qemu_domain.c 127.1 KB
Newer Older
1
/*
2
 * qemu_domain.c: QEMU domain private state
3
 *
4
 * Copyright (C) 2006-2015 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 27
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include "qemu_domain.h"
#include "qemu_command.h"
28
#include "qemu_capabilities.h"
29
#include "qemu_migration.h"
30
#include "viralloc.h"
31
#include "virlog.h"
32
#include "virerror.h"
33
#include "c-ctype.h"
34
#include "cpu/cpu.h"
35
#include "viruuid.h"
E
Eric Blake 已提交
36
#include "virfile.h"
37
#include "domain_addr.h"
38
#include "domain_event.h"
39
#include "virtime.h"
40
#include "virstoragefile.h"
41
#include "virstring.h"
42
#include "virthreadjob.h"
43
#include "viratomic.h"
44
#include "logging/log_manager.h"
45

46 47
#include "storage/storage_driver.h"

48
#include <sys/time.h>
49
#include <fcntl.h>
50

51 52 53 54
#include <libxml/xpathInternals.h>

#define VIR_FROM_THIS VIR_FROM_QEMU

55 56
VIR_LOG_INIT("qemu.qemu_domain");

57 58
#define QEMU_NAMESPACE_HREF "http://libvirt.org/schemas/domain/qemu/1.0"

59 60 61 62 63 64
VIR_ENUM_IMPL(qemuDomainJob, QEMU_JOB_LAST,
              "none",
              "query",
              "destroy",
              "suspend",
              "modify",
65
              "abort",
66
              "migration operation",
67 68 69 70 71 72 73 74 75 76
              "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",
77
              "snapshot",
78
              "start",
79 80
);

81

82 83 84
struct _qemuDomainLogContext {
    int refs;
    int writefd;
85
    int readfd; /* Only used if manager == NULL */
86
    off_t pos;
87 88 89 90
    ino_t inode; /* Only used if manager != NULL */
    unsigned char uuid[VIR_UUID_BUFLEN]; /* Only used if manager != NULL */
    char *name; /* Only used if manager != NULL */
    virLogManagerPtr manager;
91 92
};

J
Jiri Denemark 已提交
93
const char *
94
qemuDomainAsyncJobPhaseToString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
95 96 97 98 99
                                int phase ATTRIBUTE_UNUSED)
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
100 101
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
102 103
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
104
    case QEMU_ASYNC_JOB_SNAPSHOT:
105
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
106 107 108 109 110 111 112 113 114
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        ; /* fall through */
    }

    return "none";
}

int
115
qemuDomainAsyncJobPhaseFromString(qemuDomainAsyncJob job,
J
Jiri Denemark 已提交
116 117 118 119 120 121 122 123
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
124 125
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
126 127
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
128
    case QEMU_ASYNC_JOB_SNAPSHOT:
129
    case QEMU_ASYNC_JOB_START:
J
Jiri Denemark 已提交
130 131 132 133 134 135 136 137 138 139 140
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        ; /* fall through */
    }

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

141

142
void qemuDomainEventQueue(virQEMUDriverPtr driver,
143
                          virObjectEventPtr event)
144
{
145 146
    if (event)
        virObjectEventStateQueue(driver->domainEventState, event);
147 148 149
}


150 151 152 153 154 155 156 157
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

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

158
    if (virCondInit(&priv->job.asyncCond) < 0) {
159
        virCondDestroy(&priv->job.cond);
160 161 162
        return -1;
    }

163 164 165 166 167 168 169 170 171
    return 0;
}

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

    job->active = QEMU_JOB_NONE;
172
    job->owner = 0;
173
    job->ownerAPI = NULL;
J
Jiri Denemark 已提交
174
    job->started = 0;
175 176 177 178 179 180 181 182
}

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

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
183
    job->asyncOwner = 0;
184
    job->asyncOwnerAPI = NULL;
J
Jiri Denemark 已提交
185
    job->asyncStarted = 0;
J
Jiri Denemark 已提交
186
    job->phase = 0;
187
    job->mask = QEMU_JOB_DEFAULT_MASK;
188
    job->dump_memory_only = false;
189
    job->abortJob = false;
190
    job->spiceMigrated = false;
J
Jiri Denemark 已提交
191
    VIR_FREE(job->current);
192 193
}

194 195 196 197 198 199 200 201
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
                        struct qemuDomainJobObj *job)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
202
    job->owner = priv->job.owner;
203
    job->asyncJob = priv->job.asyncJob;
204
    job->asyncOwner = priv->job.asyncOwner;
J
Jiri Denemark 已提交
205
    job->phase = priv->job.phase;
206 207 208 209 210

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

211 212 213
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
J
Jiri Denemark 已提交
214
    VIR_FREE(priv->job.current);
215
    VIR_FREE(priv->job.completed);
216 217
    virCondDestroy(&priv->job.cond);
    virCondDestroy(&priv->job.asyncCond);
218 219
}

220
static bool
221
qemuDomainTrackJob(qemuDomainJob job)
222 223 224 225
{
    return (QEMU_DOMAIN_TRACK_JOBS & JOB_MASK(job)) != 0;
}

226

J
Jiri Denemark 已提交
227 228 229 230 231 232 233 234 235 236 237
int
qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
{
    unsigned long long now;

    if (!jobInfo->started)
        return 0;

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

238 239 240 241 242 243
    if (now < jobInfo->started) {
        VIR_WARN("Async job starts in the future");
        jobInfo->started = 0;
        return 0;
    }

J
Jiri Denemark 已提交
244 245 246 247
    jobInfo->timeElapsed = now - jobInfo->started;
    return 0;
}

248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269
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;
    }

    jobInfo->status.downtime = now - jobInfo->stopped;
    jobInfo->status.downtime_set = true;
    return 0;
}

J
Jiri Denemark 已提交
270 271 272 273
int
qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                        virDomainJobInfoPtr info)
{
274
    info->type = jobInfo->type;
J
Jiri Denemark 已提交
275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
    info->timeElapsed = jobInfo->timeElapsed;
    info->timeRemaining = jobInfo->timeRemaining;

    info->memTotal = jobInfo->status.ram_total;
    info->memRemaining = jobInfo->status.ram_remaining;
    info->memProcessed = jobInfo->status.ram_transferred;

    info->fileTotal = jobInfo->status.disk_total;
    info->fileRemaining = jobInfo->status.disk_remaining;
    info->fileProcessed = jobInfo->status.disk_transferred;

    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)
{
    qemuMonitorMigrationStatus *status = &jobInfo->status;
    virTypedParameterPtr par = NULL;
    int maxpar = 0;
    int npar = 0;

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

309 310 311 312 313 314 315
    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
316 317 318 319 320 321 322 323 324 325 326 327
    if (jobInfo->type == VIR_DOMAIN_JOB_BOUNDED &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_TIME_REMAINING,
                                jobInfo->timeRemaining) < 0)
        goto error;

    if (status->downtime_set &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME,
                                status->downtime) < 0)
        goto error;

328 329 330 331 332 333 334 335
    if (status->downtime_set &&
        jobInfo->timeDeltaSet &&
        status->downtime > jobInfo->timeDelta &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DOWNTIME_NET,
                                status->downtime - jobInfo->timeDelta) < 0)
        goto error;

336 337 338 339 340 341
    if (status->setup_time_set &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_SETUP_TIME,
                                status->setup_time) < 0)
        goto error;

J
Jiri Denemark 已提交
342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_TOTAL,
                                status->ram_total +
                                status->disk_total) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_PROCESSED,
                                status->ram_transferred +
                                status->disk_transferred) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DATA_REMAINING,
                                status->ram_remaining +
                                status->disk_remaining) < 0)
        goto error;

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_TOTAL,
                                status->ram_total) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_PROCESSED,
                                status->ram_transferred) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_REMAINING,
                                status->ram_remaining) < 0)
        goto error;

367 368 369 370 371 372
    if (status->ram_bps &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_MEMORY_BPS,
                                status->ram_bps) < 0)
        goto error;

J
Jiri Denemark 已提交
373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
    if (status->ram_duplicate_set) {
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_CONSTANT,
                                    status->ram_duplicate) < 0 ||
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL,
                                    status->ram_normal) < 0 ||
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_MEMORY_NORMAL_BYTES,
                                    status->ram_normal_bytes) < 0)
            goto error;
    }

    if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_TOTAL,
                                status->disk_total) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_PROCESSED,
                                status->disk_transferred) < 0 ||
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_REMAINING,
                                status->disk_remaining) < 0)
        goto error;

397 398 399 400 401 402
    if (status->disk_bps &&
        virTypedParamsAddULLong(&par, &npar, &maxpar,
                                VIR_DOMAIN_JOB_DISK_BPS,
                                status->disk_bps) < 0)
        goto error;

J
Jiri Denemark 已提交
403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432
    if (status->xbzrle_set) {
        if (virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE,
                                    status->xbzrle_cache_size) < 0 ||
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_BYTES,
                                    status->xbzrle_bytes) < 0 ||
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_PAGES,
                                    status->xbzrle_pages) < 0 ||
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_CACHE_MISSES,
                                    status->xbzrle_cache_miss) < 0 ||
            virTypedParamsAddULLong(&par, &npar, &maxpar,
                                    VIR_DOMAIN_JOB_COMPRESSION_OVERFLOW,
                                    status->xbzrle_overflow) < 0)
            goto error;
    }

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

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


433 434 435 436 437 438 439 440
static virClassPtr qemuDomainDiskPrivateClass;

static int
qemuDomainDiskPrivateOnceInit(void)
{
    qemuDomainDiskPrivateClass = virClassNew(virClassForObject(),
                                             "qemuDomainDiskPrivate",
                                             sizeof(qemuDomainDiskPrivate),
441
                                             NULL);
442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
    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;
}


465 466
static void *
qemuDomainObjPrivateAlloc(void)
467 468 469 470 471 472
{
    qemuDomainObjPrivatePtr priv;

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

473 474 475
    if (qemuDomainObjInitJob(priv) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to init qemu driver mutexes"));
476
        goto error;
477
    }
478

479 480 481
    if (virCondInit(&priv->unplugFinished) < 0)
        goto error;

482
    if (!(priv->devs = virChrdevAlloc()))
483 484
        goto error;

485
    priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
486

487
    return priv;
488

489
 error:
490 491
    VIR_FREE(priv);
    return NULL;
492 493
}

494 495
static void
qemuDomainObjPrivateFree(void *data)
496 497 498
{
    qemuDomainObjPrivatePtr priv = data;

499
    virObjectUnref(priv->qemuCaps);
500

501
    virCgroupFree(&priv->cgroup);
502
    virDomainPCIAddressSetFree(priv->pciaddrs);
J
Ján Tomko 已提交
503
    virDomainCCWAddressSetFree(priv->ccwaddrs);
504
    virDomainVirtioSerialAddrSetFree(priv->vioserialaddrs);
505
    virDomainChrSourceDefFree(priv->monConfig);
506
    qemuDomainObjFreeJob(priv);
507
    VIR_FREE(priv->vcpupids);
508
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
509
    VIR_FREE(priv->origname);
510

511
    virCondDestroy(&priv->unplugFinished);
512
    virStringFreeList(priv->qemuDevices);
513
    virChrdevFree(priv->devs);
514

