You need to sign in or sign up before continuing.
qemu_process.c 205.1 KB
Newer Older
1
/*
2
 * qemu_process.c: QEMU process management
3
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16
 *
 * 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
17
 * License along with this library.  If not, see
O
Osier Yang 已提交
18
 * <http://www.gnu.org/licenses/>.
19 20 21 22 23 24 25 26 27
 *
 */

#include <config.h>

#include <fcntl.h>
#include <unistd.h>
#include <signal.h>
#include <sys/stat.h>
R
Roman Bogorodskiy 已提交
28 29 30 31 32 33
#if defined(__linux__)
# include <linux/capability.h>
#elif defined(__FreeBSD__)
# include <sys/param.h>
# include <sys/cpuset.h>
#endif
34 35

#include "qemu_process.h"
36
#include "qemu_processpriv.h"
37
#include "qemu_alias.h"
38
#include "qemu_domain.h"
39
#include "qemu_domain_address.h"
40 41 42 43 44 45
#include "qemu_cgroup.h"
#include "qemu_capabilities.h"
#include "qemu_monitor.h"
#include "qemu_command.h"
#include "qemu_hostdev.h"
#include "qemu_hotplug.h"
46
#include "qemu_migration.h"
47
#include "qemu_interface.h"
48

49
#include "cpu/cpu.h"
50
#include "datatypes.h"
51
#include "virlog.h"
52
#include "virerror.h"
53
#include "viralloc.h"
54
#include "virhook.h"
E
Eric Blake 已提交
55
#include "virfile.h"
56
#include "virpidfile.h"
57
#include "virhostcpu.h"
58
#include "domain_audit.h"
59
#include "domain_nwfilter.h"
60
#include "locking/domain_lock.h"
61
#include "network/bridge_driver.h"
62
#include "viruuid.h"
63
#include "virprocess.h"
64
#include "virtime.h"
A
Ansis Atteka 已提交
65
#include "virnetdevtap.h"
66
#include "virnetdevopenvswitch.h"
67
#include "virnetdevmidonet.h"
68
#include "virbitmap.h"
69
#include "viratomic.h"
70
#include "virnuma.h"
71
#include "virstring.h"
72
#include "virhostdev.h"
J
John Ferlan 已提交
73
#include "secret_util.h"
74
#include "storage/storage_driver.h"
75
#include "configmake.h"
76
#include "nwfilter_conf.h"
77
#include "netdev_bandwidth_conf.h"
78 79 80

#define VIR_FROM_THIS VIR_FROM_QEMU

81 82
VIR_LOG_INIT("qemu.qemu_process");

83
/**
84
 * qemuProcessRemoveDomainStatus
85 86 87 88 89 90
 *
 * remove all state files of a domain from statedir
 *
 * Returns 0 on success
 */
static int
91
qemuProcessRemoveDomainStatus(virQEMUDriverPtr driver,
92 93 94 95
                              virDomainObjPtr vm)
{
    char ebuf[1024];
    char *file = NULL;
96
    qemuDomainObjPrivatePtr priv = vm->privateData;
97 98
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;
99

100
    if (virAsprintf(&file, "%s/%s.xml", cfg->stateDir, vm->def->name) < 0)
101
        goto cleanup;
102 103 104 105 106 107

    if (unlink(file) < 0 && errno != ENOENT && errno != ENOTDIR)
        VIR_WARN("Failed to remove domain XML for %s: %s",
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));
    VIR_FREE(file);

108 109 110
    if (priv->pidfile &&
        unlink(priv->pidfile) < 0 &&
        errno != ENOENT)
111 112 113
        VIR_WARN("Failed to remove PID file for %s: %s",
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));

114
    ret = 0;
115
 cleanup:
116 117
    virObjectUnref(cfg);
    return ret;
118 119 120 121
}


/* XXX figure out how to remove this */
122
extern virQEMUDriverPtr qemu_driver;
123

D
Daniel P. Berrange 已提交
124 125 126 127 128 129 130
/*
 * This is a callback registered with a qemuAgentPtr instance,
 * and to be invoked when the agent console hits an end of file
 * condition, or error, thus indicating VM shutdown should be
 * performed
 */
static void
131
qemuProcessHandleAgentEOF(qemuAgentPtr agent,
D
Daniel P. Berrange 已提交
132 133 134 135 136 137
                          virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv;

    VIR_DEBUG("Received EOF from agent on %p '%s'", vm, vm->def->name);

138
    virObjectLock(vm);
D
Daniel P. Berrange 已提交
139 140

    priv = vm->privateData;
141 142 143 144 145 146 147 148 149 150 151

    if (!priv->agent) {
        VIR_DEBUG("Agent freed already");
        goto unlock;
    }

    if (priv->beingDestroyed) {
        VIR_DEBUG("Domain is being destroyed, agent EOF is expected");
        goto unlock;
    }

152
    qemuAgentClose(agent);
153
    priv->agent = NULL;
154
    priv->agentError = false;
D
Daniel P. Berrange 已提交
155

156
    virObjectUnlock(vm);
157 158
    return;

159
 unlock:
160 161
    virObjectUnlock(vm);
    return;
D
Daniel P. Berrange 已提交
162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178
}


/*
 * This is invoked when there is some kind of error
 * parsing data to/from the agent. The VM can continue
 * to run, but no further agent commands will be
 * allowed
 */
static void
qemuProcessHandleAgentError(qemuAgentPtr agent ATTRIBUTE_UNUSED,
                            virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv;

    VIR_DEBUG("Received error from agent on %p '%s'", vm, vm->def->name);

179
    virObjectLock(vm);
D
Daniel P. Berrange 已提交
180 181 182 183 184

    priv = vm->privateData;

    priv->agentError = true;

185
    virObjectUnlock(vm);
D
Daniel P. Berrange 已提交
186 187 188 189 190
}

static void qemuProcessHandleAgentDestroy(qemuAgentPtr agent,
                                          virDomainObjPtr vm)
{
191 192
    VIR_DEBUG("Received destroy agent=%p vm=%p", agent, vm);

193
    virObjectUnref(vm);
D
Daniel P. Berrange 已提交
194 195 196 197 198 199 200 201 202 203
}


static qemuAgentCallbacks agentCallbacks = {
    .destroy = qemuProcessHandleAgentDestroy,
    .eofNotify = qemuProcessHandleAgentEOF,
    .errorNotify = qemuProcessHandleAgentError,
};


204
int
205
qemuConnectAgent(virQEMUDriverPtr driver, virDomainObjPtr vm)
D
Daniel P. Berrange 已提交
206 207 208
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    qemuAgentPtr agent = NULL;
209
    virDomainChrDefPtr config = qemuFindAgentConfig(vm->def);
D
Daniel P. Berrange 已提交
210 211 212 213

    if (!config)
        return 0;

214 215 216 217 218 219 220 221 222
    if (priv->agent)
        return 0;

    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_VSERPORT_CHANGE) &&
        config->state != VIR_DOMAIN_CHR_DEVICE_STATE_CONNECTED) {
        VIR_DEBUG("Deferring connecting to guest agent");
        return 0;
    }

D
Daniel P. Berrange 已提交
223 224 225 226 227 228 229 230 231
    if (virSecurityManagerSetDaemonSocketLabel(driver->securityManager,
                                               vm->def) < 0) {
        VIR_ERROR(_("Failed to set security context for agent for %s"),
                  vm->def->name);
        goto cleanup;
    }

    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted while the agent is active */
232
    virObjectRef(vm);
D
Daniel P. Berrange 已提交
233

234
    virObjectUnlock(vm);
D
Daniel P. Berrange 已提交
235 236

    agent = qemuAgentOpen(vm,
237
                          config->source,
D
Daniel P. Berrange 已提交
238 239
                          &agentCallbacks);

240
    virObjectLock(vm);
D
Daniel P. Berrange 已提交
241

242 243 244 245 246 247 248
    if (agent == NULL)
        virObjectUnref(vm);

    if (!virDomainObjIsActive(vm)) {
        qemuAgentClose(agent);
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest crashed while connecting to the guest agent"));
249
        return -1;
250 251
    }

D
Daniel P. Berrange 已提交
252 253 254 255
    if (virSecurityManagerClearSocketLabel(driver->securityManager,
                                           vm->def) < 0) {
        VIR_ERROR(_("Failed to clear security context for agent for %s"),
                  vm->def->name);
256
        qemuAgentClose(agent);
D
Daniel P. Berrange 已提交
257 258 259 260
        goto cleanup;
    }

    priv->agent = agent;
261
    if (!priv->agent)
D
Daniel P. Berrange 已提交
262 263
        VIR_INFO("Failed to connect agent for %s", vm->def->name);

264
 cleanup:
265 266 267 268 269 270 271
    if (!priv->agent) {
        VIR_WARN("Cannot connect to QEMU guest agent for %s", vm->def->name);
        priv->agentError = true;
        virResetLastError();
    }

    return 0;
D
Daniel P. Berrange 已提交
272 273 274
}


275
/*
276
 * This is a callback registered with a qemuMonitorPtr instance,
277 278 279 280 281 282
 * and to be invoked when the monitor console hits an end of file
 * condition, or error, thus indicating VM shutdown should be
 * performed
 */
static void
qemuProcessHandleMonitorEOF(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
283 284
                            virDomainObjPtr vm,
                            void *opaque)
285
{
286
    virQEMUDriverPtr driver = opaque;
287
    qemuDomainObjPrivatePtr priv;
288
    struct qemuProcessEvent *processEvent;
289

290
    virObjectLock(vm);
291

292
    VIR_DEBUG("Received EOF on %p '%s'", vm, vm->def->name);
293

294
    priv = vm->privateData;
295 296
    if (priv->beingDestroyed) {
        VIR_DEBUG("Domain is being destroyed, EOF is expected");
297
        goto cleanup;
298 299
    }

300
    if (VIR_ALLOC(processEvent) < 0)
301
        goto cleanup;
302

303 304
    processEvent->eventType = QEMU_PROCESS_EVENT_MONITOR_EOF;
    processEvent->vm = vm;
305

306 307 308 309 310
    virObjectRef(vm);
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        ignore_value(virObjectUnref(vm));
        VIR_FREE(processEvent);
        goto cleanup;
311
    }
312

313 314 315 316
    /* We don't want this EOF handler to be called over and over while the
     * thread is waiting for a job.
     */
    qemuMonitorUnregister(mon);
317

318
 cleanup:
319
    virObjectUnlock(vm);
320 321 322 323 324 325 326 327 328 329 330
}


/*
 * This is invoked when there is some kind of error
 * parsing data to/from the monitor. The VM can continue
 * to run, but no further monitor commands will be
 * allowed
 */
static void
qemuProcessHandleMonitorError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
331 332
                              virDomainObjPtr vm,
                              void *opaque)
333
{
334
    virQEMUDriverPtr driver = opaque;
335
    virObjectEventPtr event = NULL;
336 337 338

    VIR_DEBUG("Received error on %p '%s'", vm, vm->def->name);

339
    virObjectLock(vm);
340

341
    ((qemuDomainObjPrivatePtr) vm->privateData)->monError = true;
342
    event = virDomainEventControlErrorNewFromObj(vm);
343
    qemuDomainEventQueue(driver, event);
344

345
    virObjectUnlock(vm);
346 347 348
}


349
virDomainDiskDefPtr
350 351 352
qemuProcessFindDomainDiskByAlias(virDomainObjPtr vm,
                                 const char *alias)
{
353
    size_t i;
354

355
    alias = qemuAliasDiskDriveSkipPrefix(alias);
356 357 358 359 360 361 362 363 364

    for (i = 0; i < vm->def->ndisks; i++) {
        virDomainDiskDefPtr disk;

        disk = vm->def->disks[i];
        if (disk->info.alias != NULL && STREQ(disk->info.alias, alias))
            return disk;
    }

365 366 367
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("no disk found with alias %s"),
                   alias);
368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
    return NULL;
}

static int
qemuProcessGetVolumeQcowPassphrase(virConnectPtr conn,
                                   virDomainDiskDefPtr disk,
                                   char **secretRet,
                                   size_t *secretLen)
{
    char *passphrase;
    unsigned char *data;
    size_t size;
    int ret = -1;
    virStorageEncryptionPtr enc;

383
    if (!disk->src->encryption) {
384 385
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("disk %s does not have any encryption information"),
386
                       disk->src->path);
387 388
        return -1;
    }
389
    enc = disk->src->encryption;
390 391

    if (!conn) {
392 393
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("cannot find secrets without a connection"));
394 395 396 397
        goto cleanup;
    }

    if (conn->secretDriver == NULL ||
398 399
        conn->secretDriver->secretLookupByUUID == NULL ||
        conn->secretDriver->secretGetValue == NULL) {
400 401
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("secret storage not supported"));
402 403 404 405 406 407 408
        goto cleanup;
    }

    if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW ||
        enc->nsecrets != 1 ||
        enc->secrets[0]->type !=
        VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) {
409
        virReportError(VIR_ERR_XML_ERROR,
410 411
                       _("invalid <encryption> for volume %s"),
                       virDomainDiskGetSource(disk));
412 413 414
        goto cleanup;
    }

J
John Ferlan 已提交
415 416 417
    if (virSecretGetSecretString(conn, &enc->secrets[0]->seclookupdef,
                                 VIR_SECRET_USAGE_TYPE_VOLUME,
                                 &data, &size) < 0)
418 419 420 421 422
        goto cleanup;

    if (memchr(data, '\0', size) != NULL) {
        memset(data, 0, size);
        VIR_FREE(data);
423 424
        virReportError(VIR_ERR_XML_ERROR,
                       _("format='qcow' passphrase for %s must not contain a "
425
                         "'\\0'"), virDomainDiskGetSource(disk));
426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
        goto cleanup;
    }

    if (VIR_ALLOC_N(passphrase, size + 1) < 0) {
        memset(data, 0, size);
        VIR_FREE(data);
        goto cleanup;
    }
    memcpy(passphrase, data, size);
    passphrase[size] = '\0';

    memset(data, 0, size);
    VIR_FREE(data);

    *secretRet = passphrase;
    *secretLen = size;

    ret = 0;

445
 cleanup:
446 447 448 449 450 451 452 453 454
    return ret;
}

static int
qemuProcessFindVolumeQcowPassphrase(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                    virConnectPtr conn,
                                    virDomainObjPtr vm,
                                    const char *path,
                                    char **secretRet,
455 456
                                    size_t *secretLen,
                                    void *opaque ATTRIBUTE_UNUSED)
457 458 459 460
{
    virDomainDiskDefPtr disk;
    int ret = -1;

461
    virObjectLock(vm);
462 463 464 465
    if (!(disk = virDomainDiskByName(vm->def, path, true))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("no disk found with path %s"),
                       path);
466
        goto cleanup;
467
    }
468 469 470

    ret = qemuProcessGetVolumeQcowPassphrase(conn, disk, secretRet, secretLen);

471
 cleanup:
472
    virObjectUnlock(vm);
473 474 475 476 477 478
    return ret;
}


static int
qemuProcessHandleReset(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
479 480
                       virDomainObjPtr vm,
                       void *opaque)
481
{
482
    virQEMUDriverPtr driver = opaque;
483
    virObjectEventPtr event;
484
    qemuDomainObjPrivatePtr priv;
485
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
486

487
    virObjectLock(vm);
488

489
    event = virDomainEventRebootNewFromObj(vm);
490 491 492
    priv = vm->privateData;
    if (priv->agent)
        qemuAgentNotifyEvent(priv->agent, QEMU_AGENT_EVENT_RESET);
493

494
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
495
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
496

497
    virObjectUnlock(vm);
498

499
    qemuDomainEventQueue(driver, event);
500

501
    virObjectUnref(cfg);
502 503 504 505
    return 0;
}


506 507 508 509 510 511 512 513 514 515 516
/*
 * Since we have the '-no-shutdown' flag set, the
 * QEMU process will currently have guest OS shutdown
 * and the CPUS stopped. To fake the reboot, we thus
 * want todo a reset of the virtual hardware, followed
 * by restart of the CPUs. This should result in the
 * guest OS booting up again
 */
static void
qemuProcessFakeReboot(void *opaque)
{
517
    virQEMUDriverPtr driver = qemu_driver;
518 519
    virDomainObjPtr vm = opaque;
    qemuDomainObjPrivatePtr priv = vm->privateData;
520
    virObjectEventPtr event = NULL;
521
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
522
    virDomainRunningReason reason = VIR_DOMAIN_RUNNING_BOOTED;
523
    int ret = -1, rc;
524

525
    VIR_DEBUG("vm=%p", vm);
526
    virObjectLock(vm);
527
    if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0)
528 529 530
        goto cleanup;

    if (!virDomainObjIsActive(vm)) {
531 532
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
533 534 535
        goto endjob;
    }

536
    qemuDomainObjEnterMonitor(driver, vm);
537 538 539
    rc = qemuMonitorSystemReset(priv->mon);

    if (qemuDomainObjExitMonitor(driver, vm) < 0)
540 541
        goto endjob;

542
    if (rc < 0)
543 544
        goto endjob;

545 546 547
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_CRASHED)
        reason = VIR_DOMAIN_RUNNING_CRASHED;

548
    if (qemuProcessStartCPUs(driver, vm, NULL,
549
                             reason,
550
                             QEMU_ASYNC_JOB_NONE) < 0) {
551
        if (virGetLastError() == NULL)
552 553
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("resume operation failed"));
554 555
        goto endjob;
    }
556
    priv->gotShutdown = false;
557
    event = virDomainEventLifecycleNewFromObj(vm,
558 559 560
                                     VIR_DOMAIN_EVENT_RESUMED,
                                     VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);

561
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
562 563 564 565
        VIR_WARN("Unable to save status on vm %s after state change",
                 vm->def->name);
    }

566 567
    ret = 0;

568
 endjob:
569
    qemuDomainObjEndJob(driver, vm);
570

571
 cleanup:
572 573
    if (ret == -1)
        ignore_value(qemuProcessKill(vm, VIR_QEMU_PROCESS_KILL_FORCE));
M
Michal Privoznik 已提交
574
    virDomainObjEndAPI(&vm);
575
    qemuDomainEventQueue(driver, event);
576
    virObjectUnref(cfg);
577 578 579
}


580
void
581
qemuProcessShutdownOrReboot(virQEMUDriverPtr driver,
582
                            virDomainObjPtr vm)
583
{
584 585 586
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (priv->fakeReboot) {
587
        qemuDomainSetFakeReboot(driver, vm, false);
588
        virObjectRef(vm);
589 590 591 592 593
        virThread th;
        if (virThreadCreate(&th,
                            false,
                            qemuProcessFakeReboot,
                            vm) < 0) {
594
            VIR_ERROR(_("Failed to create reboot thread, killing domain"));
595
            ignore_value(qemuProcessKill(vm, VIR_QEMU_PROCESS_KILL_NOWAIT));
596
            virObjectUnref(vm);
597 598
        }
    } else {
599
        ignore_value(qemuProcessKill(vm, VIR_QEMU_PROCESS_KILL_NOWAIT));
600
    }
601
}
602

603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623

static int
qemuProcessHandleEvent(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                       virDomainObjPtr vm,
                       const char *eventName,
                       long long seconds,
                       unsigned int micros,
                       const char *details,
                       void *opaque)
{
    virQEMUDriverPtr driver = opaque;
    virObjectEventPtr event = NULL;

    VIR_DEBUG("vm=%p", vm);

    virObjectLock(vm);
    event = virDomainQemuMonitorEventNew(vm->def->id, vm->def->name,
                                         vm->def->uuid, eventName,
                                         seconds, micros, details);

    virObjectUnlock(vm);
624
    qemuDomainEventQueue(driver, event);
625 626 627 628 629

    return 0;
}


630 631
static int
qemuProcessHandleShutdown(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
632 633
                          virDomainObjPtr vm,
                          void *opaque)
634
{
635
    virQEMUDriverPtr driver = opaque;
636
    qemuDomainObjPrivatePtr priv;
637
    virObjectEventPtr event = NULL;
638
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
639

640 641
    VIR_DEBUG("vm=%p", vm);

642
    virObjectLock(vm);
643 644 645 646 647 648

    priv = vm->privateData;
    if (priv->gotShutdown) {
        VIR_DEBUG("Ignoring repeated SHUTDOWN event from domain %s",
                  vm->def->name);
        goto unlock;
649 650 651 652
    } else if (!virDomainObjIsActive(vm)) {
        VIR_DEBUG("Ignoring SHUTDOWN event from inactive domain %s",
                  vm->def->name);
        goto unlock;
653 654 655 656 657 658 659 660
    }
    priv->gotShutdown = true;

    VIR_DEBUG("Transitioned guest %s to shutdown state",
              vm->def->name);
    virDomainObjSetState(vm,
                         VIR_DOMAIN_SHUTDOWN,
                         VIR_DOMAIN_SHUTDOWN_UNKNOWN);
661
    event = virDomainEventLifecycleNewFromObj(vm,
662 663 664
                                     VIR_DOMAIN_EVENT_SHUTDOWN,
                                     VIR_DOMAIN_EVENT_SHUTDOWN_FINISHED);

665
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
666 667 668 669
        VIR_WARN("Unable to save status on vm %s after state change",
                 vm->def->name);
    }

670 671 672
    if (priv->agent)
        qemuAgentNotifyEvent(priv->agent, QEMU_AGENT_EVENT_SHUTDOWN);

673 674
    qemuProcessShutdownOrReboot(driver, vm);

675
 unlock:
676
    virObjectUnlock(vm);
677
    qemuDomainEventQueue(driver, event);
678
    virObjectUnref(cfg);
679

680 681 682 683 684 685
    return 0;
}


static int
qemuProcessHandleStop(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
686 687
                      virDomainObjPtr vm,
                      void *opaque)
688
{
689
    virQEMUDriverPtr driver = opaque;
690
    virObjectEventPtr event = NULL;
691 692
    virDomainPausedReason reason = VIR_DOMAIN_PAUSED_UNKNOWN;
    virDomainEventSuspendedDetailType detail = VIR_DOMAIN_EVENT_SUSPENDED_PAUSED;
693
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
694

695
    virObjectLock(vm);
J
Jiri Denemark 已提交
696
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
697
        qemuDomainObjPrivatePtr priv = vm->privateData;
698

699
        if (priv->gotShutdown) {
700 701
            VIR_DEBUG("Ignoring STOP event after SHUTDOWN");
            goto unlock;
702 703
        }

704
        if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_OUT) {
705 706 707 708 709 710 711 712
            if (priv->job.current->stats.status ==
                        QEMU_MONITOR_MIGRATION_STATUS_POSTCOPY) {
                reason = VIR_DOMAIN_PAUSED_POSTCOPY;
                detail = VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY;
            } else {
                reason = VIR_DOMAIN_PAUSED_MIGRATION;
                detail = VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED;
            }
713 714 715 716
        }

        VIR_DEBUG("Transitioned guest %s to paused state, reason %s",
                  vm->def->name, virDomainPausedReasonTypeToString(reason));
717

718 719 720
        if (priv->job.current)
            ignore_value(virTimeMillisNow(&priv->job.current->stopped));

721 722 723 724
        if (priv->signalStop)
            virDomainObjBroadcast(vm);

        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, reason);
725
        event = virDomainEventLifecycleNewFromObj(vm,
726 727
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  detail);
728

729 730 731 732 733
        VIR_FREE(priv->lockState);
        if (virDomainLockProcessPause(driver->lockManager, vm, &priv->lockState) < 0)
            VIR_WARN("Unable to release lease on %s", vm->def->name);
        VIR_DEBUG("Preserving lock state '%s'", NULLSTR(priv->lockState));

