(def->os.machine, "virt") ||
            STRPREFIX(def->os.machine, "virt-")) {
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
        }
        break;
1702

1703
    case VIR_ARCH_PPC64:
1704
    case VIR_ARCH_PPC64LE:
1705 1706 1707
        addPCIRoot = true;
        addDefaultUSBKBD = true;
        addDefaultUSBMouse = true;
1708 1709 1710 1711 1712
        /* For pSeries guests, the firmware provides the same
         * functionality as the pvpanic device, so automatically
         * add the definition if not already present */
        if (STRPREFIX(def->os.machine, "pseries"))
            addPanicDevice = true;
1713 1714
        break;

1715 1716 1717 1718 1719 1720 1721
    case VIR_ARCH_ALPHA:
    case VIR_ARCH_PPC:
    case VIR_ARCH_PPCEMB:
    case VIR_ARCH_SH4:
    case VIR_ARCH_SH4EB:
        addPCIRoot = true;
        break;
1722 1723 1724 1725 1726 1727
    case VIR_ARCH_S390:
        addDefaultUSB = false;
        break;
    case VIR_ARCH_S390X:
        addDefaultUSB = false;
        break;
1728 1729 1730 1731 1732 1733

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

1734 1735 1736 1737
    default:
        break;
    }

1738
    if (addDefaultUSB &&
1739 1740
        virDomainControllerFind(def, VIR_DOMAIN_CONTROLLER_TYPE_USB, 0) < 0 &&
        virDomainDefAddUSBController(def, 0, usbModel) < 0)
1741
        goto cleanup;
1742

1743 1744 1745
    if (addImplicitSATA &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_SATA, 0, -1) < 0)
1746
        goto cleanup;
1747

1748 1749 1750
    /* NB: any machine that sets addPCIRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
     */
1751 1752 1753 1754
    if (addPCIRoot &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
            VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT) < 0)
1755
        goto cleanup;
1756

1757 1758 1759
    /* 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
1760 1761 1762
     *
     * NB: any machine that sets addPCIeRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773
     */
    if (addPCIeRoot) {
        if (virDomainDefMaybeAddController(
                def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
                VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT) < 0 ||
            virDomainDefMaybeAddController(
                def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 1,
                VIR_DOMAIN_CONTROLLER_MODEL_DMI_TO_PCI_BRIDGE) < 0 ||
            virDomainDefMaybeAddController(
                def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 2,
                VIR_DOMAIN_CONTROLLER_MODEL_PCI_BRIDGE) < 0) {
1774
            goto cleanup;
1775 1776
        }
    }
1777

1778
    if (addDefaultMemballoon && !def->memballoon) {
1779 1780
        virDomainMemballoonDefPtr memballoon;
        if (VIR_ALLOC(memballoon) < 0)
1781
            goto cleanup;
1782 1783 1784 1785 1786

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

1787 1788 1789 1790 1791
    if (addDefaultUSBKBD &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_KBD,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
1792
        goto cleanup;
1793 1794 1795 1796 1797 1798

    if (addDefaultUSBMouse &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
1799
        goto cleanup;
1800

D
Dmitry Andreev 已提交
1801 1802 1803 1804 1805 1806 1807
    if (addPanicDevice) {
        size_t j;
        for (j = 0; j < def->npanics; j++) {
            if (def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT ||
                def->panics[j]->model == VIR_DOMAIN_PANIC_MODEL_PSERIES)
                break;
        }
1808

D
Dmitry Andreev 已提交
1809 1810 1811 1812 1813 1814 1815 1816 1817
        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;
            }
        }
1818 1819
    }

1820 1821 1822 1823 1824 1825
    ret = 0;
 cleanup:
    return ret;
}


A
Andrea Bolognani 已提交
1826 1827 1828 1829 1830 1831 1832 1833 1834 1835
/**
 * qemuDomainDefEnableDefaultFeatures:
 * @def: domain definition
 *
 * Make sure that features that should be enabled by default are actually
 * enabled and configure default values related to those features.
 */
static void
qemuDomainDefEnableDefaultFeatures(virDomainDefPtr def)
{
1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849
    switch (def->os.arch) {
    case VIR_ARCH_ARMV7L:
    case VIR_ARCH_AARCH64:
        if (STREQ(def->os.machine, "virt") ||
            STRPREFIX(def->os.machine, "virt-")) {
            /* GIC is always available to ARM virt machines */
            def->features[VIR_DOMAIN_FEATURE_GIC] = VIR_TRISTATE_SWITCH_ON;
        }
        break;

    default:
        break;
    }

1850
    /* Use the default GIC version if no version was specified */
A
Andrea Bolognani 已提交
1851 1852
    if (def->features[VIR_DOMAIN_FEATURE_GIC] == VIR_TRISTATE_SWITCH_ON &&
        def->gic_version == VIR_GIC_VERSION_NONE)
1853
        def->gic_version = VIR_GIC_VERSION_DEFAULT;
A
Andrea Bolognani 已提交
1854 1855 1856
}


1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876
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;
}


1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898
static void
qemuDomainRecheckInternalPaths(virDomainDefPtr def,
                               virQEMUDriverConfigPtr cfg,
                               unsigned int flags)
{
    size_t i = 0;

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

        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
            graphics->data.vnc.socket &&
            STRPREFIX(graphics->data.vnc.socket, cfg->libDir)) {
            if (flags & VIR_DOMAIN_DEF_PARSE_INACTIVE)
                VIR_FREE(graphics->data.vnc.socket);
            else
                graphics->data.vnc.socketAutogenerated = true;
        }
    }
}


1899 1900 1901
static int
qemuDomainDefPostParse(virDomainDefPtr def,
                       virCapsPtr caps,
1902
                       unsigned int parseFlags,
1903 1904 1905
                       void *opaque)
{
    virQEMUDriverPtr driver = opaque;
1906
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1907 1908 1909 1910 1911 1912 1913 1914 1915
    virQEMUCapsPtr qemuCaps = NULL;
    int ret = -1;

    if (def->os.bootloader || def->os.bootloaderArgs) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bootloader is not supported by QEMU"));
        return ret;
    }

1916 1917 1918 1919 1920 1921
    if (!def->os.machine) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("missing machine type"));
        return ret;
    }

1922 1923 1924 1925 1926 1927 1928 1929 1930
    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;
    }

1931 1932 1933 1934 1935
    /* check for emulator and create a default one if needed */
    if (!def->emulator &&
        !(def->emulator = virDomainDefGetDefaultEmulator(def, caps)))
        return ret;

1936 1937 1938
    if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache,
                                            def->emulator)))
        goto cleanup;
1939 1940 1941 1942

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

1943 1944 1945
    if (qemuCanonicalizeMachine(def, qemuCaps) < 0)
        goto cleanup;

A
Andrea Bolognani 已提交
1946 1947
    qemuDomainDefEnableDefaultFeatures(def);

1948 1949
    qemuDomainRecheckInternalPaths(def, cfg, parseFlags);

1950 1951 1952
    if (virSecurityManagerVerify(driver->securityManager, def) < 0)
        goto cleanup;

1953 1954 1955
    ret = 0;
 cleanup:
    virObjectUnref(qemuCaps);
1956
    virObjectUnref(cfg);
1957
    return ret;
1958 1959
}

1960
static const char *
1961 1962
qemuDomainDefaultNetModel(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
1963
{
1964
    if (ARCH_IS_S390(def->os.arch))
1965 1966
        return "virtio";

1967 1968
    if (def->os.arch == VIR_ARCH_ARMV7L ||
        def->os.arch == VIR_ARCH_AARCH64) {
1969 1970 1971
        if (STREQ(def->os.machine, "versatilepb"))
            return "smc91c111";

1972 1973
        if (STREQ(def->os.machine, "virt") ||
            STRPREFIX(def->os.machine, "virt-"))
1974 1975
            return "virtio";

1976 1977 1978 1979 1980
        /* Incomplete. vexpress (and a few others) use this, but not all
         * arm boards */
        return "lan9118";
    }

1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992
    /* 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 */
1993 1994
    return "rtl8139";
}
1995

1996 1997
static int
qemuDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
1998
                             const virDomainDef *def,
1999
                             virCapsPtr caps ATTRIBUTE_UNUSED,
2000
                             unsigned int parseFlags,
2001
                             void *opaque)
2002
{
2003
    virQEMUDriverPtr driver = opaque;
2004
    virQEMUCapsPtr qemuCaps = NULL;
2005
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2006
    int ret = -1;
2007

P
Pavel Fedin 已提交
2008
    qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache, def->emulator);
2009

2010
    if (dev->type == VIR_DOMAIN_DEVICE_NET &&
2011 2012
        dev->data.net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
        !dev->data.net->model) {
2013
        if (VIR_STRDUP(dev->data.net->model,
2014
                       qemuDomainDefaultNetModel(def, qemuCaps)) < 0)
2015
            goto cleanup;
2016
    }
2017

2018 2019 2020 2021
    /* set default disk types and drivers */
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
        virDomainDiskDefPtr disk = dev->data.disk;

2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049
        /* 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;
2050 2051 2052
        }
    }

2053 2054 2055 2056
    /* 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 &&
2057
        ARCH_IS_S390(def->os.arch))
2058 2059
        dev->data.chr->targetType = VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_VIRTIO;

2060 2061 2062 2063 2064
    /* 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 &&
2065
        ARCH_IS_S390(def->os.arch))
2066 2067
        dev->data.controller->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NONE;

2068 2069 2070 2071 2072 2073 2074
    /* 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;