515 516
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
517
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
518 519
        qemuMonitorClose(priv->mon);
    }
D
Daniel P. Berrange 已提交
520 521 522 523
    if (priv->agent) {
        VIR_ERROR(_("Unexpected QEMU agent still active during domain deletion"));
        qemuAgentClose(priv->agent);
    }
524
    VIR_FREE(priv->cleanupCallbacks);
525
    virBitmapFree(priv->autoNodeset);
526
    virBitmapFree(priv->autoCpuset);
527 528 529 530
    VIR_FREE(priv);
}


531
static int
532 533
qemuDomainObjPrivateXMLFormat(virBufferPtr buf,
                              virDomainObjPtr vm)
534
{
535
    qemuDomainObjPrivatePtr priv = vm->privateData;
536
    const char *monitorpath;
537
    qemuDomainJob job;
538 539 540

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
541
        switch (priv->monConfig->type) {
542
        case VIR_DOMAIN_CHR_TYPE_UNIX:
543
            monitorpath = priv->monConfig->data.nix.path;
544 545 546
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
547
            monitorpath = priv->monConfig->data.file.path;
548 549 550
            break;
        }

551
        virBufferEscapeString(buf, "<monitor path='%s'", monitorpath);
552 553
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
554
        virBufferAsprintf(buf, " type='%s'/>\n",
555
                          virDomainChrTypeToString(priv->monConfig->type));
556 557 558 559
    }


    if (priv->nvcpupids) {
560
        size_t i;
561 562
        virBufferAddLit(buf, "<vcpus>\n");
        virBufferAdjustIndent(buf, 2);
563
        for (i = 0; i < priv->nvcpupids; i++)
564 565 566
            virBufferAsprintf(buf, "<vcpu pid='%d'/>\n", priv->vcpupids[i]);
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</vcpus>\n");
567 568
    }

569
    if (priv->qemuCaps) {
570
        size_t i;
571 572
        virBufferAddLit(buf, "<qemuCaps>\n");
        virBufferAdjustIndent(buf, 2);
573
        for (i = 0; i < QEMU_CAPS_LAST; i++) {
574
            if (virQEMUCapsGet(priv->qemuCaps, i)) {
575
                virBufferAsprintf(buf, "<flag name='%s'/>\n",
576
                                  virQEMUCapsTypeToString(i));
577 578
            }
        }
579 580
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</qemuCaps>\n");
581 582
    }

583
    if (priv->lockState)
584
        virBufferAsprintf(buf, "<lockstate>%s</lockstate>\n", priv->lockState);
585

586 587 588 589
    job = priv->job.active;
    if (!qemuDomainTrackJob(job))
        priv->job.active = QEMU_JOB_NONE;

590
    if (priv->job.active || priv->job.asyncJob) {
591
        virBufferAsprintf(buf, "<job type='%s' async='%s'",
592 593
                          qemuDomainJobTypeToString(priv->job.active),
                          qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
J
Jiri Denemark 已提交
594 595 596 597 598
        if (priv->job.phase) {
            virBufferAsprintf(buf, " phase='%s'",
                              qemuDomainAsyncJobPhaseToString(
                                    priv->job.asyncJob, priv->job.phase));
        }
599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619
        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");
        }
620
    }
621
    priv->job.active = job;
622

623
    if (priv->fakeReboot)
624
        virBufferAddLit(buf, "<fakereboot/>\n");
625

626 627
    if (priv->qemuDevices && *priv->qemuDevices) {
        char **tmp = priv->qemuDevices;
628 629
        virBufferAddLit(buf, "<devices>\n");
        virBufferAdjustIndent(buf, 2);
630
        while (*tmp) {
631
            virBufferAsprintf(buf, "<device alias='%s'/>\n", *tmp);
632 633
            tmp++;
        }
634 635
        virBufferAdjustIndent(buf, -2);
        virBufferAddLit(buf, "</devices>\n");
636 637
    }

638 639 640 641 642 643 644 645 646 647
    if (priv->autoNodeset) {
        char *nodeset = virBitmapFormat(priv->autoNodeset);

        if (!nodeset)
            return -1;

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

648 649 650
    return 0;
}

651
static int
652
qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt,
653
                             virDomainObjPtr vm,
654
                             virDomainDefParserConfigPtr config)
655
{
656
    qemuDomainObjPrivatePtr priv = vm->privateData;
657
    virQEMUDriverPtr driver = config->priv;
658
    char *monitorpath;
659
    char *tmp = NULL;
660 661
    int n;
    size_t i;
662
    xmlNodePtr *nodes = NULL;
663
    virQEMUCapsPtr qemuCaps = NULL;
664
    virCapsPtr caps = NULL;
665

666
    if (VIR_ALLOC(priv->monConfig) < 0)
667 668 669 670
        goto error;

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
671 672
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("no monitor path"));
673 674 675 676 677
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
678
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
679
    else
680
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
681 682
    VIR_FREE(tmp);

E
Eric Blake 已提交
683 684
    priv->monJSON = virXPathBoolean("count(./monitor[@json = '1']) > 0",
                                    ctxt) > 0;
685

686
    switch (priv->monConfig->type) {
687
    case VIR_DOMAIN_CHR_TYPE_PTY:
688
        priv->monConfig->data.file.path = monitorpath;
689 690
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
691
        priv->monConfig->data.nix.path = monitorpath;
692 693 694
        break;
    default:
        VIR_FREE(monitorpath);
695 696 697
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unsupported monitor type '%s'"),
                       virDomainChrTypeToString(priv->monConfig->type));
698 699 700 701 702 703 704 705
        goto error;
    }

    n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes);
    if (n < 0)
        goto error;
    if (n) {
        priv->nvcpupids = n;
706
        if (VIR_REALLOC_N(priv->vcpupids, priv->nvcpupids) < 0)
707 708
            goto error;

709
        for (i = 0; i < n; i++) {
710 711 712 713 714 715 716 717 718 719 720 721
            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);
    }
722

723
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
724 725
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("failed to parse qemu capabilities flags"));
726 727 728
        goto error;
    }
    if (n > 0) {
729
        if (!(qemuCaps = virQEMUCapsNew()))
730 731
            goto error;

732
        for (i = 0; i < n; i++) {
733 734
            char *str = virXMLPropString(nodes[i], "name");
            if (str) {
735
                int flag = virQEMUCapsTypeFromString(str);
736
                if (flag < 0) {
737 738
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("Unknown qemu capabilities flag %s"), str);
739
                    VIR_FREE(str);
740 741
                    goto error;
                }
742
                VIR_FREE(str);
743
                virQEMUCapsSet(qemuCaps, flag);
744 745 746
            }
        }

747
        priv->qemuCaps = qemuCaps;
748
        qemuCaps = NULL;
749 750 751
    }
    VIR_FREE(nodes);

752
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
753

754 755 756 757
    if ((tmp = virXPathString("string(./job[1]/@type)", ctxt))) {
        int type;

        if ((type = qemuDomainJobTypeFromString(tmp)) < 0) {
758 759
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown job type %s"), tmp);
760 761 762 763 764 765 766 767 768 769 770
            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) {
771 772
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown async job type %s"), tmp);
773 774 775 776 777
            VIR_FREE(tmp);
            goto error;
        }
        VIR_FREE(tmp);
        priv->job.asyncJob = async;
J
Jiri Denemark 已提交
778 779 780 781

        if ((tmp = virXPathString("string(./job[1]/@phase)", ctxt))) {
            priv->job.phase = qemuDomainAsyncJobPhaseFromString(async, tmp);
            if (priv->job.phase < 0) {
782 783
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Unknown job phase %s"), tmp);
J
Jiri Denemark 已提交
784 785 786 787 788
                VIR_FREE(tmp);
                goto error;
            }
            VIR_FREE(tmp);
        }
789 790
    }

791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813
    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);

814 815
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836
    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);

837 838 839 840 841 842 843 844 845 846 847 848 849 850 851
    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);

852 853
    return 0;

854
 error:
855
    virDomainChrSourceDefFree(priv->monConfig);
856 857
    priv->monConfig = NULL;
    VIR_FREE(nodes);
858
    VIR_FREE(tmp);
859 860
    virStringFreeList(priv->qemuDevices);
    priv->qemuDevices = NULL;
861
    virObjectUnref(qemuCaps);
862
    virObjectUnref(caps);
863 864 865 866
    return -1;
}


867 868 869
virDomainXMLPrivateDataCallbacks virQEMUDriverPrivateDataCallbacks = {
    .alloc = qemuDomainObjPrivateAlloc,
    .free = qemuDomainObjPrivateFree,
870
    .diskNew = qemuDomainDiskPrivateNew,
871 872 873 874 875
    .parse = qemuDomainObjPrivateXMLParse,
    .format = qemuDomainObjPrivateXMLFormat,
};


876 877 878 879 880
static void
qemuDomainDefNamespaceFree(void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;

881
    qemuDomainCmdlineDefFree(cmd);
882 883 884
}

static int
P
Philipp Hahn 已提交
885 886
qemuDomainDefNamespaceParse(xmlDocPtr xml ATTRIBUTE_UNUSED,
                            xmlNodePtr root ATTRIBUTE_UNUSED,
887 888 889 890
                            xmlXPathContextPtr ctxt,
                            void **data)
{
    qemuDomainCmdlineDefPtr cmd = NULL;
P
Philipp Hahn 已提交
891
    bool uses_qemu_ns = false;
892
    xmlNodePtr *nodes = NULL;
893 894
    int n;
    size_t i;
895

P
Philipp Hahn 已提交
896
    if (xmlXPathRegisterNs(ctxt, BAD_CAST "qemu", BAD_CAST QEMU_NAMESPACE_HREF) < 0) {
897 898 899
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Failed to register xml namespace '%s'"),
                       QEMU_NAMESPACE_HREF);
900 901 902
        return -1;
    }

903
    if (VIR_ALLOC(cmd) < 0)
904 905 906 907 908 909
        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
910
    uses_qemu_ns |= n > 0;
911 912

    if (n && VIR_ALLOC_N(cmd->args, n) < 0)
913
        goto error;
914 915 916 917

    for (i = 0; i < n; i++) {
        cmd->args[cmd->num_args] = virXMLPropString(nodes[i], "value");
        if (cmd->args[cmd->num_args] == NULL) {
918 919
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu command-line argument specified"));
920 921 922 923 924 925 926 927 928 929 930
            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
931
    uses_qemu_ns |= n > 0;
932 933

    if (n && VIR_ALLOC_N(cmd->env_name, n) < 0)
934
        goto error;
935 936

    if (n && VIR_ALLOC_N(cmd->env_value, n) < 0)
937
        goto error;
938 939 940 941 942 943

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

        tmp = virXMLPropString(nodes[i], "name");
        if (tmp == NULL) {
944 945
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("No qemu environment name specified"));
946 947 948
            goto error;
        }
        if (tmp[0] == '\0') {
949 950
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Empty qemu environment name specified"));
951 952 953
            goto error;
        }
        if (!c_isalpha(tmp[0]) && tmp[0] != '_') {
954 955
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must begin with a letter or underscore"));
956 957 958
            goto error;
        }
        if (strspn(tmp, "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789_") != strlen(tmp)) {
959 960
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Invalid environment name, it must contain only alphanumerics and underscore"));
961 962 963 964 965 966 967 968 969 970 971 972
            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
973 974 975 976
    if (uses_qemu_ns)
        *data = cmd;
    else
        VIR_FREE(cmd);
977 978 979

    return 0;