734
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
735 736 737
            VIR_WARN("Unable to save status on vm %s after state change",
                     vm->def->name);
        }
738
    }
739

740
 unlock:
741
    virObjectUnlock(vm);
742
    qemuDomainEventQueue(driver, event);
743
    virObjectUnref(cfg);
744 745 746 747 748

    return 0;
}


749 750
static int
qemuProcessHandleResume(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
751 752
                        virDomainObjPtr vm,
                        void *opaque)
753
{
754
    virQEMUDriverPtr driver = opaque;
755
    virObjectEventPtr event = NULL;
756
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
757

758
    virObjectLock(vm);
759 760 761 762 763 764 765 766 767 768 769 770 771
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED) {
        qemuDomainObjPrivatePtr priv = vm->privateData;

        if (priv->gotShutdown) {
            VIR_DEBUG("Ignoring RESUME event after SHUTDOWN");
            goto unlock;
        }

        VIR_DEBUG("Transitioned guest %s out of paused into resumed state",
                  vm->def->name);

        virDomainObjSetState(vm, VIR_DOMAIN_RUNNING,
                                 VIR_DOMAIN_RUNNING_UNPAUSED);
772
        event = virDomainEventLifecycleNewFromObj(vm,
773 774 775 776
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);

        VIR_DEBUG("Using lock state '%s' on resume event", NULLSTR(priv->lockState));
777
        if (virDomainLockProcessResume(driver->lockManager, cfg->uri,
778 779 780 781 782 783 784 785 786
                                       vm, priv->lockState) < 0) {
            /* Don't free priv->lockState on error, because we need
             * to make sure we have state still present if the user
             * tries to resume again
             */
            goto unlock;
        }
        VIR_FREE(priv->lockState);

787
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
788 789 790 791 792
            VIR_WARN("Unable to save status on vm %s after state change",
                     vm->def->name);
        }
    }

793
 unlock:
794
    virObjectUnlock(vm);
795
    qemuDomainEventQueue(driver, event);
796
    virObjectUnref(cfg);
797 798 799
    return 0;
}

800 801 802
static int
qemuProcessHandleRTCChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                           virDomainObjPtr vm,
803 804
                           long long offset,
                           void *opaque)
805
{
806
    virQEMUDriverPtr driver = opaque;
807
    virObjectEventPtr event = NULL;
808
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
809

810
    virObjectLock(vm);
811

812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828
    if (vm->def->clock.offset == VIR_DOMAIN_CLOCK_OFFSET_VARIABLE) {
        /* when a basedate is manually given on the qemu commandline
         * rather than simply "-rtc base=utc", the offset sent by qemu
         * in this event is *not* the new offset from UTC, but is
         * instead the new offset from the *original basedate* +
         * uptime. For example, if the original offset was 3600 and
         * the guest clock has been advanced by 10 seconds, qemu will
         * send "10" in the event - this means that the new offset
         * from UTC is 3610, *not* 10. If the guest clock is advanced
         * by another 10 seconds, qemu will now send "20" - i.e. each
         * event is the sum of the most recent change and all previous
         * changes since the domain was started. Fortunately, we have
         * saved the initial offset in "adjustment0", so to arrive at
         * the proper new "adjustment", we just add the most recent
         * offset to adjustment0.
         */
        offset += vm->def->clock.data.variable.adjustment0;
829
        vm->def->clock.data.variable.adjustment = offset;
830

831
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
832 833 834 835
           VIR_WARN("unable to save domain status with RTC change");
    }

    event = virDomainEventRTCChangeNewFromObj(vm, offset);
836

837
    virObjectUnlock(vm);
838

839
    qemuDomainEventQueue(driver, event);
840
    virObjectUnref(cfg);
841 842 843 844 845 846 847
    return 0;
}


static int
qemuProcessHandleWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm,
848 849
                          int action,
                          void *opaque)
850
{
851
    virQEMUDriverPtr driver = opaque;
852 853
    virObjectEventPtr watchdogEvent = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
854
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
855

856
    virObjectLock(vm);
857 858 859
    watchdogEvent = virDomainEventWatchdogNewFromObj(vm, action);

    if (action == VIR_DOMAIN_EVENT_WATCHDOG_PAUSE &&
J
Jiri Denemark 已提交
860
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
861
        qemuDomainObjPrivatePtr priv = vm->privateData;
862 863
        VIR_DEBUG("Transitioned guest %s to paused state due to watchdog", vm->def->name);

J
Jiri Denemark 已提交
864
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_WATCHDOG);
865
        lifecycleEvent = virDomainEventLifecycleNewFromObj(vm,
866 867 868
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  VIR_DOMAIN_EVENT_SUSPENDED_WATCHDOG);

869 870 871 872 873
        VIR_FREE(priv->lockState);
        if (virDomainLockProcessPause(driver->lockManager, vm, &priv->lockState) < 0)
            VIR_WARN("Unable to release lease on %s", vm->def->name);
        VIR_DEBUG("Preserving lock state '%s'", NULLSTR(priv->lockState));

874
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
875 876 877
            VIR_WARN("Unable to save status on vm %s after watchdog event",
                     vm->def->name);
        }
878 879 880
    }

    if (vm->def->watchdog->action == VIR_DOMAIN_WATCHDOG_ACTION_DUMP) {
C
Chen Fan 已提交
881 882 883 884 885
        struct qemuProcessEvent *processEvent;
        if (VIR_ALLOC(processEvent) == 0) {
            processEvent->eventType = QEMU_PROCESS_EVENT_WATCHDOG;
            processEvent->action = VIR_DOMAIN_WATCHDOG_ACTION_DUMP;
            processEvent->vm = vm;
W
Wen Congyang 已提交
886 887 888
            /* Hold an extra reference because we can't allow 'vm' to be
             * deleted before handling watchdog event is finished.
             */
889
            virObjectRef(vm);
C
Chen Fan 已提交
890
            if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
891
                if (!virObjectUnref(vm))
892
                    vm = NULL;
C
Chen Fan 已提交
893
                VIR_FREE(processEvent);
W
Wen Congyang 已提交
894
            }
895
        }
896 897
    }

898
    if (vm)
899
        virObjectUnlock(vm);
900 901
    qemuDomainEventQueue(driver, watchdogEvent);
    qemuDomainEventQueue(driver, lifecycleEvent);
902

903
    virObjectUnref(cfg);
904 905 906 907 908 909 910 911 912
    return 0;
}


static int
qemuProcessHandleIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virDomainObjPtr vm,
                         const char *diskAlias,
                         int action,
913 914
                         const char *reason,
                         void *opaque)
915
{
916
    virQEMUDriverPtr driver = opaque;
917 918 919
    virObjectEventPtr ioErrorEvent = NULL;
    virObjectEventPtr ioErrorEvent2 = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
920 921 922
    const char *srcPath;
    const char *devAlias;
    virDomainDiskDefPtr disk;
923
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
924

925
    virObjectLock(vm);
926 927 928
    disk = qemuProcessFindDomainDiskByAlias(vm, diskAlias);

    if (disk) {
929
        srcPath = virDomainDiskGetSource(disk);
930 931 932 933 934 935 936 937 938 939
        devAlias = disk->info.alias;
    } else {
        srcPath = "";
        devAlias = "";
    }

    ioErrorEvent = virDomainEventIOErrorNewFromObj(vm, srcPath, devAlias, action);
    ioErrorEvent2 = virDomainEventIOErrorReasonNewFromObj(vm, srcPath, devAlias, action, reason);

    if (action == VIR_DOMAIN_EVENT_IO_ERROR_PAUSE &&
J
Jiri Denemark 已提交
940
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
941
        qemuDomainObjPrivatePtr priv = vm->privateData;
942 943
        VIR_DEBUG("Transitioned guest %s to paused state due to IO error", vm->def->name);

944 945 946
        if (priv->signalIOError)
            virDomainObjBroadcast(vm);

J
Jiri Denemark 已提交
947
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_IOERROR);
948
        lifecycleEvent = virDomainEventLifecycleNewFromObj(vm,
949 950 951
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  VIR_DOMAIN_EVENT_SUSPENDED_IOERROR);

952 953 954 955 956
        VIR_FREE(priv->lockState);
        if (virDomainLockProcessPause(driver->lockManager, vm, &priv->lockState) < 0)
            VIR_WARN("Unable to release lease on %s", vm->def->name);
        VIR_DEBUG("Preserving lock state '%s'", NULLSTR(priv->lockState));

957
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
958 959
            VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name);
    }
960
    virObjectUnlock(vm);
961

962 963 964
    qemuDomainEventQueue(driver, ioErrorEvent);
    qemuDomainEventQueue(driver, ioErrorEvent2);
    qemuDomainEventQueue(driver, lifecycleEvent);
965
    virObjectUnref(cfg);
966 967 968
    return 0;
}

969 970 971 972 973
static int
qemuProcessHandleBlockJob(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm,
                          const char *diskAlias,
                          int type,
974 975
                          int status,
                          void *opaque)
976
{
977
    virQEMUDriverPtr driver = opaque;
978
    struct qemuProcessEvent *processEvent = NULL;
979
    virDomainDiskDefPtr disk;
980
    qemuDomainDiskPrivatePtr diskPriv;
981
    char *data = NULL;
982

983
    virObjectLock(vm);
984

985 986
    VIR_DEBUG("Block job for device %s (domain: %p,%s) type %d status %d",
              diskAlias, vm, vm->def->name, type, status);
987

988
    if (!(disk = qemuProcessFindDomainDiskByAlias(vm, diskAlias)))
989
        goto error;
990
    diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
991

992
    if (diskPriv->blockJobSync) {
993
        /* We have a SYNC API waiting for this event, dispatch it back */
994 995
        diskPriv->blockJobType = type;
        diskPriv->blockJobStatus = status;
996
        virDomainObjBroadcast(vm);
997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008
    } else {
        /* there is no waiting SYNC API, dispatch the update to a thread */
        if (VIR_ALLOC(processEvent) < 0)
            goto error;

        processEvent->eventType = QEMU_PROCESS_EVENT_BLOCK_JOB;
        if (VIR_STRDUP(data, diskAlias) < 0)
            goto error;
        processEvent->data = data;
        processEvent->vm = vm;
        processEvent->action = type;
        processEvent->status = status;
1009

1010 1011 1012 1013 1014
        virObjectRef(vm);
        if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
            ignore_value(virObjectUnref(vm));
            goto error;
        }
1015 1016
    }

1017
 cleanup:
1018
    virObjectUnlock(vm);
1019
    return 0;
1020 1021 1022 1023 1024
 error:
    if (processEvent)
        VIR_FREE(processEvent->data);
    VIR_FREE(processEvent);
    goto cleanup;
1025
}
1026

1027

1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039
static int
qemuProcessHandleGraphics(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm,
                          int phase,
                          int localFamily,
                          const char *localNode,
                          const char *localService,
                          int remoteFamily,
                          const char *remoteNode,
                          const char *remoteService,
                          const char *authScheme,
                          const char *x509dname,
1040 1041
                          const char *saslUsername,
                          void *opaque)
1042
{
1043
    virQEMUDriverPtr driver = opaque;
1044
    virObjectEventPtr event;
1045 1046 1047
    virDomainEventGraphicsAddressPtr localAddr = NULL;
    virDomainEventGraphicsAddressPtr remoteAddr = NULL;
    virDomainEventGraphicsSubjectPtr subject = NULL;
1048
    size_t i;
1049 1050

    if (VIR_ALLOC(localAddr) < 0)
1051
        goto error;
1052
    localAddr->family = localFamily;
1053 1054 1055
    if (VIR_STRDUP(localAddr->service, localService) < 0 ||
        VIR_STRDUP(localAddr->node, localNode) < 0)
        goto error;
1056 1057

    if (VIR_ALLOC(remoteAddr) < 0)
1058
        goto error;
1059
    remoteAddr->family = remoteFamily;
1060 1061 1062
    if (VIR_STRDUP(remoteAddr->service, remoteService) < 0 ||
        VIR_STRDUP(remoteAddr->node, remoteNode) < 0)
        goto error;
1063 1064

    if (VIR_ALLOC(subject) < 0)
1065
        goto error;
1066 1067
    if (x509dname) {
        if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0)
1068
            goto error;
1069
        subject->nidentity++;
1070 1071 1072
        if (VIR_STRDUP(subject->identities[subject->nidentity-1].type, "x509dname") < 0 ||
            VIR_STRDUP(subject->identities[subject->nidentity-1].name, x509dname) < 0)
            goto error;
1073 1074 1075
    }
    if (saslUsername) {
        if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0)
1076
            goto error;
1077
        subject->nidentity++;
1078 1079 1080
        if (VIR_STRDUP(subject->identities[subject->nidentity-1].type, "saslUsername") < 0 ||
            VIR_STRDUP(subject->identities[subject->nidentity-1].name, saslUsername) < 0)
            goto error;
1081 1082
    }

1083
    virObjectLock(vm);
1084
    event = virDomainEventGraphicsNewFromObj(vm, phase, localAddr, remoteAddr, authScheme, subject);
1085
    virObjectUnlock(vm);
1086

1087
    qemuDomainEventQueue(driver, event);
1088 1089 1090

    return 0;

1091
 error:
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102
    if (localAddr) {
        VIR_FREE(localAddr->service);
        VIR_FREE(localAddr->node);
        VIR_FREE(localAddr);
    }
    if (remoteAddr) {
        VIR_FREE(remoteAddr->service);
        VIR_FREE(remoteAddr->node);
        VIR_FREE(remoteAddr);
    }
    if (subject) {
1103
        for (i = 0; i < subject->nidentity; i++) {
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113
            VIR_FREE(subject->identities[i].type);
            VIR_FREE(subject->identities[i].name);
        }
        VIR_FREE(subject->identities);
        VIR_FREE(subject);
    }

    return -1;
}

1114 1115 1116 1117
static int
qemuProcessHandleTrayChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                            virDomainObjPtr vm,
                            const char *devAlias,
1118 1119
                            int reason,
                            void *opaque)
1120
{
1121
    virQEMUDriverPtr driver = opaque;
1122
    virObjectEventPtr event = NULL;
1123
    virDomainDiskDefPtr disk;
1124
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1125

1126
    virObjectLock(vm);
1127 1128 1129 1130 1131 1132
    disk = qemuProcessFindDomainDiskByAlias(vm, devAlias);

    if (disk) {
        event = virDomainEventTrayChangeNewFromObj(vm,
                                                   devAlias,
                                                   reason);
1133 1134 1135 1136 1137 1138
        /* Update disk tray status */
        if (reason == VIR_DOMAIN_EVENT_TRAY_CHANGE_OPEN)
            disk->tray_status = VIR_DOMAIN_DISK_TRAY_OPEN;
        else if (reason == VIR_DOMAIN_EVENT_TRAY_CHANGE_CLOSE)
            disk->tray_status = VIR_DOMAIN_DISK_TRAY_CLOSED;

1139
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1140 1141 1142
            VIR_WARN("Unable to save status on vm %s after tray moved event",
                     vm->def->name);
        }
1143 1144

        virDomainObjBroadcast(vm);
1145 1146
    }

1147
    virObjectUnlock(vm);
1148
    qemuDomainEventQueue(driver, event);
1149
    virObjectUnref(cfg);
1150 1151 1152
    return 0;
}

O
Osier Yang 已提交
1153 1154
static int
qemuProcessHandlePMWakeup(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1155 1156
                          virDomainObjPtr vm,
                          void *opaque)
O
Osier Yang 已提交
1157
{
1158
    virQEMUDriverPtr driver = opaque;
1159 1160
    virObjectEventPtr event = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
1161
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
O
Osier Yang 已提交
1162

1163
    virObjectLock(vm);
O
Osier Yang 已提交
1164 1165
    event = virDomainEventPMWakeupNewFromObj(vm);

1166 1167 1168 1169 1170 1171 1172 1173 1174
    /* Don't set domain status back to running if it wasn't paused
     * from guest side, otherwise it can just cause confusion.
     */
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PMSUSPENDED) {
        VIR_DEBUG("Transitioned guest %s from pmsuspended to running "
                  "state due to QMP wakeup event", vm->def->name);

        virDomainObjSetState(vm, VIR_DOMAIN_RUNNING,
                             VIR_DOMAIN_RUNNING_WAKEUP);
1175
        lifecycleEvent = virDomainEventLifecycleNewFromObj(vm,
1176 1177 1178
                                                  VIR_DOMAIN_EVENT_STARTED,
                                                  VIR_DOMAIN_EVENT_STARTED_WAKEUP);

1179
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1180 1181 1182 1183 1184
            VIR_WARN("Unable to save status on vm %s after wakeup event",
                     vm->def->name);
        }
    }

1185
    virObjectUnlock(vm);
1186 1187
    qemuDomainEventQueue(driver, event);
    qemuDomainEventQueue(driver, lifecycleEvent);
1188
    virObjectUnref(cfg);
O
Osier Yang 已提交
1189 1190
    return 0;
}
1191

O
Osier Yang 已提交
1192 1193
static int
qemuProcessHandlePMSuspend(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1194 1195
                           virDomainObjPtr vm,
                           void *opaque)
O
Osier Yang 已提交
1196
{
1197
    virQEMUDriverPtr driver = opaque;
1198 1199
    virObjectEventPtr event = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
1200
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
O
Osier Yang 已提交
1201

1202
    virObjectLock(vm);
O
Osier Yang 已提交
1203 1204
    event = virDomainEventPMSuspendNewFromObj(vm);

1205
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
1206
        qemuDomainObjPrivatePtr priv = vm->privateData;
1207 1208 1209 1210 1211
        VIR_DEBUG("Transitioned guest %s to pmsuspended state due to "
                  "QMP suspend event", vm->def->name);

        virDomainObjSetState(vm, VIR_DOMAIN_PMSUSPENDED,
                             VIR_DOMAIN_PMSUSPENDED_UNKNOWN);
J
Jiri Denemark 已提交
1212
        lifecycleEvent =
1213
            virDomainEventLifecycleNewFromObj(vm,
J
Jiri Denemark 已提交
1214 1215
                                     VIR_DOMAIN_EVENT_PMSUSPENDED,
                                     VIR_DOMAIN_EVENT_PMSUSPENDED_MEMORY);
1216

1217
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1218 1219 1220
            VIR_WARN("Unable to save status on vm %s after suspend event",
                     vm->def->name);
        }
1221 1222 1223

        if (priv->agent)
            qemuAgentNotifyEvent(priv->agent, QEMU_AGENT_EVENT_SUSPEND);
1224 1225
    }

1226
    virObjectUnlock(vm);
O
Osier Yang 已提交
1227

1228 1229
    qemuDomainEventQueue(driver, event);
    qemuDomainEventQueue(driver, lifecycleEvent);
1230
    virObjectUnref(cfg);
O
Osier Yang 已提交
1231 1232 1233
    return 0;
}

1234 1235 1236
static int
qemuProcessHandleBalloonChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
1237 1238
                               unsigned long long actual,
                               void *opaque)
1239
{
1240
    virQEMUDriverPtr driver = opaque;
1241
    virObjectEventPtr event = NULL;
1242
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1243

1244
    virObjectLock(vm);
1245 1246 1247 1248 1249 1250
    event = virDomainEventBalloonChangeNewFromObj(vm, actual);

    VIR_DEBUG("Updating balloon from %lld to %lld kb",
              vm->def->mem.cur_balloon, actual);
    vm->def->mem.cur_balloon = actual;

1251
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
1252 1253
        VIR_WARN("unable to save domain status with balloon change");

1254
    virObjectUnlock(vm);
1255

1256
    qemuDomainEventQueue(driver, event);
1257
    virObjectUnref(cfg);
1258 1259 1260
    return 0;
}

1261 1262
static int
qemuProcessHandlePMSuspendDisk(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1263 1264
                               virDomainObjPtr vm,
                               void *opaque)
1265
{
1266
    virQEMUDriverPtr driver = opaque;
1267 1268
    virObjectEventPtr event = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
1269
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1270

1271
    virObjectLock(vm);
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281
    event = virDomainEventPMSuspendDiskNewFromObj(vm);

    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
        qemuDomainObjPrivatePtr priv = vm->privateData;
        VIR_DEBUG("Transitioned guest %s to pmsuspended state due to "
                  "QMP suspend_disk event", vm->def->name);

        virDomainObjSetState(vm, VIR_DOMAIN_PMSUSPENDED,
                             VIR_DOMAIN_PMSUSPENDED_UNKNOWN);
        lifecycleEvent =
1282
            virDomainEventLifecycleNewFromObj(vm,
1283 1284 1285
                                     VIR_DOMAIN_EVENT_PMSUSPENDED,
                                     VIR_DOMAIN_EVENT_PMSUSPENDED_DISK);

1286
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1287 1288 1289 1290 1291 1292 1293 1294
            VIR_WARN("Unable to save status on vm %s after suspend event",
                     vm->def->name);
        }

        if (priv->agent)
            qemuAgentNotifyEvent(priv->agent, QEMU_AGENT_EVENT_SUSPEND);
    }

1295
    virObjectUnlock(vm);
1296

1297 1298
    qemuDomainEventQueue(driver, event);
    qemuDomainEventQueue(driver, lifecycleEvent);
1299 1300
    virObjectUnref(cfg);

1301 1302 1303
    return 0;
}

1304

1305 1306
static int
qemuProcessHandleGuestPanic(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1307 1308
                            virDomainObjPtr vm,
                            void *opaque)
1309
{
1310
    virQEMUDriverPtr driver = opaque;
1311 1312 1313
    struct qemuProcessEvent *processEvent;

    virObjectLock(vm);
1314
    if (VIR_ALLOC(processEvent) < 0)
1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329
        goto cleanup;

    processEvent->eventType = QEMU_PROCESS_EVENT_GUESTPANIC;
    processEvent->action = vm->def->onCrash;
    processEvent->vm = vm;
    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted before handling guest panic event is finished.
     */
    virObjectRef(vm);
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        if (!virObjectUnref(vm))
            vm = NULL;
        VIR_FREE(processEvent);
    }

1330
 cleanup:
1331
    if (vm)
1332
        virObjectUnlock(vm);
1333 1334 1335 1336 1337

    return 0;
}


1338
int
1339 1340
qemuProcessHandleDeviceDeleted(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
1341 1342
                               const char *devAlias,
                               void *opaque)
1343
{
1344
    virQEMUDriverPtr driver = opaque;
1345 1346
    struct qemuProcessEvent *processEvent = NULL;
    char *data;
1347 1348 1349 1350 1351 1352

    virObjectLock(vm);

    VIR_DEBUG("Device %s removed from domain %p %s",
              devAlias, vm, vm->def->name);

1353 1354
    if (qemuDomainSignalDeviceRemoval(vm, devAlias,
                                      QEMU_DOMAIN_UNPLUGGING_DEVICE_STATUS_OK))
1355
        goto cleanup;
1356

1357 1358
    if (VIR_ALLOC(processEvent) < 0)
        goto error;
1359

1360 1361 1362 1363 1364
    processEvent->eventType = QEMU_PROCESS_EVENT_DEVICE_DELETED;
    if (VIR_STRDUP(data, devAlias) < 0)
        goto error;
    processEvent->data = data;
    processEvent->vm = vm;
1365

1366 1367 1368 1369 1370
    virObjectRef(vm);
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        ignore_value(virObjectUnref(vm));
        goto error;
    }
1371