2075 2076 2077
    /* clear auto generated unix socket path for inactive definitions */
    if ((parseFlags & VIR_DOMAIN_DEF_PARSE_INACTIVE) &&
        dev->type == VIR_DOMAIN_DEVICE_CHR &&
2078 2079 2080
        dev->data.chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_CHANNEL &&
        dev->data.chr->targetType == VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO &&
        dev->data.chr->source.type == VIR_DOMAIN_CHR_TYPE_UNIX &&
2081 2082 2083 2084 2085 2086
        dev->data.chr->source.data.nix.path &&
        STRPREFIX(dev->data.chr->source.data.nix.path, cfg->channelTargetDir)) {
        /*
         * If the address is generated by us (starts with our
         * channel dir), we should not keep it in the persistent
         * XML.  If libvirt is the one who generated it, users
2087
         * shouldn't care about that.  If they do, they are
2088 2089 2090
         * supposed to set it themselves.
         */
        VIR_FREE(dev->data.chr->source.data.nix.path);
2091 2092
    }

2093 2094 2095 2096 2097 2098 2099 2100 2101 2102
    /* 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;
    }

2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117
    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 {
2118
            dev->data.video->vgamem = QEMU_QXL_VGAMEM_DEFAULT;
2119 2120 2121
        }
    }

2122 2123 2124 2125 2126 2127 2128 2129 2130
    if (dev->type == VIR_DOMAIN_DEVICE_PANIC &&
        dev->data.panic->model == VIR_DOMAIN_PANIC_MODEL_DEFAULT) {
        if (ARCH_IS_PPC64(def->os.arch) &&
            STRPREFIX(def->os.machine, "pseries"))
            dev->data.panic->model = VIR_DOMAIN_PANIC_MODEL_PSERIES;
        else
            dev->data.panic->model = VIR_DOMAIN_PANIC_MODEL_ISA;
    }

2131 2132 2133 2134

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

2135 2136 2137
        if (cont->type == VIR_DOMAIN_CONTROLLER_TYPE_PCI) {
            if (cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS &&
                !qemuDomainMachineIsI440FX(def)) {
2138 2139 2140 2141 2142
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("pci-expander-bus controllers are only supported "
                                 "on 440fx-based machinetypes"));
                goto cleanup;
            }
2143 2144 2145 2146 2147 2148 2149
            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;
            }
2150 2151 2152 2153 2154 2155

            /* 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.
             */
2156 2157 2158
            if ((cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCI_EXPANDER_BUS ||
                 cont->model == VIR_DOMAIN_CONTROLLER_MODEL_PCIE_EXPANDER_BUS) &&
                (int) virDomainNumaGetNodeCount(def->numa)
2159 2160
                <= cont->opts.pciopts.numaNode) {
                virReportError(VIR_ERR_XML_ERROR,
2161
                               _("%s with index %d is "
2162 2163 2164
                                 "configured for a NUMA node (%d) "
                                 "not present in the domain's "
                                 "<cpu><numa> array (%zu)"),
2165
                               virDomainControllerModelPCITypeToString(cont->model),
2166 2167 2168 2169 2170 2171 2172
                               cont->idx, cont->opts.pciopts.numaNode,
                               virDomainNumaGetNodeCount(def->numa));
                goto cleanup;
            }
        }
    }

2173 2174
    ret = 0;

2175
 cleanup:
2176
    virObjectUnref(qemuCaps);
2177 2178
    virObjectUnref(cfg);
    return ret;
2179 2180 2181 2182 2183
}


virDomainDefParserConfig virQEMUDriverDomainDefParserConfig = {
    .devicesPostParseCallback = qemuDomainDeviceDefPostParse,
2184
    .domainPostParseCallback = qemuDomainDefPostParse,
2185 2186
    .features = VIR_DOMAIN_DEF_FEATURE_MEMORY_HOTPLUG |
                VIR_DOMAIN_DEF_FEATURE_OFFLINE_VCPUPIN
2187 2188 2189
};


2190
static void
2191
qemuDomainObjSaveJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
2192
{
2193 2194 2195
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    if (virDomainObjIsActive(obj)) {
2196
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj, driver->caps) < 0)
2197
            VIR_WARN("Failed to save status on vm %s", obj->def->name);
2198
    }
2199

2200
    virObjectUnref(cfg);
2201 2202
}

J
Jiri Denemark 已提交
2203
void
2204
qemuDomainObjSetJobPhase(virQEMUDriverPtr driver,
J
Jiri Denemark 已提交
2205 2206 2207 2208
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
2209
    unsigned long long me = virThreadSelfID();
J
Jiri Denemark 已提交
2210 2211 2212 2213

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

2214 2215 2216 2217 2218
    VIR_DEBUG("Setting '%s' phase to '%s'",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob, phase));

    if (priv->job.asyncOwner && me != priv->job.asyncOwner) {
2219
        VIR_WARN("'%s' async job is owned by thread %llu",
2220 2221 2222 2223
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }

J
Jiri Denemark 已提交
2224
    priv->job.phase = phase;
2225
    priv->job.asyncOwner = me;
J
Jiri Denemark 已提交
2226 2227 2228
    qemuDomainObjSaveJob(driver, obj);
}

2229
void
2230 2231
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
2232 2233 2234
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

2235 2236 2237 2238 2239 2240 2241
    if (!priv->job.asyncJob)
        return;

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

void
2242
qemuDomainObjDiscardAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
2243 2244 2245 2246 2247 2248
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
2249
    qemuDomainObjSaveJob(driver, obj);
2250 2251
}

2252 2253 2254 2255 2256 2257 2258 2259 2260
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()) {
2261
        VIR_WARN("'%s' async job is owned by thread %llu",
2262 2263 2264 2265 2266 2267
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }
    priv->job.asyncOwner = 0;
}

2268
static bool
2269
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
2270 2271
{
    return !priv->job.asyncJob || (priv->job.mask & JOB_MASK(job)) != 0;
2272 2273
}

2274
bool
2275
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
2276 2277 2278 2279
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

2280 2281 2282
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

2283
/*
2284
 * obj must be locked before calling
2285
 */
2286
static int ATTRIBUTE_NONNULL(1)
2287
qemuDomainObjBeginJobInternal(virQEMUDriverPtr driver,
2288
                              virDomainObjPtr obj,
2289 2290
                              qemuDomainJob job,
                              qemuDomainAsyncJob asyncJob)
2291 2292
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
2293
    unsigned long long now;
2294
    unsigned long long then;
2295
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
J
Jiri Denemark 已提交
2296
    bool async = job == QEMU_JOB_ASYNC;
2297
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2298
    const char *blocker = NULL;
J
Jiri Denemark 已提交
2299
    int ret = -1;
J
Jiri Denemark 已提交
2300 2301
    unsigned long long duration = 0;
    unsigned long long asyncDuration = 0;
J
Jiri Denemark 已提交
2302
    const char *jobStr;
2303

J
Jiri Denemark 已提交
2304 2305 2306 2307 2308 2309 2310 2311 2312
    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));
2313

2314 2315
    if (virTimeMillisNow(&now) < 0) {
        virObjectUnref(cfg);
2316
        return -1;
2317 2318
    }

2319
    priv->jobs_queued++;
J
Jiri Denemark 已提交
2320
    then = now + QEMU_JOB_WAIT_TIME;
2321

2322
 retry:
2323 2324
    if (cfg->maxQueuedJobs &&
        priv->jobs_queued > cfg->maxQueuedJobs) {
2325 2326 2327
        goto error;
    }

2328
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
2329
        VIR_DEBUG("Waiting for async job (vm=%p name=%s)", obj, obj->def->name);
2330
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->parent.lock, then) < 0)
2331 2332 2333
            goto error;
    }

2334
    while (priv->job.active) {
2335
        VIR_DEBUG("Waiting for job (vm=%p name=%s)", obj, obj->def->name);
2336
        if (virCondWaitUntil(&priv->job.cond, &obj->parent.lock, then) < 0)
2337
            goto error;
2338
    }
2339 2340 2341

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
2342
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
2343 2344
        goto retry;

2345
    qemuDomainObjResetJob(priv);
2346

J
Jiri Denemark 已提交
2347 2348
    ignore_value(virTimeMillisNow(&now));

2349
    if (job != QEMU_JOB_ASYNC) {
2350
        VIR_DEBUG("Started job: %s (async=%s vm=%p name=%s)",
2351
                   qemuDomainJobTypeToString(job),
2352 2353
                  qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                  obj, obj->def->name);
2354
        priv->job.active = job;
2355
        priv->job.owner = virThreadSelfID();
2356
        priv->job.ownerAPI = virThreadJobGet();
J
Jiri Denemark 已提交
2357
        priv->job.started = now;
2358
    } else {
2359 2360 2361
        VIR_DEBUG("Started async job: %s (vm=%p name=%s)",
                  qemuDomainAsyncJobTypeToString(asyncJob),
                  obj, obj->def->name);
2362
        qemuDomainObjResetAsyncJob(priv);
J
Jiri Denemark 已提交
2363 2364
        if (VIR_ALLOC(priv->job.current) < 0)
            goto cleanup;
2365
        priv->job.asyncJob = asyncJob;
2366
        priv->job.asyncOwner = virThreadSelfID();
2367
        priv->job.asyncOwnerAPI = virThreadJobGet();
J
Jiri Denemark 已提交
2368
        priv->job.asyncStarted = now;
J
Jiri Denemark 已提交
2369
        priv->job.current->started = now;
2370
    }
2371

2372 2373
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
2374

2375
    virObjectUnref(cfg);
2376
    return 0;
2377

2378
 error:
J
Jiri Denemark 已提交
2379 2380 2381 2382 2383 2384
    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;