980
 error:
981 982 983 984 985 986 987 988 989 990
    VIR_FREE(nodes);
    qemuDomainDefNamespaceFree(cmd);
    return -1;
}

static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;
991
    size_t i;
992 993 994 995

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

996 997 998
    virBufferAddLit(buf, "<qemu:commandline>\n");
    virBufferAdjustIndent(buf, 2);

999
    for (i = 0; i < cmd->num_args; i++)
1000
        virBufferEscapeString(buf, "<qemu:arg value='%s'/>\n",
1001 1002
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
1003
        virBufferAsprintf(buf, "<qemu:env name='%s'", cmd->env_name[i]);
1004 1005 1006 1007 1008
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }

1009 1010
    virBufferAdjustIndent(buf, -2);
    virBufferAddLit(buf, "</qemu:commandline>\n");
1011 1012 1013 1014 1015 1016 1017 1018 1019 1020
    return 0;
}

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


1021 1022 1023 1024 1025 1026
virDomainXMLNamespace virQEMUDriverDomainXMLNamespace = {
    .parse = qemuDomainDefNamespaceParse,
    .free = qemuDomainDefNamespaceFree,
    .format = qemuDomainDefNamespaceFormatXML,
    .href = qemuDomainDefNamespaceHref,
};
1027

1028

1029 1030 1031
static int
qemuDomainDefPostParse(virDomainDefPtr def,
                       virCapsPtr caps,
1032
                       void *opaque)
1033
{
1034 1035
    virQEMUDriverPtr driver = opaque;
    virQEMUCapsPtr qemuCaps = NULL;
1036
    bool addDefaultUSB = true;
1037
    bool addImplicitSATA = false;
1038
    bool addPCIRoot = false;
L
Laine Stump 已提交
1039
    bool addPCIeRoot = false;
1040
    bool addDefaultMemballoon = true;
1041 1042
    bool addDefaultUSBKBD = false;
    bool addDefaultUSBMouse = false;
1043
    bool addPanicDevice = false;
1044
    int ret = -1;
1045

J
Ján Tomko 已提交
1046 1047 1048
    if (def->os.bootloader || def->os.bootloaderArgs) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("bootloader is not supported by QEMU"));
1049
        return ret;
J
Ján Tomko 已提交
1050 1051
    }

1052 1053 1054
    /* check for emulator and create a default one if needed */
    if (!def->emulator &&
        !(def->emulator = virDomainDefGetDefaultEmulator(def, caps)))
1055 1056 1057
        return ret;


P
Pavel Fedin 已提交
1058
    qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache, def->emulator);
1059

1060 1061 1062 1063 1064 1065
    /* 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
1066
        if (STREQ(def->os.machine, "isapc")) {
1067
            addDefaultUSB = false;
1068
            break;
1069
        }
L
Laine Stump 已提交
1070 1071
        if (STRPREFIX(def->os.machine, "pc-q35") ||
            STREQ(def->os.machine, "q35")) {
1072 1073 1074 1075
            addPCIeRoot = true;
            addDefaultUSB = false;
            addImplicitSATA = true;
            break;
L
Laine Stump 已提交
1076
        }
1077 1078
        if (!STRPREFIX(def->os.machine, "pc-0.") &&
            !STRPREFIX(def->os.machine, "pc-1.") &&
1079
            !STRPREFIX(def->os.machine, "pc-i440") &&
1080
            STRNEQ(def->os.machine, "pc") &&
1081 1082 1083 1084 1085
            !STRPREFIX(def->os.machine, "rhel"))
            break;
        addPCIRoot = true;
        break;

1086
    case VIR_ARCH_ARMV7L:
1087
    case VIR_ARCH_AARCH64:
1088 1089 1090 1091 1092 1093 1094
        addDefaultUSB = false;
        addDefaultMemballoon = false;
        if (STREQ(def->os.machine, "virt") ||
            STRPREFIX(def->os.machine, "virt-")) {
            addPCIeRoot = virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_GPEX);
        }
        break;
1095

1096
    case VIR_ARCH_PPC64:
1097
    case VIR_ARCH_PPC64LE:
1098 1099 1100
        addPCIRoot = true;
        addDefaultUSBKBD = true;
        addDefaultUSBMouse = true;
1101 1102 1103 1104 1105
        /* 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;
1106 1107
        break;

1108 1109 1110 1111 1112 1113 1114
    case VIR_ARCH_ALPHA:
    case VIR_ARCH_PPC:
    case VIR_ARCH_PPCEMB:
    case VIR_ARCH_SH4:
    case VIR_ARCH_SH4EB:
        addPCIRoot = true;
        break;
1115 1116 1117 1118 1119 1120
    case VIR_ARCH_S390:
        addDefaultUSB = false;
        break;
    case VIR_ARCH_S390X:
        addDefaultUSB = false;
        break;
1121 1122 1123 1124 1125 1126

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

1127 1128 1129 1130
    default:
        break;
    }

1131 1132 1133
    if (addDefaultUSB &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_USB, 0, -1) < 0)
1134
        goto cleanup;
1135

1136 1137 1138
    if (addImplicitSATA &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_SATA, 0, -1) < 0)
1139
        goto cleanup;
1140

1141 1142 1143
    /* NB: any machine that sets addPCIRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
     */
1144 1145 1146 1147
    if (addPCIRoot &&
        virDomainDefMaybeAddController(
            def, VIR_DOMAIN_CONTROLLER_TYPE_PCI, 0,
            VIR_DOMAIN_CONTROLLER_MODEL_PCI_ROOT) < 0)
1148
        goto cleanup;
1149

1150 1151 1152
    /* 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
1153 1154 1155
     *
     * NB: any machine that sets addPCIeRoot to true must also return
     * true from the function qemuDomainSupportsPCI().
1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
     */
    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) {
1167
            goto cleanup;
1168 1169
        }
    }
1170

1171
    if (addDefaultMemballoon && !def->memballoon) {
1172 1173
        virDomainMemballoonDefPtr memballoon;
        if (VIR_ALLOC(memballoon) < 0)
1174
            goto cleanup;
1175 1176 1177 1178 1179

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

1180 1181 1182 1183 1184
    if (addDefaultUSBKBD &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_KBD,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
1185
        goto cleanup;
1186 1187 1188 1189 1190 1191

    if (addDefaultUSBMouse &&
        def->ngraphics > 0 &&
        virDomainDefMaybeAddInput(def,
                                  VIR_DOMAIN_INPUT_TYPE_MOUSE,
                                  VIR_DOMAIN_INPUT_BUS_USB) < 0)
1192
        goto cleanup;
1193

D
Dmitry Andreev 已提交
1194 1195 1196 1197 1198 1199 1200
    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;
        }
1201

D
Dmitry Andreev 已提交
1202 1203 1204 1205 1206 1207 1208 1209 1210
        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;
            }
        }
1211 1212
    }

1213 1214 1215 1216
    ret = 0;
 cleanup:
    virObjectUnref(qemuCaps);
    return ret;
1217 1218
}

1219
static const char *
1220 1221
qemuDomainDefaultNetModel(const virDomainDef *def,
                          virQEMUCapsPtr qemuCaps)
1222
{
1223
    if (ARCH_IS_S390(def->os.arch))
1224 1225
        return "virtio";

1226 1227
    if (def->os.arch == VIR_ARCH_ARMV7L ||
        def->os.arch == VIR_ARCH_AARCH64) {
1228 1229 1230
        if (STREQ(def->os.machine, "versatilepb"))
            return "smc91c111";

1231 1232 1233
        if (STREQ(def->os.machine, "virt"))
            return "virtio";

1234 1235 1236 1237 1238
        /* Incomplete. vexpress (and a few others) use this, but not all
         * arm boards */
        return "lan9118";
    }

1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
    /* 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 */
1251 1252
    return "rtl8139";
}
1253

1254 1255
static int
qemuDomainDeviceDefPostParse(virDomainDeviceDefPtr dev,
1256
                             const virDomainDef *def,
1257
                             virCapsPtr caps ATTRIBUTE_UNUSED,
1258
                             void *opaque)
1259
{
1260
    virQEMUDriverPtr driver = opaque;
1261
    virQEMUCapsPtr qemuCaps = NULL;
1262
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1263
    int ret = -1;
1264

P
Pavel Fedin 已提交
1265
    qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache, def->emulator);
1266

1267
    if (dev->type == VIR_DOMAIN_DEVICE_NET &&
1268 1269
        dev->data.net->type != VIR_DOMAIN_NET_TYPE_HOSTDEV &&
        !dev->data.net->model) {
1270
        if (VIR_STRDUP(dev->data.net->model,
1271
                       qemuDomainDefaultNetModel(def, qemuCaps)) < 0)
1272
            goto cleanup;
1273
    }
1274

1275 1276 1277 1278
    /* set default disk types and drivers */
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
        virDomainDiskDefPtr disk = dev->data.disk;

1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306
        /* 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;
1307 1308 1309
        }
    }

1310 1311 1312 1313
    /* 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 &&
1314
        ARCH_IS_S390(def->os.arch))
1315 1316
        dev->data.chr->targetType = VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_VIRTIO;

1317 1318 1319 1320 1321
    /* 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 &&
1322
        ARCH_IS_S390(def->os.arch))
1323 1324
        dev->data.controller->model = VIR_DOMAIN_CONTROLLER_MODEL_USB_NONE;

1325 1326 1327 1328 1329 1330 1331
    /* 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;

1332 1333 1334 1335 1336
    /* auto generate unix socket path */
    if (dev->type == VIR_DOMAIN_DEVICE_CHR &&
        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 &&
1337
        !dev->data.chr->source.data.nix.path) {
1338 1339 1340 1341 1342 1343
        if (virAsprintf(&dev->data.chr->source.data.nix.path,
                        "%s/domain-%s/%s",
                        cfg->channelTargetDir, def->name,
                        dev->data.chr->target.name ? dev->data.chr->target.name
                        : "unknown.sock") < 0)
            goto cleanup;
1344

1345 1346 1347
        dev->data.chr->source.data.nix.listen = true;
    }

1348 1349 1350 1351 1352 1353 1354 1355 1356 1357
    /* 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;
    }

1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372
    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 {
1373
            dev->data.video->vgamem = QEMU_QXL_VGAMEM_DEFAULT;
1374 1375 1376
        }
    }

1377 1378 1379 1380 1381 1382 1383 1384 1385
    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;
    }

1386 1387
    ret = 0;

1388
 cleanup:
1389
    virObjectUnref(qemuCaps);
1390 1391
    virObjectUnref(cfg);
    return ret;
1392 1393 1394 1395 1396
}


virDomainDefParserConfig virQEMUDriverDomainDefParserConfig = {
    .devicesPostParseCallback = qemuDomainDeviceDefPostParse,
1397
    .domainPostParseCallback = qemuDomainDefPostParse,
1398 1399 1400
};


1401
static void
1402
qemuDomainObjSaveJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
1403
{
1404 1405 1406
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

    if (virDomainObjIsActive(obj)) {
1407
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj) < 0)
1408
            VIR_WARN("Failed to save status on vm %s", obj->def->name);
1409
    }
1410