1372
 cleanup:
1373 1374
    virObjectUnlock(vm);
    return 0;
1375 1376 1377 1378 1379
 error:
    if (processEvent)
        VIR_FREE(processEvent->data);
    VIR_FREE(processEvent);
    goto cleanup;
1380 1381 1382
}


1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450
/**
 *
 * Meaning of fields reported by the event according to the ACPI standard:
 * @source:
 *  0x00 - 0xff: Notification values, as passed at the request time
 *  0x100: Operating System Shutdown Processing
 *  0x103: Ejection processing
 *  0x200: Insertion processing
 *  other values are reserved
 *
 * @status:
 *   general values
 *     0x00: success
 *     0x01: non-specific failure
 *     0x02: unrecognized notify code
 *     0x03 - 0x7f: reserved
 *     other values are specific to the notification type
 *
 *   for the 0x100 source the following additional codes are standardized
 *     0x80: OS Shutdown request denied
 *     0x81: OS Shutdown in progress
 *     0x82: OS Shutdown completed
 *     0x83: OS Graceful shutdown not supported
 *     other values are reserved
 *
 * Other fields and semantics are specific to the qemu handling of the event.
 *  - @alias may be NULL for successful unplug operations
 *  - @slotType describes the device type a bit more closely, currently the
 *    only known value is 'DIMM'
 *  - @slot describes the specific device
 *
 *  Note that qemu does not emit the event for all the documented sources or
 *  devices.
 */
static int
qemuProcessHandleAcpiOstInfo(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                             virDomainObjPtr vm,
                             const char *alias,
                             const char *slotType,
                             const char *slot,
                             unsigned int source,
                             unsigned int status,
                             void *opaque)
{
    virQEMUDriverPtr driver = opaque;
    virObjectEventPtr event = NULL;

    virObjectLock(vm);

    VIR_DEBUG("ACPI OST info for device %s domain %p %s. "
              "slotType='%s' slot='%s' source=%u status=%u",
              NULLSTR(alias), vm, vm->def->name, slotType, slot, source, status);

    /* handle memory unplug failure */
    if (STREQ(slotType, "DIMM") && alias && status == 1) {
        qemuDomainSignalDeviceRemoval(vm, alias,
                                      QEMU_DOMAIN_UNPLUGGING_DEVICE_STATUS_GUEST_REJECTED);

        event = virDomainEventDeviceRemovalFailedNewFromObj(vm, alias);
    }

    virObjectUnlock(vm);
    qemuDomainEventQueue(driver, event);

    return 0;
}


1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491
static int
qemuProcessHandleNicRxFilterChanged(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                    virDomainObjPtr vm,
                                    const char *devAlias,
                                    void *opaque)
{
    virQEMUDriverPtr driver = opaque;
    struct qemuProcessEvent *processEvent = NULL;
    char *data;

    virObjectLock(vm);

    VIR_DEBUG("Device %s RX Filter changed in domain %p %s",
              devAlias, vm, vm->def->name);

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

    processEvent->eventType = QEMU_PROCESS_EVENT_NIC_RX_FILTER_CHANGED;
    if (VIR_STRDUP(data, devAlias) < 0)
        goto error;
    processEvent->data = data;
    processEvent->vm = vm;

    virObjectRef(vm);
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        ignore_value(virObjectUnref(vm));
        goto error;
    }

 cleanup:
    virObjectUnlock(vm);
    return 0;
 error:
    if (processEvent)
        VIR_FREE(processEvent->data);
    VIR_FREE(processEvent);
    goto cleanup;
}


1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534
static int
qemuProcessHandleSerialChanged(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
                               const char *devAlias,
                               bool connected,
                               void *opaque)
{
    virQEMUDriverPtr driver = opaque;
    struct qemuProcessEvent *processEvent = NULL;
    char *data;

    virObjectLock(vm);

    VIR_DEBUG("Serial port %s state changed to '%d' in domain %p %s",
              devAlias, connected, vm, vm->def->name);

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

    processEvent->eventType = QEMU_PROCESS_EVENT_SERIAL_CHANGED;
    if (VIR_STRDUP(data, devAlias) < 0)
        goto error;
    processEvent->data = data;
    processEvent->action = connected;
    processEvent->vm = vm;

    virObjectRef(vm);
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        ignore_value(virObjectUnref(vm));
        goto error;
    }

 cleanup:
    virObjectUnlock(vm);
    return 0;
 error:
    if (processEvent)
        VIR_FREE(processEvent->data);
    VIR_FREE(processEvent);
    goto cleanup;
}


1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553
static int
qemuProcessHandleSpiceMigrated(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
                               void *opaque ATTRIBUTE_UNUSED)
{
    qemuDomainObjPrivatePtr priv;

    virObjectLock(vm);

    VIR_DEBUG("Spice migration completed for domain %p %s",
              vm, vm->def->name);

    priv = vm->privateData;
    if (priv->job.asyncJob != QEMU_ASYNC_JOB_MIGRATION_OUT) {
        VIR_DEBUG("got SPICE_MIGRATE_COMPLETED event without a migration job");
        goto cleanup;
    }

    priv->job.spiceMigrated = true;
1554
    virDomainObjBroadcast(vm);
1555 1556 1557 1558 1559 1560 1561

 cleanup:
    virObjectUnlock(vm);
    return 0;
}


1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576
static int
qemuProcessHandleMigrationStatus(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                 virDomainObjPtr vm,
                                 int status,
                                 void *opaque ATTRIBUTE_UNUSED)
{
    qemuDomainObjPrivatePtr priv;

    virObjectLock(vm);

    VIR_DEBUG("Migration of domain %p %s changed state to %s",
              vm, vm->def->name,
              qemuMonitorMigrationStatusTypeToString(status));

    priv = vm->privateData;
1577
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_NONE) {
1578 1579 1580 1581
        VIR_DEBUG("got MIGRATION event without a migration job");
        goto cleanup;
    }

1582
    priv->job.current->stats.status = status;
1583 1584 1585 1586 1587 1588 1589 1590
    virDomainObjBroadcast(vm);

 cleanup:
    virObjectUnlock(vm);
    return 0;
}


1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619
static int
qemuProcessHandleMigrationPass(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
                               int pass,
                               void *opaque)
{
    virQEMUDriverPtr driver = opaque;
    qemuDomainObjPrivatePtr priv;

    virObjectLock(vm);

    VIR_DEBUG("Migrating domain %p %s, iteration %d",
              vm, vm->def->name, pass);

    priv = vm->privateData;
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_NONE) {
        VIR_DEBUG("got MIGRATION_PASS event without a migration job");
        goto cleanup;
    }

    qemuDomainEventQueue(driver,
                         virDomainEventMigrationIterationNewFromObj(vm, pass));

 cleanup:
    virObjectUnlock(vm);
    return 0;
}


1620 1621
static qemuMonitorCallbacks monitorCallbacks = {
    .eofNotify = qemuProcessHandleMonitorEOF,
1622
    .errorNotify = qemuProcessHandleMonitorError,
1623
    .diskSecretLookup = qemuProcessFindVolumeQcowPassphrase,
1624
    .domainEvent = qemuProcessHandleEvent,
1625 1626
    .domainShutdown = qemuProcessHandleShutdown,
    .domainStop = qemuProcessHandleStop,
1627
    .domainResume = qemuProcessHandleResume,
1628 1629 1630 1631 1632
    .domainReset = qemuProcessHandleReset,
    .domainRTCChange = qemuProcessHandleRTCChange,
    .domainWatchdog = qemuProcessHandleWatchdog,
    .domainIOError = qemuProcessHandleIOError,
    .domainGraphics = qemuProcessHandleGraphics,
1633
    .domainBlockJob = qemuProcessHandleBlockJob,
1634
    .domainTrayChange = qemuProcessHandleTrayChange,
O
Osier Yang 已提交
1635
    .domainPMWakeup = qemuProcessHandlePMWakeup,
O
Osier Yang 已提交
1636
    .domainPMSuspend = qemuProcessHandlePMSuspend,
1637
    .domainBalloonChange = qemuProcessHandleBalloonChange,
1638
    .domainPMSuspendDisk = qemuProcessHandlePMSuspendDisk,
1639
    .domainGuestPanic = qemuProcessHandleGuestPanic,
1640
    .domainDeviceDeleted = qemuProcessHandleDeviceDeleted,
1641
    .domainNicRxFilterChanged = qemuProcessHandleNicRxFilterChanged,
1642
    .domainSerialChange = qemuProcessHandleSerialChanged,
1643
    .domainSpiceMigrated = qemuProcessHandleSpiceMigrated,
1644
    .domainMigrationStatus = qemuProcessHandleMigrationStatus,
1645
    .domainMigrationPass = qemuProcessHandleMigrationPass,
1646
    .domainAcpiOstInfo = qemuProcessHandleAcpiOstInfo,
1647 1648
};

1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661
static void
qemuProcessMonitorReportLogError(qemuMonitorPtr mon,
                                 const char *msg,
                                 void *opaque);


static void
qemuProcessMonitorLogFree(void *opaque)
{
    qemuDomainLogContextPtr logCtxt = opaque;
    qemuDomainLogContextFree(logCtxt);
}

1662
static int
1663
qemuConnectMonitor(virQEMUDriverPtr driver, virDomainObjPtr vm, int asyncJob,
1664
                   qemuDomainLogContextPtr logCtxt)
1665 1666 1667
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret = -1;
1668
    qemuMonitorPtr mon = NULL;
1669

1670
    if (virSecurityManagerSetDaemonSocketLabel(driver->securityManager,
1671
                                               vm->def) < 0) {
1672 1673
        VIR_ERROR(_("Failed to set security context for monitor for %s"),
                  vm->def->name);
1674
        return -1;
1675 1676 1677
    }

    /* Hold an extra reference because we can't allow 'vm' to be
C
Chen Hanxiao 已提交
1678
     * deleted until the monitor gets its own reference. */
1679
    virObjectRef(vm);
1680

1681
    ignore_value(virTimeMillisNow(&priv->monStart));
1682
    virObjectUnlock(vm);
1683 1684 1685 1686

    mon = qemuMonitorOpen(vm,
                          priv->monConfig,
                          priv->monJSON,
1687 1688
                          &monitorCallbacks,
                          driver);
1689

1690 1691 1692 1693 1694 1695 1696
    if (mon && logCtxt) {
        qemuDomainLogContextRef(logCtxt);
        qemuMonitorSetDomainLog(mon,
                                qemuProcessMonitorReportLogError,
                                logCtxt,
                                qemuProcessMonitorLogFree);
    }
1697

1698
    virObjectLock(vm);
M
Michal Privoznik 已提交
1699
    virObjectUnref(vm);
1700
    priv->monStart = 0;
1701

M
Michal Privoznik 已提交
1702
    if (!virDomainObjIsActive(vm)) {
1703
        qemuMonitorClose(mon);
1704
        mon = NULL;
1705 1706 1707
    }
    priv->mon = mon;

1708
    if (virSecurityManagerClearSocketLabel(driver->securityManager, vm->def) < 0) {
1709 1710
        VIR_ERROR(_("Failed to clear security context for monitor for %s"),
                  vm->def->name);
1711
        return -1;
1712 1713 1714 1715
    }

    if (priv->mon == NULL) {
        VIR_INFO("Failed to connect monitor for %s", vm->def->name);
1716
        return -1;
1717 1718 1719
    }


1720
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
1721
        return -1;
1722

1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736
    if (qemuMonitorSetCapabilities(priv->mon) < 0)
        goto cleanup;

    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MONITOR_JSON) &&
        virQEMUCapsProbeQMP(priv->qemuCaps, priv->mon) < 0)
        goto cleanup;

    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATION_EVENT) &&
        qemuMonitorSetMigrationCapability(priv->mon,
                                          QEMU_MONITOR_MIGRATION_CAPS_EVENTS,
                                          true) < 0) {
        VIR_DEBUG("Cannot enable migration events; clearing capability");
        virQEMUCapsClear(priv->qemuCaps, QEMU_CAPS_MIGRATION_EVENT);
    }
1737

1738 1739 1740 1741 1742
    ret = 0;

 cleanup:
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
1743 1744 1745
    return ret;
}

1746 1747 1748

/**
 * qemuProcessReadLog: Read log file of a qemu VM
1749
 * @logCtxt: the domain log context
1750
 * @msg: pointer to buffer to store the read messages in
1751 1752
 *
 * Reads log of a qemu VM. Skips messages not produced by qemu or irrelevant
1753
 * messages. Returns returns 0 on success or -1 on error
1754
 */
1755
static int
1756
qemuProcessReadLog(qemuDomainLogContextPtr logCtxt, char **msg)
1757
{
1758 1759
    char *buf;
    ssize_t got;
1760
    char *eol;
1761
    char *filter_next;
1762

1763
    if ((got = qemuDomainLogContextRead(logCtxt, &buf)) < 0)
1764
        return -1;
1765

1766 1767 1768 1769 1770 1771 1772
    /* Filter out debug messages from intermediate libvirt process */
    filter_next = buf;
    while ((eol = strchr(filter_next, '\n'))) {
        *eol = '\0';
        if (virLogProbablyLogMessage(filter_next) ||
            STRPREFIX(filter_next, "char device redirected to")) {
            size_t skip = (eol + 1) - filter_next;
1773
            memmove(filter_next, eol + 1, buf + got - eol);
1774 1775 1776 1777
            got -= skip;
        } else {
            filter_next = eol + 1;
            *eol = '\n';
1778 1779
        }
    }
1780
    filter_next = NULL; /* silence false coverity warning */
1781

1782 1783
    if (got > 0 &&
        buf[got - 1] == '\n') {
1784 1785
        buf[got - 1] = '\0';
        got--;
1786
    }
1787
    ignore_value(VIR_REALLOC_N_QUIET(buf, got + 1));
1788 1789 1790
    *msg = buf;
    return 0;
}
1791 1792


1793 1794
static int
qemuProcessReportLogError(qemuDomainLogContextPtr logCtxt,
1795 1796 1797 1798
                          const char *msgprefix)
{
    char *logmsg = NULL;

1799
    if (qemuProcessReadLog(logCtxt, &logmsg) < 0)
1800 1801 1802
        return -1;

    virResetLastError();
1803 1804 1805 1806 1807
    if (virStringIsEmpty(logmsg))
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", msgprefix);
    else
        virReportError(VIR_ERR_INTERNAL_ERROR, _("%s: %s"), msgprefix, logmsg);

1808 1809
    VIR_FREE(logmsg);
    return 0;
1810 1811 1812
}


1813 1814 1815 1816 1817 1818 1819 1820 1821 1822
static void
qemuProcessMonitorReportLogError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                 const char *msg,
                                 void *opaque)
{
    qemuDomainLogContextPtr logCtxt = opaque;
    qemuProcessReportLogError(logCtxt, msg);
}


1823
static int
1824 1825 1826
qemuProcessLookupPTYs(virDomainDefPtr def,
                      virQEMUCapsPtr qemuCaps,
                      virDomainChrDefPtr *devices,
1827
                      int count,
1828
                      virHashTablePtr info)
1829
{
1830
    size_t i;
1831

1832
    for (i = 0; i < count; i++) {
1833
        virDomainChrDefPtr chr = devices[i];
1834 1835
        bool chardevfmt = virQEMUCapsSupportsChardev(def, qemuCaps, chr);

1836
        if (chr->source->type == VIR_DOMAIN_CHR_TYPE_PTY) {
C
Cole Robinson 已提交
1837
            char id[32];
1838
            qemuMonitorChardevInfoPtr entry;
1839

C
Cole Robinson 已提交
1840
            if (snprintf(id, sizeof(id), "%s%s",
1841
                         chardevfmt ? "char" : "",
1842 1843 1844 1845
                         chr->info.alias) >= sizeof(id)) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("failed to format device alias "
                                 "for PTY retrieval"));
1846
                return -1;
1847
            }
1848

1849 1850
            entry = virHashLookup(info, id);
            if (!entry || !entry->ptyPath) {
1851
                if (chr->source->data.file.path == NULL) {
1852 1853 1854
                    /* neither the log output nor 'info chardev' had a
                     * pty path for this chardev, report an error
                     */
1855 1856
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("no assigned pty for device %s"), id);
1857 1858 1859 1860 1861 1862 1863 1864 1865
                    return -1;
                } else {
                    /* 'info chardev' had no pty path for this chardev,
                     * but the log output had, so we're fine
                     */
                    continue;
                }
            }

1866 1867
            VIR_FREE(chr->source->data.file.path);
            if (VIR_STRDUP(chr->source->data.file.path, entry->ptyPath) < 0)
1868 1869 1870 1871 1872 1873 1874
                return -1;
        }
    }

    return 0;
}

1875 1876
static int
qemuProcessFindCharDevicePTYsMonitor(virDomainObjPtr vm,
1877
                                     virQEMUCapsPtr qemuCaps,
1878
                                     virHashTablePtr info)
1879
{
1880
    size_t i = 0;
C
Cole Robinson 已提交
1881

1882 1883
    if (qemuProcessLookupPTYs(vm->def, qemuCaps,
                              vm->def->serials, vm->def->nserials,
1884
                              info) < 0)
1885 1886
        return -1;

1887 1888
    if (qemuProcessLookupPTYs(vm->def, qemuCaps,
                              vm->def->parallels, vm->def->nparallels,
1889
                              info) < 0)
1890
        return -1;
1891

1892 1893
    if (qemuProcessLookupPTYs(vm->def, qemuCaps,
                              vm->def->channels, vm->def->nchannels,
1894
                              info) < 0)
1895
        return -1;
1896 1897 1898 1899
    /* For historical reasons, console[0] can be just an alias
     * for serial[0]. That's why we need to update it as well. */
    if (vm->def->nconsoles) {
        virDomainChrDefPtr chr = vm->def->consoles[0];
1900

1901 1902 1903 1904 1905
        if (vm->def->nserials &&
            chr->deviceType == VIR_DOMAIN_CHR_DEVICE_TYPE_CONSOLE &&
            chr->targetType == VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_SERIAL) {
            /* yes, the first console is just an alias for serials[0] */
            i = 1;
1906 1907
            if (virDomainChrSourceDefCopy(chr->source,
                                          ((vm->def->serials[0])->source)) < 0)
1908 1909 1910 1911
                return -1;
        }
    }

1912 1913
    if (qemuProcessLookupPTYs(vm->def, qemuCaps,
                              vm->def->consoles + i, vm->def->nconsoles - i,
1914
                              info) < 0)
1915
        return -1;
1916 1917 1918 1919 1920

    return 0;
}


1921
static int
1922 1923 1924 1925
qemuProcessRefreshChannelVirtioState(virQEMUDriverPtr driver,
                                     virDomainObjPtr vm,
                                     virHashTablePtr info,
                                     int booted)
1926 1927
{
    size_t i;
1928
    int agentReason = VIR_CONNECT_DOMAIN_EVENT_AGENT_LIFECYCLE_REASON_CHANNEL;
1929
    qemuMonitorChardevInfoPtr entry;
1930
    virObjectEventPtr event = NULL;
1931 1932
    char id[32];

1933 1934 1935
    if (booted)
        agentReason = VIR_CONNECT_DOMAIN_EVENT_AGENT_LIFECYCLE_REASON_DOMAIN_STARTED;

1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951
    for (i = 0; i < vm->def->nchannels; i++) {
        virDomainChrDefPtr chr = vm->def->channels[i];
        if (chr->targetType == VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO) {
            if (snprintf(id, sizeof(id), "char%s",
                         chr->info.alias) >= sizeof(id)) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("failed to format device alias "
                                 "for PTY retrieval"));
                return -1;
            }

            /* port state not reported */
            if (!(entry = virHashLookup(info, id)) ||
                !entry->state)
                continue;

1952 1953 1954 1955 1956 1957
            if (entry->state != VIR_DOMAIN_CHR_DEVICE_STATE_DEFAULT &&
                STREQ_NULLABLE(chr->target.name, "org.qemu.guest_agent.0") &&
                (event = virDomainEventAgentLifecycleNewFromObj(vm, entry->state,
                                                                agentReason)))
                qemuDomainEventQueue(driver, event);

1958 1959 1960 1961 1962 1963 1964 1965
            chr->state = entry->state;
        }
    }

    return 0;
}


1966 1967
int
qemuRefreshVirtioChannelState(virQEMUDriverPtr driver,
1968 1969
                              virDomainObjPtr vm,
                              qemuDomainAsyncJob asyncJob)
1970 1971 1972 1973 1974
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virHashTablePtr info = NULL;
    int ret = -1;

1975 1976 1977
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;

1978
    ret = qemuMonitorGetChardevInfo(priv->mon, &info);
1979 1980
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
1981 1982 1983 1984

    if (ret < 0)
        goto cleanup;

1985
    ret = qemuProcessRefreshChannelVirtioState(driver, vm, info, false);
1986 1987 1988 1989 1990 1991

 cleanup:
    virHashFree(info);
    return ret;
}

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
static void
qemuRefreshRTC(virQEMUDriverPtr driver,
               virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    time_t now, then;
    struct tm thenbits;
    long localOffset;
    int rv;

    if (vm->def->clock.offset != VIR_DOMAIN_CLOCK_OFFSET_VARIABLE)
        return;

    memset(&thenbits, 0, sizeof(thenbits));
    qemuDomainObjEnterMonitor(driver, vm);
    now = time(NULL);
    rv = qemuMonitorGetRTCTime(priv->mon, &thenbits);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        rv = -1;

    if (rv < 0)
        return;

    thenbits.tm_isdst = -1;
    if ((then = mktime(&thenbits)) == (time_t) -1) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Unable to convert time"));
        return;
    }

    /* Thing is, @now is in local TZ but @then in UTC. */
    if (virTimeLocalOffsetFromUTC(&localOffset) < 0)
        return;

    vm->def->clock.data.variable.adjustment = then - now + localOffset;
}
2028

2029
int
2030 2031 2032 2033 2034 2035 2036 2037 2038
qemuProcessRefreshBalloonState(virQEMUDriverPtr driver,
                               virDomainObjPtr vm,
                               int asyncJob)
{
    unsigned long long balloon;
    int rc;

    /* if no ballooning is available, the current size equals to the current
     * full memory size */
2039
    if (!virDomainDefHasMemballoon(vm->def)) {
2040
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
2041 2042 2043 2044 2045 2046 2047
        return 0;
    }

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

    rc = qemuMonitorGetBalloonInfo(qemuDomainGetMonitor(vm), &balloon);
2048
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || rc < 0)
2049 2050 2051 2052 2053 2054 2055 2056
        return -1;

    vm->def->mem.cur_balloon = balloon;

    return 0;
}


2057
static int
2058
qemuProcessWaitForMonitor(virQEMUDriverPtr driver,
C
Cole Robinson 已提交
2059
                          virDomainObjPtr vm,
2060
                          int asyncJob,
2061
                          virQEMUCapsPtr qemuCaps,
2062
                          qemuDomainLogContextPtr logCtxt)
2063 2064
{
    int ret = -1;
2065
    virHashTablePtr info = NULL;
2066
    qemuDomainObjPrivatePtr priv;
2067 2068

    VIR_DEBUG("Connect monitor to %p '%s'", vm, vm->def->name);
2069
    if (qemuConnectMonitor(driver, vm, asyncJob, logCtxt) < 0)
2070 2071 2072 2073 2074 2075
        goto cleanup;