2385
    VIR_WARN("Cannot start job (%s, %s) for domain %s; "
J
Jiri Denemark 已提交
2386 2387
             "current job is (%s, %s) owned by (%llu %s, %llu %s) "
             "for (%llus, %llus)",
2388 2389 2390 2391 2392
             qemuDomainJobTypeToString(job),
             qemuDomainAsyncJobTypeToString(asyncJob),
             obj->def->name,
             qemuDomainJobTypeToString(priv->job.active),
             qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
2393
             priv->job.owner, NULLSTR(priv->job.ownerAPI),
J
Jiri Denemark 已提交
2394 2395
             priv->job.asyncOwner, NULLSTR(priv->job.asyncOwnerAPI),
             duration / 1000, asyncDuration / 1000);
2396 2397 2398 2399 2400

    if (nested || qemuDomainNestedJobAllowed(priv, job))
        blocker = priv->job.ownerAPI;
    else
        blocker = priv->job.asyncOwnerAPI;
2401

2402 2403
    ret = -1;
    if (errno == ETIMEDOUT) {
2404 2405 2406 2407 2408 2409 2410 2411
        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"));
        }
2412 2413 2414
        ret = -2;
    } else if (cfg->maxQueuedJobs &&
               priv->jobs_queued > cfg->maxQueuedJobs) {
2415 2416 2417 2418 2419 2420 2421 2422 2423 2424
        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"));
        }
2425 2426
        ret = -2;
    } else {
2427
        virReportSystemError(errno, "%s", _("cannot acquire job mutex"));
2428
    }
J
Jiri Denemark 已提交
2429 2430

 cleanup:
2431
    priv->jobs_queued--;
2432
    virObjectUnref(cfg);
2433
    return ret;
2434 2435 2436
}

/*
2437
 * obj must be locked before calling
2438 2439 2440 2441
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
2442
 * Successful calls must be followed by EndJob eventually
2443
 */
2444
int qemuDomainObjBeginJob(virQEMUDriverPtr driver,
2445
                          virDomainObjPtr obj,
2446
                          qemuDomainJob job)
2447
{
2448 2449 2450 2451 2452
    if (qemuDomainObjBeginJobInternal(driver, obj, job,
                                      QEMU_ASYNC_JOB_NONE) < 0)
        return -1;
    else
        return 0;
2453 2454
}

2455
int qemuDomainObjBeginAsyncJob(virQEMUDriverPtr driver,
2456
                               virDomainObjPtr obj,
2457
                               qemuDomainAsyncJob asyncJob)
2458
{
2459 2460 2461 2462 2463
    if (qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_ASYNC,
                                      asyncJob) < 0)
        return -1;
    else
        return 0;
2464 2465
}

2466
int
2467 2468
qemuDomainObjBeginNestedJob(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
2469
                            qemuDomainAsyncJob asyncJob)
2470 2471 2472 2473 2474 2475 2476 2477 2478 2479
{
    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()) {
2480
        VIR_WARN("This thread doesn't seem to be the async job owner: %llu",
2481 2482 2483 2484 2485 2486 2487 2488
                 priv->job.asyncOwner);
    }

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

2489

2490
/*
2491
 * obj must be locked and have a reference before calling
2492 2493 2494 2495
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 */
2496 2497
void
qemuDomainObjEndJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
2498 2499
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
2500
    qemuDomainJob job = priv->job.active;
2501

2502 2503
    priv->jobs_queued--;

2504
    VIR_DEBUG("Stopping job: %s (async=%s vm=%p name=%s)",
2505
              qemuDomainJobTypeToString(job),
2506 2507
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
2508

2509
    qemuDomainObjResetJob(priv);
2510 2511
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
2512
    virCondSignal(&priv->job.cond);
2513 2514
}

2515
void
2516
qemuDomainObjEndAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
2517 2518
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
2519

2520 2521
    priv->jobs_queued--;

2522 2523 2524
    VIR_DEBUG("Stopping async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
2525

2526
    qemuDomainObjResetAsyncJob(priv);
2527
    qemuDomainObjSaveJob(driver, obj);
2528 2529 2530
    virCondBroadcast(&priv->job.asyncCond);
}

2531 2532 2533 2534 2535
void
qemuDomainObjAbortAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

2536 2537 2538
    VIR_DEBUG("Requesting abort of async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
2539

2540 2541
    priv->job.abortJob = true;
    virDomainObjBroadcast(obj);
2542 2543
}

2544 2545 2546 2547 2548 2549 2550 2551 2552
/*
 * 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
 */
2553
static int
2554
qemuDomainObjEnterMonitorInternal(virQEMUDriverPtr driver,
2555
                                  virDomainObjPtr obj,
2556
                                  qemuDomainAsyncJob asyncJob)
2557 2558 2559
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

2560
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
2561 2562 2563
        int ret;
        if ((ret = qemuDomainObjBeginNestedJob(driver, obj, asyncJob)) < 0)
            return ret;
2564
        if (!virDomainObjIsActive(obj)) {
2565 2566
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
2567
            qemuDomainObjEndJob(driver, obj);
2568 2569
            return -1;
        }
2570 2571 2572
    } 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");
2573 2574
    }

2575 2576
    VIR_DEBUG("Entering monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
2577
    virObjectLock(priv->mon);
2578
    virObjectRef(priv->mon);
2579
    ignore_value(virTimeMillisNow(&priv->monStart));
2580
    virObjectUnlock(obj);
2581 2582

    return 0;
2583 2584
}

2585
static void ATTRIBUTE_NONNULL(1)
2586
qemuDomainObjExitMonitorInternal(virQEMUDriverPtr driver,
2587
                                 virDomainObjPtr obj)
2588 2589
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
2590
    bool hasRefs;
2591

2592
    hasRefs = virObjectUnref(priv->mon);
2593

2594
    if (hasRefs)
2595
        virObjectUnlock(priv->mon);
2596

2597
    virObjectLock(obj);
2598 2599
    VIR_DEBUG("Exited monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
2600

2601
    priv->monStart = 0;
2602
    if (!hasRefs)
2603
        priv->mon = NULL;
2604

J
Jiri Denemark 已提交
2605 2606
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjEndJob(driver, obj);
2607 2608
}

2609
void qemuDomainObjEnterMonitor(virQEMUDriverPtr driver,
2610
                               virDomainObjPtr obj)
2611
{
2612
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, obj,
2613
                                                   QEMU_ASYNC_JOB_NONE));
2614 2615
}

2616
/* obj must NOT be locked before calling
2617 2618
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
2619 2620 2621 2622 2623 2624
 *
 * 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.
2625
 */
2626 2627
int qemuDomainObjExitMonitor(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
2628
{
2629
    qemuDomainObjExitMonitorInternal(driver, obj);
2630
    if (!virDomainObjIsActive(obj)) {
2631 2632 2633
        if (!virGetLastError())
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
2634 2635 2636
        return -1;
    }
    return 0;
2637
}
2638 2639

/*
2640
 * obj must be locked before calling
2641 2642
 *
 * To be called immediately before any QEMU monitor API call.
2643
 * Must have already either called qemuDomainObjBeginJob()
2644 2645 2646 2647 2648
 * 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
2649 2650 2651
 * 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).
2652 2653
 */
int
2654
qemuDomainObjEnterMonitorAsync(virQEMUDriverPtr driver,
2655
                               virDomainObjPtr obj,
2656
                               qemuDomainAsyncJob asyncJob)
2657
{
2658
    return qemuDomainObjEnterMonitorInternal(driver, obj, asyncJob);
2659 2660
}

D
Daniel P. Berrange 已提交
2661

2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673
/**
 * qemuDomainGetAgent:
 * @vm: domain object
 *
 * Returns the agent pointer of @vm;
 */
qemuAgentPtr
qemuDomainGetAgent(virDomainObjPtr vm)
{
    return (((qemuDomainObjPrivatePtr)(vm->privateData))->agent);
}

D
Daniel P. Berrange 已提交
2674

2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685
/*
 * 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
 */
void
qemuDomainObjEnterAgent(virDomainObjPtr obj)
D
Daniel P. Berrange 已提交
2686 2687 2688
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

2689 2690
    VIR_DEBUG("Entering agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
2691
    virObjectLock(priv->agent);
2692
    virObjectRef(priv->agent);
D
Daniel P. Berrange 已提交
2693
    ignore_value(virTimeMillisNow(&priv->agentStart));
2694
    virObjectUnlock(obj);
D
Daniel P. Berrange 已提交
2695 2696
}

2697 2698 2699 2700 2701 2702 2703

/* obj must NOT be locked before calling
 *
 * Should be paired with an earlier qemuDomainObjEnterAgent() call
 */
void
qemuDomainObjExitAgent(virDomainObjPtr obj)
D
Daniel P. Berrange 已提交
2704 2705
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
2706
    bool hasRefs;
D
Daniel P. Berrange 已提交
2707

2708
    hasRefs = virObjectUnref(priv->agent);
D
Daniel P. Berrange 已提交
2709

2710
    if (hasRefs)
2711
        virObjectUnlock(priv->agent);
D
Daniel P. Berrange 已提交
2712

2713
    virObjectLock(obj);
2714 2715
    VIR_DEBUG("Exited agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
D
Daniel P. Berrange 已提交
2716 2717

    priv->agentStart = 0;
2718
    if (!hasRefs)
D
Daniel P. Berrange 已提交
2719 2720 2721
        priv->agent = NULL;
}

2722
void qemuDomainObjEnterRemote(virDomainObjPtr obj)
2723
{
2724 2725
    VIR_DEBUG("Entering remote (vm=%p name=%s)",
              obj, obj->def->name);
2726
    virObjectUnlock(obj);
2727 2728
}