1411
    virObjectUnref(cfg);
1412 1413
}

J
Jiri Denemark 已提交
1414
void
1415
qemuDomainObjSetJobPhase(virQEMUDriverPtr driver,
J
Jiri Denemark 已提交
1416 1417 1418 1419
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
1420
    unsigned long long me = virThreadSelfID();
J
Jiri Denemark 已提交
1421 1422 1423 1424

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

1425 1426 1427 1428 1429
    VIR_DEBUG("Setting '%s' phase to '%s'",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob, phase));

    if (priv->job.asyncOwner && me != priv->job.asyncOwner) {
1430
        VIR_WARN("'%s' async job is owned by thread %llu",
1431 1432 1433 1434
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }

J
Jiri Denemark 已提交
1435
    priv->job.phase = phase;
1436
    priv->job.asyncOwner = me;
J
Jiri Denemark 已提交
1437 1438 1439
    qemuDomainObjSaveJob(driver, obj);
}

1440
void
1441 1442
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
1443 1444 1445
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

1446 1447 1448 1449 1450 1451 1452
    if (!priv->job.asyncJob)
        return;

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

void
1453
qemuDomainObjDiscardAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
1454 1455 1456 1457 1458 1459
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
1460
    qemuDomainObjSaveJob(driver, obj);
1461 1462
}

1463 1464 1465 1466 1467 1468 1469 1470 1471
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()) {
1472
        VIR_WARN("'%s' async job is owned by thread %llu",
1473 1474 1475 1476 1477 1478
                 qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                 priv->job.asyncOwner);
    }
    priv->job.asyncOwner = 0;
}

1479
static bool
1480
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
1481 1482
{
    return !priv->job.asyncJob || (priv->job.mask & JOB_MASK(job)) != 0;
1483 1484
}

1485
bool
1486
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, qemuDomainJob job)
1487 1488 1489 1490
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

1491 1492 1493
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

1494
/*
1495
 * obj must be locked before calling
1496
 */
1497
static int ATTRIBUTE_NONNULL(1)
1498
qemuDomainObjBeginJobInternal(virQEMUDriverPtr driver,
1499
                              virDomainObjPtr obj,
1500 1501
                              qemuDomainJob job,
                              qemuDomainAsyncJob asyncJob)
1502 1503
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
1504
    unsigned long long now;
1505
    unsigned long long then;
1506
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
J
Jiri Denemark 已提交
1507
    bool async = job == QEMU_JOB_ASYNC;
1508
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1509
    const char *blocker = NULL;
J
Jiri Denemark 已提交
1510
    int ret = -1;
J
Jiri Denemark 已提交
1511 1512
    unsigned long long duration = 0;
    unsigned long long asyncDuration = 0;
J
Jiri Denemark 已提交
1513
    const char *jobStr;
1514

J
Jiri Denemark 已提交
1515 1516 1517 1518 1519 1520 1521 1522 1523
    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));
1524

1525 1526
    if (virTimeMillisNow(&now) < 0) {
        virObjectUnref(cfg);
1527
        return -1;
1528 1529
    }

1530
    priv->jobs_queued++;
J
Jiri Denemark 已提交
1531
    then = now + QEMU_JOB_WAIT_TIME;
1532

1533
 retry:
1534 1535
    if (cfg->maxQueuedJobs &&
        priv->jobs_queued > cfg->maxQueuedJobs) {
1536 1537 1538
        goto error;
    }

1539
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
1540
        VIR_DEBUG("Waiting for async job (vm=%p name=%s)", obj, obj->def->name);
1541
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->parent.lock, then) < 0)
1542 1543 1544
            goto error;
    }

1545
    while (priv->job.active) {
1546
        VIR_DEBUG("Waiting for job (vm=%p name=%s)", obj, obj->def->name);
1547
        if (virCondWaitUntil(&priv->job.cond, &obj->parent.lock, then) < 0)
1548
            goto error;
1549
    }
1550 1551 1552

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
1553
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
1554 1555
        goto retry;

1556
    qemuDomainObjResetJob(priv);
1557

J
Jiri Denemark 已提交
1558 1559
    ignore_value(virTimeMillisNow(&now));

1560
    if (job != QEMU_JOB_ASYNC) {
1561
        VIR_DEBUG("Started job: %s (async=%s vm=%p name=%s)",
1562
                   qemuDomainJobTypeToString(job),
1563 1564
                  qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                  obj, obj->def->name);
1565
        priv->job.active = job;
1566
        priv->job.owner = virThreadSelfID();
1567
        priv->job.ownerAPI = virThreadJobGet();
J
Jiri Denemark 已提交
1568
        priv->job.started = now;
1569
    } else {
1570 1571 1572
        VIR_DEBUG("Started async job: %s (vm=%p name=%s)",
                  qemuDomainAsyncJobTypeToString(asyncJob),
                  obj, obj->def->name);
1573
        qemuDomainObjResetAsyncJob(priv);
J
Jiri Denemark 已提交
1574 1575
        if (VIR_ALLOC(priv->job.current) < 0)
            goto cleanup;
1576
        priv->job.asyncJob = asyncJob;
1577
        priv->job.asyncOwner = virThreadSelfID();
1578
        priv->job.asyncOwnerAPI = virThreadJobGet();
J
Jiri Denemark 已提交
1579
        priv->job.asyncStarted = now;
J
Jiri Denemark 已提交
1580
        priv->job.current->started = now;
1581
    }
1582

1583 1584
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
1585

1586
    virObjectUnref(cfg);
1587
    return 0;
1588

1589
 error:
J
Jiri Denemark 已提交
1590 1591 1592 1593 1594 1595
    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;

1596
    VIR_WARN("Cannot start job (%s, %s) for domain %s; "
J
Jiri Denemark 已提交
1597 1598
             "current job is (%s, %s) owned by (%llu %s, %llu %s) "
             "for (%llus, %llus)",
1599 1600 1601 1602 1603
             qemuDomainJobTypeToString(job),
             qemuDomainAsyncJobTypeToString(asyncJob),
             obj->def->name,
             qemuDomainJobTypeToString(priv->job.active),
             qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
1604
             priv->job.owner, NULLSTR(priv->job.ownerAPI),
J
Jiri Denemark 已提交
1605 1606
             priv->job.asyncOwner, NULLSTR(priv->job.asyncOwnerAPI),
             duration / 1000, asyncDuration / 1000);
1607 1608 1609 1610 1611

    if (nested || qemuDomainNestedJobAllowed(priv, job))
        blocker = priv->job.ownerAPI;
    else
        blocker = priv->job.asyncOwnerAPI;
1612

1613 1614
    ret = -1;
    if (errno == ETIMEDOUT) {
1615 1616 1617 1618 1619 1620 1621 1622
        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"));
        }
1623 1624 1625
        ret = -2;
    } else if (cfg->maxQueuedJobs &&
               priv->jobs_queued > cfg->maxQueuedJobs) {
1626 1627 1628 1629 1630 1631 1632 1633 1634 1635
        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"));
        }
1636 1637
        ret = -2;
    } else {
1638
        virReportSystemError(errno, "%s", _("cannot acquire job mutex"));
1639
    }
J
Jiri Denemark 已提交
1640 1641

 cleanup:
1642
    priv->jobs_queued--;
1643
    virObjectUnref(cfg);
1644
    return ret;
1645 1646 1647
}

/*
1648
 * obj must be locked before calling
1649 1650 1651 1652
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
1653
 * Successful calls must be followed by EndJob eventually
1654
 */
1655
int qemuDomainObjBeginJob(virQEMUDriverPtr driver,
1656
                          virDomainObjPtr obj,
1657
                          qemuDomainJob job)
1658
{
1659 1660 1661 1662 1663
    if (qemuDomainObjBeginJobInternal(driver, obj, job,
                                      QEMU_ASYNC_JOB_NONE) < 0)
        return -1;
    else
        return 0;
1664 1665
}

1666
int qemuDomainObjBeginAsyncJob(virQEMUDriverPtr driver,
1667
                               virDomainObjPtr obj,
1668
                               qemuDomainAsyncJob asyncJob)
1669
{
1670 1671 1672 1673 1674
    if (qemuDomainObjBeginJobInternal(driver, obj, QEMU_JOB_ASYNC,
                                      asyncJob) < 0)
        return -1;
    else
        return 0;
1675 1676
}

1677
static int ATTRIBUTE_RETURN_CHECK
1678 1679
qemuDomainObjBeginNestedJob(virQEMUDriverPtr driver,
                            virDomainObjPtr obj,
1680
                            qemuDomainAsyncJob asyncJob)
1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
{
    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()) {
1691
        VIR_WARN("This thread doesn't seem to be the async job owner: %llu",
1692 1693 1694 1695 1696 1697 1698 1699
                 priv->job.asyncOwner);
    }

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

1700

1701
/*
1702
 * obj must be locked and have a reference before calling
1703 1704 1705 1706
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 */
1707 1708
void
qemuDomainObjEndJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
1709 1710
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
1711
    qemuDomainJob job = priv->job.active;
1712

1713 1714
    priv->jobs_queued--;

1715
    VIR_DEBUG("Stopping job: %s (async=%s vm=%p name=%s)",
1716
              qemuDomainJobTypeToString(job),
1717 1718
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
1719

1720
    qemuDomainObjResetJob(priv);
1721 1722
    if (qemuDomainTrackJob(job))
        qemuDomainObjSaveJob(driver, obj);
1723
    virCondSignal(&priv->job.cond);
1724 1725
}

1726
void
1727
qemuDomainObjEndAsyncJob(virQEMUDriverPtr driver, virDomainObjPtr obj)
1728 1729
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
1730

1731 1732
    priv->jobs_queued--;

1733 1734 1735
    VIR_DEBUG("Stopping async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
1736

1737
    qemuDomainObjResetAsyncJob(priv);
1738
    qemuDomainObjSaveJob(driver, obj);
1739 1740 1741
    virCondBroadcast(&priv->job.asyncCond);
}

1742 1743 1744 1745 1746
void
qemuDomainObjAbortAsyncJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

1747 1748 1749
    VIR_DEBUG("Requesting abort of async job: %s (vm=%p name=%s)",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              obj, obj->def->name);
1750

1751 1752
    priv->job.abortJob = true;
    virDomainObjBroadcast(obj);
1753 1754
}

1755 1756 1757 1758 1759 1760 1761 1762 1763
/*
 * 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
 */
1764
static int
1765
qemuDomainObjEnterMonitorInternal(virQEMUDriverPtr driver,
1766
                                  virDomainObjPtr obj,
1767
                                  qemuDomainAsyncJob asyncJob)
1768 1769 1770
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

1771
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
1772 1773 1774
        int ret;
        if ((ret = qemuDomainObjBeginNestedJob(driver, obj, asyncJob)) < 0)
            return ret;
1775
        if (!virDomainObjIsActive(obj)) {
1776 1777
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
1778
            qemuDomainObjEndJob(driver, obj);
1779 1780
            return -1;
        }
1781 1782 1783
    } 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");
1784 1785
    }

1786 1787
    VIR_DEBUG("Entering monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
1788
    virObjectLock(priv->mon);
1789
    virObjectRef(priv->mon);
1790
    ignore_value(virTimeMillisNow(&priv->monStart));
1791
    virObjectUnlock(obj);
1792 1793

    return 0;
1794 1795
}

1796
static void ATTRIBUTE_NONNULL(1)
1797
qemuDomainObjExitMonitorInternal(virQEMUDriverPtr driver,
1798
                                 virDomainObjPtr obj)
1799 1800
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
1801
    bool hasRefs;
1802