    /* Try to get the pty path mappings again via the monitor. This is much more
     * reliable if it's available.
     * Note that the monitor itself can be on a pty, so we still need to try the
     * log output method. */
2076
    priv = vm->privateData;
2077 2078
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;
2079 2080
    ret = qemuMonitorGetChardevInfo(priv->mon, &info);
    VIR_DEBUG("qemuMonitorGetChardevInfo returned %i", ret);
2081 2082 2083
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;

2084 2085 2086 2087 2088
    if (ret == 0) {
        if ((ret = qemuProcessFindCharDevicePTYsMonitor(vm, qemuCaps,
                                                        info)) < 0)
            goto cleanup;

2089 2090
        if ((ret = qemuProcessRefreshChannelVirtioState(driver, vm, info,
                                                        true)) < 0)
2091 2092
            goto cleanup;
    }
2093

2094
 cleanup:
2095
    virHashFree(info);
2096

2097 2098
    if (logCtxt && kill(vm->pid, 0) == -1 && errno == ESRCH) {
        qemuProcessReportLogError(logCtxt,
2099
                                  _("process exited while connecting to monitor"));
2100 2101 2102 2103 2104 2105
        ret = -1;
    }

    return ret;
}

2106

2107 2108 2109 2110 2111 2112
static int
qemuProcessDetectIOThreadPIDs(virQEMUDriverPtr driver,
                              virDomainObjPtr vm,
                              int asyncJob)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2113
    qemuMonitorIOThreadInfoPtr *iothreads = NULL;
2114 2115 2116 2117
    int niothreads = 0;
    int ret = -1;
    size_t i;

2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_IOTHREAD)) {
        /* The following check is because at one time a domain could
         * define iothreadids and start the domain - only failing the
         * capability check when attempting to add a disk. Because the
         * iothreads and [n]iothreadids were left untouched other code
         * assumed it could use the ->thread_id value to make thread_id
         * based adjustments (e.g. pinning, scheduling) which while
         * succeeding would execute on the calling thread.
         */
        if (vm->def->niothreadids) {
            for (i = 0; i < vm->def->niothreadids; i++) {
                /* Check if the domain had defined any iothreadid elements
                 * and supply a VIR_INFO indicating that it's being removed.
                 */
                if (!vm->def->iothreadids[i]->autofill)
                    VIR_INFO("IOThreads not supported, remove iothread id '%u'",
                             vm->def->iothreadids[i]->iothread_id);
                virDomainIOThreadIDDefFree(vm->def->iothreadids[i]);
            }
            /* Remove any trace */
            VIR_FREE(vm->def->iothreadids);
            vm->def->niothreadids = 0;
            vm->def->iothreads = 0;
        }
2142
        return 0;
2143
    }
2144

2145 2146 2147 2148
    /* Get the list of IOThreads from qemu */
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;
    niothreads = qemuMonitorGetIOThreads(priv->mon, &iothreads);
2149 2150
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        goto cleanup;
J
John Ferlan 已提交
2151
    if (niothreads < 0)
2152 2153
        goto cleanup;

2154
    if (niothreads != vm->def->niothreadids) {
2155 2156
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("got wrong number of IOThread pids from QEMU monitor. "
2157 2158
                         "got %d, wanted %zu"),
                       niothreads, vm->def->niothreadids);
2159 2160 2161
        goto cleanup;
    }

2162 2163 2164 2165 2166 2167
    /* Nothing to do */
    if (niothreads == 0) {
        ret = 0;
        goto cleanup;
    }

2168 2169 2170
    for (i = 0; i < niothreads; i++) {
        virDomainIOThreadIDDefPtr iothrid;

2171 2172
        if (!(iothrid = virDomainIOThreadIDFind(vm->def,
                                                iothreads[i]->iothread_id))) {
2173
            virReportError(VIR_ERR_INTERNAL_ERROR,
2174 2175
                           _("iothread %d not found"),
                           iothreads[i]->iothread_id);
2176 2177 2178 2179
            goto cleanup;
        }
        iothrid->thread_id = iothreads[i]->thread_id;
    }
2180 2181 2182 2183 2184 2185

    ret = 0;

 cleanup:
    if (iothreads) {
        for (i = 0; i < niothreads; i++)
2186
            VIR_FREE(iothreads[i]);
2187 2188 2189 2190 2191
        VIR_FREE(iothreads);
    }
    return ret;
}

2192 2193 2194 2195 2196

/*
 * To be run between fork/exec of QEMU only
 */
static int
2197
qemuProcessInitCpuAffinity(virDomainObjPtr vm)
2198 2199 2200 2201
{
    int ret = -1;
    virBitmapPtr cpumap = NULL;
    virBitmapPtr cpumapToSet = NULL;
2202
    qemuDomainObjPrivatePtr priv = vm->privateData;
2203

2204 2205 2206 2207 2208 2209
    if (!vm->pid) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Cannot setup CPU affinity until process is started"));
        return -1;
    }

2210 2211
    if (vm->def->placement_mode == VIR_DOMAIN_CPU_PLACEMENT_MODE_AUTO) {
        VIR_DEBUG("Set CPU affinity with advisory nodeset from numad");
2212
        cpumapToSet = priv->autoCpuset;
2213
    } else {
2214
        VIR_DEBUG("Set CPU affinity with specified cpuset");
O
Osier Yang 已提交
2215
        if (vm->def->cpumask) {
H
Hu Tao 已提交
2216
            cpumapToSet = vm->def->cpumask;
O
Osier Yang 已提交
2217 2218 2219 2220 2221
        } else {
            /* You may think this is redundant, but we can't assume libvirtd
             * itself is running on all pCPUs, so we need to explicitly set
             * the spawned QEMU instance to all pCPUs if no map is given in
             * its config file */
2222 2223 2224 2225
            int hostcpus;

            /* setaffinity fails if you set bits for CPUs which
             * aren't present, so we have to limit ourselves */
2226
            if ((hostcpus = virHostCPUGetCount()) < 0)
2227 2228 2229 2230 2231 2232 2233 2234
                goto cleanup;

            if (hostcpus > QEMUD_CPUMASK_LEN)
                hostcpus = QEMUD_CPUMASK_LEN;

            if (!(cpumap = virBitmapNew(hostcpus)))
                goto cleanup;

2235
            virBitmapSetAll(cpumap);
2236 2237

            cpumapToSet = cpumap;
O
Osier Yang 已提交
2238
        }
2239 2240
    }

2241
    if (virProcessSetAffinity(vm->pid, cpumapToSet) < 0)
2242
        goto cleanup;
2243

2244 2245
    ret = 0;

2246
 cleanup:
2247
    virBitmapFree(cpumap);
2248
    return ret;
2249 2250
}

2251 2252
/* set link states to down on interfaces at qemu start */
static int
2253 2254 2255
qemuProcessSetLinkStates(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         qemuDomainAsyncJob asyncJob)
2256 2257 2258
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDefPtr def = vm->def;
2259
    size_t i;
2260 2261 2262 2263 2264
    int ret = -1;
    int rv;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
2265 2266 2267

    for (i = 0; i < def->nnets; i++) {
        if (def->nets[i]->linkstate == VIR_DOMAIN_NET_INTERFACE_LINK_STATE_DOWN) {
2268 2269 2270
            if (!def->nets[i]->info.alias) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing alias for network device"));
2271
                goto cleanup;
2272 2273
            }

2274 2275
            VIR_DEBUG("Setting link state: %s", def->nets[i]->info.alias);

2276
            if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_NETDEV)) {
2277
                virReportError(VIR_ERR_OPERATION_UNSUPPORTED, "%s",
2278
                               _("Setting of link state is not supported by this qemu"));
2279
                goto cleanup;
2280 2281
            }

2282 2283 2284 2285
            rv = qemuMonitorSetLink(priv->mon,
                                    def->nets[i]->info.alias,
                                    VIR_DOMAIN_NET_INTERFACE_LINK_STATE_DOWN);
            if (rv < 0) {
2286
                virReportError(VIR_ERR_OPERATION_FAILED,
2287 2288 2289
                               _("Couldn't set link state on interface: %s"),
                               def->nets[i]->info.alias);
                goto cleanup;
2290 2291 2292 2293
            }
        }
    }

2294 2295 2296 2297 2298
    ret = 0;

 cleanup:
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
2299 2300 2301
    return ret;
}

2302

2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 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 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410
/**
 * qemuProcessSetupPid:
 *
 * This function sets resource properities (affinity, cgroups,
 * scheduler) for any PID associated with a domain.  It should be used
 * to set up emulator PIDs as well as vCPU and I/O thread pids to
 * ensure they are all handled the same way.
 *
 * Returns 0 on success, -1 on error.
 */
static int
qemuProcessSetupPid(virDomainObjPtr vm,
                    pid_t pid,
                    virCgroupThreadName nameval,
                    int id,
                    virBitmapPtr cpumask,
                    unsigned long long period,
                    long long quota,
                    virDomainThreadSchedParamPtr sched)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainNumatuneMemMode mem_mode;
    virCgroupPtr cgroup = NULL;
    virBitmapPtr use_cpumask;
    char *mem_mask = NULL;
    int ret = -1;

    if ((period || quota) &&
        !virCgroupHasController(priv->cgroup, VIR_CGROUP_CONTROLLER_CPU)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("cgroup cpu is required for scheduler tuning"));
        goto cleanup;
    }

    /* Infer which cpumask shall be used. */
    if (cpumask)
        use_cpumask = cpumask;
    else if (vm->def->placement_mode == VIR_DOMAIN_CPU_PLACEMENT_MODE_AUTO)
        use_cpumask = priv->autoCpuset;
    else
        use_cpumask = vm->def->cpumask;

    /*
     * If CPU cgroup controller is not initialized here, then we need
     * neither period nor quota settings.  And if CPUSET controller is
     * not initialized either, then there's nothing to do anyway.
     */
    if (virCgroupHasController(priv->cgroup, VIR_CGROUP_CONTROLLER_CPU) ||
        virCgroupHasController(priv->cgroup, VIR_CGROUP_CONTROLLER_CPUSET)) {

        if (virDomainNumatuneGetMode(vm->def->numa, -1, &mem_mode) == 0 &&
            mem_mode == VIR_DOMAIN_NUMATUNE_MEM_STRICT &&
            virDomainNumatuneMaybeFormatNodeset(vm->def->numa,
                                                priv->autoNodeset,
                                                &mem_mask, -1) < 0)
            goto cleanup;

        if (virCgroupNewThread(priv->cgroup, nameval, id, true, &cgroup) < 0)
            goto cleanup;

        if (virCgroupHasController(priv->cgroup, VIR_CGROUP_CONTROLLER_CPUSET)) {
            if (use_cpumask &&
                qemuSetupCgroupCpusetCpus(cgroup, use_cpumask) < 0)
                goto cleanup;

            /*
             * Don't setup cpuset.mems for the emulator, they need to
             * be set up after initialization in order for kvm
             * allocations to succeed.
             */
            if (nameval != VIR_CGROUP_THREAD_EMULATOR &&
                mem_mask && virCgroupSetCpusetMems(cgroup, mem_mask) < 0)
                goto cleanup;

        }

        if ((period || quota) &&
            qemuSetupCgroupVcpuBW(cgroup, period, quota) < 0)
            goto cleanup;

        /* Move the thread to the sub dir */
        if (virCgroupAddTask(cgroup, pid) < 0)
            goto cleanup;

    }

    /* Setup legacy affinity. */
    if (use_cpumask && virProcessSetAffinity(pid, use_cpumask) < 0)
        goto cleanup;

    /* Set scheduler type and priority. */
    if (sched &&
        virProcessSetScheduler(pid, sched->policy, sched->priority) < 0)
        goto cleanup;

    ret = 0;
 cleanup:
    VIR_FREE(mem_mask);
    if (cgroup) {
        if (ret < 0)
            virCgroupRemove(cgroup);
        virCgroupFree(&cgroup);
    }

    return ret;
}


2411
static int
2412
qemuProcessSetupEmulator(virDomainObjPtr vm)
2413
{
2414 2415 2416 2417 2418
    return qemuProcessSetupPid(vm, vm->pid, VIR_CGROUP_THREAD_EMULATOR,
                               0, vm->def->cputune.emulatorpin,
                               vm->def->cputune.emulator_period,
                               vm->def->cputune.emulator_quota,
                               NULL);
2419 2420
}

2421

2422 2423
static int
qemuProcessInitPasswords(virConnectPtr conn,
2424
                         virQEMUDriverPtr driver,
2425 2426
                         virDomainObjPtr vm,
                         int asyncJob)
2427 2428 2429
{
    int ret = 0;
    qemuDomainObjPrivatePtr priv = vm->privateData;
2430
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2431
    size_t i;
2432 2433
    char *alias = NULL;
    char *secret = NULL;
2434

2435
    for (i = 0; i < vm->def->ngraphics; ++i) {
2436 2437
        virDomainGraphicsDefPtr graphics = vm->def->graphics[i];
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
2438 2439
            ret = qemuDomainChangeGraphicsPasswords(driver, vm,
                                                    VIR_DOMAIN_GRAPHICS_TYPE_VNC,
2440
                                                    &graphics->data.vnc.auth,
2441 2442
                                                    cfg->vncPassword,
                                                    asyncJob);
2443
        } else if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
2444 2445
            ret = qemuDomainChangeGraphicsPasswords(driver, vm,
                                                    VIR_DOMAIN_GRAPHICS_TYPE_SPICE,
2446
                                                    &graphics->data.spice.auth,
2447 2448
                                                    cfg->spicePassword,
                                                    asyncJob);
2449 2450
        }

2451 2452 2453
        if (ret < 0)
            goto cleanup;
    }
2454

2455 2456
    for (i = 0; i < vm->def->ndisks; i++) {
        size_t secretLen;
2457

2458 2459 2460
        if (!vm->def->disks[i]->src->encryption ||
            !virDomainDiskGetSource(vm->def->disks[i]))
            continue;
2461

2462 2463 2464 2465 2466 2467
        if (vm->def->disks[i]->src->encryption->format !=
            VIR_STORAGE_ENCRYPTION_FORMAT_DEFAULT &&
            vm->def->disks[i]->src->encryption->format !=
            VIR_STORAGE_ENCRYPTION_FORMAT_QCOW)
            continue;

2468 2469 2470 2471 2472
        VIR_FREE(secret);
        if (qemuProcessGetVolumeQcowPassphrase(conn,
                                               vm->def->disks[i],
                                               &secret, &secretLen) < 0)
            goto cleanup;
2473

2474
        VIR_FREE(alias);
2475
        if (!(alias = qemuAliasFromDisk(vm->def->disks[i])))
2476 2477 2478 2479 2480 2481 2482 2483
            goto cleanup;
        if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
            goto cleanup;
        ret = qemuMonitorSetDrivePassphrase(priv->mon, alias, secret);
        if (qemuDomainObjExitMonitor(driver, vm) < 0)
            ret = -1;
        if (ret < 0)
            goto cleanup;
2484 2485
    }

2486
 cleanup:
2487 2488
    VIR_FREE(alias);
    VIR_FREE(secret);
2489
    virObjectUnref(cfg);
2490 2491 2492 2493 2494 2495 2496 2497 2498 2499
    return ret;
}


static int
qemuProcessPrepareChardevDevice(virDomainDefPtr def ATTRIBUTE_UNUSED,
                                virDomainChrDefPtr dev,
                                void *opaque ATTRIBUTE_UNUSED)
{
    int fd;
2500
    if (dev->source->type != VIR_DOMAIN_CHR_TYPE_FILE)
2501 2502
        return 0;

2503
    if ((fd = open(dev->source->data.file.path,
2504 2505 2506
                   O_CREAT | O_APPEND, S_IRUSR|S_IWUSR)) < 0) {
        virReportSystemError(errno,
                             _("Unable to pre-create chardev file '%s'"),
2507
                             dev->source->data.file.path);
2508 2509 2510 2511 2512 2513 2514 2515 2516
        return -1;
    }

    VIR_FORCE_CLOSE(fd);

    return 0;
}


2517 2518 2519 2520 2521
static int
qemuProcessCleanupChardevDevice(virDomainDefPtr def ATTRIBUTE_UNUSED,
                                virDomainChrDefPtr dev,
                                void *opaque ATTRIBUTE_UNUSED)
{
2522 2523 2524 2525
    if (dev->source->type == VIR_DOMAIN_CHR_TYPE_UNIX &&
        dev->source->data.nix.listen &&
        dev->source->data.nix.path)
        unlink(dev->source->data.nix.path);
2526 2527 2528 2529 2530

    return 0;
}


2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566
/**
 * Loads and update video memory size for video devices according to QEMU
 * process as the QEMU will silently update the values that we pass to QEMU
 * through command line.  We need to load these updated values and store them
 * into the status XML.
 *
 * We will fail if for some reason the values cannot be loaded from QEMU because
 * its mandatory to get the correct video memory size to status XML to not break
 * migration.
 */
static int
qemuProcessUpdateVideoRamSize(virQEMUDriverPtr driver,
                              virDomainObjPtr vm,
                              int asyncJob)
{
    int ret = -1;
    ssize_t i;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainVideoDefPtr video = NULL;
    virQEMUDriverConfigPtr cfg = NULL;

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

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

        switch (video->type) {
        case VIR_DOMAIN_VIDEO_TYPE_VGA:
            if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_VGA_VGAMEM)) {
                if (qemuMonitorUpdateVideoMemorySize(priv->mon, video, "VGA") < 0)
                    goto error;
            }
            break;
        case VIR_DOMAIN_VIDEO_TYPE_QXL:
            if (i == 0) {
2567
                if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QXL_VGAMEM) &&
2568 2569
                    qemuMonitorUpdateVideoMemorySize(priv->mon, video,
                                                     "qxl-vga") < 0)
2570
                        goto error;
2571

2572
                if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QXL_VRAM64) &&
2573 2574 2575
                    qemuMonitorUpdateVideoVram64Size(priv->mon, video,
                                                     "qxl-vga") < 0)
                    goto error;
2576
            } else {
2577 2578 2579 2580 2581 2582 2583 2584
                if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QXL_VGAMEM) &&
                    qemuMonitorUpdateVideoMemorySize(priv->mon, video,
                                                     "qxl") < 0)
                        goto error;

                if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QXL_VRAM64) &&
                    qemuMonitorUpdateVideoVram64Size(priv->mon, video,
                                                     "qxl") < 0)
2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603
                        goto error;
            }
            break;
        case VIR_DOMAIN_VIDEO_TYPE_VMVGA:
            if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_VMWARE_SVGA_VGAMEM)) {
                if (qemuMonitorUpdateVideoMemorySize(priv->mon, video,
                                                     "vmware-svga") < 0)
                    goto error;
            }
            break;
        case VIR_DOMAIN_VIDEO_TYPE_CIRRUS:
        case VIR_DOMAIN_VIDEO_TYPE_XEN:
        case VIR_DOMAIN_VIDEO_TYPE_VBOX:
        case VIR_DOMAIN_VIDEO_TYPE_LAST:
            break;
        }

    }

2604 2605
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
2606 2607

    cfg = virQEMUDriverGetConfig(driver);
2608
    ret = virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps);
2609 2610 2611 2612 2613
    virObjectUnref(cfg);

    return ret;

 error:
2614
    ignore_value(qemuDomainObjExitMonitor(driver, vm));
2615 2616 2617 2618
    return -1;
}


2619 2620 2621
struct qemuProcessHookData {
    virConnectPtr conn;
    virDomainObjPtr vm;
2622
    virQEMUDriverPtr driver;
2623
    virQEMUDriverConfigPtr cfg;
2624 2625 2626 2627 2628
};

static int qemuProcessHook(void *data)
{
    struct qemuProcessHookData *h = data;
2629
    qemuDomainObjPrivatePtr priv = h->vm->privateData;
2630
    int ret = -1;
2631
    int fd;
2632 2633 2634
    virBitmapPtr nodeset = NULL;
    virDomainNumatuneMemMode mode;

2635 2636 2637 2638 2639
    /* This method cannot use any mutexes, which are not
     * protected across fork()
     */

    virSecurityManagerPostFork(h->driver->securityManager);
2640 2641 2642 2643 2644

    /* Some later calls want pid present */
    h->vm->pid = getpid();

    VIR_DEBUG("Obtaining domain lock");
2645 2646 2647 2648 2649 2650 2651
    /*
     * Since we're going to leak the returned FD to QEMU,
     * we need to make sure it gets a sensible label.
     * This mildly sucks, because there could be other
     * sockets the lock driver opens that we don't want
     * labelled. So far we're ok though.
     */
2652
    if (virSecurityManagerSetSocketLabel(h->driver->securityManager, h->vm->def) < 0)
2653
        goto cleanup;
2654
    if (virDomainLockProcessStart(h->driver->lockManager,
2655
                                  h->cfg->uri,
2656
                                  h->vm,
J
Ján Tomko 已提交
2657
                                  /* QEMU is always paused initially */
2658 2659
                                  true,
                                  &fd) < 0)
2660
        goto cleanup;
2661
    if (virSecurityManagerClearSocketLabel(h->driver->securityManager, h->vm->def) < 0)
2662
        goto cleanup;
2663

2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674
    if (virDomainNumatuneGetMode(h->vm->def->numa, -1, &mode) == 0) {
        if (mode == VIR_DOMAIN_NUMATUNE_MEM_STRICT &&
            h->cfg->cgroupControllers & (1 << VIR_CGROUP_CONTROLLER_CPUSET) &&
            virCgroupControllerAvailable(VIR_CGROUP_CONTROLLER_CPUSET)) {
            /* Use virNuma* API iff necessary. Once set and child is exec()-ed,
             * there's no way for us to change it. Rely on cgroups (if available
             * and enabled in the config) rather than virNuma*. */
            VIR_DEBUG("Relying on CGroups for memory binding");
        } else {
            nodeset = virDomainNumatuneGetNodeset(h->vm->def->numa,
                                                  priv->autoNodeset, -1);
2675

2676 2677 2678
            if (virNumaSetupMemoryPolicy(mode, nodeset) < 0)
                goto cleanup;
        }
2679
    }
2680

2681 2682
    ret = 0;

2683
 cleanup:
2684
    virObjectUnref(h->cfg);
2685 2686
    VIR_DEBUG("Hook complete ret=%d", ret);
    return ret;
2687 2688 2689
}

int
2690 2691
qemuProcessPrepareMonitorChr(virDomainChrSourceDefPtr monConfig,
                             const char *domainDir)
2692 2693 2694 2695
{
    monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monConfig->data.nix.listen = true;

2696 2697
    if (virAsprintf(&monConfig->data.nix.path, "%s/monitor.sock",
                    domainDir) < 0)
2698 2699
        return -1;
    return 0;
2700 2701 2702
}


2703
/*
2704 2705
 * Precondition: vm must be locked, and a job must be active.
 * This method will call {Enter,Exit}Monitor
2706
 */
E
Eric Blake 已提交
2707
int
2708
qemuProcessStartCPUs(virQEMUDriverPtr driver, virDomainObjPtr vm,
2709
                     virConnectPtr conn, virDomainRunningReason reason,
2710
                     qemuDomainAsyncJob asyncJob)
2711
{
2712
    int ret = -1;
2713
    qemuDomainObjPrivatePtr priv = vm->privateData;
2714
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2715

2716
    /* Bring up netdevs before starting CPUs */
2717
    if (qemuInterfaceStartDevices(vm->def) < 0)
2718 2719
       goto cleanup;