2729
void qemuDomainObjExitRemote(virDomainObjPtr obj)
2730
{
2731
    virObjectLock(obj);
2732 2733
    VIR_DEBUG("Exited remote (vm=%p name=%s)",
              obj, obj->def->name);
2734
}
2735 2736


2737 2738 2739 2740 2741 2742 2743 2744
virDomainDefPtr
qemuDomainDefCopy(virQEMUDriverPtr driver,
                  virDomainDefPtr src,
                  unsigned int flags)
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virDomainDefPtr ret = NULL;
    virCapsPtr caps = NULL;
2745
    char *xml = NULL;
2746 2747 2748 2749 2750 2751 2752 2753 2754 2755

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

    xml = virBufferContentAndReset(&buf);

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

    if (!(ret = virDomainDefParseString(xml, caps, driver->xmlopt,
2756
                                        VIR_DOMAIN_DEF_PARSE_INACTIVE)))
2757 2758
        goto cleanup;

2759
 cleanup:
2760 2761 2762 2763 2764
    VIR_FREE(xml);
    virObjectUnref(caps);
    return ret;
}

2765
int
2766
qemuDomainDefFormatBuf(virQEMUDriverPtr driver,
2767 2768 2769
                       virDomainDefPtr def,
                       unsigned int flags,
                       virBuffer *buf)
2770
{
2771
    int ret = -1;
2772
    virCPUDefPtr cpu = NULL;
2773
    virCPUDefPtr def_cpu = def->cpu;
2774 2775
    virDomainControllerDefPtr *controllers = NULL;
    int ncontrollers = 0;
2776 2777 2778 2779
    virCapsPtr caps = NULL;

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;
2780 2781

    /* Update guest CPU requirements according to host CPU */
2782 2783 2784
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) &&
        def_cpu &&
        (def_cpu->mode != VIR_CPU_MODE_CUSTOM || def_cpu->model)) {
2785 2786
        if (!caps->host.cpu ||
            !caps->host.cpu->model) {
2787 2788
            virReportError(VIR_ERR_OPERATION_FAILED,
                           "%s", _("cannot get host CPU capabilities"));
2789 2790 2791
            goto cleanup;
        }

2792
        if (!(cpu = virCPUDefCopy(def_cpu)) ||
2793
            cpuUpdate(cpu, caps->host.cpu) < 0)
2794 2795 2796 2797
            goto cleanup;
        def->cpu = cpu;
    }

2798
    if ((flags & VIR_DOMAIN_XML_MIGRATABLE)) {
2799
        size_t i;
2800
        int toremove = 0;
2801
        virDomainControllerDefPtr usb = NULL, pci = NULL;
2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819

        /* 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];
            }
        }
        if (usb && usb->idx == 0 && usb->model == -1) {
            VIR_DEBUG("Removing default USB controller from domain '%s'"
                      " for migration compatibility", def->name);
2820
            toremove++;
2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838
        } 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
2839
            VIR_DEBUG("Removing default pci-root from domain '%s'"
2840
                      " for migration compatibility", def->name);
2841
            toremove++;
2842 2843 2844 2845
        } else {
            pci = NULL;
        }

2846
        if (toremove) {
2847 2848
            controllers = def->controllers;
            ncontrollers = def->ncontrollers;
2849
            if (VIR_ALLOC_N(def->controllers, ncontrollers - toremove) < 0) {
2850 2851 2852 2853 2854 2855
                controllers = NULL;
                goto cleanup;
            }

            def->ncontrollers = 0;
            for (i = 0; i < ncontrollers; i++) {
2856
                if (controllers[i] != usb && controllers[i] != pci)
2857 2858 2859
                    def->controllers[def->ncontrollers++] = controllers[i];
            }
        }
2860 2861


2862 2863
    }

2864
    ret = virDomainDefFormatInternal(def, driver->caps,
2865 2866
                                     virDomainDefFormatConvertXMLFlags(flags),
                                     buf);
2867

2868
 cleanup:
2869 2870
    def->cpu = def_cpu;
    virCPUDefFree(cpu);
2871 2872 2873 2874 2875
    if (controllers) {
        VIR_FREE(def->controllers);
        def->controllers = controllers;
        def->ncontrollers = ncontrollers;
    }
2876
    virObjectUnref(caps);
2877 2878
    return ret;
}
2879

2880
char *qemuDomainDefFormatXML(virQEMUDriverPtr driver,
2881
                             virDomainDefPtr def,
2882
                             unsigned int flags)
2883 2884 2885
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

2886
    if (qemuDomainDefFormatBuf(driver, def, flags, &buf) < 0) {
2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899
        virBufferFreeAndReset(&buf);
        return NULL;
    }

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

    return virBufferContentAndReset(&buf);
}

2900
char *qemuDomainFormatXML(virQEMUDriverPtr driver,
2901
                          virDomainObjPtr vm,
2902
                          unsigned int flags)
2903 2904 2905 2906 2907 2908 2909 2910
{
    virDomainDefPtr def;

    if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef)
        def = vm->newDef;
    else
        def = vm->def;

2911
    return qemuDomainDefFormatXML(driver, def, flags);
2912 2913
}

2914
char *
2915
qemuDomainDefFormatLive(virQEMUDriverPtr driver,
2916
                        virDomainDefPtr def,
2917 2918
                        bool inactive,
                        bool compatible)
2919 2920 2921 2922 2923
{
    unsigned int flags = QEMU_DOMAIN_FORMAT_LIVE_FLAGS;

    if (inactive)
        flags |= VIR_DOMAIN_XML_INACTIVE;
2924 2925
    if (compatible)
        flags |= VIR_DOMAIN_XML_MIGRATABLE;
2926

2927
    return qemuDomainDefFormatXML(driver, def, flags);
2928 2929
}

2930

2931
void qemuDomainObjTaint(virQEMUDriverPtr driver,
2932
                        virDomainObjPtr obj,
2933
                        virDomainTaintFlags taint,
2934
                        qemuDomainLogContextPtr logCtxt)
2935
{
2936
    virErrorPtr orig_err = NULL;
2937
    bool closeLog = false;
2938

2939 2940 2941 2942 2943 2944 2945 2946 2947
    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));
2948 2949 2950 2951 2952

        /* We don't care about errors logging taint info, so
         * preserve original error, and clear any error that
         * is raised */
        orig_err = virSaveLastError();
2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970
        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)
2971
            virResetLastError();
2972 2973
        if (closeLog)
            qemuDomainLogContextFree(logCtxt);
2974 2975 2976 2977
        if (orig_err) {
            virSetError(orig_err);
            virFreeError(orig_err);
        }
2978 2979 2980 2981
    }
}


2982
void qemuDomainObjCheckTaint(virQEMUDriverPtr driver,
2983
                             virDomainObjPtr obj,
2984
                             qemuDomainLogContextPtr logCtxt)
2985
{
2986
    size_t i;
2987
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2988
    qemuDomainObjPrivatePtr priv = obj->privateData;
2989

2990
    if (virQEMUDriverIsPrivileged(driver) &&
2991 2992 2993
        (!cfg->clearEmulatorCapabilities ||
         cfg->user == 0 ||
         cfg->group == 0))
2994
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
2995

2996
    if (priv->hookRun)
2997
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOOK, logCtxt);
2998

2999 3000 3001
    if (obj->def->namespaceData) {
        qemuDomainCmdlineDefPtr qemucmd = obj->def->namespaceData;
        if (qemucmd->num_args || qemucmd->num_env)
3002
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logCtxt);
3003 3004
    }

3005
    if (obj->def->cpu && obj->def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
3006
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOST_CPU, logCtxt);
3007

3008
    for (i = 0; i < obj->def->ndisks; i++)
3009
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logCtxt);
3010

3011 3012
    for (i = 0; i < obj->def->nhostdevs; i++)
        qemuDomainObjCheckHostdevTaint(driver, obj, obj->def->hostdevs[i],
3013
                                       logCtxt);
3014

3015
    for (i = 0; i < obj->def->nnets; i++)
3016
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logCtxt);
3017

3018
    if (obj->def->os.dtb)
3019
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_DTB, logCtxt);
3020

3021
    virObjectUnref(cfg);
3022 3023 3024
}


3025
void qemuDomainObjCheckDiskTaint(virQEMUDriverPtr driver,
3026
                                 virDomainObjPtr obj,
3027
                                 virDomainDiskDefPtr disk,
3028
                                 qemuDomainLogContextPtr logCtxt)
3029
{
3030
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3031
    int format = virDomainDiskGetFormat(disk);
3032

3033
    if ((!format || format == VIR_STORAGE_FILE_AUTO) &&
3034
        cfg->allowDiskFormatProbing)
3035
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_DISK_PROBING, logCtxt);
3036

3037 3038
    if (disk->rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
3039
                           logCtxt);
3040

3041 3042 3043 3044
    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,
3045
                           logCtxt);
3046

3047
    virObjectUnref(cfg);
3048 3049 3050
}


3051 3052 3053
void qemuDomainObjCheckHostdevTaint(virQEMUDriverPtr driver,
                                    virDomainObjPtr obj,
                                    virDomainHostdevDefPtr hostdev,
3054
                                    qemuDomainLogContextPtr logCtxt)
3055 3056 3057 3058 3059 3060
{
    virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi;

    if (hostdev->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_SCSI &&
        scsisrc->rawio == VIR_TRISTATE_BOOL_YES)
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
3061
                               logCtxt);
3062 3063 3064
}