1803
    hasRefs = virObjectUnref(priv->mon);
1804

1805
    if (hasRefs)
1806
        virObjectUnlock(priv->mon);
1807

1808
    virObjectLock(obj);
1809 1810
    VIR_DEBUG("Exited monitor (mon=%p vm=%p name=%s)",
              priv->mon, obj, obj->def->name);
1811

1812
    priv->monStart = 0;
1813
    if (!hasRefs)
1814
        priv->mon = NULL;
1815

1816 1817 1818 1819 1820
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED) {
        qemuDomainObjResetJob(priv);
        qemuDomainObjSaveJob(driver, obj);
        virCondSignal(&priv->job.cond);
    }
1821 1822
}

1823
void qemuDomainObjEnterMonitor(virQEMUDriverPtr driver,
1824
                               virDomainObjPtr obj)
1825
{
1826
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, obj,
1827
                                                   QEMU_ASYNC_JOB_NONE));
1828 1829
}

1830
/* obj must NOT be locked before calling
1831 1832
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
1833 1834 1835 1836 1837 1838
 *
 * 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.
1839
 */
1840 1841
int qemuDomainObjExitMonitor(virQEMUDriverPtr driver,
                             virDomainObjPtr obj)
1842
{
1843
    qemuDomainObjExitMonitorInternal(driver, obj);
1844
    if (!virDomainObjIsActive(obj)) {
1845 1846 1847
        if (!virGetLastError())
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("domain is no longer running"));
1848 1849 1850
        return -1;
    }
    return 0;
1851
}
1852 1853

/*
1854
 * obj must be locked before calling
1855 1856
 *
 * To be called immediately before any QEMU monitor API call.
1857
 * Must have already either called qemuDomainObjBeginJob()
1858 1859 1860 1861 1862
 * 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
1863 1864 1865
 * 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).
1866 1867
 */
int
1868
qemuDomainObjEnterMonitorAsync(virQEMUDriverPtr driver,
1869
                               virDomainObjPtr obj,
1870
                               qemuDomainAsyncJob asyncJob)
1871
{
1872
    return qemuDomainObjEnterMonitorInternal(driver, obj, asyncJob);
1873 1874
}

D
Daniel P. Berrange 已提交
1875 1876


1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887
/*
 * 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
1888 1889 1890
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

1891 1892
    VIR_DEBUG("Entering agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
1893
    virObjectLock(priv->agent);
1894
    virObjectRef(priv->agent);
D
Daniel P. Berrange 已提交
1895
    ignore_value(virTimeMillisNow(&priv->agentStart));
1896
    virObjectUnlock(obj);
D
Daniel P. Berrange 已提交
1897 1898
}

1899 1900 1901 1902 1903 1904 1905

/* 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
1906 1907
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
1908
    bool hasRefs;
D
Daniel P. Berrange 已提交
1909

1910
    hasRefs = virObjectUnref(priv->agent);
D
Daniel P. Berrange 已提交
1911

1912
    if (hasRefs)
1913
        virObjectUnlock(priv->agent);
D
Daniel P. Berrange 已提交
1914

1915
    virObjectLock(obj);
1916 1917
    VIR_DEBUG("Exited agent (agent=%p vm=%p name=%s)",
              priv->agent, obj, obj->def->name);
D
Daniel P. Berrange 已提交
1918 1919

    priv->agentStart = 0;
1920
    if (!hasRefs)
D
Daniel P. Berrange 已提交
1921 1922 1923
        priv->agent = NULL;
}

1924
void qemuDomainObjEnterRemote(virDomainObjPtr obj)
1925
{
1926 1927
    VIR_DEBUG("Entering remote (vm=%p name=%s)",
              obj, obj->def->name);
1928
    virObjectUnlock(obj);
1929 1930
}

1931
void qemuDomainObjExitRemote(virDomainObjPtr obj)
1932
{
1933
    virObjectLock(obj);
1934 1935
    VIR_DEBUG("Exited remote (vm=%p name=%s)",
              obj, obj->def->name);
1936
}
1937 1938


1939 1940 1941 1942 1943 1944 1945 1946
virDomainDefPtr
qemuDomainDefCopy(virQEMUDriverPtr driver,
                  virDomainDefPtr src,
                  unsigned int flags)
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    virDomainDefPtr ret = NULL;
    virCapsPtr caps = NULL;
1947
    char *xml = NULL;
1948 1949 1950 1951 1952 1953 1954 1955 1956 1957

    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,
1958
                                        VIR_DOMAIN_DEF_PARSE_INACTIVE)))
1959 1960
        goto cleanup;

1961
 cleanup:
1962 1963 1964 1965 1966
    VIR_FREE(xml);
    virObjectUnref(caps);
    return ret;
}

1967
int
1968
qemuDomainDefFormatBuf(virQEMUDriverPtr driver,
1969 1970 1971
                       virDomainDefPtr def,
                       unsigned int flags,
                       virBuffer *buf)
1972
{
1973
    int ret = -1;
1974
    virCPUDefPtr cpu = NULL;
1975
    virCPUDefPtr def_cpu = def->cpu;
1976 1977
    virDomainControllerDefPtr *controllers = NULL;
    int ncontrollers = 0;
1978 1979 1980 1981
    virCapsPtr caps = NULL;

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;
1982 1983

    /* Update guest CPU requirements according to host CPU */
1984 1985 1986
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) &&
        def_cpu &&
        (def_cpu->mode != VIR_CPU_MODE_CUSTOM || def_cpu->model)) {
1987 1988
        if (!caps->host.cpu ||
            !caps->host.cpu->model) {
1989 1990
            virReportError(VIR_ERR_OPERATION_FAILED,
                           "%s", _("cannot get host CPU capabilities"));
1991 1992 1993
            goto cleanup;
        }

1994
        if (!(cpu = virCPUDefCopy(def_cpu)) ||
1995
            cpuUpdate(cpu, caps->host.cpu) < 0)
1996 1997 1998 1999
            goto cleanup;
        def->cpu = cpu;
    }

2000
    if ((flags & VIR_DOMAIN_XML_MIGRATABLE)) {
2001
        size_t i;
2002
        int toremove = 0;
2003
        virDomainControllerDefPtr usb = NULL, pci = NULL;
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021

        /* 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);
2022
            toremove++;
2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040
        } 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
2041
            VIR_DEBUG("Removing default pci-root from domain '%s'"
2042
                      " for migration compatibility", def->name);
2043
            toremove++;
2044 2045 2046 2047
        } else {
            pci = NULL;
        }

2048
        if (toremove) {
2049 2050
            controllers = def->controllers;
            ncontrollers = def->ncontrollers;
2051
            if (VIR_ALLOC_N(def->controllers, ncontrollers - toremove) < 0) {
2052 2053 2054 2055 2056 2057
                controllers = NULL;
                goto cleanup;
            }

            def->ncontrollers = 0;
            for (i = 0; i < ncontrollers; i++) {
2058
                if (controllers[i] != usb && controllers[i] != pci)
2059 2060 2061
                    def->controllers[def->ncontrollers++] = controllers[i];
            }
        }
2062 2063


2064 2065
    }

2066 2067 2068
    ret = virDomainDefFormatInternal(def,
                                     virDomainDefFormatConvertXMLFlags(flags),
                                     buf);
2069

2070
 cleanup:
2071 2072
    def->cpu = def_cpu;
    virCPUDefFree(cpu);
2073 2074 2075 2076 2077
    if (controllers) {
        VIR_FREE(def->controllers);
        def->controllers = controllers;
        def->ncontrollers = ncontrollers;
    }
2078
    virObjectUnref(caps);
2079 2080
    return ret;
}
2081

2082
char *qemuDomainDefFormatXML(virQEMUDriverPtr driver,
2083
                             virDomainDefPtr def,
2084
                             unsigned int flags)
2085 2086 2087
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

2088
    if (qemuDomainDefFormatBuf(driver, def, flags, &buf) < 0) {
2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101
        virBufferFreeAndReset(&buf);
        return NULL;
    }

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

    return virBufferContentAndReset(&buf);
}

2102
char *qemuDomainFormatXML(virQEMUDriverPtr driver,
2103
                          virDomainObjPtr vm,
2104
                          unsigned int flags)
2105 2106 2107 2108 2109 2110 2111 2112
{
    virDomainDefPtr def;

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

2113
    return qemuDomainDefFormatXML(driver, def, flags);
2114 2115
}

2116
char *
2117
qemuDomainDefFormatLive(virQEMUDriverPtr driver,
2118
                        virDomainDefPtr def,
2119 2120
                        bool inactive,
                        bool compatible)
2121 2122 2123 2124 2125
{
    unsigned int flags = QEMU_DOMAIN_FORMAT_LIVE_FLAGS;

    if (inactive)
        flags |= VIR_DOMAIN_XML_INACTIVE;
2126 2127
    if (compatible)
        flags |= VIR_DOMAIN_XML_MIGRATABLE;
2128

2129
    return qemuDomainDefFormatXML(driver, def, flags);
2130 2131
}

2132

2133
void qemuDomainObjTaint(virQEMUDriverPtr driver,
2134
                        virDomainObjPtr obj,
2135
                        virDomainTaintFlags taint,
2136
                        qemuDomainLogContextPtr logCtxt)
2137
{
2138
    virErrorPtr orig_err = NULL;
2139
    bool closeLog = false;
2140

2141 2142 2143 2144 2145 2146 2147 2148 2149
    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));
2150 2151 2152 2153 2154

        /* We don't care about errors logging taint info, so
         * preserve original error, and clear any error that
         * is raised */
        orig_err = virSaveLastError();
2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172
        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)
2173
            virResetLastError();
2174 2175
        if (closeLog)
            qemuDomainLogContextFree(logCtxt);
2176 2177 2178 2179
        if (orig_err) {
            virSetError(orig_err);
            virFreeError(orig_err);
        }
2180 2181 2182 2183
    }
}


2184
void qemuDomainObjCheckTaint(virQEMUDriverPtr driver,
2185
                             virDomainObjPtr obj,
2186
                             qemuDomainLogContextPtr logCtxt)
2187
{
2188
    size_t i;
2189
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2190
    qemuDomainObjPrivatePtr priv = obj->privateData;
2191

2192
    if (virQEMUDriverIsPrivileged(driver) &&
2193 2194 2195
        (!cfg->clearEmulatorCapabilities ||
         cfg->user == 0 ||
         cfg->group == 0))
2196
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logCtxt);
2197

2198
    if (priv->hookRun)
2199
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOOK, logCtxt);
2200

2201 2202 2203
    if (obj->def->namespaceData) {
        qemuDomainCmdlineDefPtr qemucmd = obj->def->namespaceData;
        if (qemucmd->num_args || qemucmd->num_env)
2204
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logCtxt);
2205 2206
    }

2207
    if (obj->def->cpu && obj->def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
2208
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HOST_CPU, logCtxt);
2209

2210
    for (i = 0; i < obj->def->ndisks; i++)
2211
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logCtxt);
2212

2213 2214
    for (i = 0; i < obj->def->nhostdevs; i++)
        qemuDomainObjCheckHostdevTaint(driver, obj, obj->def->hostdevs[i],
2215
                                       logCtxt);
2216