2720
    VIR_DEBUG("Using lock state '%s'", NULLSTR(priv->lockState));
2721
    if (virDomainLockProcessResume(driver->lockManager, cfg->uri,
2722
                                   vm, priv->lockState) < 0) {
2723 2724 2725 2726
        /* Don't free priv->lockState on error, because we need
         * to make sure we have state still present if the user
         * tries to resume again
         */
2727
        goto cleanup;
2728 2729 2730
    }
    VIR_FREE(priv->lockState);

2731 2732
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto release;
J
Jiri Denemark 已提交
2733

2734
    ret = qemuMonitorStartCPUs(priv->mon, conn);
2735 2736
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
2737 2738 2739 2740 2741

    if (ret < 0)
        goto release;

    virDomainObjSetState(vm, VIR_DOMAIN_RUNNING, reason);
2742

2743
 cleanup:
2744
    virObjectUnref(cfg);
2745
    return ret;
2746 2747 2748 2749 2750 2751

 release:
    if (virDomainLockProcessPause(driver->lockManager, vm, &priv->lockState) < 0)
        VIR_WARN("Unable to release lease on %s", vm->def->name);
    VIR_DEBUG("Preserving lock state '%s'", NULLSTR(priv->lockState));
    goto cleanup;
2752 2753 2754
}


2755 2756
int qemuProcessStopCPUs(virQEMUDriverPtr driver,
                        virDomainObjPtr vm,
2757
                        virDomainPausedReason reason,
2758
                        qemuDomainAsyncJob asyncJob)
2759
{
2760
    int ret = -1;
2761 2762
    qemuDomainObjPrivatePtr priv = vm->privateData;

2763
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
2764

2765 2766
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;
J
Jiri Denemark 已提交
2767

2768
    ret = qemuMonitorStopCPUs(priv->mon);
2769 2770
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
2771 2772 2773 2774

    if (ret < 0)
        goto cleanup;

2775 2776 2777
    /* de-activate netdevs after stopping CPUs */
    ignore_value(qemuInterfaceStopDevices(vm->def));

2778 2779 2780
    if (priv->job.current)
        ignore_value(virTimeMillisNow(&priv->job.current->stopped));

2781 2782 2783 2784
    virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, reason);
    if (virDomainLockProcessPause(driver->lockManager, vm, &priv->lockState) < 0)
        VIR_WARN("Unable to release lease on %s", vm->def->name);
    VIR_DEBUG("Preserving lock state '%s'", NULLSTR(priv->lockState));
J
Jiri Denemark 已提交
2785

2786
 cleanup:
2787 2788 2789 2790 2791
    return ret;
}



2792 2793 2794
static int
qemuProcessNotifyNets(virDomainDefPtr def)
{
2795
    size_t i;
2796

2797 2798
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
2799 2800 2801 2802 2803 2804 2805 2806
        /* keep others from trying to use the macvtap device name, but
         * don't return error if this happens, since that causes the
         * domain to be unceremoniously killed, which would be *very*
         * impolite.
         */
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT)
           ignore_value(virNetDevMacVLanReserveName(net->ifname, false));

2807
        if (networkNotifyActualDevice(def, net) < 0)
2808 2809 2810 2811 2812
            return -1;
    }
    return 0;
}

2813
static int
2814
qemuProcessFiltersInstantiate(virDomainDefPtr def)
2815
{
2816
    size_t i;
2817

2818
    for (i = 0; i < def->nnets; i++) {
2819 2820
        virDomainNetDefPtr net = def->nets[i];
        if ((net->filter) && (net->ifname)) {
2821
            if (virDomainConfNWFilterInstantiate(def->uuid, net) < 0)
J
Ján Tomko 已提交
2822
                return 1;
2823 2824 2825
        }
    }

J
Ján Tomko 已提交
2826
    return 0;
2827 2828
}

2829
static int
2830
qemuProcessUpdateState(virQEMUDriverPtr driver, virDomainObjPtr vm)
2831 2832 2833
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainState state;
2834
    virDomainPausedReason reason;
2835
    virDomainState newState = VIR_DOMAIN_NOSTATE;
2836
    int oldReason;
2837
    int newReason;
2838
    bool running;
2839
    char *msg = NULL;
2840 2841
    int ret;

2842
    qemuDomainObjEnterMonitor(driver, vm);
2843
    ret = qemuMonitorGetStatus(priv->mon, &running, &reason);
2844 2845
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
2846

2847
    if (ret < 0)
2848 2849
        return -1;

2850
    state = virDomainObjGetState(vm, &oldReason);
2851

2852 2853 2854 2855 2856 2857 2858 2859
    if (running &&
        (state == VIR_DOMAIN_SHUTOFF ||
         (state == VIR_DOMAIN_PAUSED &&
          oldReason == VIR_DOMAIN_PAUSED_STARTING_UP))) {
        newState = VIR_DOMAIN_RUNNING;
        newReason = VIR_DOMAIN_RUNNING_BOOTED;
        ignore_value(VIR_STRDUP_QUIET(msg, "finished booting"));
    } else if (state == VIR_DOMAIN_PAUSED && running) {
2860 2861
        newState = VIR_DOMAIN_RUNNING;
        newReason = VIR_DOMAIN_RUNNING_UNPAUSED;
2862
        ignore_value(VIR_STRDUP_QUIET(msg, "was unpaused"));
2863
    } else if (state == VIR_DOMAIN_RUNNING && !running) {
2864 2865 2866
        if (reason == VIR_DOMAIN_PAUSED_SHUTTING_DOWN) {
            newState = VIR_DOMAIN_SHUTDOWN;
            newReason = VIR_DOMAIN_SHUTDOWN_UNKNOWN;
2867
            ignore_value(VIR_STRDUP_QUIET(msg, "shutdown"));
2868
        } else if (reason == VIR_DOMAIN_PAUSED_CRASHED) {
2869 2870 2871
            newState = VIR_DOMAIN_CRASHED;
            newReason = VIR_DOMAIN_CRASHED_PANICKED;
            ignore_value(VIR_STRDUP_QUIET(msg, "crashed"));
2872 2873 2874
        } else {
            newState = VIR_DOMAIN_PAUSED;
            newReason = reason;
S
Stefan Berger 已提交
2875 2876
            ignore_value(virAsprintf(&msg, "was paused (%s)",
                                 virDomainPausedReasonTypeToString(reason)));
2877 2878 2879 2880 2881 2882 2883
        }
    }

    if (newState != VIR_DOMAIN_NOSTATE) {
        VIR_DEBUG("Domain %s %s while its monitor was disconnected;"
                  " changing state to %s (%s)",
                  vm->def->name,
2884
                  NULLSTR(msg),
2885 2886 2887 2888
                  virDomainStateTypeToString(newState),
                  virDomainStateReasonToString(newState, newReason));
        VIR_FREE(msg);
        virDomainObjSetState(vm, newState, newReason);
2889 2890 2891 2892 2893
    }

    return 0;
}

2894
static int
2895 2896 2897 2898 2899
qemuProcessRecoverMigrationIn(virQEMUDriverPtr driver,
                              virDomainObjPtr vm,
                              virConnectPtr conn,
                              qemuMigrationJobPhase phase,
                              virDomainState state,
2900
                              int reason)
2901
{
2902 2903 2904 2905 2906
    bool postcopy = (state == VIR_DOMAIN_PAUSED &&
                     reason == VIR_DOMAIN_PAUSED_POSTCOPY_FAILED) ||
                    (state == VIR_DOMAIN_RUNNING &&
                     reason == VIR_DOMAIN_RUNNING_POSTCOPY);

2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922
    switch (phase) {
    case QEMU_MIGRATION_PHASE_NONE:
    case QEMU_MIGRATION_PHASE_PERFORM2:
    case QEMU_MIGRATION_PHASE_BEGIN3:
    case QEMU_MIGRATION_PHASE_PERFORM3:
    case QEMU_MIGRATION_PHASE_PERFORM3_DONE:
    case QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED:
    case QEMU_MIGRATION_PHASE_CONFIRM3:
    case QEMU_MIGRATION_PHASE_LAST:
        /* N/A for incoming migration */
        break;

    case QEMU_MIGRATION_PHASE_PREPARE:
        VIR_DEBUG("Killing unfinished incoming migration for domain %s",
                  vm->def->name);
        return -1;
2923

2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938
    case QEMU_MIGRATION_PHASE_FINISH2:
        /* source domain is already killed so let's just resume the domain
         * and hope we are all set */
        VIR_DEBUG("Incoming migration finished, resuming domain %s",
                  vm->def->name);
        if (qemuProcessStartCPUs(driver, vm, conn,
                                 VIR_DOMAIN_RUNNING_UNPAUSED,
                                 QEMU_ASYNC_JOB_NONE) < 0) {
            VIR_WARN("Could not resume domain %s", vm->def->name);
        }
        break;

    case QEMU_MIGRATION_PHASE_FINISH3:
        /* migration finished, we started resuming the domain but didn't
         * confirm success or failure yet; killing it seems safest unless
2939 2940 2941 2942
         * we already started guest CPUs or we were in post-copy mode */
        if (postcopy) {
            qemuMigrationPostcopyFailed(driver, vm);
        } else if (state != VIR_DOMAIN_RUNNING) {
2943
            VIR_DEBUG("Killing migrated domain %s", vm->def->name);
2944
            return -1;
2945 2946 2947
        }
        break;
    }
2948

2949 2950
    return 0;
}
2951

2952 2953 2954 2955 2956 2957 2958 2959
static int
qemuProcessRecoverMigrationOut(virQEMUDriverPtr driver,
                               virDomainObjPtr vm,
                               virConnectPtr conn,
                               qemuMigrationJobPhase phase,
                               virDomainState state,
                               int reason)
{
2960 2961 2962 2963
    bool postcopy = state == VIR_DOMAIN_PAUSED &&
                    (reason == VIR_DOMAIN_PAUSED_POSTCOPY ||
                     reason == VIR_DOMAIN_PAUSED_POSTCOPY_FAILED);

2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980
    switch (phase) {
    case QEMU_MIGRATION_PHASE_NONE:
    case QEMU_MIGRATION_PHASE_PREPARE:
    case QEMU_MIGRATION_PHASE_FINISH2:
    case QEMU_MIGRATION_PHASE_FINISH3:
    case QEMU_MIGRATION_PHASE_LAST:
        /* N/A for outgoing migration */
        break;

    case QEMU_MIGRATION_PHASE_BEGIN3:
        /* nothing happened so far, just forget we were about to migrate the
         * domain */
        break;

    case QEMU_MIGRATION_PHASE_PERFORM2:
    case QEMU_MIGRATION_PHASE_PERFORM3:
        /* migration is still in progress, let's cancel it and resume the
2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993
         * domain; however we can only do that before migration enters
         * post-copy mode
         */
        if (postcopy) {
            qemuMigrationPostcopyFailed(driver, vm);
        } else {
            VIR_DEBUG("Cancelling unfinished migration of domain %s",
                      vm->def->name);
            if (qemuMigrationCancel(driver, vm) < 0) {
                VIR_WARN("Could not cancel ongoing migration of domain %s",
                         vm->def->name);
            }
            goto resume;
2994
        }
2995
        break;
2996

2997 2998
    case QEMU_MIGRATION_PHASE_PERFORM3_DONE:
        /* migration finished but we didn't have a chance to get the result
2999 3000
         * of Finish3 step; third party needs to check what to do next; in
         * post-copy mode we can use PAUSED_POSTCOPY_FAILED state for this
3001
         */
3002 3003
        if (postcopy)
            qemuMigrationPostcopyFailed(driver, vm);
3004
        break;
3005

3006
    case QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED:
3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018
        /* Finish3 failed, we need to resume the domain, but once we enter
         * post-copy mode there's no way back, so let's just mark the domain
         * as broken in that case
         */
        if (postcopy) {
            qemuMigrationPostcopyFailed(driver, vm);
        } else {
            VIR_DEBUG("Resuming domain %s after failed migration",
                      vm->def->name);
            goto resume;
        }
        break;
3019

3020 3021 3022 3023
    case QEMU_MIGRATION_PHASE_CONFIRM3:
        /* migration completed, we need to kill the domain here */
        return -1;
    }
3024

3025
    return 0;
3026

3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037
 resume:
    /* resume the domain but only if it was paused as a result of
     * migration
     */
    if (state == VIR_DOMAIN_PAUSED &&
        (reason == VIR_DOMAIN_PAUSED_MIGRATION ||
         reason == VIR_DOMAIN_PAUSED_UNKNOWN)) {
        if (qemuProcessStartCPUs(driver, vm, conn,
                                 VIR_DOMAIN_RUNNING_UNPAUSED,
                                 QEMU_ASYNC_JOB_NONE) < 0) {
            VIR_WARN("Could not resume domain %s", vm->def->name);
3038 3039 3040 3041 3042
        }
    }
    return 0;
}

3043
static int
3044
qemuProcessRecoverJob(virQEMUDriverPtr driver,
3045 3046 3047 3048
                      virDomainObjPtr vm,
                      virConnectPtr conn,
                      const struct qemuDomainJobObj *job)
{
3049
    qemuDomainObjPrivatePtr priv = vm->privateData;
3050 3051 3052 3053 3054 3055 3056
    virDomainState state;
    int reason;

    state = virDomainObjGetState(vm, &reason);

    switch (job->asyncJob) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
3057 3058 3059 3060 3061
        if (qemuProcessRecoverMigrationOut(driver, vm, conn, job->phase,
                                           state, reason) < 0)
            return -1;
        break;

3062
    case QEMU_ASYNC_JOB_MIGRATION_IN:
3063 3064
        if (qemuProcessRecoverMigrationIn(driver, vm, conn, job->phase,
                                          state, reason) < 0)
3065
            return -1;
3066 3067 3068 3069
        break;

    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
3070
    case QEMU_ASYNC_JOB_SNAPSHOT:
3071
        qemuDomainObjEnterMonitor(driver, vm);
3072
        ignore_value(qemuMonitorMigrateCancel(priv->mon));
3073 3074
        if (qemuDomainObjExitMonitor(driver, vm) < 0)
            return -1;
3075
        /* resume the domain but only if it was paused as a result of
3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091
         * running a migration-to-file operation.  Although we are
         * recovering an async job, this function is run at startup
         * and must resume things using sync monitor connections.  */
         if (state == VIR_DOMAIN_PAUSED &&
             ((job->asyncJob == QEMU_ASYNC_JOB_DUMP &&
               reason == VIR_DOMAIN_PAUSED_DUMP) ||
              (job->asyncJob == QEMU_ASYNC_JOB_SAVE &&
               reason == VIR_DOMAIN_PAUSED_SAVE) ||
              (job->asyncJob == QEMU_ASYNC_JOB_SNAPSHOT &&
               reason == VIR_DOMAIN_PAUSED_SNAPSHOT) ||
              reason == VIR_DOMAIN_PAUSED_UNKNOWN)) {
             if (qemuProcessStartCPUs(driver, vm, conn,
                                      VIR_DOMAIN_RUNNING_UNPAUSED,
                                      QEMU_ASYNC_JOB_NONE) < 0) {
                 VIR_WARN("Could not resume domain '%s' after migration to file",
                          vm->def->name);
3092 3093 3094 3095
            }
        }
        break;

3096 3097 3098 3099
    case QEMU_ASYNC_JOB_START:
        /* Already handled in VIR_DOMAIN_PAUSED_STARTING_UP check. */
        break;

3100 3101 3102 3103 3104 3105 3106 3107
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        break;
    }

    if (!virDomainObjIsActive(vm))
        return -1;

3108 3109 3110 3111
    /* In case any special handling is added for job type that has been ignored
     * before, QEMU_DOMAIN_TRACK_JOBS (from qemu_domain.h) needs to be updated
     * for the job to be properly tracked in domain state XML.
     */
3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131
    switch (job->active) {
    case QEMU_JOB_QUERY:
        /* harmless */
        break;

    case QEMU_JOB_DESTROY:
        VIR_DEBUG("Domain %s should have already been destroyed",
                  vm->def->name);
        return -1;

    case QEMU_JOB_SUSPEND:
        /* mostly harmless */
        break;

    case QEMU_JOB_MODIFY:
        /* XXX depending on the command we may be in an inconsistent state and
         * we should probably fall back to "monitor error" state and refuse to
         */
        break;

3132
    case QEMU_JOB_MIGRATION_OP:
3133
    case QEMU_JOB_ABORT:
3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144
    case QEMU_JOB_ASYNC:
    case QEMU_JOB_ASYNC_NESTED:
        /* async job was already handled above */
    case QEMU_JOB_NONE:
    case QEMU_JOB_LAST:
        break;
    }

    return 0;
}

3145 3146 3147 3148 3149 3150
static int
qemuProcessUpdateDevices(virQEMUDriverPtr driver,
                         virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDeviceDef dev;
3151
    const char **qemuDevices;
3152 3153 3154 3155 3156 3157 3158 3159 3160
    char **old;
    char **tmp;
    int ret = -1;

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

    old = priv->qemuDevices;
    priv->qemuDevices = NULL;
3161
    if (qemuDomainUpdateDeviceList(driver, vm, QEMU_ASYNC_JOB_NONE) < 0)
3162 3163
        goto cleanup;

3164
    qemuDevices = (const char **) priv->qemuDevices;
3165 3166
    if ((tmp = old)) {
        while (*tmp) {
3167
            if (!virStringListHasString(qemuDevices, *tmp) &&
3168 3169 3170 3171
                virDomainDefFindDevice(vm->def, *tmp, &dev, false) == 0 &&
                qemuDomainRemoveDevice(driver, vm, &dev) < 0) {
                goto cleanup;
            }
3172 3173 3174 3175 3176
            tmp++;
        }
    }
    ret = 0;

3177
 cleanup:
3178
    virStringListFree(old);
3179 3180 3181
    return ret;
}

3182 3183 3184 3185 3186 3187 3188
static int
qemuDomainPerfRestart(virDomainObjPtr vm)
{
    size_t i;
    virDomainDefPtr def = vm->def;
    qemuDomainObjPrivatePtr priv = vm->privateData;

3189
    if (!(priv->perf = virPerfNew()))
3190 3191 3192
        return -1;

    for (i = 0; i < VIR_PERF_EVENT_LAST; i++) {
3193 3194
        if (def->perf.events[i] &&
            def->perf.events[i] == VIR_TRISTATE_BOOL_YES) {
3195 3196 3197

            /* Failure to re-enable the perf event should not be fatal */
            if (virPerfEventEnable(priv->perf, i, vm->pid) < 0)
3198
                def->perf.events[i] = VIR_TRISTATE_BOOL_NO;
3199 3200 3201 3202 3203 3204
        }
    }

    return 0;
}

3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227

static void
qemuProcessReconnectCheckMemAliasOrderMismatch(virDomainObjPtr vm)
{
    size_t i;
    int aliasidx;
    virDomainDefPtr def = vm->def;
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (!virDomainDefHasMemoryHotplug(def) || def->nmems == 0)
        return;

    for (i = 0; i < def->nmems; i++) {
        aliasidx = qemuDomainDeviceAliasIndex(&def->mems[i]->info, "dimm");

        if (def->mems[i]->info.addr.dimm.slot != aliasidx) {
            priv->memAliasOrderMismatch = true;
            break;
        }
    }
}


3228 3229
struct qemuProcessReconnectData {
    virConnectPtr conn;
3230
    virQEMUDriverPtr driver;
3231
    virDomainObjPtr obj;
3232 3233 3234 3235
};
/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
S
Stefan Berger 已提交
3236 3237 3238 3239
 *
 * We own the virConnectPtr we are passed here - whoever started
 * this thread function has increased the reference counter to it
 * so that we now have to close it.
3240
 *
3241
 * This function also inherits a locked and ref'd domain object.
3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253
 *
 * This function needs to:
 * 1. Enter job
 * 1. just before monitor reconnect do lightweight MonitorEnter
 *    (increase VM refcount and unlock VM)
 * 2. reconnect to monitor
 * 3. do lightweight MonitorExit (lock VM)
 * 4. continue reconnect process
 * 5. EndJob
 *
 * We can't do normal MonitorEnter & MonitorExit because these two lock the
 * monitor lock, which does not exists in this early phase.
3254 3255
 */
static void
3256
qemuProcessReconnect(void *opaque)
3257 3258
{
    struct qemuProcessReconnectData *data = opaque;
3259
    virQEMUDriverPtr driver = data->driver;
3260
    virDomainObjPtr obj = data->obj;
3261 3262
    qemuDomainObjPrivatePtr priv;
    virConnectPtr conn = data->conn;
3263
    struct qemuDomainJobObj oldjob;
3264 3265
    int state;
    int reason;
3266
    virQEMUDriverConfigPtr cfg;
3267
    size_t i;
3268
    unsigned int stopFlags = 0;
3269
    bool jobStarted = false;
3270
    virCapsPtr caps = NULL;
3271

3272 3273
    VIR_FREE(data);

3274
    qemuDomainObjRestoreJob(obj, &oldjob);
3275 3276
    if (oldjob.asyncJob == QEMU_ASYNC_JOB_MIGRATION_IN)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;
3277

3278 3279 3280
    cfg = virQEMUDriverGetConfig(driver);
    priv = obj->privateData;

3281 3282 3283
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto error;

3284 3285 3286 3287
    if (qemuDomainObjBeginJob(driver, obj, QEMU_JOB_MODIFY) < 0)
        goto error;
    jobStarted = true;

3288 3289 3290
    /* XXX If we ever gonna change pid file pattern, come up with
     * some intelligence here to deal with old paths. */
    if (!(priv->pidfile = virPidFileBuildPath(cfg->stateDir, obj->def->name)))
3291
        goto error;
3292

J
John Ferlan 已提交
3293 3294 3295 3296
    /* Restore the masterKey */
    if (qemuDomainMasterKeyReadFile(priv) < 0)
        goto error;

3297
    virNWFilterReadLockFilterUpdates();
3298 3299 3300 3301

    VIR_DEBUG("Reconnect monitor to %p '%s'", obj, obj->def->name);

    /* XXX check PID liveliness & EXE path */
3302
    if (qemuConnectMonitor(driver, obj, QEMU_ASYNC_JOB_NONE, NULL) < 0)
3303 3304
        goto error;

3305
    if (qemuHostdevUpdateActiveDomainDevices(driver, obj->def) < 0)
3306 3307
        goto error;

3308
    if (qemuConnectCgroup(driver, obj) < 0)
3309 3310
        goto error;

3311 3312 3313
    if (qemuDomainPerfRestart(obj) < 0)
        goto error;

3314
    /* XXX: Need to change as long as lock is introduced for
3315
     * qemu_driver->sharedDevices.
3316 3317
     */
    for (i = 0; i < obj->def->ndisks; i++) {
3318 3319
        virDomainDeviceDef dev;

3320
        if (virStorageTranslateDiskSourcePool(conn, obj->def->disks[i]) < 0)
3321
            goto error;
3322

3323 3324 3325 3326 3327 3328
        /* XXX we should be able to restore all data from XML in the future.
         * This should be the only place that calls qemuDomainDetermineDiskChain
         * with @report_broken == false to guarantee best-effort domain
         * reconnect */
        if (qemuDomainDetermineDiskChain(driver, obj, obj->def->disks[i],
                                         true, false) < 0)
3329 3330
            goto error;

3331 3332 3333
        dev.type = VIR_DOMAIN_DEVICE_DISK;
        dev.data.disk = obj->def->disks[i];
        if (qemuAddSharedDevice(driver, &dev, obj->def->name) < 0)
3334 3335 3336
            goto error;
    }

3337 3338 3339
    if (qemuProcessUpdateState(driver, obj) < 0)
        goto error;