3065
void qemuDomainObjCheckNetTaint(virQEMUDriverPtr driver,
3066
                                virDomainObjPtr obj,
3067
                                virDomainNetDefPtr net,
3068
                                qemuDomainLogContextPtr logCtxt)
3069
{
3070 3071 3072 3073 3074 3075
    /* 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)
3076
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logCtxt);
3077
}
3078 3079


3080 3081 3082 3083 3084 3085 3086 3087 3088 3089
qemuDomainLogContextPtr qemuDomainLogContextNew(virQEMUDriverPtr driver,
                                                virDomainObjPtr vm,
                                                qemuDomainLogContextMode mode)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainLogContextPtr ctxt = NULL;

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

3090
    VIR_DEBUG("Context new %p stdioLogD=%d", ctxt, cfg->stdioLogD);
3091 3092 3093 3094
    ctxt->writefd = -1;
    ctxt->readfd = -1;
    virAtomicIntSet(&ctxt->refs, 1);

3095 3096 3097
    if (virAsprintf(&ctxt->path, "%s/%s.log", cfg->logDir, vm->def->name) < 0)
        goto error;

3098 3099 3100 3101
    if (cfg->stdioLogD) {
        ctxt->manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver));
        if (!ctxt->manager)
            goto error;
3102

3103 3104 3105 3106
        ctxt->writefd = virLogManagerDomainOpenLogFile(ctxt->manager,
                                                       "qemu",
                                                       vm->def->uuid,
                                                       vm->def->name,
3107
                                                       ctxt->path,
3108 3109 3110 3111 3112 3113
                                                       0,
                                                       &ctxt->inode,
                                                       &ctxt->pos);
        if (ctxt->writefd < 0)
            goto error;
    } else {
3114
        if ((ctxt->writefd = open(ctxt->path, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
3115
            virReportSystemError(errno, _("failed to create logfile %s"),
3116
                                 ctxt->path);
3117 3118
            goto error;
        }
3119
        if (virSetCloseExec(ctxt->writefd) < 0) {
3120
            virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
3121
                                 ctxt->path);
3122 3123 3124
            goto error;
        }

3125 3126 3127 3128 3129 3130 3131
        /* 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"),
3132
                                 ctxt->path);
3133 3134 3135 3136
            goto error;
        }

        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START) {
3137
            if ((ctxt->readfd = open(ctxt->path, O_RDONLY, S_IRUSR | S_IWUSR)) < 0) {
3138
                virReportSystemError(errno, _("failed to open logfile %s"),
3139
                                     ctxt->path);
3140 3141 3142 3143
                goto error;
            }
            if (virSetCloseExec(ctxt->readfd) < 0) {
                virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
3144
                                     ctxt->path);
3145 3146 3147 3148
                goto error;
            }
        }

3149 3150
        if ((ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END)) < 0) {
            virReportSystemError(errno, _("failed to seek in log file %s"),
3151
                                 ctxt->path);
3152 3153 3154 3155
            goto error;
        }
    }

3156
 cleanup:
3157 3158 3159 3160 3161
    virObjectUnref(cfg);
    return ctxt;

 error:
    qemuDomainLogContextFree(ctxt);
3162 3163
    ctxt = NULL;
    goto cleanup;
3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177
}


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;
3178 3179
    if (!ctxt->manager &&
        lseek(ctxt->writefd, 0, SEEK_END) < 0) {
3180
        virReportSystemError(errno, "%s",
3181
                             _("Unable to seek to end of domain logfile"));
3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201
        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)
{
3202 3203 3204 3205
    VIR_DEBUG("Context read %p manager=%p inode=%llu pos=%llu",
              ctxt, ctxt->manager,
              (unsigned long long)ctxt->inode,
              (unsigned long long)ctxt->pos);
3206
    char *buf;
3207 3208 3209
    size_t buflen;
    if (ctxt->manager) {
        buf = virLogManagerDomainReadLogFile(ctxt->manager,
3210
                                             ctxt->path,
3211 3212 3213 3214 3215 3216 3217 3218 3219
                                             ctxt->inode,
                                             ctxt->pos,
                                             1024 * 128,
                                             0);
        if (!buf)
            return -1;
        buflen = strlen(buf);
    } else {
        ssize_t got;
3220

3221
        buflen = 1024 * 128;
3222

3223 3224
        /* Best effort jump to start of messages */
        ignore_value(lseek(ctxt->readfd, ctxt->pos, SEEK_SET));
3225

3226 3227
        if (VIR_ALLOC_N(buf, buflen) < 0)
            return -1;
3228

3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241
        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;
    }
3242 3243 3244

    *msg = buf;

3245
    return buflen;
3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256
}


int qemuDomainLogContextGetWriteFD(qemuDomainLogContextPtr ctxt)
{
    return ctxt->writefd;
}


void qemuDomainLogContextMarkPosition(qemuDomainLogContextPtr ctxt)
{
3257 3258
    if (ctxt->manager)
        virLogManagerDomainGetLogFilePosition(ctxt->manager,
3259
                                              ctxt->path,
3260 3261 3262 3263 3264
                                              0,
                                              &ctxt->inode,
                                              &ctxt->pos);
    else
        ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END);
3265 3266 3267 3268 3269
}


void qemuDomainLogContextRef(qemuDomainLogContextPtr ctxt)
{
3270
    VIR_DEBUG("Context ref %p", ctxt);
3271 3272 3273 3274
    virAtomicIntInc(&ctxt->refs);
}


3275 3276 3277 3278 3279 3280
virLogManagerPtr qemuDomainLogContextGetManager(qemuDomainLogContextPtr ctxt)
{
    return ctxt->manager;
}


3281 3282 3283 3284 3285 3286 3287 3288
void qemuDomainLogContextFree(qemuDomainLogContextPtr ctxt)
{
    bool lastRef;

    if (!ctxt)
        return;

    lastRef = virAtomicIntDecAndTest(&ctxt->refs);
3289
    VIR_DEBUG("Context free %p lastref=%d", ctxt, lastRef);
3290 3291 3292
    if (!lastRef)
        return;

3293
    virLogManagerFree(ctxt->manager);
3294
    VIR_FREE(ctxt->path);
3295 3296 3297 3298 3299 3300
    VIR_FORCE_CLOSE(ctxt->writefd);
    VIR_FORCE_CLOSE(ctxt->readfd);
    VIR_FREE(ctxt);
}


3301 3302
/* Locate an appropriate 'qemu-img' binary.  */
const char *
3303
qemuFindQemuImgBinary(virQEMUDriverPtr driver)
3304
{
3305 3306
    if (!driver->qemuImgBinary)
        virReportError(VIR_ERR_INTERNAL_ERROR,
3307
                       "%s", _("unable to find qemu-img"));
3308 3309 3310 3311 3312 3313 3314

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
                                virDomainSnapshotObjPtr snapshot,
3315
                                virCapsPtr caps,
3316 3317 3318 3319 3320 3321 3322 3323 3324
                                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);
3325
    newxml = virDomainSnapshotDefFormat(
3326
        uuidstr, snapshot->def, caps,
3327 3328
        virDomainDefFormatConvertXMLFlags(QEMU_DOMAIN_FORMAT_LIVE_FLAGS),
        1);
3329
    if (newxml == NULL)
3330 3331
        return -1;

3332
    if (virAsprintf(&snapDir, "%s/%s", snapshotDir, vm->def->name) < 0)
3333 3334 3335 3336 3337 3338 3339
        goto cleanup;
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
        goto cleanup;
    }

3340
    if (virAsprintf(&snapFile, "%s/%s.xml", snapDir, snapshot->def->name) < 0)
3341 3342
        goto cleanup;

J
Ján Tomko 已提交
3343
    ret = virXMLSaveFile(snapFile, NULL, "snapshot-edit", newxml);
3344

3345
 cleanup:
3346 3347 3348 3349 3350 3351 3352 3353
    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.  */
3354
static int
3355
qemuDomainSnapshotForEachQcow2Raw(virQEMUDriverPtr driver,
3356 3357 3358 3359 3360
                                  virDomainDefPtr def,
                                  const char *name,
                                  const char *op,
                                  bool try_all,
                                  int ndisks)
3361 3362
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
3363
    size_t i;
3364 3365 3366 3367 3368 3369 3370 3371 3372
    bool skipped = false;

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

    qemuimgarg[2] = op;
3373
    qemuimgarg[3] = name;
3374

3375
    for (i = 0; i < ndisks; i++) {
3376
        /* FIXME: we also need to handle LVM here */
3377
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
3378 3379 3380
            int format = virDomainDiskGetFormat(def->disks[i]);

            if (format > 0 && format != VIR_STORAGE_FILE_QCOW2) {
3381 3382 3383 3384 3385
                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",
3386
                             def->disks[i]->dst);
3387 3388
                    skipped = true;
                    continue;
3389 3390 3391 3392 3393
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
3394
                }
3395 3396 3397 3398
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("Disk device '%s' does not support"
                                 " snapshotting"),
                               def->disks[i]->dst);
3399 3400 3401
                return -1;
            }

3402
            qemuimgarg[4] = virDomainDiskGetSource(def->disks[i]);
3403 3404 3405 3406

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
3407
                             def->disks[i]->dst);
3408 3409
                    skipped = true;
                    continue;
3410 3411 3412 3413 3414
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
3415 3416 3417 3418 3419 3420 3421 3422 3423
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

3424 3425 3426
/* 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
3427
qemuDomainSnapshotForEachQcow2(virQEMUDriverPtr driver,
3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443
                               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);
}