2217
    for (i = 0; i < obj->def->nnets; i++)
2218
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logCtxt);
2219

2220
    if (obj->def->os.dtb)
2221
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_DTB, logCtxt);
2222

2223
    virObjectUnref(cfg);
2224 2225 2226
}


2227
void qemuDomainObjCheckDiskTaint(virQEMUDriverPtr driver,
2228
                                 virDomainObjPtr obj,
2229
                                 virDomainDiskDefPtr disk,
2230
                                 qemuDomainLogContextPtr logCtxt)
2231
{
2232
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2233
    int format = virDomainDiskGetFormat(disk);
2234

2235
    if ((!format || format == VIR_STORAGE_FILE_AUTO) &&
2236
        cfg->allowDiskFormatProbing)
2237
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_DISK_PROBING, logCtxt);
2238

2239 2240
    if (disk->rawio == VIR_TRISTATE_BOOL_YES)
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES,
2241
                           logCtxt);
2242

2243 2244 2245 2246
    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,
2247
                           logCtxt);
2248

2249
    virObjectUnref(cfg);
2250 2251 2252
}


2253 2254 2255
void qemuDomainObjCheckHostdevTaint(virQEMUDriverPtr driver,
                                    virDomainObjPtr obj,
                                    virDomainHostdevDefPtr hostdev,
2256
                                    qemuDomainLogContextPtr logCtxt)
2257 2258 2259 2260 2261 2262
{
    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,
2263
                               logCtxt);
2264 2265 2266
}


2267
void qemuDomainObjCheckNetTaint(virQEMUDriverPtr driver,
2268
                                virDomainObjPtr obj,
2269
                                virDomainNetDefPtr net,
2270
                                qemuDomainLogContextPtr logCtxt)
2271
{
2272 2273 2274 2275 2276 2277
    /* 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)
2278
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logCtxt);
2279
}
2280 2281


2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292
qemuDomainLogContextPtr qemuDomainLogContextNew(virQEMUDriverPtr driver,
                                                virDomainObjPtr vm,
                                                qemuDomainLogContextMode mode)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    qemuDomainLogContextPtr ctxt = NULL;
    char *logfile = NULL;

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

2293
    VIR_DEBUG("Context new %p stdioLogD=%d", ctxt, cfg->stdioLogD);
2294 2295 2296 2297
    ctxt->writefd = -1;
    ctxt->readfd = -1;
    virAtomicIntSet(&ctxt->refs, 1);

2298 2299 2300 2301
    if (cfg->stdioLogD) {
        ctxt->manager = virLogManagerNew(virQEMUDriverIsPrivileged(driver));
        if (!ctxt->manager)
            goto error;
2302

2303 2304
        if (VIR_STRDUP(ctxt->name, vm->def->name) < 0)
            goto error;
2305

2306
        memcpy(ctxt->uuid, vm->def->uuid, VIR_UUID_BUFLEN);
2307

2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322
        ctxt->writefd = virLogManagerDomainOpenLogFile(ctxt->manager,
                                                       "qemu",
                                                       vm->def->uuid,
                                                       vm->def->name,
                                                       0,
                                                       &ctxt->inode,
                                                       &ctxt->pos);
        if (ctxt->writefd < 0)
            goto error;
    } else {
        if (virAsprintf(&logfile, "%s/%s.log", cfg->logDir, vm->def->name) < 0)
            goto error;

        if ((ctxt->writefd = open(logfile, O_WRONLY | O_CREAT | O_APPEND, S_IRUSR | S_IWUSR)) < 0) {
            virReportSystemError(errno, _("failed to create logfile %s"),
2323 2324 2325
                                 logfile);
            goto error;
        }
2326
        if (virSetCloseExec(ctxt->writefd) < 0) {
2327 2328 2329 2330 2331
            virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
                                 logfile);
            goto error;
        }

2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355
        /* 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"),
                                 logfile);
            goto error;
        }

        if (mode == QEMU_DOMAIN_LOG_CONTEXT_MODE_START) {
            if ((ctxt->readfd = open(logfile, O_RDONLY, S_IRUSR | S_IWUSR)) < 0) {
                virReportSystemError(errno, _("failed to open logfile %s"),
                                     logfile);
                goto error;
            }
            if (virSetCloseExec(ctxt->readfd) < 0) {
                virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
                                     logfile);
                goto error;
            }
        }

2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383
        if ((ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END)) < 0) {
            virReportSystemError(errno, _("failed to seek in log file %s"),
                                 logfile);
            goto error;
        }
    }

    virObjectUnref(cfg);
    return ctxt;

 error:
    virObjectUnref(cfg);
    qemuDomainLogContextFree(ctxt);
    return NULL;
}


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;
2384 2385
    if (!ctxt->manager &&
        lseek(ctxt->writefd, 0, SEEK_END) < 0) {
2386
        virReportSystemError(errno, "%s",
2387
                             _("Unable to seek to end of domain logfile"));
2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407
        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)
{
2408 2409 2410 2411
    VIR_DEBUG("Context read %p manager=%p inode=%llu pos=%llu",
              ctxt, ctxt->manager,
              (unsigned long long)ctxt->inode,
              (unsigned long long)ctxt->pos);
2412
    char *buf;
2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427
    size_t buflen;
    if (ctxt->manager) {
        buf = virLogManagerDomainReadLogFile(ctxt->manager,
                                             "qemu",
                                             ctxt->uuid,
                                             ctxt->name,
                                             ctxt->inode,
                                             ctxt->pos,
                                             1024 * 128,
                                             0);
        if (!buf)
            return -1;
        buflen = strlen(buf);
    } else {
        ssize_t got;
2428

2429
        buflen = 1024 * 128;
2430

2431 2432
        /* Best effort jump to start of messages */
        ignore_value(lseek(ctxt->readfd, ctxt->pos, SEEK_SET));
2433

2434 2435
        if (VIR_ALLOC_N(buf, buflen) < 0)
            return -1;
2436

2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449
        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;
    }
2450 2451 2452

    *msg = buf;

2453
    return buflen;
2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464
}


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


void qemuDomainLogContextMarkPosition(qemuDomainLogContextPtr ctxt)
{
2465 2466 2467 2468 2469 2470 2471 2472 2473 2474
    if (ctxt->manager)
        virLogManagerDomainGetLogFilePosition(ctxt->manager,
                                              "qemu",
                                              ctxt->uuid,
                                              ctxt->name,
                                              0,
                                              &ctxt->inode,
                                              &ctxt->pos);
    else
        ctxt->pos = lseek(ctxt->writefd, 0, SEEK_END);
2475 2476 2477 2478 2479
}


void qemuDomainLogContextRef(qemuDomainLogContextPtr ctxt)
{
2480
    VIR_DEBUG("Context ref %p", ctxt);
2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492
    virAtomicIntInc(&ctxt->refs);
}


void qemuDomainLogContextFree(qemuDomainLogContextPtr ctxt)
{
    bool lastRef;

    if (!ctxt)
        return;

    lastRef = virAtomicIntDecAndTest(&ctxt->refs);
2493
    VIR_DEBUG("Context free %p lastref=%d", ctxt, lastRef);
2494 2495 2496
    if (!lastRef)
        return;

2497 2498
    virLogManagerFree(ctxt->manager);
    VIR_FREE(ctxt->name);
2499 2500 2501 2502 2503 2504
    VIR_FORCE_CLOSE(ctxt->writefd);
    VIR_FORCE_CLOSE(ctxt->readfd);
    VIR_FREE(ctxt);
}


2505 2506
/* Locate an appropriate 'qemu-img' binary.  */
const char *
2507
qemuFindQemuImgBinary(virQEMUDriverPtr driver)
2508
{
2509 2510 2511
    if (!driver->qemuImgBinary)
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("unable to find kvm-img or qemu-img"));
2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
                                virDomainSnapshotObjPtr snapshot,
                                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);
2528 2529 2530 2531
    newxml = virDomainSnapshotDefFormat(
        uuidstr, snapshot->def,
        virDomainDefFormatConvertXMLFlags(QEMU_DOMAIN_FORMAT_LIVE_FLAGS),
        1);
2532
    if (newxml == NULL)
2533 2534
        return -1;

2535
    if (virAsprintf(&snapDir, "%s/%s", snapshotDir, vm->def->name) < 0)
2536 2537 2538 2539 2540 2541 2542
        goto cleanup;
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
        goto cleanup;
    }

2543
    if (virAsprintf(&snapFile, "%s/%s.xml", snapDir, snapshot->def->name) < 0)
2544 2545
        goto cleanup;

J
Ján Tomko 已提交
2546
    ret = virXMLSaveFile(snapFile, NULL, "snapshot-edit", newxml);
2547

2548
 cleanup:
2549 2550 2551 2552 2553 2554 2555 2556
    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.  */
2557
static int
2558
qemuDomainSnapshotForEachQcow2Raw(virQEMUDriverPtr driver,
2559 2560 2561 2562 2563
                                  virDomainDefPtr def,
                                  const char *name,
                                  const char *op,
                                  bool try_all,
                                  int ndisks)
2564 2565
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
2566
    size_t i;
2567 2568 2569 2570 2571 2572 2573 2574 2575
    bool skipped = false;

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

    qemuimgarg[2] = op;
2576
    qemuimgarg[3] = name;
2577

2578
    for (i = 0; i < ndisks; i++) {
2579
        /* FIXME: we also need to handle LVM here */
2580
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
2581 2582 2583
            int format = virDomainDiskGetFormat(def->disks[i]);

            if (format > 0 && format != VIR_STORAGE_FILE_QCOW2) {
2584 2585 2586 2587 2588
                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",
2589
                             def->disks[i]->dst);
2590 2591
                    skipped = true;
                    continue;
2592 2593 2594 2595 2596
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
2597
                }
2598 2599 2600 2601
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("Disk device '%s' does not support"
                                 " snapshotting"),
                               def->disks[i]->dst);
2602 2603 2604
                return -1;
            }

2605
            qemuimgarg[4] = virDomainDiskGetSource(def->disks[i]);
2606 2607 2608 2609

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
2610
                             def->disks[i]->dst);
2611 2612
                    skipped = true;
                    continue;
2613 2614 2615 2616 2617
                } else if (STREQ(op, "-c") && i) {
                    /* We must roll back partial creation by deleting
                     * all earlier snapshots.  */
                    qemuDomainSnapshotForEachQcow2Raw(driver, def, name,
                                                      "-d", false, i);
2618 2619 2620 2621 2622 2623 2624 2625 2626
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

2627 2628 2629
/* 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
2630
qemuDomainSnapshotForEachQcow2(virQEMUDriverPtr driver,
2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646
                               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);
}

2647 2648
/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
2649
qemuDomainSnapshotDiscard(virQEMUDriverPtr driver,
2650 2651 2652 2653 2654 2655 2656 2657 2658
                          virDomainObjPtr vm,
                          virDomainSnapshotObjPtr snap,
                          bool update_current,
                          bool metadata_only)
{
    char *snapFile = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;
    virDomainSnapshotObjPtr parentsnap = NULL;
2659
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2660 2661 2662 2663 2664 2665 2666 2667 2668

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
                goto cleanup;
        } else {
            priv = vm->privateData;
2669
            qemuDomainObjEnterMonitor(driver, vm);
2670 2671
            /* we continue on even in the face of error */
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
2672
            ignore_value(qemuDomainObjExitMonitor(driver, vm));
2673 2674 2675
        }
    }