3340
    state = virDomainObjGetState(obj, &reason);
3341 3342 3343
    if (state == VIR_DOMAIN_SHUTOFF ||
        (state == VIR_DOMAIN_PAUSED &&
         reason == VIR_DOMAIN_PAUSED_STARTING_UP)) {
3344 3345 3346 3347 3348
        VIR_DEBUG("Domain '%s' wasn't fully started yet, killing it",
                  obj->def->name);
        goto error;
    }

3349 3350 3351
    /* If upgrading from old libvirtd we won't have found any
     * caps in the domain status, so re-query them
     */
3352
    if (!priv->qemuCaps &&
3353 3354
        !(priv->qemuCaps = virQEMUCapsCacheLookupCopy(caps,
                                                      driver->qemuCapsCache,
3355 3356
                                                      obj->def->emulator,
                                                      obj->def->os.machine)))
3357 3358
        goto error;

3359
    /* In case the domain shutdown while we were not running,
3360
     * we need to finish the shutdown process. And we need to do it after
3361
     * we have virQEMUCaps filled in.
3362
     */
3363 3364 3365 3366 3367 3368
    if (state == VIR_DOMAIN_SHUTDOWN ||
        (state == VIR_DOMAIN_PAUSED &&
         reason == VIR_DOMAIN_PAUSED_SHUTTING_DOWN)) {
        VIR_DEBUG("Finishing shutdown sequence for domain %s",
                  obj->def->name);
        qemuProcessShutdownOrReboot(driver, obj);
3369
        goto cleanup;
3370 3371
    }

3372
    if ((qemuDomainAssignAddresses(obj->def, priv->qemuCaps, obj, false)) < 0)
3373
        goto error;
3374

3375 3376 3377 3378 3379 3380
    /* if domain requests security driver we haven't loaded, report error, but
     * do not kill the domain
     */
    ignore_value(virSecurityManagerCheckAllLabel(driver->securityManager,
                                                 obj->def));

3381 3382 3383
    if (qemuDomainRefreshVcpuInfo(driver, obj, QEMU_ASYNC_JOB_NONE, true) < 0)
        goto error;

3384
    if (virSecurityManagerReserveLabel(driver->securityManager, obj->def, obj->pid) < 0)
3385 3386
        goto error;

3387 3388 3389
    if (qemuProcessNotifyNets(obj->def) < 0)
        goto error;

3390
    if (qemuProcessFiltersInstantiate(obj->def))
3391 3392
        goto error;

3393
    if (qemuProcessRefreshDisks(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
3394 3395
        goto error;

3396
    if (qemuRefreshVirtioChannelState(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
3397 3398
        goto error;

3399 3400 3401
    /* If querying of guest's RTC failed, report error, but do not kill the domain. */
    qemuRefreshRTC(driver, obj);

3402 3403 3404
    if (qemuProcessRefreshBalloonState(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
        goto error;

3405 3406 3407
    if (qemuProcessRecoverJob(driver, obj, conn, &oldjob) < 0)
        goto error;

3408 3409 3410
    if (qemuProcessUpdateDevices(driver, obj) < 0)
        goto error;

3411 3412
    qemuProcessReconnectCheckMemAliasOrderMismatch(obj);

3413 3414
    if (qemuConnectAgent(driver, obj) < 0)
        goto error;
3415

3416
    /* update domain state XML with possibly updated state in virDomainObj */
3417
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj, driver->caps) < 0)
3418 3419
        goto error;

3420 3421
    /* Run an hook to allow admins to do some magic */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
3422
        char *xml = qemuDomainDefFormatXML(driver, obj->def, 0);
3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436
        int hookret;

        hookret = virHookCall(VIR_HOOK_DRIVER_QEMU, obj->def->name,
                              VIR_HOOK_QEMU_OP_RECONNECT, VIR_HOOK_SUBOP_BEGIN,
                              NULL, xml, NULL);
        VIR_FREE(xml);

        /*
         * If the script raised an error abort the launch
         */
        if (hookret < 0)
            goto error;
    }

3437
    if (virAtomicIntInc(&driver->nactive) == 1 && driver->inhibitCallback)
3438 3439
        driver->inhibitCallback(true, driver->inhibitOpaque);

3440 3441 3442 3443 3444 3445 3446 3447
 cleanup:
    if (jobStarted)
        qemuDomainObjEndJob(driver, obj);
    if (!virDomainObjIsActive(obj))
        qemuDomainRemoveInactive(driver, obj);
    virDomainObjEndAPI(&obj);
    virObjectUnref(conn);
    virObjectUnref(cfg);
3448
    virObjectUnref(caps);
3449 3450
    virNWFilterUnlockFilterUpdates();
    return;
3451

3452
 error:
3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466
    if (virDomainObjIsActive(obj)) {
        /* We can't get the monitor back, so must kill the VM
         * to remove danger of it ending up running twice if
         * user tries to start it again later
         */
        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_NO_SHUTDOWN)) {
            /* If we couldn't get the monitor and qemu supports
             * no-shutdown, we can safely say that the domain
             * crashed ... */
            state = VIR_DOMAIN_SHUTOFF_CRASHED;
        } else {
            /* ... but if it doesn't we can't say what the state
             * really is and FAILED means "failed to start" */
            state = VIR_DOMAIN_SHUTOFF_UNKNOWN;
3467
        }
3468 3469 3470 3471 3472
        /* If BeginJob failed, we jumped here without a job, let's hope another
         * thread didn't have a chance to start playing with the domain yet
         * (it's all we can do anyway).
         */
        qemuProcessStop(driver, obj, state, QEMU_ASYNC_JOB_NONE, stopFlags);
3473
    }
3474
    goto cleanup;
3475 3476
}

3477 3478
static int
qemuProcessReconnectHelper(virDomainObjPtr obj,
3479 3480 3481 3482 3483 3484
                           void *opaque)
{
    virThread thread;
    struct qemuProcessReconnectData *src = opaque;
    struct qemuProcessReconnectData *data;

3485
    /* If the VM was inactive, we don't need to reconnect */
3486 3487 3488
    if (!obj->pid)
        return 0;

3489
    if (VIR_ALLOC(data) < 0)
3490
        return -1;
3491 3492

    memcpy(data, src, sizeof(*data));
3493
    data->obj = obj;
3494

3495 3496
    /* this lock and reference will be eventually transferred to the thread
     * that handles the reconnect */
3497
    virObjectLock(obj);
3498
    virObjectRef(obj);
3499

3500 3501 3502
    /* Since we close the connection later on, we have to make sure that the
     * threads we start see a valid connection throughout their lifetime. We
     * simply increase the reference counter here.
S
Stefan Berger 已提交
3503
     */
3504
    virObjectRef(data->conn);
S
Stefan Berger 已提交
3505

3506
    if (virThreadCreate(&thread, false, qemuProcessReconnect, data) < 0) {
3507 3508 3509
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not create thread. QEMU initialization "
                         "might be incomplete"));
3510 3511 3512 3513 3514 3515 3516
        /* We can't spawn a thread and thus connect to monitor. Kill qemu.
         * It's safe to call qemuProcessStop without a job here since there
         * is no thread that could be doing anything else with the same domain
         * object.
         */
        qemuProcessStop(src->driver, obj, VIR_DOMAIN_SHUTOFF_FAILED,
                        QEMU_ASYNC_JOB_NONE, 0);
3517
        qemuDomainRemoveInactive(src->driver, obj);
3518

M
Michal Privoznik 已提交
3519
        virDomainObjEndAPI(&obj);
3520 3521 3522 3523
        virObjectUnref(data->conn);
        VIR_FREE(data);
        return -1;
    }
3524

3525
    return 0;
3526 3527 3528 3529 3530 3531 3532 3533 3534
}

/**
 * qemuProcessReconnectAll
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
void
3535
qemuProcessReconnectAll(virConnectPtr conn, virQEMUDriverPtr driver)
3536
{
3537
    struct qemuProcessReconnectData data = {.conn = conn, .driver = driver};
3538
    virDomainObjListForEach(driver->domains, qemuProcessReconnectHelper, &data);
3539 3540
}

3541
static int
3542
qemuProcessVNCAllocatePorts(virQEMUDriverPtr driver,
3543 3544
                            virDomainGraphicsDefPtr graphics,
                            bool allocate)
3545 3546 3547
{
    unsigned short port;

3548 3549 3550 3551 3552 3553 3554
    if (!allocate) {
        if (graphics->data.vnc.autoport)
            graphics->data.vnc.port = 5900;

        return 0;
    }

3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568
    if (graphics->data.vnc.autoport) {
        if (virPortAllocatorAcquire(driver->remotePorts, &port) < 0)
            return -1;
        graphics->data.vnc.port = port;
    }

    if (graphics->data.vnc.websocket == -1) {
        if (virPortAllocatorAcquire(driver->webSocketPorts, &port) < 0)
            return -1;
        graphics->data.vnc.websocket = port;
    }

    return 0;
}
3569

3570
static int
3571
qemuProcessSPICEAllocatePorts(virQEMUDriverPtr driver,
3572 3573
                              virDomainGraphicsDefPtr graphics,
                              bool allocate)
3574
{
3575
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3576 3577
    unsigned short port = 0;
    unsigned short tlsPort;
3578
    size_t i;
3579
    int defaultMode = graphics->data.spice.defaultMode;
3580
    int ret = -1;
3581 3582 3583 3584 3585 3586

    bool needTLSPort = false;
    bool needPort = false;

    if (graphics->data.spice.autoport) {
        /* check if tlsPort or port need allocation */
3587
        for (i = 0; i < VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_LAST; i++) {
3588 3589 3590 3591 3592 3593 3594 3595 3596 3597
            switch (graphics->data.spice.channels[i]) {
            case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_SECURE:
                needTLSPort = true;
                break;

            case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_INSECURE:
                needPort = true;
                break;

            case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_ANY:
3598
                /* default mode will be used */
3599 3600 3601
                break;
            }
        }
3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616
        switch (defaultMode) {
        case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_SECURE:
            needTLSPort = true;
            break;

        case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_INSECURE:
            needPort = true;
            break;

        case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_ANY:
            if (cfg->spiceTLS)
                needTLSPort = true;
            needPort = true;
            break;
        }
3617 3618
    }

3619 3620 3621 3622 3623 3624 3625
    if (!allocate) {
        if (needPort || graphics->data.spice.port == -1)
            graphics->data.spice.port = 5901;

        if (needTLSPort || graphics->data.spice.tlsPort == -1)
            graphics->data.spice.tlsPort = 5902;

3626 3627
        ret = 0;
        goto cleanup;
3628 3629
    }

3630
    if (needPort || graphics->data.spice.port == -1) {
3631
        if (virPortAllocatorAcquire(driver->remotePorts, &port) < 0)
3632
            goto cleanup;
3633 3634

        graphics->data.spice.port = port;
3635 3636 3637

        if (!graphics->data.spice.autoport)
            graphics->data.spice.portReserved = true;
3638 3639
    }

3640 3641
    if (needTLSPort || graphics->data.spice.tlsPort == -1) {
        if (!cfg->spiceTLS) {
3642 3643 3644
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Auto allocation of spice TLS port requested "
                             "but spice TLS is disabled in qemu.conf"));
3645
            goto cleanup;
3646
        }
3647 3648

        if (virPortAllocatorAcquire(driver->remotePorts, &tlsPort) < 0)
3649
            goto cleanup;
3650 3651

        graphics->data.spice.tlsPort = tlsPort;
3652 3653 3654

        if (!graphics->data.spice.autoport)
            graphics->data.spice.tlsPortReserved = true;
3655 3656
    }

3657
    ret = 0;
3658

3659
 cleanup:
3660
    virPortAllocatorRelease(driver->remotePorts, port);
3661 3662
    virObjectUnref(cfg);
    return ret;
3663 3664 3665
}


3666 3667 3668
static int
qemuValidateCpuCount(virDomainDefPtr def,
                     virQEMUCapsPtr qemuCaps)
3669
{
3670
    unsigned int maxCpus = virQEMUCapsGetMachineMaxCpus(qemuCaps, def->os.machine);
3671

3672 3673 3674 3675 3676 3677
    if (virDomainDefGetVcpus(def) == 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Domain requires at least 1 vCPU"));
        return -1;
    }

3678 3679 3680 3681
    if (maxCpus > 0 && virDomainDefGetVcpusMax(def) > maxCpus) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Maximum CPUs greater than specified machine type limit"));
        return -1;
3682 3683
    }

3684
    return 0;
3685 3686
}

3687 3688

static bool
3689 3690 3691
qemuProcessVerifyGuestCPU(virQEMUDriverPtr driver,
                          virDomainObjPtr vm,
                          int asyncJob)
3692 3693 3694 3695 3696
{
    virDomainDefPtr def = vm->def;
    virArch arch = def->os.arch;
    virCPUDataPtr guestcpu = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;
3697
    int rc;
3698
    bool ret = false;
J
Ján Tomko 已提交
3699
    size_t i;
3700 3701 3702 3703

    switch (arch) {
    case VIR_ARCH_I686:
    case VIR_ARCH_X86_64:
3704 3705
        if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
            return false;
3706
        rc = qemuMonitorGetGuestCPU(priv->mon, arch, &guestcpu);
3707 3708
        if (qemuDomainObjExitMonitor(driver, vm) < 0)
            return false;
3709

3710 3711 3712 3713
        if (rc < 0) {
            if (rc == -2)
                break;

3714
            goto cleanup;
3715
        }
3716

J
Ján Tomko 已提交
3717
        if (def->features[VIR_DOMAIN_FEATURE_PVSPINLOCK] == VIR_TRISTATE_SWITCH_ON) {
3718
            if (!virCPUDataCheckFeature(guestcpu, VIR_CPU_x86_KVM_PV_UNHALT)) {
3719 3720 3721 3722 3723
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("host doesn't support paravirtual spinlocks"));
                goto cleanup;
            }
        }
J
Ján Tomko 已提交
3724

3725 3726 3727 3728 3729 3730
        for (i = 0; i < VIR_DOMAIN_HYPERV_LAST; i++) {
            if (def->hyperv_features[i] == VIR_TRISTATE_SWITCH_ON) {
                char *cpuFeature;
                if (virAsprintf(&cpuFeature, "__kvm_hv_%s",
                                virDomainHypervTypeToString(i)) < 0)
                    goto cleanup;
3731
                if (!virCPUDataCheckFeature(guestcpu, cpuFeature)) {
3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749
                    switch ((virDomainHyperv) i) {
                    case VIR_DOMAIN_HYPERV_RELAXED:
                    case VIR_DOMAIN_HYPERV_VAPIC:
                    case VIR_DOMAIN_HYPERV_SPINLOCKS:
                        VIR_WARN("host doesn't support hyperv '%s' feature",
                                 virDomainHypervTypeToString(i));
                        break;
                    case VIR_DOMAIN_HYPERV_VPINDEX:
                    case VIR_DOMAIN_HYPERV_RUNTIME:
                    case VIR_DOMAIN_HYPERV_SYNIC:
                    case VIR_DOMAIN_HYPERV_STIMER:
                    case VIR_DOMAIN_HYPERV_RESET:
                    case VIR_DOMAIN_HYPERV_VENDOR_ID:
                        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                                       _("host doesn't support hyperv '%s' feature"),
                                       virDomainHypervTypeToString(i));
                        goto cleanup;
                        break;
3750 3751

                    /* coverity[dead_error_begin] */
3752 3753 3754 3755 3756 3757
                    case VIR_DOMAIN_HYPERV_LAST:
                        break;
                    }
                }
            }
        }
J
Ján Tomko 已提交
3758

3759 3760 3761
        if (def->cpu && def->cpu->mode != VIR_CPU_MODE_HOST_PASSTHROUGH) {
            for (i = 0; i < def->cpu->nfeatures; i++) {
                virCPUFeatureDefPtr feature = &def->cpu->features[i];
J
Ján Tomko 已提交
3762

3763 3764 3765 3766
                if (feature->policy != VIR_CPU_FEATURE_REQUIRE)
                    continue;

                if (STREQ(feature->name, "invtsc") &&
3767
                    !virCPUDataCheckFeature(guestcpu, feature->name)) {
3768 3769 3770 3771
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                                   _("host doesn't support invariant TSC"));
                    goto cleanup;
                }
J
Ján Tomko 已提交
3772 3773
            }
        }
3774 3775 3776 3777 3778 3779 3780 3781
        break;

    default:
        break;
    }

    ret = true;

3782
 cleanup:
3783 3784 3785 3786 3787
    cpuDataFree(guestcpu);
    return ret;
}


3788 3789
static int
qemuPrepareNVRAM(virQEMUDriverConfigPtr cfg,
3790
                 virDomainObjPtr vm)
3791 3792 3793 3794
{
    int ret = -1;
    int srcFD = -1;
    int dstFD = -1;
3795
    virDomainLoaderDefPtr loader = vm->def->os.loader;
3796
    bool created = false;
3797 3798
    const char *master_nvram_path;
    ssize_t r;
3799

3800
    if (!loader || !loader->nvram || virFileExists(loader->nvram))
3801 3802
        return 0;

3803 3804 3805
    master_nvram_path = loader->templt;
    if (!loader->templt) {
        size_t i;
3806 3807 3808
        for (i = 0; i < cfg->nfirmwares; i++) {
            if (STREQ(cfg->firmwares[i]->name, loader->path)) {
                master_nvram_path = cfg->firmwares[i]->nvram;
3809
                break;
3810 3811
            }
        }
3812
    }
3813

3814 3815 3816 3817 3818 3819
    if (!master_nvram_path) {
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("unable to find any master var store for "
                         "loader: %s"), loader->path);
        goto cleanup;
    }
3820

3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837
    if ((srcFD = virFileOpenAs(master_nvram_path, O_RDONLY,
                               0, -1, -1, 0)) < 0) {
        virReportSystemError(-srcFD,
                             _("Failed to open file '%s'"),
                             master_nvram_path);
        goto cleanup;
    }
    if ((dstFD = virFileOpenAs(loader->nvram,
                               O_WRONLY | O_CREAT | O_EXCL,
                               S_IRUSR | S_IWUSR,
                               cfg->user, cfg->group, 0)) < 0) {
        virReportSystemError(-dstFD,
                             _("Failed to create file '%s'"),
                             loader->nvram);
        goto cleanup;
    }
    created = true;
3838

3839 3840
    do {
        char buf[1024];
3841

3842
        if ((r = saferead(srcFD, buf, sizeof(buf))) < 0) {
3843
            virReportSystemError(errno,
3844
                                 _("Unable to read from file '%s'"),
3845 3846 3847
                                 master_nvram_path);
            goto cleanup;
        }
3848 3849

        if (safewrite(dstFD, buf, r) < 0) {
3850
            virReportSystemError(errno,
3851
                                 _("Unable to write to file '%s'"),
3852 3853 3854
                                 loader->nvram);
            goto cleanup;
        }
3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867
    } while (r);

    if (VIR_CLOSE(srcFD) < 0) {
        virReportSystemError(errno,
                             _("Unable to close file '%s'"),
                             master_nvram_path);
        goto cleanup;
    }
    if (VIR_CLOSE(dstFD) < 0) {
        virReportSystemError(errno,
                             _("Unable to close file '%s'"),
                             loader->nvram);
        goto cleanup;
3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884
    }

    ret = 0;
 cleanup:
    /* We successfully generated the nvram path, but failed to
     * copy the file content. Roll back. */
    if (ret < 0) {
        if (created)
            unlink(loader->nvram);
    }

    VIR_FORCE_CLOSE(srcFD);
    VIR_FORCE_CLOSE(dstFD);
    return ret;
}


3885 3886 3887
static void
qemuLogOperation(virDomainObjPtr vm,
                 const char *msg,
3888 3889
                 virCommandPtr cmd,
                 qemuDomainLogContextPtr logCtxt)
3890 3891 3892 3893 3894
{
    char *timestamp;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int qemuVersion = virQEMUCapsGetVersion(priv->qemuCaps);
    const char *package = virQEMUCapsGetPackage(priv->qemuCaps);
3895
    char *hostname = virGetHostname();
3896 3897

    if ((timestamp = virTimeStringNow()) == NULL)
3898
        goto cleanup;
3899

3900 3901
    if (qemuDomainLogContextWrite(logCtxt,
                                  "%s: %s %s, qemu version: %d.%d.%d%s, hostname: %s\n",
3902 3903 3904 3905
                                  timestamp, msg, VIR_LOG_VERSION_STRING,
                                  (qemuVersion / 1000000) % 1000,
                                  (qemuVersion / 1000) % 1000,
                                  qemuVersion % 1000,
3906 3907
                                  package ? package : "",
                                  hostname ? hostname : "") < 0)
3908
        goto cleanup;
3909

3910 3911 3912 3913 3914
    if (cmd) {
        char *args = virCommandToString(cmd);
        qemuDomainLogContextWrite(logCtxt, "%s\n", args);
        VIR_FREE(args);
    }
3915 3916

 cleanup:
3917
    VIR_FREE(hostname);
3918 3919 3920
    VIR_FREE(timestamp);
}

3921 3922 3923 3924 3925 3926 3927

void
qemuProcessIncomingDefFree(qemuProcessIncomingDefPtr inc)
{
    if (!inc)
        return;

3928
    VIR_FREE(inc->address);
3929
    VIR_FREE(inc->launchURI);
3930
    VIR_FREE(inc->deferredURI);
3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941
    VIR_FREE(inc);
}


/*
 * This function does not copy @path, the caller is responsible for keeping
 * the @path pointer valid during the lifetime of the allocated
 * qemuProcessIncomingDef structure.
 */
qemuProcessIncomingDefPtr
qemuProcessIncomingDefNew(virQEMUCapsPtr qemuCaps,
3942
                          const char *listenAddress,
3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954
                          const char *migrateFrom,
                          int fd,
                          const char *path)
{
    qemuProcessIncomingDefPtr inc = NULL;

    if (qemuMigrationCheckIncoming(qemuCaps, migrateFrom) < 0)
        return NULL;

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

3955 3956 3957
    if (VIR_STRDUP(inc->address, listenAddress) < 0)
        goto error;

3958 3959 3960 3961
    inc->launchURI = qemuMigrationIncomingURI(migrateFrom, fd);
    if (!inc->launchURI)
        goto error;

3962 3963 3964 3965 3966 3967
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_INCOMING_DEFER)) {
        inc->deferredURI = inc->launchURI;
        if (VIR_STRDUP(inc->launchURI, "defer") < 0)
            goto error;
    }

3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978
    inc->fd = fd;
    inc->path = path;

    return inc;

 error:
    qemuProcessIncomingDefFree(inc);
    return NULL;
}


3979 3980 3981 3982 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
/*
 * This function starts a new QEMU_ASYNC_JOB_START async job. The user is
 * responsible for calling qemuProcessEndJob to stop this job and for passing
 * QEMU_ASYNC_JOB_START as @asyncJob argument to any function requiring this
 * parameter between qemuProcessBeginJob and qemuProcessEndJob.
 */
int
qemuProcessBeginJob(virQEMUDriverPtr driver,
                    virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

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

    qemuDomainObjSetAsyncJobMask(vm, QEMU_JOB_NONE);
    priv->job.current->type = VIR_DOMAIN_JOB_UNBOUNDED;

    return 0;
}


void
qemuProcessEndJob(virQEMUDriverPtr driver,
                  virDomainObjPtr vm)
{
    qemuDomainObjEndAsyncJob(driver, vm);
}