3444 3445
/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
3446
qemuDomainSnapshotDiscard(virQEMUDriverPtr driver,
3447 3448 3449 3450 3451 3452 3453 3454 3455
                          virDomainObjPtr vm,
                          virDomainSnapshotObjPtr snap,
                          bool update_current,
                          bool metadata_only)
{
    char *snapFile = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;
    virDomainSnapshotObjPtr parentsnap = NULL;
3456
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3457 3458 3459 3460 3461 3462 3463 3464 3465

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
                goto cleanup;
        } else {
            priv = vm->privateData;
3466
            qemuDomainObjEnterMonitor(driver, vm);
3467 3468
            /* we continue on even in the face of error */
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
3469
            ignore_value(qemuDomainObjExitMonitor(driver, vm));
3470 3471 3472
        }
    }

3473
    if (virAsprintf(&snapFile, "%s/%s/%s.xml", cfg->snapshotDir,
3474
                    vm->def->name, snap->def->name) < 0)
3475 3476 3477 3478
        goto cleanup;

    if (snap == vm->current_snapshot) {
        if (update_current && snap->def->parent) {
3479
            parentsnap = virDomainSnapshotFindByName(vm->snapshots,
3480 3481 3482 3483 3484 3485
                                                     snap->def->parent);
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
                         snap->def->parent);
            } else {
                parentsnap->def->current = true;
3486
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap, driver->caps,
3487
                                                    cfg->snapshotDir) < 0) {
3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499
                    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);
3500
    virDomainSnapshotObjListRemove(vm->snapshots, snap);
3501 3502 3503

    ret = 0;

3504
 cleanup:
3505
    VIR_FREE(snapFile);
3506
    virObjectUnref(cfg);
3507 3508 3509 3510
    return ret;
}

/* Hash iterator callback to discard multiple snapshots.  */
3511 3512 3513
int qemuDomainSnapshotDiscardAll(void *payload,
                                 const void *name ATTRIBUTE_UNUSED,
                                 void *data)
3514 3515
{
    virDomainSnapshotObjPtr snap = payload;
3516
    virQEMUSnapRemovePtr curr = data;
3517 3518 3519 3520 3521 3522 3523 3524
    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;
3525
    return 0;
3526 3527 3528
}

int
3529
qemuDomainSnapshotDiscardAllMetadata(virQEMUDriverPtr driver,
3530 3531
                                     virDomainObjPtr vm)
{
3532
    virQEMUSnapRemove rem;
3533 3534 3535 3536 3537

    rem.driver = driver;
    rem.vm = vm;
    rem.metadata_only = true;
    rem.err = 0;
3538 3539
    virDomainSnapshotForEach(vm->snapshots, qemuDomainSnapshotDiscardAll,
                             &rem);
3540 3541 3542 3543 3544

    return rem.err;
}

/*
3545
 * The caller must hold a lock the vm.
3546 3547
 */
void
3548
qemuDomainRemoveInactive(virQEMUDriverPtr driver,
3549 3550
                         virDomainObjPtr vm)
{
3551
    bool haveJob = true;
3552
    char *snapDir;
3553 3554 3555 3556 3557 3558 3559 3560
    virQEMUDriverConfigPtr cfg;

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

    cfg = virQEMUDriverGetConfig(driver);
3561

3562 3563 3564
    if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0)
        haveJob = false;

3565 3566 3567 3568 3569
    /* 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);
    }
3570
    else if (virAsprintf(&snapDir, "%s/%s", cfg->snapshotDir,
3571 3572
                         vm->def->name) < 0) {
        VIR_WARN("unable to remove snapshot directory %s/%s",
3573
                 cfg->snapshotDir, vm->def->name);
3574 3575 3576 3577 3578
    } else {
        if (rmdir(snapDir) < 0 && errno != ENOENT)
            VIR_WARN("unable to remove snapshot directory %s", snapDir);
        VIR_FREE(snapDir);
    }
3579 3580 3581

    virObjectRef(vm);

3582
    virDomainObjListRemove(driver->domains, vm);
3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595
    /*
     * 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);
3596
    virObjectUnref(cfg);
3597 3598

    if (haveJob)
3599
        qemuDomainObjEndJob(driver, vm);
3600 3601

    virObjectUnref(vm);
3602
}
3603 3604

void
3605
qemuDomainSetFakeReboot(virQEMUDriverPtr driver,
3606 3607 3608 3609
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
3610
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3611 3612

    if (priv->fakeReboot == value)
3613
        goto cleanup;
3614 3615 3616

    priv->fakeReboot = value;

3617
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
3618
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
3619

3620
 cleanup:
3621
    virObjectUnref(cfg);
3622
}
M
Michal Privoznik 已提交
3623

3624 3625 3626
static int
qemuDomainCheckRemoveOptionalDisk(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
3627
                                  size_t diskIndex)
3628 3629
{
    char uuid[VIR_UUID_STRING_BUFLEN];
3630
    virObjectEventPtr event = NULL;
3631
    virDomainDiskDefPtr disk = vm->def->disks[diskIndex];
3632
    const char *src = virDomainDiskGetSource(disk);
3633 3634 3635 3636 3637

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

    VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
              "due to inaccessible source '%s'",
3638
              disk->dst, vm->def->name, uuid, src);
3639 3640 3641 3642

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

3643
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
3644 3645
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
3646
        ignore_value(virDomainDiskSetSource(disk, NULL));
3647
    } else {
3648
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
3649 3650
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_DROP_MISSING_ON_START);
3651 3652
        virDomainDiskRemove(vm->def, diskIndex);
        virDomainDiskDefFree(disk);
3653 3654
    }

3655
    qemuDomainEventQueue(driver, event);
3656 3657 3658 3659

    return 0;
}

3660 3661 3662
static int
qemuDomainCheckDiskStartupPolicy(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
3663
                                 size_t diskIndex,
3664 3665
                                 bool cold_boot)
{
3666
    int startupPolicy = vm->def->disks[diskIndex]->startupPolicy;
3667
    int device = vm->def->disks[diskIndex]->device;
3668

3669
    switch ((virDomainStartupPolicy) startupPolicy) {
3670
        case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
3671 3672 3673 3674 3675 3676 3677
            /* 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)
                goto error;
3678 3679 3680 3681 3682 3683
            break;

        case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
            goto error;

        case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
3684
            if (cold_boot)
3685 3686 3687 3688 3689 3690 3691 3692 3693
                goto error;
            break;

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

3694
    if (qemuDomainCheckRemoveOptionalDisk(driver, vm, diskIndex) < 0)
3695
        goto error;
3696 3697 3698

    return 0;

3699
 error:
3700 3701 3702
    return -1;
}

3703

M
Michal Privoznik 已提交
3704
int
3705
qemuDomainCheckDiskPresence(virQEMUDriverPtr driver,
M
Michal Privoznik 已提交
3706
                            virDomainObjPtr vm,
3707
                            bool cold_boot)
M
Michal Privoznik 已提交
3708 3709
{
    int ret = -1;
3710
    size_t i;
M
Michal Privoznik 已提交
3711

3712
    VIR_DEBUG("Checking for disk presence");
3713
    for (i = vm->def->ndisks; i > 0; i--) {
3714 3715
        size_t idx = i - 1;
        virDomainDiskDefPtr disk = vm->def->disks[idx];
3716
        virStorageFileFormat format = virDomainDiskGetFormat(disk);
M
Michal Privoznik 已提交
3717

3718
        if (virStorageSourceIsEmpty(disk->src))
3719 3720 3721 3722 3723
            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 */
3724
        if (virStorageSourceIsLocalStorage(disk->src) &&
3725
            format >= VIR_STORAGE_FILE_NONE &&
3726
            format < VIR_STORAGE_FILE_BACKING &&
3727
            virFileExists(virDomainDiskGetSource(disk)))
M
Michal Privoznik 已提交
3728 3729
            continue;

3730
        if (qemuDomainDetermineDiskChain(driver, vm, disk, true, true) >= 0)
M
Michal Privoznik 已提交
3731
            continue;
3732

3733
        if (disk->startupPolicy &&
3734
            qemuDomainCheckDiskStartupPolicy(driver, vm, idx,
3735 3736 3737
                                             cold_boot) >= 0) {
            virResetLastError();
            continue;
M
Michal Privoznik 已提交
3738 3739
        }

3740
        goto error;
M
Michal Privoznik 已提交
3741 3742 3743 3744
    }

    ret = 0;

3745
 error:
M
Michal Privoznik 已提交
3746 3747
    return ret;
}
3748 3749 3750 3751 3752 3753 3754 3755 3756 3757

/*
 * The vm must be locked when any of the following cleanup functions is
 * called.
 */
int
qemuDomainCleanupAdd(virDomainObjPtr vm,
                     qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
3758
    size_t i;
3759 3760 3761 3762 3763 3764 3765 3766 3767 3768

    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,
3769
                     priv->ncleanupCallbacks, 1) < 0)
3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780
        return -1;

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

void
qemuDomainCleanupRemove(virDomainObjPtr vm,
                        qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
3781
    size_t i;
3782 3783 3784 3785

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

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
3786 3787 3788
        if (priv->cleanupCallbacks[i] == cb)
            VIR_DELETE_ELEMENT_INPLACE(priv->cleanupCallbacks,
                                       i, priv->ncleanupCallbacks);
3789 3790 3791 3792 3793 3794 3795 3796
    }

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

void
3797
qemuDomainCleanupRun(virQEMUDriverPtr driver,
3798 3799 3800
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
3801
    size_t i;
3802 3803 3804 3805

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

    /* run cleanup callbacks in reverse order */
3806 3807
    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[priv->ncleanupCallbacks - (i + 1)])
3808 3809 3810 3811 3812 3813 3814
            priv->cleanupCallbacks[i](driver, vm);
    }