2676
    if (virAsprintf(&snapFile, "%s/%s/%s.xml", cfg->snapshotDir,
2677
                    vm->def->name, snap->def->name) < 0)
2678 2679 2680 2681
        goto cleanup;

    if (snap == vm->current_snapshot) {
        if (update_current && snap->def->parent) {
2682
            parentsnap = virDomainSnapshotFindByName(vm->snapshots,
2683 2684 2685 2686 2687 2688 2689
                                                     snap->def->parent);
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
                         snap->def->parent);
            } else {
                parentsnap->def->current = true;
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap,
2690
                                                    cfg->snapshotDir) < 0) {
2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702
                    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);
2703
    virDomainSnapshotObjListRemove(vm->snapshots, snap);
2704 2705 2706

    ret = 0;

2707
 cleanup:
2708
    VIR_FREE(snapFile);
2709
    virObjectUnref(cfg);
2710 2711 2712 2713 2714 2715 2716 2717 2718
    return ret;
}

/* Hash iterator callback to discard multiple snapshots.  */
void qemuDomainSnapshotDiscardAll(void *payload,
                                  const void *name ATTRIBUTE_UNUSED,
                                  void *data)
{
    virDomainSnapshotObjPtr snap = payload;
2719
    virQEMUSnapRemovePtr curr = data;
2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730
    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;
}

int
2731
qemuDomainSnapshotDiscardAllMetadata(virQEMUDriverPtr driver,
2732 2733
                                     virDomainObjPtr vm)
{
2734
    virQEMUSnapRemove rem;
2735 2736 2737 2738 2739

    rem.driver = driver;
    rem.vm = vm;
    rem.metadata_only = true;
    rem.err = 0;
2740 2741
    virDomainSnapshotForEach(vm->snapshots, qemuDomainSnapshotDiscardAll,
                             &rem);
2742 2743 2744 2745 2746

    return rem.err;
}

/*
2747
 * The caller must hold a lock the vm.
2748 2749
 */
void
2750
qemuDomainRemoveInactive(virQEMUDriverPtr driver,
2751 2752
                         virDomainObjPtr vm)
{
2753
    bool haveJob = true;
2754
    char *snapDir;
2755 2756 2757 2758 2759 2760 2761 2762
    virQEMUDriverConfigPtr cfg;

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

    cfg = virQEMUDriverGetConfig(driver);
2763

2764 2765 2766
    if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0)
        haveJob = false;

2767 2768 2769 2770 2771
    /* 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);
    }
2772
    else if (virAsprintf(&snapDir, "%s/%s", cfg->snapshotDir,
2773 2774
                         vm->def->name) < 0) {
        VIR_WARN("unable to remove snapshot directory %s/%s",
2775
                 cfg->snapshotDir, vm->def->name);
2776 2777 2778 2779 2780
    } else {
        if (rmdir(snapDir) < 0 && errno != ENOENT)
            VIR_WARN("unable to remove snapshot directory %s", snapDir);
        VIR_FREE(snapDir);
    }
2781 2782 2783

    virObjectRef(vm);

2784
    virDomainObjListRemove(driver->domains, vm);
2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797
    /*
     * 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);
2798
    virObjectUnref(cfg);
2799 2800

    if (haveJob)
2801
        qemuDomainObjEndJob(driver, vm);
2802 2803

    virObjectUnref(vm);
2804
}
2805 2806

void
2807
qemuDomainSetFakeReboot(virQEMUDriverPtr driver,
2808 2809 2810 2811
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2812
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2813 2814

    if (priv->fakeReboot == value)
2815
        goto cleanup;
2816 2817 2818

    priv->fakeReboot = value;

2819
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm) < 0)
2820
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
2821

2822
 cleanup:
2823
    virObjectUnref(cfg);
2824
}
M
Michal Privoznik 已提交
2825

2826 2827 2828
static int
qemuDomainCheckRemoveOptionalDisk(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
2829
                                  size_t diskIndex)
2830 2831
{
    char uuid[VIR_UUID_STRING_BUFLEN];
2832
    virObjectEventPtr event = NULL;
2833
    virDomainDiskDefPtr disk = vm->def->disks[diskIndex];
2834
    const char *src = virDomainDiskGetSource(disk);
2835 2836 2837 2838 2839

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

    VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
              "due to inaccessible source '%s'",
2840
              disk->dst, vm->def->name, uuid, src);
2841 2842 2843 2844

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

2845
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
2846 2847
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
2848
        ignore_value(virDomainDiskSetSource(disk, NULL));
2849
    } else {
2850
        event = virDomainEventDiskChangeNewFromObj(vm, src, NULL,
2851 2852
                                                   disk->info.alias,
                                                   VIR_DOMAIN_EVENT_DISK_DROP_MISSING_ON_START);
2853 2854
        virDomainDiskRemove(vm->def, diskIndex);
        virDomainDiskDefFree(disk);
2855 2856
    }

2857
    qemuDomainEventQueue(driver, event);
2858 2859 2860 2861

    return 0;
}

2862 2863 2864
static int
qemuDomainCheckDiskStartupPolicy(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm,
2865
                                 size_t diskIndex,
2866 2867
                                 bool cold_boot)
{
2868
    int startupPolicy = vm->def->disks[diskIndex]->startupPolicy;
2869
    int device = vm->def->disks[diskIndex]->device;
2870

2871
    switch ((virDomainStartupPolicy) startupPolicy) {
2872
        case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
2873 2874 2875 2876 2877 2878 2879
            /* 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;
2880 2881 2882 2883 2884 2885
            break;

        case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
            goto error;

        case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
2886
            if (cold_boot)
2887 2888 2889 2890 2891 2892 2893 2894 2895
                goto error;
            break;

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

2896
    if (qemuDomainCheckRemoveOptionalDisk(driver, vm, diskIndex) < 0)
2897
        goto error;
2898 2899 2900

    return 0;

2901
 error:
2902 2903 2904
    return -1;
}

2905

M
Michal Privoznik 已提交
2906
int
2907
qemuDomainCheckDiskPresence(virQEMUDriverPtr driver,
M
Michal Privoznik 已提交
2908
                            virDomainObjPtr vm,
2909
                            bool cold_boot)
M
Michal Privoznik 已提交
2910 2911
{
    int ret = -1;
2912
    size_t i;
M
Michal Privoznik 已提交
2913

2914
    VIR_DEBUG("Checking for disk presence");
2915
    for (i = vm->def->ndisks; i > 0; i--) {
2916 2917
        size_t idx = i - 1;
        virDomainDiskDefPtr disk = vm->def->disks[idx];
2918
        virStorageFileFormat format = virDomainDiskGetFormat(disk);
M
Michal Privoznik 已提交
2919

2920
        if (virStorageSourceIsEmpty(disk->src))
2921 2922 2923 2924 2925
            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 */
2926
        if (virStorageSourceIsLocalStorage(disk->src) &&
2927
            format >= VIR_STORAGE_FILE_NONE &&
2928
            format < VIR_STORAGE_FILE_BACKING &&
2929
            virFileExists(virDomainDiskGetSource(disk)))
M
Michal Privoznik 已提交
2930 2931
            continue;

2932
        if (qemuDomainDetermineDiskChain(driver, vm, disk, true, true) >= 0)
M
Michal Privoznik 已提交
2933
            continue;
2934

2935
        if (disk->startupPolicy &&
2936
            qemuDomainCheckDiskStartupPolicy(driver, vm, idx,
2937 2938 2939
                                             cold_boot) >= 0) {
            virResetLastError();
            continue;
M
Michal Privoznik 已提交
2940 2941
        }

2942
        goto error;
M
Michal Privoznik 已提交
2943 2944 2945 2946
    }

    ret = 0;

2947
 error:
M
Michal Privoznik 已提交
2948 2949
    return ret;
}
2950 2951 2952 2953 2954 2955 2956 2957 2958 2959

/*
 * The vm must be locked when any of the following cleanup functions is
 * called.
 */
int
qemuDomainCleanupAdd(virDomainObjPtr vm,
                     qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2960
    size_t i;
2961 2962 2963 2964 2965 2966 2967 2968 2969 2970

    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,
2971
                     priv->ncleanupCallbacks, 1) < 0)
2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982
        return -1;

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

void
qemuDomainCleanupRemove(virDomainObjPtr vm,
                        qemuDomainCleanupCallback cb)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2983
    size_t i;
2984 2985 2986 2987

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

    for (i = 0; i < priv->ncleanupCallbacks; i++) {
2988 2989 2990
        if (priv->cleanupCallbacks[i] == cb)
            VIR_DELETE_ELEMENT_INPLACE(priv->cleanupCallbacks,
                                       i, priv->ncleanupCallbacks);
2991 2992 2993 2994 2995 2996 2997 2998
    }

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

void
2999
qemuDomainCleanupRun(virQEMUDriverPtr driver,
3000 3001 3002
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
3003
    size_t i;
3004 3005 3006 3007

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

    /* run cleanup callbacks in reverse order */
3008 3009
    for (i = 0; i < priv->ncleanupCallbacks; i++) {
        if (priv->cleanupCallbacks[priv->ncleanupCallbacks - (i + 1)])
3010 3011 3012 3013 3014 3015 3016
            priv->cleanupCallbacks[i](driver, vm);
    }

    VIR_FREE(priv->cleanupCallbacks);
    priv->ncleanupCallbacks = 0;
    priv->ncleanupCallbacks_max = 0;
}
3017

3018 3019 3020
static void
qemuDomainGetImageIds(virQEMUDriverConfigPtr cfg,
                      virDomainObjPtr vm,
3021
                      virStorageSourcePtr src,
3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039
                      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;
    }

3040 3041
    if (vm && (vmlabel = virDomainDefGetSecurityLabelDef(vm->def, "dac")) &&
        vmlabel->label)
3042 3043
        virParseOwnershipIds(vmlabel->label, uid, gid);

3044
    if ((disklabel = virStorageSourceGetSecurityLabelDef(src, "dac")) &&
3045
        disklabel->label)
3046 3047 3048 3049
        virParseOwnershipIds(disklabel->label, uid, gid);
}


3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072
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;
}


3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088
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;
}


3089
int
3090
qemuDomainDetermineDiskChain(virQEMUDriverPtr driver,
3091
                             virDomainObjPtr vm,
3092
                             virDomainDiskDefPtr disk,
3093 3094
                             bool force_probe,
                             bool report_broken)
3095
{
3096 3097
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = 0;
3098 3099
    uid_t uid;
    gid_t gid;
3100

3101
    if (virStorageSourceIsEmpty(disk->src))
3102
        goto cleanup;
3103

3104
    if (disk->src->backingStore) {
3105
        if (force_probe)
3106
            virStorageSourceBackingStoreClear(disk->src);
3107
        else
3108
            goto cleanup;
3109
    }
3110

3111
    qemuDomainGetImageIds(cfg, vm, disk->src, &uid, &gid);
3112

3113
    if (virStorageFileGetMetadata(disk->src,
3114
                                  uid, gid,
3115
                                  cfg->allowDiskFormatProbing,
3116
                                  report_broken) < 0)
3117 3118
        ret = -1;