4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031
static int
qemuProcessStartHook(virQEMUDriverPtr driver,
                     virDomainObjPtr vm,
                     virHookQemuOpType op,
                     virHookSubopType subop)
{
    char *xml;
    int ret;

    if (!virHookPresent(VIR_HOOK_DRIVER_QEMU))
        return 0;

    if (!(xml = qemuDomainDefFormatXML(driver, vm->def, 0)))
        return -1;

    ret = virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name, op, subop,
                      NULL, xml, NULL);
    VIR_FREE(xml);

    return ret;
}


4032
static int
4033
qemuProcessGraphicsReservePorts(virQEMUDriverPtr driver,
4034
                                virDomainGraphicsDefPtr graphics)
4035
{
4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046
    virDomainGraphicsListenDefPtr glisten;

    if (graphics->nListens <= 0)
        return 0;

    glisten = &graphics->listens[0];

    if (glisten->type != VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_ADDRESS &&
        glisten->type != VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_NETWORK)
        return 0;

4047 4048 4049 4050 4051 4052 4053
    if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        !graphics->data.vnc.autoport) {
        if (virPortAllocatorSetUsed(driver->remotePorts,
                                    graphics->data.vnc.port,
                                    true) < 0)
            return -1;
        graphics->data.vnc.portReserved = true;
4054

4055 4056 4057
    } else if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE &&
               !graphics->data.spice.autoport) {
        if (graphics->data.spice.port > 0) {
4058
            if (virPortAllocatorSetUsed(driver->remotePorts,
4059
                                        graphics->data.spice.port,
4060
                                        true) < 0)
4061
                return -1;
4062 4063
            graphics->data.spice.portReserved = true;
        }
4064

4065 4066 4067 4068 4069 4070
        if (graphics->data.spice.tlsPort > 0) {
            if (virPortAllocatorSetUsed(driver->remotePorts,
                                        graphics->data.spice.tlsPort,
                                        true) < 0)
                return -1;
            graphics->data.spice.tlsPortReserved = true;
4071 4072 4073
        }
    }

4074 4075 4076 4077
    return 0;
}


4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115
static int
qemuProcessGraphicsAllocatePorts(virQEMUDriverPtr driver,
                                 virDomainGraphicsDefPtr graphics,
                                 bool allocate)
{
    virDomainGraphicsListenDefPtr glisten;

    if (graphics->nListens <= 0)
        return 0;

    glisten = &graphics->listens[0];

    if (glisten->type != VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_ADDRESS &&
        glisten->type != VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_NETWORK)
        return 0;

    switch (graphics->type) {
    case VIR_DOMAIN_GRAPHICS_TYPE_VNC:
        if (qemuProcessVNCAllocatePorts(driver, graphics, allocate) < 0)
            return -1;
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SPICE:
        if (qemuProcessSPICEAllocatePorts(driver, graphics, allocate) < 0)
            return -1;
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SDL:
    case VIR_DOMAIN_GRAPHICS_TYPE_RDP:
    case VIR_DOMAIN_GRAPHICS_TYPE_DESKTOP:
    case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
        break;
    }

    return 0;
}


4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142
static int
qemuProcessGraphicsSetupNetworkAddress(virDomainGraphicsListenDefPtr glisten,
                                       const char *listenAddr)
{
    int rc;

    /* TODO: reject configuration without network specified for network listen */
    if (!glisten->network) {
        if (VIR_STRDUP(glisten->address, listenAddr) < 0)
            return -1;
        return 0;
    }

    rc = networkGetNetworkAddress(glisten->network, &glisten->address);
    if (rc <= -2) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("network-based listen isn't possible, "
                         "network driver isn't present"));
        return -1;
    }
    if (rc < 0)
        return -1;

    return 0;
}


4143
static int
4144
qemuProcessGraphicsSetupListen(virQEMUDriverPtr driver,
4145 4146
                               virDomainGraphicsDefPtr graphics,
                               virDomainObjPtr vm)
4147
{
4148
    qemuDomainObjPrivatePtr priv = vm->privateData;
4149
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4150
    const char *type = virDomainGraphicsTypeToString(graphics->type);
4151
    char *listenAddr = NULL;
4152
    bool useSocket = false;
4153
    size_t i;
4154
    int ret = -1;
4155 4156 4157

    switch (graphics->type) {
    case VIR_DOMAIN_GRAPHICS_TYPE_VNC:
4158
        useSocket = cfg->vncAutoUnixSocket;
4159 4160 4161 4162
        listenAddr = cfg->vncListen;
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SPICE:
4163
        useSocket = cfg->spiceAutoUnixSocket;
4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178
        listenAddr = cfg->spiceListen;
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SDL:
    case VIR_DOMAIN_GRAPHICS_TYPE_RDP:
    case VIR_DOMAIN_GRAPHICS_TYPE_DESKTOP:
    case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
        break;
    }

    for (i = 0; i < graphics->nListens; i++) {
        virDomainGraphicsListenDefPtr glisten = &graphics->listens[i];

        switch (glisten->type) {
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_ADDRESS:
4179 4180 4181 4182 4183
            if (!glisten->address) {
                /* If there is no address specified and qemu.conf has
                 * *_auto_unix_socket set we should use unix socket as
                 * default instead of tcp listen. */
                if (useSocket) {
4184 4185
                    memset(glisten, 0, sizeof(virDomainGraphicsListenDef));
                    if (virAsprintf(&glisten->socket, "%s/%s.sock",
4186
                                    priv->libDir, type) < 0)
4187
                        goto cleanup;
4188 4189
                    glisten->fromConfig = true;
                    glisten->type = VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_SOCKET;
4190 4191
                } else if (listenAddr) {
                    if (VIR_STRDUP(glisten->address, listenAddr) < 0)
4192
                        goto cleanup;
4193 4194 4195
                    glisten->fromConfig = true;
                }
            }
4196 4197 4198 4199 4200 4201 4202 4203
            break;

        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_NETWORK:
            if (glisten->address || !listenAddr)
                continue;

            if (qemuProcessGraphicsSetupNetworkAddress(glisten,
                                                       listenAddr) < 0)
4204
                goto cleanup;
4205 4206
            break;

4207 4208 4209 4210
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_SOCKET:
            if (!glisten->socket) {
                if (virAsprintf(&glisten->socket, "%s/%s.sock",
                                priv->libDir, type) < 0)
4211
                    goto cleanup;
4212 4213 4214 4215
                glisten->autoGenerated = true;
            }
            break;

4216 4217 4218 4219 4220 4221
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_NONE:
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_LAST:
            break;
        }
    }

4222 4223 4224 4225 4226
    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
4227 4228 4229
}


4230 4231
static int
qemuProcessSetupGraphics(virQEMUDriverPtr driver,
4232 4233
                         virDomainObjPtr vm,
                         unsigned int flags)
4234
{
4235
    virDomainGraphicsDefPtr graphics;
4236
    bool allocate = !(flags & VIR_QEMU_PROCESS_START_PRETEND);
4237
    size_t i;
4238 4239
    int ret = -1;

4240 4241 4242 4243 4244 4245 4246
    for (i = 0; i < vm->def->ngraphics; i++) {
        graphics = vm->def->graphics[i];

        if (qemuProcessGraphicsSetupListen(driver, graphics, vm) < 0)
            goto cleanup;
    }

4247 4248 4249 4250 4251 4252 4253 4254
    if (allocate) {
        for (i = 0; i < vm->def->ngraphics; i++) {
            graphics = vm->def->graphics[i];

            if (qemuProcessGraphicsReservePorts(driver, graphics) < 0)
                goto cleanup;
        }
    }
4255

4256
    for (i = 0; i < vm->def->ngraphics; ++i) {
4257
        graphics = vm->def->graphics[i];
4258

4259 4260
        if (qemuProcessGraphicsAllocatePorts(driver, graphics, allocate) < 0)
            goto cleanup;
4261 4262 4263 4264 4265 4266 4267 4268 4269
    }

    ret = 0;

 cleanup:
    return ret;
}


4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302
static int
qemuProcessSetupRawIO(virQEMUDriverPtr driver,
                      virDomainObjPtr vm,
                      virCommandPtr cmd ATTRIBUTE_UNUSED)
{
    bool rawio = false;
    size_t i;
    int ret = -1;

    /* in case a certain disk is desirous of CAP_SYS_RAWIO, add this */
    for (i = 0; i < vm->def->ndisks; i++) {
        virDomainDeviceDef dev;
        virDomainDiskDefPtr disk = vm->def->disks[i];

        if (disk->rawio == VIR_TRISTATE_BOOL_YES) {
            rawio = true;
#ifndef CAP_SYS_RAWIO
            break;
#endif
        }

        dev.type = VIR_DOMAIN_DEVICE_DISK;
        dev.data.disk = disk;
        if (qemuAddSharedDevice(driver, &dev, vm->def->name) < 0)
            goto cleanup;

        if (qemuSetUnprivSGIO(&dev) < 0)
            goto cleanup;
    }

    /* If rawio not already set, check hostdevs as well */
    if (!rawio) {
        for (i = 0; i < vm->def->nhostdevs; i++) {
4303
            if (!virHostdevIsSCSIDevice(vm->def->hostdevs[i]))
4304 4305
                continue;

4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331
            virDomainHostdevSubsysSCSIPtr scsisrc =
                &vm->def->hostdevs[i]->source.subsys.u.scsi;
            if (scsisrc->rawio == VIR_TRISTATE_BOOL_YES) {
                rawio = true;
                break;
            }
        }
    }

    ret = 0;

 cleanup:
    if (rawio) {
#ifdef CAP_SYS_RAWIO
        if (ret == 0)
            virCommandAllowCap(cmd, CAP_SYS_RAWIO);
#else
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Raw I/O is not supported on this platform"));
        ret = -1;
#endif
    }
    return ret;
}


4332 4333 4334 4335 4336 4337 4338 4339 4340
static int
qemuProcessSetupBalloon(virQEMUDriverPtr driver,
                        virDomainObjPtr vm,
                        qemuDomainAsyncJob asyncJob)
{
    unsigned long long balloon = vm->def->mem.cur_balloon;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret = -1;

4341
    if (!virDomainDefHasMemballoon(vm->def))
4342 4343 4344 4345 4346
        return 0;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;

4347 4348 4349
    if (vm->def->memballoon->period)
        qemuMonitorSetMemoryStatsPeriod(priv->mon, vm->def->memballoon,
                                        vm->def->memballoon->period);
4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361
    if (qemuMonitorSetBalloon(priv->mon, balloon) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
    return ret;
}


J
Jiri Denemark 已提交
4362 4363 4364
static int
qemuProcessMakeDir(virQEMUDriverPtr driver,
                   virDomainObjPtr vm,
4365
                   const char *path)
J
Jiri Denemark 已提交
4366 4367 4368 4369 4370 4371 4372 4373
{
    int ret = -1;

    if (virFileMakePathWithMode(path, 0750) < 0) {
        virReportSystemError(errno, _("Cannot create directory '%s'"), path);
        goto cleanup;
    }

4374 4375
    if (virSecurityManagerDomainSetPathLabel(driver->securityManager,
                                             vm->def, path) < 0)
J
Jiri Denemark 已提交
4376 4377 4378 4379 4380 4381 4382 4383 4384
        goto cleanup;

    ret = 0;

 cleanup:
    return ret;
}


4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433
static void
qemuProcessStartWarnShmem(virDomainObjPtr vm)
{
    size_t i;
    bool check_shmem = false;
    bool shmem = vm->def->nshmems;

    /*
     * For vhost-user to work, the domain has to have some type of
     * shared memory configured.  We're not the proper ones to judge
     * whether shared hugepages or shm are enough and will be in the
     * future, so we'll just warn in case neither is configured.
     * Moreover failing would give the false illusion that libvirt is
     * really checking that everything works before running the domain
     * and not only we are unable to do that, but it's also not our
     * aim to do so.
     */
    for (i = 0; i < vm->def->nnets; i++) {
        if (virDomainNetGetActualType(vm->def->nets[i]) ==
                                      VIR_DOMAIN_NET_TYPE_VHOSTUSER) {
            check_shmem = true;
            break;
        }
    }

    if (!check_shmem)
        return;

    /*
     * This check is by no means complete.  We merely check
     * whether there are *some* hugepages enabled and *some* NUMA
     * nodes with shared memory access.
     */
    if (!shmem && vm->def->mem.nhugepages) {
        for (i = 0; i < virDomainNumaGetNodeCount(vm->def->numa); i++) {
            if (virDomainNumaGetNodeMemoryAccessMode(vm->def->numa, i) ==
                VIR_NUMA_MEM_ACCESS_SHARED) {
                shmem = true;
                break;
            }
        }
    }

    if (!shmem) {
        VIR_WARN("Detected vhost-user interface without any shared memory, "
                 "the interface might not be operational");
    }
}

4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465

static int
qemuProcessStartValidateGraphics(virDomainObjPtr vm)
{
    size_t i;

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

        switch (graphics->type) {
        case VIR_DOMAIN_GRAPHICS_TYPE_VNC:
        case VIR_DOMAIN_GRAPHICS_TYPE_SPICE:
            if (graphics->nListens > 1) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("QEMU does not support multiple listens for "
                                 "one graphics device."));
                return -1;
            }
            break;

        case VIR_DOMAIN_GRAPHICS_TYPE_SDL:
        case VIR_DOMAIN_GRAPHICS_TYPE_RDP:
        case VIR_DOMAIN_GRAPHICS_TYPE_DESKTOP:
        case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
            break;
        }
    }

    return 0;
}


4466 4467 4468 4469 4470 4471 4472 4473 4474 4475
static int
qemuProcessStartValidateVideo(virDomainObjPtr vm,
                              virQEMUCapsPtr qemuCaps)
{
    size_t i;
    virDomainVideoDefPtr video;

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

4476 4477 4478 4479 4480 4481 4482 4483 4484 4485
        if ((video->type == VIR_DOMAIN_VIDEO_TYPE_VGA &&
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VGA)) ||
            (video->type == VIR_DOMAIN_VIDEO_TYPE_CIRRUS &&
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_CIRRUS_VGA)) ||
            (video->type == VIR_DOMAIN_VIDEO_TYPE_VMVGA &&
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VMWARE_SVGA)) ||
            (video->type == VIR_DOMAIN_VIDEO_TYPE_QXL &&
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_QXL)) ||
            (video->type == VIR_DOMAIN_VIDEO_TYPE_VIRTIO &&
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_GPU))) {
4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("this QEMU does not support '%s' video device"),
                           virDomainVideoTypeToString(video->type));
            return -1;
        }

        if (video->accel) {
            if (video->accel->accel3d == VIR_TRISTATE_SWITCH_ON &&
                (video->type != VIR_DOMAIN_VIDEO_TYPE_VIRTIO ||
                 !virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_GPU_VIRGL))) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("%s 3d acceleration is not supported"),
                               virDomainVideoTypeToString(video->type));
                return -1;
            }
        }
    }

    return 0;
}


4508
static int
4509 4510
qemuProcessStartValidateXML(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
4511
                            virQEMUCapsPtr qemuCaps,
4512
                            virCapsPtr caps,
4513
                            unsigned int flags)
4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525
{
    /* The bits we validate here are XML configs that we previously
     * accepted. We reject them at VM startup time rather than parse
     * time so that pre-existing VMs aren't rejected and dropped from
     * the VM list when libvirt is updated.
     *
     * If back compat isn't a concern, XML validation should probably
     * be done at parse time.
     */
    if (qemuValidateCpuCount(vm->def, qemuCaps) < 0)
        return -1;

4526 4527 4528
    /* checks below should not be executed when starting a qemu process for a
     * VM that was running before (migration, snapshots, save). It's more
     * important to start such VM than keep the configuration clean */
4529 4530 4531
    if ((flags & VIR_QEMU_PROCESS_START_NEW) &&
        virDomainDefValidate(vm->def, caps, 0, driver->xmlopt) < 0)
        return -1;
4532 4533 4534

    return 0;
}
4535

4536

4537 4538 4539 4540 4541 4542
/**
 * qemuProcessStartValidate:
 * @vm: domain object
 * @qemuCaps: emulator capabilities
 * @migration: restoration of existing state
 *
4543 4544 4545 4546 4547
 * This function aggregates checks done prior to start of a VM.
 *
 * Flag VIR_QEMU_PROCESS_START_PRETEND tells, that we don't want to actually
 * start the domain but create a valid qemu command.  If some code shouldn't be
 * executed in this case, make sure to check this flag.
4548
 */
4549
static int
4550 4551
qemuProcessStartValidate(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
4552
                         virQEMUCapsPtr qemuCaps,
4553
                         virCapsPtr caps,
4554
                         unsigned int flags)
4555
{
4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574
    if (!(flags & VIR_QEMU_PROCESS_START_PRETEND)) {
        if (vm->def->virtType == VIR_DOMAIN_VIRT_KVM) {
            VIR_DEBUG("Checking for KVM availability");
            if (!virFileExists("/dev/kvm")) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("Domain requires KVM, but it is not available. "
                                 "Check that virtualization is enabled in the "
                                 "host BIOS, and host configuration is setup to "
                                 "load the kvm modules."));
                return -1;
            }
        }

        VIR_DEBUG("Checking domain and device security labels");
        if (virSecurityManagerCheckAllLabel(driver->securityManager, vm->def) < 0)
            return -1;

    }

4575
    if (qemuProcessStartValidateXML(driver, vm, qemuCaps, caps, flags) < 0)
4576 4577
        return -1;

4578 4579 4580
    if (qemuProcessStartValidateGraphics(vm) < 0)
        return -1;

4581 4582 4583
    if (qemuProcessStartValidateVideo(vm, qemuCaps) < 0)
        return -1;

4584 4585
    VIR_DEBUG("Checking for any possible (non-fatal) issues");

4586
    qemuProcessStartWarnShmem(vm);
4587

4588
    return 0;
4589 4590 4591
}


J
Jiri Denemark 已提交
4592 4593 4594 4595 4596 4597
/**
 * qemuProcessInit:
 *
 * Prepares the domain up to the point when priv->qemuCaps is initialized. The
 * function calls qemuProcessStop when needed.
 *
4598 4599 4600 4601
 * Flag VIR_QEMU_PROCESS_START_PRETEND tells, that we don't want to actually
 * start the domain but create a valid qemu command.  If some code shouldn't be
 * executed in this case, make sure to check this flag.
 *
J
Jiri Denemark 已提交
4602 4603 4604 4605 4606
 * Returns 0 on success, -1 on error.
 */
int
qemuProcessInit(virQEMUDriverPtr driver,
                virDomainObjPtr vm,
4607
                qemuDomainAsyncJob asyncJob,
4608
                bool migration,
4609
                unsigned int flags)
J
Jiri Denemark 已提交
4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    virCapsPtr caps = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int stopFlags;
    int ret = -1;

    VIR_DEBUG("vm=%p name=%s id=%d migration=%d",
              vm, vm->def->name, vm->def->id, migration);

    VIR_DEBUG("Beginning VM startup process");

    if (virDomainObjIsActive(vm)) {
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("VM is already active"));
        goto cleanup;
    }

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
4629 4630 4631 4632
        goto cleanup;

    VIR_DEBUG("Determining emulator version");
    virObjectUnref(priv->qemuCaps);
4633 4634
    if (!(priv->qemuCaps = virQEMUCapsCacheLookupCopy(caps,
                                                      driver->qemuCapsCache,
4635 4636 4637
                                                      vm->def->emulator,
                                                      vm->def->os.machine)))
        goto cleanup;
J
Jiri Denemark 已提交
4638

4639
    if (qemuProcessStartValidate(driver, vm, priv->qemuCaps, caps, flags) < 0)
4640 4641
        goto cleanup;

J
Jiri Denemark 已提交
4642 4643 4644 4645 4646
    /* Do this upfront, so any part of the startup process can add
     * runtime state to vm->def that won't be persisted. This let's us
     * report implicit runtime defaults in the XML, like vnc listen/socket
     */
    VIR_DEBUG("Setting current domain def as transient");
4647
    if (virDomainObjSetDefTransient(caps, driver->xmlopt, vm) < 0)
J
Jiri Denemark 已提交
4648 4649
        goto stop;

4650 4651 4652 4653
    if (!(flags & VIR_QEMU_PROCESS_START_PRETEND)) {
        vm->def->id = qemuDriverAllocateID(driver);
        qemuDomainSetFakeReboot(driver, vm, false);
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_STARTING_UP);
J
Jiri Denemark 已提交
4654

4655 4656
        if (virAtomicIntInc(&driver->nactive) == 1 && driver->inhibitCallback)
            driver->inhibitCallback(true, driver->inhibitOpaque);
J
Jiri Denemark 已提交
4657

4658 4659 4660 4661 4662 4663
        /* Run an early hook to set-up missing devices */
        if (qemuProcessStartHook(driver, vm,
                                 VIR_HOOK_QEMU_OP_PREPARE,
                                 VIR_HOOK_SUBOP_BEGIN) < 0)
            goto stop;
    }
J
Jiri Denemark 已提交
4664

4665
    if (qemuDomainSetPrivatePaths(driver, vm) < 0)
4666 4667
        goto cleanup;

J
Jiri Denemark 已提交
4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678
    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    virObjectUnref(caps);
    return ret;

 stop:
    stopFlags = VIR_QEMU_PROCESS_STOP_NO_RELABEL;
    if (migration)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;
4679
    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, asyncJob, stopFlags);
J
Jiri Denemark 已提交
4680 4681 4682 4683
    goto cleanup;
}


4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694
/**
 * qemuProcessNetworkPrepareDevices
 */
static int
qemuProcessNetworkPrepareDevices(virDomainDefPtr def)
{
    int ret = -1;
    size_t i;

    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
4695
        virDomainNetType actualType;
4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736

        /* If appropriate, grab a physical device from the configured
         * network's pool of devices, or resolve bridge device name
         * to the one defined in the network definition.
         */
        if (networkAllocateActualDevice(def, net) < 0)
            goto cleanup;

        actualType = virDomainNetGetActualType(net);
        if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV &&
            net->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
            /* Each type='hostdev' network device must also have a
             * corresponding entry in the hostdevs array. For netdevs
             * that are hardcoded as type='hostdev', this is already
             * done by the parser, but for those allocated from a
             * network / determined at runtime, we need to do it
             * separately.
             */
            virDomainHostdevDefPtr hostdev = virDomainNetGetActualHostdev(net);
            virDomainHostdevSubsysPCIPtr pcisrc = &hostdev->source.subsys.u.pci;

            if (virDomainHostdevFind(def, hostdev, NULL) >= 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("PCI device %04x:%02x:%02x.%x "
                                 "allocated from network %s is already "
                                 "in use by domain %s"),
                               pcisrc->addr.domain, pcisrc->addr.bus,
                               pcisrc->addr.slot, pcisrc->addr.function,
                               net->data.network.name, def->name);
                goto cleanup;
            }
            if (virDomainHostdevInsert(def, hostdev) < 0)
                goto cleanup;
        }
    }
    ret = 0;
 cleanup:
    return ret;
}


4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752
/**
 * qemuProcessSetupVcpu:
 * @vm: domain object
 * @vcpuid: id of VCPU to set defaults
 *
 * This function sets resource properties (cgroups, affinity, scheduler) for a
 * vCPU. This function expects that the vCPU is online and the vCPU pids were
 * correctly detected at the point when it's called.
 *
 * Returns 0 on success, -1 on error.
 */
int
qemuProcessSetupVcpu(virDomainObjPtr vm,
                     unsigned int vcpuid)
{
    pid_t vcpupid = qemuDomainGetVcpuPid(vm, vcpuid);
4753
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
4754