    VIR_FREE(priv->cleanupCallbacks);
    priv->ncleanupCallbacks = 0;
    priv->ncleanupCallbacks_max = 0;
}
3815

3816 3817 3818
static void
qemuDomainGetImageIds(virQEMUDriverConfigPtr cfg,
                      virDomainObjPtr vm,
3819
                      virStorageSourcePtr src,
3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837
                      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;
    }

3838 3839
    if (vm && (vmlabel = virDomainDefGetSecurityLabelDef(vm->def, "dac")) &&
        vmlabel->label)
3840 3841
        virParseOwnershipIds(vmlabel->label, uid, gid);

3842
    if ((disklabel = virStorageSourceGetSecurityLabelDef(src, "dac")) &&
3843
        disklabel->label)
3844 3845 3846 3847
        virParseOwnershipIds(disklabel->label, uid, gid);
}


3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870
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;
}


3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886
char *
qemuDomainStorageAlias(const char *device, int depth)
{
    char *alias;

    if (STRPREFIX(device, QEMU_DRIVE_HOST_PREFIX))
        device += strlen(QEMU_DRIVE_HOST_PREFIX);

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


3887
int
3888
qemuDomainDetermineDiskChain(virQEMUDriverPtr driver,
3889
                             virDomainObjPtr vm,
3890
                             virDomainDiskDefPtr disk,
3891 3892
                             bool force_probe,
                             bool report_broken)
3893
{
3894 3895
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = 0;
3896 3897
    uid_t uid;
    gid_t gid;
3898

3899
    if (virStorageSourceIsEmpty(disk->src))
3900
        goto cleanup;
3901

3902
    if (disk->src->backingStore) {
3903
        if (force_probe)
3904
            virStorageSourceBackingStoreClear(disk->src);
3905
        else
3906
            goto cleanup;
3907
    }
3908

3909
    qemuDomainGetImageIds(cfg, vm, disk->src, &uid, &gid);
3910

3911
    if (virStorageFileGetMetadata(disk->src,
3912
                                  uid, gid,
3913
                                  cfg->allowDiskFormatProbing,
3914
                                  report_broken) < 0)
3915 3916
        ret = -1;

3917
 cleanup:
3918 3919
    virObjectUnref(cfg);
    return ret;
3920
}
3921

3922

3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985
/**
 * 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;
}


3986
bool
3987
qemuDomainDiskSourceDiffers(virDomainDiskDefPtr disk,
3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002
                            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;

4003 4004 4005 4006
    /* 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)
4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019
        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;
}


4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042
/*
 * 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);
4043 4044 4045 4046
    CHECK_EQ(bus, "bus", false);
    if (STRNEQ(disk->dst, orig_disk->dst)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
4047
                       "target");
4048 4049 4050 4051
        return false;
    }
    CHECK_EQ(tray_status, "tray", true);
    CHECK_EQ(removable, "removable", true);
4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134

    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;
    }

4135 4136 4137 4138 4139 4140 4141
    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);
4142
    /* "snapshot" is a libvirt internal field and thus can be changed */
4143
    /* startupPolicy is allowed to be updated. Therefore not checked here. */
4144
    CHECK_EQ(transient, "transient", true);
4145
    CHECK_EQ(info.bootIndex, "boot order", true);
4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157
    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;
    }
4158

4159 4160 4161 4162 4163 4164
    /* 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);

4165 4166 4167 4168 4169
#undef CHECK_EQ

    return true;
}

4170 4171 4172
bool
qemuDomainDiskBlockJobIsActive(virDomainDiskDefPtr disk)
{
4173 4174
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

4175 4176 4177 4178 4179 4180 4181 4182
    if (disk->mirror) {
        virReportError(VIR_ERR_BLOCK_COPY_ACTIVE,
                       _("disk '%s' already in active block job"),
                       disk->dst);

        return true;
    }

4183
    if (diskPriv->blockjob) {
4184 4185 4186 4187 4188 4189 4190 4191 4192 4193
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("disk '%s' already in active block job"),
                       disk->dst);
        return true;
    }

    return false;
}


4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208
/**
 * 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++) {
4209 4210 4211 4212
        virDomainDiskDefPtr disk = vm->def->disks[i];
        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

        if (!copy_only && diskPriv->blockjob)
4213 4214
            return true;

4215
        if (disk->mirror && disk->mirrorJob == VIR_DOMAIN_BLOCK_JOB_TYPE_COPY)
4216 4217 4218 4219 4220 4221 4222
            return true;
    }

    return false;
}


4223 4224
int
qemuDomainUpdateDeviceList(virQEMUDriverPtr driver,
4225 4226
                           virDomainObjPtr vm,
                           int asyncJob)
4227 4228 4229
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char **aliases;
4230
    int rc;
4231 4232 4233 4234

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

4235 4236
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
4237 4238 4239 4240
    rc = qemuMonitorGetDeviceAliases(priv->mon, &aliases);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
    if (rc < 0)
4241 4242 4243 4244 4245 4246
        return -1;

    virStringFreeList(priv->qemuDevices);
    priv->qemuDevices = aliases;
    return 0;
}
4247

4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296

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;
}


4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312
bool
qemuDomainDefCheckABIStability(virQEMUDriverPtr driver,
                               virDomainDefPtr src,
                               virDomainDefPtr dst)
{
    virDomainDefPtr migratableDefSrc = NULL;
    virDomainDefPtr migratableDefDst = NULL;
    const int flags = VIR_DOMAIN_XML_SECURE | VIR_DOMAIN_XML_UPDATE_CPU | VIR_DOMAIN_XML_MIGRATABLE;
    bool ret = false;

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

    ret = virDomainDefCheckABIStability(migratableDefSrc, migratableDefDst);

4313
 cleanup:
4314 4315 4316 4317
    virDomainDefFree(migratableDefSrc);
    virDomainDefFree(migratableDefDst);
    return ret;
}
4318 4319

bool
4320
qemuDomainAgentAvailable(virDomainObjPtr vm,
4321 4322
                         bool reportError)
{
4323 4324
    qemuDomainObjPrivatePtr priv = vm->privateData;

4325 4326 4327 4328 4329 4330 4331
    if (virDomainObjGetState(vm, NULL) != VIR_DOMAIN_RUNNING) {
        if (reportError) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
        }
        return false;
    }
4332 4333 4334 4335 4336 4337 4338 4339 4340
    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) {
4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352
        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;
4353 4354 4355 4356
        }
    }
    return true;
}
4357

4358

4359
static unsigned long long
4360
qemuDomainGetMemorySizeAlignment(virDomainDefPtr def)
4361
{
4362 4363 4364 4365 4366
    /* 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;

4367 4368 4369 4370 4371 4372 4373
    /* Align memory size. QEMU requires rounding to next 4KiB block.
     * We'll take the "traditional" path and round it to 1MiB*/

    return 1024;
}


4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388
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;
}


4389 4390 4391
int
qemuDomainAlignMemorySizes(virDomainDefPtr def)
{
4392 4393
    unsigned long long maxmemkb = virMemoryMaxValue(false) >> 10;
    unsigned long long maxmemcapped = virMemoryMaxValue(true) >> 10;
4394
    unsigned long long initialmem = 0;
4395
    unsigned long long mem;
4396
    unsigned long long align = qemuDomainGetMemorySizeAlignment(def);
4397 4398 4399 4400 4401
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    /* align NUMA cell sizes if relevant */
    for (i = 0; i < ncells; i++) {
4402 4403
        mem = VIR_ROUND_UP(virDomainNumaGetNodeMemorySize(def->numa, i), align);
        initialmem += mem;
4404 4405 4406 4407 4408 4409 4410

        if (mem > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("memory size of NUMA node '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
4411
        virDomainNumaSetNodeMemorySize(def->numa, i, mem);
4412 4413
    }

4414 4415 4416 4417 4418
    /* 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);

4419 4420 4421 4422 4423 4424
    if (initialmem > maxmemcapped) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("initial memory size overflowed after alignment"));
        return -1;
    }

4425
    virDomainDefSetMemoryInitial(def, initialmem);
4426

4427
    def->mem.max_memory = VIR_ROUND_UP(def->mem.max_memory, align);
4428 4429 4430 4431 4432
    if (def->mem.max_memory > maxmemkb) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("maximum memory size overflowed after alignment"));
        return -1;
    }
4433

4434
    /* Align memory module sizes */
4435 4436
    for (i = 0; i < def->nmems; i++) {
        align = qemuDomainGetMemoryModuleSizeAlignment(def, def->mems[i]);
4437
        def->mems[i]->size = VIR_ROUND_UP(def->mems[i]->size, align);
4438 4439 4440 4441 4442 4443 4444

        if (def->mems[i]->size > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("size of memory module '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
4445
    }
4446

4447 4448
    return 0;
}
4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459


/**
 * 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
4460 4461
qemuDomainMemoryDeviceAlignSize(virDomainDefPtr def,
                                virDomainMemoryDefPtr mem)
4462
{
4463
    mem->size = VIR_ROUND_UP(mem->size, qemuDomainGetMemorySizeAlignment(def));
4464
}
4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477


/**
 * qemuDomainGetMonitor:
 * @vm: domain object
 *
 * Returns the monitor pointer corresponding to the domain object @vm.
 */
qemuMonitorPtr
qemuDomainGetMonitor(virDomainObjPtr vm)
{
    return ((qemuDomainObjPrivatePtr) vm->privateData)->mon;
}
4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493


/**
 * 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;
4494 4495
    bool asynchronous = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKJOB_ASYNC);
    bool synchronous = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKJOB_SYNC);
4496