3119
 cleanup:
3120 3121
    virObjectUnref(cfg);
    return ret;
3122
}
3123

3124

3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157
bool
qemuDomainDiskSourceDiffers(virConnectPtr conn,
                            virDomainDiskDefPtr disk,
                            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;

    if (qemuGetDriveSourceString(disk->src, conn, &diskSrc) < 0 ||
        qemuGetDriveSourceString(origDisk->src, conn, &origDiskSrc) < 0)
        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;
}


3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180
/*
 * 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);
3181 3182 3183 3184 3185 3186 3187 3188 3189
    CHECK_EQ(bus, "bus", false);
    if (STRNEQ(disk->dst, orig_disk->dst)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("cannot modify field '%s' of the disk"),
                       "bus");
        return false;
    }
    CHECK_EQ(tray_status, "tray", true);
    CHECK_EQ(removable, "removable", true);
3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272

    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;
    }

3273 3274 3275 3276 3277 3278 3279 3280
    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);
    CHECK_EQ(snapshot, "snapshot", true);
3281
    /* startupPolicy is allowed to be updated. Therefore not checked here. */
3282
    CHECK_EQ(transient, "transient", true);
3283
    CHECK_EQ(info.bootIndex, "boot order", true);
3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295
    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;
    }
3296 3297 3298 3299 3300 3301

#undef CHECK_EQ

    return true;
}

3302 3303 3304
bool
qemuDomainDiskBlockJobIsActive(virDomainDiskDefPtr disk)
{
3305 3306
    qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

3307 3308 3309 3310 3311 3312 3313 3314
    if (disk->mirror) {
        virReportError(VIR_ERR_BLOCK_COPY_ACTIVE,
                       _("disk '%s' already in active block job"),
                       disk->dst);

        return true;
    }

3315
    if (diskPriv->blockjob) {
3316 3317 3318 3319 3320 3321 3322 3323 3324 3325
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("disk '%s' already in active block job"),
                       disk->dst);
        return true;
    }

    return false;
}


3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340
/**
 * 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++) {
3341 3342 3343 3344
        virDomainDiskDefPtr disk = vm->def->disks[i];
        qemuDomainDiskPrivatePtr diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);

        if (!copy_only && diskPriv->blockjob)
3345 3346
            return true;

3347
        if (disk->mirror && disk->mirrorJob == VIR_DOMAIN_BLOCK_JOB_TYPE_COPY)
3348 3349 3350 3351 3352 3353 3354
            return true;
    }

    return false;
}


3355 3356
int
qemuDomainUpdateDeviceList(virQEMUDriverPtr driver,
3357 3358
                           virDomainObjPtr vm,
                           int asyncJob)
3359 3360 3361
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char **aliases;
3362
    int rc;
3363 3364 3365 3366

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

3367 3368
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
3369 3370 3371 3372
    rc = qemuMonitorGetDeviceAliases(priv->mon, &aliases);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
    if (rc < 0)
3373 3374 3375 3376 3377 3378
        return -1;

    virStringFreeList(priv->qemuDevices);
    priv->qemuDevices = aliases;
    return 0;
}
3379

3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428

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;
}


3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444
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);

3445
 cleanup:
3446 3447 3448 3449
    virDomainDefFree(migratableDefSrc);
    virDomainDefFree(migratableDefDst);
    return ret;
}
3450 3451

bool
3452
qemuDomainAgentAvailable(virDomainObjPtr vm,
3453 3454
                         bool reportError)
{
3455 3456
    qemuDomainObjPrivatePtr priv = vm->privateData;

3457 3458 3459 3460 3461 3462 3463
    if (virDomainObjGetState(vm, NULL) != VIR_DOMAIN_RUNNING) {
        if (reportError) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("domain is not running"));
        }
        return false;
    }
3464 3465 3466 3467 3468 3469 3470 3471 3472
    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) {
3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484
        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;
3485 3486 3487 3488
        }
    }
    return true;
}
3489

3490

3491
static unsigned long long
3492
qemuDomainGetMemorySizeAlignment(virDomainDefPtr def)
3493
{
3494 3495 3496 3497 3498
    /* 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;

3499 3500 3501 3502 3503 3504 3505
    /* Align memory size. QEMU requires rounding to next 4KiB block.
     * We'll take the "traditional" path and round it to 1MiB*/

    return 1024;
}


3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520
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;
}


3521 3522 3523
int
qemuDomainAlignMemorySizes(virDomainDefPtr def)
{
3524 3525
    unsigned long long maxmemkb = virMemoryMaxValue(false) >> 10;
    unsigned long long maxmemcapped = virMemoryMaxValue(true) >> 10;
3526
    unsigned long long initialmem = 0;
3527
    unsigned long long mem;
3528
    unsigned long long align = qemuDomainGetMemorySizeAlignment(def);
3529 3530 3531 3532 3533
    size_t ncells = virDomainNumaGetNodeCount(def->numa);
    size_t i;

    /* align NUMA cell sizes if relevant */
    for (i = 0; i < ncells; i++) {
3534 3535
        mem = VIR_ROUND_UP(virDomainNumaGetNodeMemorySize(def->numa, i), align);
        initialmem += mem;
3536 3537 3538 3539 3540 3541 3542

        if (mem > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("memory size of NUMA node '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
3543
        virDomainNumaSetNodeMemorySize(def->numa, i, mem);
3544 3545
    }

3546 3547 3548 3549 3550
    /* 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);

3551 3552 3553 3554 3555 3556
    if (initialmem > maxmemcapped) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("initial memory size overflowed after alignment"));
        return -1;
    }

3557
    virDomainDefSetMemoryInitial(def, initialmem);
3558

3559
    def->mem.max_memory = VIR_ROUND_UP(def->mem.max_memory, align);
3560 3561 3562 3563 3564
    if (def->mem.max_memory > maxmemkb) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("maximum memory size overflowed after alignment"));
        return -1;
    }
3565

3566
    /* Align memory module sizes */
3567 3568
    for (i = 0; i < def->nmems; i++) {
        align = qemuDomainGetMemoryModuleSizeAlignment(def, def->mems[i]);
3569
        def->mems[i]->size = VIR_ROUND_UP(def->mems[i]->size, align);
3570 3571 3572 3573 3574 3575 3576

        if (def->mems[i]->size > maxmemkb) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("size of memory module '%zu' overflowed after "
                             "alignment"), i);
            return -1;
        }
3577
    }
3578

3579 3580
    return 0;
}
3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591


/**
 * 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
3592 3593
qemuDomainMemoryDeviceAlignSize(virDomainDefPtr def,
                                virDomainMemoryDefPtr mem)
3594
{
3595
    mem->size = VIR_ROUND_UP(mem->size, qemuDomainGetMemorySizeAlignment(def));
3596
}
3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609


/**
 * qemuDomainGetMonitor:
 * @vm: domain object
 *
 * Returns the monitor pointer corresponding to the domain object @vm.
 */
qemuMonitorPtr
qemuDomainGetMonitor(virDomainObjPtr vm)
{
    return ((qemuDomainObjPrivatePtr) vm->privateData)->mon;
}
3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625


/**
 * 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;
3626 3627
    bool asynchronous = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKJOB_ASYNC);
    bool synchronous = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKJOB_SYNC);
3628

3629
    if (!synchronous && !asynchronous) {
3630 3631 3632 3633 3634 3635
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("block jobs not supported with this QEMU binary"));
        return -1;
    }

    if (modern)
3636
        *modern = asynchronous;
3637 3638 3639

    return 0;
}
3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668


/**
 * 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.
 */
virDomainChrSourceDefPtr
qemuFindAgentConfig(virDomainDefPtr def)
{
    virDomainChrSourceDefPtr config = NULL;
    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;

        if (STREQ_NULLABLE(channel->target.name, "org.qemu.guest_agent.0")) {
            config = &channel->source;
            break;
        }
    }

    return config;
}
3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687


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"));
}
3688 3689


3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707
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;
}


3708 3709 3710 3711 3712 3713
bool
qemuDomainMachineIsS390CCW(const virDomainDef *def)
{
    return STRPREFIX(def->os.machine, "s390-ccw");
}

3714

3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761
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;
        }

3762 3763 3764 3765 3766 3767 3768
        if (virDomainNumaGetNodeCount(def->numa) != 0) {
            if (mem->targetNode == -1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("target NUMA node needs to be specifed for "
                                 "memory device"));
                return -1;
            }
3769 3770
        }

3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796
        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;
}


3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824
/**
 * 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;
3825 3826 3827

        if (qemuDomainDefValidateMemoryHotplugDevice(mem, def) < 0)
            return -1;
3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846
    }

    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;
    }

3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857
    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;
        }
3858 3859 3860 3861 3862 3863 3864 3865 3866
    }

    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;
    }

3867
    for (i = 0; i < def->nmems; i++) {
3868 3869
        hotplugMemory += def->mems[i]->size;

3870 3871 3872 3873 3874 3875
        /* already existing devices don't need to be checked on hotplug */
        if (!mem &&
            qemuDomainDefValidateMemoryHotplugDevice(def->mems[i], def) < 0)
            return -1;
    }

3876 3877 3878 3879 3880 3881 3882 3883 3884 3885
    if (hotplugMemory > hotplugSpace) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("memory device total size exceeds hotplug space"));
        return -1;
    }

    return 0;
}


3886 3887 3888 3889 3890 3891 3892 3893 3894 3895
bool
qemuDomainMachineHasBuiltinIDE(const virDomainDef *def)
{
    return qemuDomainMachineIsI440FX(def) ||
        STREQ(def->os.machine, "malta") ||
        STREQ(def->os.machine, "sun4u") ||
        STREQ(def->os.machine, "g3beige");
}


3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954
/**
 * 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 */
    if (!vm->def->memballoon ||
        vm->def->memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_NONE) {
        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;

3955 3956
        vm->def->mem.cur_balloon = balloon;
    }
3957 3958 3959

    return 0;
}
3960 3961 3962 3963 3964 3965 3966 3967


/**
 * qemuDomainGetMlockLimitBytes:
 *
 * @def: domain definition
 *
 * Returns the size of the memory in bytes that needs to be set as
3968
 * RLIMIT_MEMLOCK for the QEMU process.
3969 3970 3971 3972 3973 3974 3975 3976
 * 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
qemuDomainGetMlockLimitBytes(virDomainDefPtr def)
{
    unsigned long long memKB;

3977 3978 3979 3980 3981 3982
    /* prefer the hard limit */
    if (virMemoryLimitIsSet(def->mem.hard_limit)) {
        memKB = def->mem.hard_limit;
        goto done;
    }

3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060
    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;
    }

4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078
    /* 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.
     */
4079
    memKB = virDomainDefGetMemoryActual(def) + 1024 * 1024;
4080

4081
 done:
4082 4083
    return memKB << 10;
}
4084 4085 4086 4087 4088


/**
 * @def: domain definition
 *
4089 4090 4091
 * Returns true if the locked memory limit needs to be set or updated because
 * of domain configuration, VFIO passthrough devices or architecture-specific
 * requirements.
4092 4093 4094 4095 4096 4097 4098 4099 4100
 * */
bool
qemuDomainRequiresMlock(virDomainDefPtr def)
{
    size_t i;

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

4101 4102 4103 4104
    /* ppc64 domains need to lock some memory even when VFIO is not used */
    if (ARCH_IS_PPC64(def->os.arch))
        return true;

4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115
    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;
}