4497
    if (!synchronous && !asynchronous) {
4498 4499 4500 4501 4502 4503
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("block jobs not supported with this QEMU binary"));
        return -1;
    }

    if (modern)
4504
        *modern = asynchronous;
4505 4506 4507

    return 0;
}
4508 4509 4510 4511 4512 4513 4514 4515 4516


/**
 * 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.
 */
4517
virDomainChrDefPtr
4518 4519 4520 4521 4522 4523 4524 4525 4526 4527
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;

4528 4529
        if (STREQ_NULLABLE(channel->target.name, "org.qemu.guest_agent.0"))
            return channel;
4530 4531
    }

4532
    return NULL;
4533
}
4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552


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"));
}
4553 4554


4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572
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;
}


4573 4574 4575 4576 4577 4578
bool
qemuDomainMachineIsS390CCW(const virDomainDef *def)
{
    return STRPREFIX(def->os.machine, "s390-ccw");
}

4579

4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626
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;
        }

4627 4628 4629
        if (virDomainNumaGetNodeCount(def->numa) != 0) {
            if (mem->targetNode == -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
Y
Yuri Chornoivan 已提交
4630
                               _("target NUMA node needs to be specified for "
4631 4632 4633
                                 "memory device"));
                return -1;
            }
4634 4635
        }

4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661
        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:
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("invalid memory device type"));
        return -1;
    }

    return 0;
}


4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689
/**
 * 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;
4690 4691 4692

        if (qemuDomainDefValidateMemoryHotplugDevice(mem, def) < 0)
            return -1;
4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711
    }

    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;
    }

4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722
    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;
        }
4723 4724 4725 4726 4727 4728 4729 4730 4731
    }

    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;
    }

4732
    for (i = 0; i < def->nmems; i++) {
4733 4734
        hotplugMemory += def->mems[i]->size;

4735 4736 4737 4738 4739 4740
        /* already existing devices don't need to be checked on hotplug */
        if (!mem &&
            qemuDomainDefValidateMemoryHotplugDevice(def->mems[i], def) < 0)
            return -1;
    }

4741 4742 4743 4744 4745 4746 4747 4748 4749 4750
    if (hotplugMemory > hotplugSpace) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory device total size exceeds hotplug space"));
        return -1;
    }

    return 0;
}


4751 4752 4753 4754 4755 4756 4757 4758 4759 4760
bool
qemuDomainMachineHasBuiltinIDE(const virDomainDef *def)
{
    return qemuDomainMachineIsI440FX(def) ||
        STREQ(def->os.machine, "malta") ||
        STREQ(def->os.machine, "sun4u") ||
        STREQ(def->os.machine, "g3beige");
}


4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784
/**
 * 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 */
4785
    if (!virDomainDefHasMemballoon(vm->def)) {
4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818
        vm->def->mem.cur_balloon = virDomainDefGetMemoryActual(vm->def);
        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;

4819 4820
        vm->def->mem.cur_balloon = balloon;
    }
4821 4822 4823

    return 0;
}
4824 4825 4826


/**
4827
 * qemuDomainGetMemLockLimitBytes:
4828 4829 4830 4831
 *
 * @def: domain definition
 *
 * Returns the size of the memory in bytes that needs to be set as
4832
 * RLIMIT_MEMLOCK for the QEMU process.
4833 4834 4835 4836
 * 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
4837
qemuDomainGetMemLockLimitBytes(virDomainDefPtr def)
4838 4839 4840
{
    unsigned long long memKB;

4841 4842 4843 4844 4845 4846
    /* prefer the hard limit */
    if (virMemoryLimitIsSet(def->mem.hard_limit)) {
        memKB = def->mem.hard_limit;
        goto done;
    }

4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924
    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;
            }
        }

        memory = virDomainDefGetMemoryActual(def);

        if (def->mem.max_memory)
            maxMemory = def->mem.max_memory;
        else
            maxMemory = memory;

        /* baseLimit := maxMemory / 128                                  (a)
         *              + 4 MiB * #PHBs + 8 MiB                          (b)
         *
         * (a) is the hash table
         *
         * (b) is accounting for the 32-bit DMA window - it could be either the
         * KVM accelerated TCE tables for emulated devices, or the VFIO
         * userspace view. The 4 MiB per-PHB (including the default one) covers
         * a 2GiB DMA window: default is 1GiB, but it's possible it'll be
         * increased to help performance. The 8 MiB extra should be plenty for
         * the TCE table index for any reasonable number of PHBs and several
         * spapr-vlan or spapr-vscsi devices (512kB + a tiny bit each) */
        baseLimit = maxMemory / 128 +
                    4096 * nPCIHostBridges +
                    8192;

        /* 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;
    }

4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942
    /* 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.
     */
4943
    memKB = virDomainDefGetMemoryActual(def) + 1024 * 1024;
4944

4945
 done:
4946 4947
    return memKB << 10;
}
4948 4949 4950 4951 4952


/**
 * @def: domain definition
 *
4953 4954 4955
 * Returns true if the locked memory limit needs to be set or updated because
 * of domain configuration, VFIO passthrough devices or architecture-specific
 * requirements.
4956 4957
 * */
bool
4958
qemuDomainRequiresMemLock(virDomainDefPtr def)
4959 4960 4961 4962 4963 4964
{
    size_t i;

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

4965 4966 4967 4968
    /* ppc64 domains need to lock some memory even when VFIO is not used */
    if (ARCH_IS_PPC64(def->os.arch))
        return true;

4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979
    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;
    }

    return false;
}
4980

4981 4982 4983 4984 4985 4986 4987
/**
 * qemuDomainAdjustMaxMemLock:
 * @vm: domain
 *
 * Adjust the memory locking limit for the QEMU process associated to @vm, in
 * order to comply with VFIO or architecture requirements.
 *
4988 4989 4990 4991
 * 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.
4992 4993 4994 4995 4996 4997 4998 4999 5000
 *
 * Returns: 0 on success, <0 on failure
 */
int
qemuDomainAdjustMaxMemLock(virDomainObjPtr vm)
{
    unsigned long long bytes = 0;
    int ret = -1;

5001
    if (qemuDomainRequiresMemLock(vm->def)) {
5002 5003 5004 5005 5006 5007 5008 5009
        /* 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;
        }
5010
        bytes = qemuDomainGetMemLockLimitBytes(vm->def);
5011 5012 5013 5014 5015 5016
    } 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;
    }
5017 5018 5019 5020 5021 5022 5023 5024 5025 5026

    /* 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;
}
5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040

/**
 * qemuDomainHasVcpuPids:
 * @vm: Domain object
 *
 * Returns true if we were able to successfully detect vCPU pids for the VM.
 */
bool
qemuDomainHasVcpuPids(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    return priv->nvcpupids > 0;
}
5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060


/**
 * 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,
                     unsigned int vcpu)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (vcpu >= priv->nvcpupids)
        return 0;

    return priv->vcpupids[vcpu];
}
5061 5062 5063 5064 5065 5066 5067 5068 5069 5070


/**
 * qemuDomainDetectVcpuPids:
 * @driver: qemu driver data
 * @vm: domain object
 * @asyncJob: current asynchronous job type
 *
 * Updates vCPU thread ids in the private data of @vm.
 *
5071
 * Returns number of detected vCPU threads on success, -1 on error and reports
5072
 * an appropriate error, -2 if the domain doesn't exist any more.
5073 5074 5075 5076 5077 5078 5079
 */
int
qemuDomainDetectVcpuPids(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         int asyncJob)
{
    pid_t *cpupids = NULL;
5080
    int ncpupids = 0;
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
    qemuDomainObjPrivatePtr priv = vm->privateData;

    /*
     * 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.
     */
5111 5112
    if (vm->def->virtType == VIR_DOMAIN_VIRT_QEMU)
        goto done;
5113 5114 5115 5116

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
    ncpupids = qemuMonitorGetCPUInfo(priv->mon, &cpupids);
5117 5118
    if (qemuDomainObjExitMonitor(driver, vm) < 0) {
        VIR_FREE(cpupids);
5119
        return -2;
5120 5121
    }
    /* failure to get the VCPU <-> PID mapping or to execute the query
5122 5123 5124 5125
     * command will not be treated fatal as some versions of qemu don't
     * support this command */
    if (ncpupids <= 0) {
        virResetLastError();
5126 5127
        ncpupids = 0;
        goto done;
5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138
    }

    if (ncpupids != virDomainDefGetVcpus(vm->def)) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("got wrong number of vCPU pids from QEMU monitor. "
                         "got %d, wanted %d"),
                       ncpupids, virDomainDefGetVcpus(vm->def));
        VIR_FREE(cpupids);
        return -1;
    }

5139 5140
 done:
    VIR_FREE(priv->vcpupids);
5141 5142
    priv->nvcpupids = ncpupids;
    priv->vcpupids = cpupids;
5143
    return ncpupids;
5144
}
5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173


bool
qemuDomainSupportsNicdev(virDomainDefPtr def,
                         virQEMUCapsPtr qemuCaps,
                         virDomainNetDefPtr net)
{
    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE))
        return false;

    /* 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)
{
    if (!qemuDomainSupportsNicdev(def, qemuCaps, net))
        return false;
    return virQEMUCapsGet(qemuCaps, QEMU_CAPS_NETDEV);
}
J
John Ferlan 已提交
5174 5175 5176 5177 5178 5179 5180


int
qemuDomainNetVLAN(virDomainNetDefPtr def)
{
    return qemuDomainDeviceAliasIndex(&def->info, "net");
}
P
Peter Krempa 已提交
5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196


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;
}