4755 4756 4757 4758 4759
    return qemuProcessSetupPid(vm, vcpupid, VIR_CGROUP_THREAD_VCPU,
                               vcpuid, vcpu->cpumask,
                               vm->def->cputune.period,
                               vm->def->cputune.quota,
                               &vcpu->sched);
4760 4761 4762 4763 4764 4765
}


static int
qemuProcessSetupVcpus(virDomainObjPtr vm)
{
4766
    virDomainVcpuDefPtr vcpu;
4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811
    unsigned int maxvcpus = virDomainDefGetVcpusMax(vm->def);
    size_t i;

    if ((vm->def->cputune.period || vm->def->cputune.quota) &&
        !virCgroupHasController(((qemuDomainObjPrivatePtr) vm->privateData)->cgroup,
                                VIR_CGROUP_CONTROLLER_CPU)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("cgroup cpu is required for scheduler tuning"));
        return -1;
    }

    if (!qemuDomainHasVcpuPids(vm)) {
        /* If any CPU has custom affinity that differs from the
         * VM default affinity, we must reject it */
        for (i = 0; i < maxvcpus; i++) {
            vcpu = virDomainDefGetVcpu(vm->def, i);

            if (!vcpu->online)
                continue;

            if (vcpu->cpumask &&
                !virBitmapEqual(vm->def->cpumask, vcpu->cpumask)) {
                virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                                _("cpu affinity is not supported"));
                return -1;
            }
        }

        return 0;
    }

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

        if (!vcpu->online)
            continue;

        if (qemuProcessSetupVcpu(vm, i) < 0)
            return -1;
    }

    return 0;
}


4812 4813 4814 4815 4816
int
qemuProcessSetupIOThread(virDomainObjPtr vm,
                         virDomainIOThreadIDDefPtr iothread)
{

4817 4818
    return qemuProcessSetupPid(vm, iothread->thread_id,
                               VIR_CGROUP_THREAD_IOTHREAD,
4819
                               iothread->iothread_id,
4820
                               iothread->cpumask,
4821 4822
                               vm->def->cputune.iothread_period,
                               vm->def->cputune.iothread_quota,
4823
                               &iothread->sched);
4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842
}


static int
qemuProcessSetupIOThreads(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->niothreadids; i++) {
        virDomainIOThreadIDDefPtr info = vm->def->iothreadids[i];

        if (qemuProcessSetupIOThread(vm, info) < 0)
            return -1;
    }

    return 0;
}


4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854
static int
qemuProcessValidateHotpluggableVcpus(virDomainDefPtr def)
{
    virDomainVcpuDefPtr vcpu;
    virDomainVcpuDefPtr subvcpu;
    qemuDomainVcpuPrivatePtr vcpupriv;
    unsigned int maxvcpus = virDomainDefGetVcpusMax(def);
    size_t i = 0;
    size_t j;
    virBitmapPtr ordermap = NULL;
    int ret = -1;

4855
    if (!(ordermap = virBitmapNew(maxvcpus + 1)))
4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871
        goto cleanup;

    /* validate:
     * - all hotpluggable entities to be hotplugged have the correct data
     * - vcpus belonging to a hotpluggable entity share configuration
     * - order of the hotpluggable entities is unique
     */
    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(def, i);
        vcpupriv = QEMU_DOMAIN_VCPU_PRIVATE(vcpu);

        /* skip over hotpluggable entities  */
        if (vcpupriv->vcpus == 0)
            continue;

        if (vcpu->order != 0) {
4872
            if (virBitmapIsBitSet(ordermap, vcpu->order)) {
4873
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
4874
                               _("duplicate vcpu order '%u'"), vcpu->order);
4875 4876 4877
                goto cleanup;
            }

4878 4879 4880 4881 4882 4883
            if (virBitmapSetBit(ordermap, vcpu->order)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("vcpu order '%u' exceeds vcpu count"),
                               vcpu->order);
                goto cleanup;
            }
4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950
        }

        for (j = i + 1; j < (i + vcpupriv->vcpus); j++) {
            subvcpu = virDomainDefGetVcpu(def, j);
            if (subvcpu->hotpluggable != vcpu->hotpluggable ||
                subvcpu->online != vcpu->online ||
                subvcpu->order != vcpu->order) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("vcpus '%zu' and '%zu' are in the same hotplug "
                                 "group but differ in configuration"), i, j);
                goto cleanup;
            }
        }

        if (vcpu->online && vcpu->hotpluggable == VIR_TRISTATE_BOOL_YES) {
            if ((vcpupriv->socket_id == -1 && vcpupriv->core_id == -1 &&
                 vcpupriv->thread_id == -1) ||
                !vcpupriv->type) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("vcpu '%zu' is missing hotplug data"), i);
                goto cleanup;
            }
        }
    }

    ret = 0;
 cleanup:
    virBitmapFree(ordermap);
    return ret;
}


static int
qemuDomainHasHotpluggableStartupVcpus(virDomainDefPtr def)
{
    size_t maxvcpus = virDomainDefGetVcpusMax(def);
    virDomainVcpuDefPtr vcpu;
    size_t i;

    for (i = 0; i < maxvcpus; i++) {
        vcpu = virDomainDefGetVcpu(def, i);

        if (vcpu->online && vcpu->hotpluggable == VIR_TRISTATE_BOOL_YES)
            return true;
    }

    return false;
}


static int
qemuProcessVcpusSortOrder(const void *a,
                          const void *b)
{
    virDomainVcpuDefPtr vcpua = *((virDomainVcpuDefPtr *)a);
    virDomainVcpuDefPtr vcpub = *((virDomainVcpuDefPtr *)b);

    return vcpua->order - vcpub->order;
}


static int
qemuProcessSetupHotpluggableVcpus(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm,
                                  qemuDomainAsyncJob asyncJob)
{
    unsigned int maxvcpus = virDomainDefGetVcpusMax(vm->def);
4951 4952
    qemuDomainObjPrivatePtr priv = vm->privateData;
    qemuCgroupEmulatorAllNodesDataPtr emulatorCgroup = NULL;
4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984
    virDomainVcpuDefPtr vcpu;
    qemuDomainVcpuPrivatePtr vcpupriv;
    virJSONValuePtr vcpuprops = NULL;
    size_t i;
    int ret = -1;
    int rc;

    virDomainVcpuDefPtr *bootHotplug = NULL;
    size_t nbootHotplug = 0;

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

        if (vcpu->hotpluggable == VIR_TRISTATE_BOOL_YES && vcpu->online &&
            vcpupriv->vcpus != 0) {
            if (virAsprintf(&vcpupriv->alias, "vcpu%zu", i) < 0)
                goto cleanup;

            if (VIR_APPEND_ELEMENT(bootHotplug, nbootHotplug, vcpu) < 0)
                goto cleanup;
        }
    }

    if (nbootHotplug == 0) {
        ret = 0;
        goto cleanup;
    }

    qsort(bootHotplug, nbootHotplug, sizeof(*bootHotplug),
          qemuProcessVcpusSortOrder);

4985 4986 4987
    if (qemuCgroupEmulatorAllNodesAllow(priv->cgroup, &emulatorCgroup) < 0)
        goto cleanup;

4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011
    for (i = 0; i < nbootHotplug; i++) {
        vcpu = bootHotplug[i];

        if (!(vcpuprops = qemuBuildHotpluggableCPUProps(vcpu)))
            goto cleanup;

        if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
            goto cleanup;

        rc = qemuMonitorAddDeviceArgs(qemuDomainGetMonitor(vm), vcpuprops);
        vcpuprops = NULL;

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

        if (rc < 0)
            goto cleanup;

        virJSONValueFree(vcpuprops);
    }

    ret = 0;

 cleanup:
5012
    qemuCgroupEmulatorAllNodesRestore(emulatorCgroup);
5013 5014 5015 5016 5017 5018
    VIR_FREE(bootHotplug);
    virJSONValueFree(vcpuprops);
    return ret;
}


5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059
static int
qemuProcessUpdateGuestCPU(virDomainDefPtr def,
                          virQEMUCapsPtr qemuCaps,
                          virCapsPtr caps,
                          unsigned int flags)
{
    int ret = -1;
    size_t nmodels = 0;
    char **models = NULL;

    if (!def->cpu)
        return 0;

    /* nothing to do if only topology part of CPU def is used */
    if (def->cpu->mode == VIR_CPU_MODE_CUSTOM && !def->cpu->model)
        return 0;

    /* Old libvirt added host CPU model to host-model CPUs for migrations,
     * while new libvirt just turns host-model into custom mode. We need
     * to fix the mode to maintain backward compatibility and to avoid
     * the CPU model to be replaced in virCPUUpdate.
     */
    if (!(flags & VIR_QEMU_PROCESS_START_NEW) &&
        ARCH_IS_X86(def->os.arch) &&
        def->cpu->mode == VIR_CPU_MODE_HOST_MODEL &&
        def->cpu->model) {
        def->cpu->mode = VIR_CPU_MODE_CUSTOM;
    }

    if (!virQEMUCapsIsCPUModeSupported(qemuCaps, caps, def->virtType,
                                       def->cpu->mode)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("CPU mode '%s' for %s %s domain on %s host is not "
                         "supported by hypervisor"),
                       virCPUModeTypeToString(def->cpu->mode),
                       virArchToString(def->os.arch),
                       virDomainVirtTypeToString(def->virtType),
                       virArchToString(caps->host.arch));
        return -1;
    }

5060 5061 5062
    if (virCPUConvertLegacy(caps->host.arch, def->cpu) < 0)
        return -1;

5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079
    /* nothing to update for host-passthrough */
    if (def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
        return 0;

    /* custom CPUs in TCG mode don't need to be compared to host CPU */
    if (def->virtType != VIR_DOMAIN_VIRT_QEMU ||
        def->cpu->mode != VIR_CPU_MODE_CUSTOM) {
        if (virCPUCompare(caps->host.arch, virQEMUCapsGetHostModel(qemuCaps),
                          def->cpu, true) < 0)
            return -1;
    }

    if (virCPUUpdate(def->os.arch, def->cpu,
                     virQEMUCapsGetHostModel(qemuCaps)) < 0)
        goto cleanup;

    if (virQEMUCapsGetCPUDefinitions(qemuCaps, &models, &nmodels) < 0 ||
5080 5081
        virCPUTranslate(def->os.arch, def->cpu,
                        (const char **) models, nmodels) < 0)
5082 5083 5084 5085 5086 5087
        goto cleanup;

    def->cpu->fallback = VIR_CPU_FALLBACK_FORBID;
    ret = 0;

 cleanup:
5088
    virStringListFreeCount(models, nmodels);
5089 5090 5091 5092
    return ret;
}


5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134
/**
 * qemuProcessPrepareDomain
 *
 * This function groups all code that modifies only live XML of a domain which
 * is about to start and it's the only place to do those modifications.
 *
 * Flag VIR_QEMU_PROCESS_START_PRETEND tells, that we don't want to actually
 * start the domain but create a valid qemu command.  If some code shouldn't be
 * executed in this case, make sure to check this flag.
 *
 * TODO: move all XML modification from qemuBuildCommandLine into this function
 */
int
qemuProcessPrepareDomain(virConnectPtr conn,
                         virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         unsigned int flags)
{
    int ret = -1;
    size_t i;
    char *nodeset = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCapsPtr caps;

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

    if (!(flags & VIR_QEMU_PROCESS_START_PRETEND)) {
        /* If you are using a SecurityDriver with dynamic labelling,
           then generate a security label for isolation */
        VIR_DEBUG("Generating domain security label (if required)");
        if (virSecurityManagerGenLabel(driver->securityManager, vm->def) < 0) {
            virDomainAuditSecurityLabel(vm, false);
            goto cleanup;
        }
        virDomainAuditSecurityLabel(vm, true);

        /* Get the advisory nodeset from numad if 'placement' of
         * either <vcpu> or <numatune> is 'auto'.
         */
        if (virDomainDefNeedsPlacementAdvice(vm->def)) {
            nodeset = virNumaGetAutoPlacementAdvice(virDomainDefGetVcpus(vm->def),
5135
                                                    virDomainDefGetMemoryTotal(vm->def));
5136 5137 5138 5139 5140
            if (!nodeset)
                goto cleanup;

            VIR_DEBUG("Nodeset returned from numad: %s", nodeset);

5141
            if (virBitmapParse(nodeset, &priv->autoNodeset,
5142 5143 5144 5145 5146 5147 5148 5149 5150
                               VIR_DOMAIN_CPUMASK_LEN) < 0)
                goto cleanup;

            if (!(priv->autoCpuset = virCapabilitiesGetCpusForNodemask(caps,
                                                                       priv->autoNodeset)))
                goto cleanup;
        }
    }

5151 5152 5153 5154 5155 5156 5157
    /*
     * Normally PCI addresses are assigned in the virDomainCreate
     * or virDomainDefine methods. We might still need to assign
     * some here to cope with the question of upgrades. Regardless
     * we also need to populate the PCI address set cache for later
     * use in hotplug
     */
5158
    VIR_DEBUG("Assigning domain PCI addresses");
5159 5160
    if ((qemuDomainAssignAddresses(vm->def, priv->qemuCaps, vm,
                                   !!(flags & VIR_QEMU_PROCESS_START_NEW))) < 0)
5161
        goto cleanup;
5162

5163 5164 5165
    if (qemuAssignDeviceAliases(vm->def, priv->qemuCaps) < 0)
        goto cleanup;

5166
    VIR_DEBUG("Setting graphics devices");
5167 5168 5169
    if (qemuProcessSetupGraphics(driver, vm, flags) < 0)
        goto cleanup;

5170 5171 5172 5173 5174
    /* Drop possibly missing disks from the definition. This function
     * also resolves source pool/volume into a path and it needs to
     * happen after the def is copied and aliases are set. */
    if (qemuDomainCheckDiskPresence(conn, driver, vm, flags) < 0)
        goto cleanup;
5175

5176 5177 5178 5179
    VIR_DEBUG("Create domain masterKey");
    if (qemuDomainMasterKeyCreate(vm) < 0)
        goto cleanup;

5180 5181 5182 5183 5184
    VIR_DEBUG("Prepare chardev source backends for TLS");
    qemuDomainPrepareChardevSource(vm->def, driver);

    VIR_DEBUG("Add secrets to disks, hostdevs, and chardevs");
    if (qemuDomainSecretPrepare(conn, driver, vm) < 0)
5185 5186
        goto cleanup;

5187 5188 5189 5190 5191 5192
    for (i = 0; i < vm->def->nchannels; i++) {
        if (qemuDomainPrepareChannel(vm->def->channels[i],
                                     priv->channelTargetDir) < 0)
            goto cleanup;
    }

5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204
    if (VIR_ALLOC(priv->monConfig) < 0)
        goto cleanup;

    VIR_DEBUG("Preparing monitor state");
    if (qemuProcessPrepareMonitorChr(priv->monConfig, priv->libDir) < 0)
        goto cleanup;

    priv->monJSON = virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MONITOR_JSON);
    priv->monError = false;
    priv->monStart = 0;
    priv->gotShutdown = false;

5205 5206 5207 5208
    VIR_DEBUG("Updating guest CPU definition");
    if (qemuProcessUpdateGuestCPU(vm->def, priv->qemuCaps, caps, flags) < 0)
        goto cleanup;

5209 5210 5211 5212 5213 5214 5215 5216
    ret = 0;
 cleanup:
    VIR_FREE(nodeset);
    virObjectUnref(caps);
    return ret;
}


J
Jiri Denemark 已提交
5217
/**
5218
 * qemuProcessPrepareHost
J
Jiri Denemark 已提交
5219
 *
5220 5221 5222
 * This function groups all code that modifies host system (which also may
 * update live XML) to prepare environment for a domain which is about to start
 * and it's the only place to do those modifications.
J
Jiri Denemark 已提交
5223
 *
5224
 * TODO: move all host modification from qemuBuildCommandLine into this function
J
Jiri Denemark 已提交
5225 5226
 */
int
5227 5228 5229
qemuProcessPrepareHost(virQEMUDriverPtr driver,
                       virDomainObjPtr vm,
                       bool incoming)
5230
{
5231
    int ret = -1;
5232
    unsigned int hostdev_flags = 0;
5233 5234 5235
    size_t i;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
5236

5237
    if (qemuPrepareNVRAM(cfg, vm) < 0)
J
Jiri Denemark 已提交
5238
        goto cleanup;
5239

5240 5241 5242 5243 5244
    /* network devices must be "prepared" before hostdevs, because
     * setting up a network device might create a new hostdev that
     * will need to be setup.
     */
    VIR_DEBUG("Preparing network devices");
5245
    if (qemuProcessNetworkPrepareDevices(vm->def) < 0)
J
Jiri Denemark 已提交
5246
        goto cleanup;
5247

5248
    /* Must be run before security labelling */
5249
    VIR_DEBUG("Preparing host devices");
5250 5251
    if (!cfg->relaxedACS)
        hostdev_flags |= VIR_HOSTDEV_STRICT_ACS_CHECK;
J
Jiri Denemark 已提交
5252
    if (!incoming)
5253
        hostdev_flags |= VIR_HOSTDEV_COLD_BOOT;
5254 5255
    if (qemuHostdevPrepareDomainDevices(driver, vm->def, priv->qemuCaps,
                                        hostdev_flags) < 0)
J
Jiri Denemark 已提交
5256
        goto cleanup;
5257

5258
    VIR_DEBUG("Preparing chr devices");
5259 5260 5261 5262
    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuProcessPrepareChardevDevice,
                               NULL) < 0)
J
Jiri Denemark 已提交
5263
        goto cleanup;
5264

5265
    if (vm->def->mem.nhugepages) {
5266 5267 5268 5269
        for (i = 0; i < cfg->nhugetlbfs; i++) {
            char *hugepagePath = qemuGetHugepagePath(&cfg->hugetlbfs[i]);

            if (!hugepagePath)
J
Jiri Denemark 已提交
5270
                goto cleanup;
5271 5272 5273 5274 5275 5276

            if (virSecurityManagerSetHugepages(driver->securityManager,
                                               vm->def, hugepagePath) < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               "%s", _("Unable to set huge path in security driver"));
                VIR_FREE(hugepagePath);
J
Jiri Denemark 已提交
5277
                goto cleanup;
5278 5279
            }
            VIR_FREE(hugepagePath);
5280 5281 5282
        }
    }

5283 5284
    /* Ensure no historical cgroup for this VM is lying around bogus
     * settings */
5285
    VIR_DEBUG("Ensuring no historical cgroup is lying around");
5286
    qemuRemoveCgroup(vm);
5287

5288
    if (virFileMakePath(cfg->logDir) < 0) {
5289 5290
        virReportSystemError(errno,
                             _("cannot create log directory %s"),
5291
                             cfg->logDir);
J
Jiri Denemark 已提交
5292
        goto cleanup;
5293 5294
    }

5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317
    VIR_FREE(priv->pidfile);
    if (!(priv->pidfile = virPidFileBuildPath(cfg->stateDir, vm->def->name))) {
        virReportSystemError(errno,
                             "%s", _("Failed to build pidfile path."));
        goto cleanup;
    }

    if (unlink(priv->pidfile) < 0 &&
        errno != ENOENT) {
        virReportSystemError(errno,
                             _("Cannot remove stale PID file %s"),
                             priv->pidfile);
        goto cleanup;
    }

    /*
     * Create all per-domain directories in order to make sure domain
     * with any possible seclabels can access it.
     */
    if (qemuProcessMakeDir(driver, vm, priv->libDir) < 0 ||
        qemuProcessMakeDir(driver, vm, priv->channelTargetDir) < 0)
        goto cleanup;

5318 5319
    VIR_DEBUG("Write domain masterKey");
    if (qemuDomainWriteMasterKeyFile(driver, vm) < 0)
J
John Ferlan 已提交
5320 5321
        goto cleanup;

5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
    return ret;
}


/**
 * qemuProcessLaunch:
 *
 * Launch a new QEMU process with stopped virtual CPUs.
 *
 * The caller is supposed to call qemuProcessStop with appropriate
 * flags in case of failure.
 *
 * Returns 0 on success,
 *        -1 on error which happened before devices were labeled and thus
 *           there is no need to restore them,
 *        -2 on error requesting security labels to be restored.
 */
int
qemuProcessLaunch(virConnectPtr conn,
                  virQEMUDriverPtr driver,
                  virDomainObjPtr vm,
                  qemuDomainAsyncJob asyncJob,
                  qemuProcessIncomingDefPtr incoming,
                  virDomainSnapshotObjPtr snapshot,
                  virNetDevVPortProfileOp vmop,
                  unsigned int flags)
{
    int ret = -1;
    int rv;
    int logfile = -1;
    qemuDomainLogContextPtr logCtxt = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCommandPtr cmd = NULL;
    struct qemuProcessHookData hookData;
    virQEMUDriverConfigPtr cfg;
    virCapsPtr caps = NULL;
    size_t nnicindexes = 0;
    int *nicindexes = NULL;
Q
Qiaowei Ren 已提交
5363
    size_t i;
5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379

    VIR_DEBUG("vm=%p name=%s id=%d asyncJob=%d "
              "incoming.launchURI=%s incoming.deferredURI=%s "
              "incoming.fd=%d incoming.path=%s "
              "snapshot=%p vmop=%d flags=0x%x",
              vm, vm->def->name, vm->def->id, asyncJob,
              NULLSTR(incoming ? incoming->launchURI : NULL),
              NULLSTR(incoming ? incoming->deferredURI : NULL),
              incoming ? incoming->fd : -1,
              NULLSTR(incoming ? incoming->path : NULL),
              snapshot, vmop, flags);

    /* Okay, these are just internal flags,
     * but doesn't hurt to check */
    virCheckFlags(VIR_QEMU_PROCESS_START_COLD |
                  VIR_QEMU_PROCESS_START_PAUSED |
5380 5381
                  VIR_QEMU_PROCESS_START_AUTODESTROY |
                  VIR_QEMU_PROCESS_START_NEW, -1);
5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393

    cfg = virQEMUDriverGetConfig(driver);

    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;
    /* We don't increase cfg's reference counter here. */
    hookData.cfg = cfg;

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

5394
    VIR_DEBUG("Creating domain log file");
5395 5396
    if (!(logCtxt = qemuDomainLogContextNew(driver, vm,
                                            QEMU_DOMAIN_LOG_CONTEXT_MODE_START)))
J
Jiri Denemark 已提交
5397
        goto cleanup;
5398
    logfile = qemuDomainLogContextGetWriteFD(logCtxt);
5399

5400
    VIR_DEBUG("Building emulator command line");
5401
    if (!(cmd = qemuBuildCommandLine(driver,
5402 5403
                                     qemuDomainLogContextGetManager(logCtxt),
                                     vm->def, priv->monConfig,
E
Eric Blake 已提交
5404
                                     priv->monJSON, priv->qemuCaps,
5405 5406
                                     incoming ? incoming->launchURI : NULL,
                                     snapshot, vmop,
J
Ján Tomko 已提交
5407
                                     false,
5408
                                     qemuCheckFips(),
5409
                                     priv->autoNodeset,
5410
                                     &nnicindexes, &nicindexes,
5411
                                     priv->libDir)))
J
Jiri Denemark 已提交
5412
        goto cleanup;
5413

5414 5415
    if (incoming && incoming->fd != -1)
        virCommandPassFD(cmd, incoming->fd, 0);
5416

5417
    /* now that we know it is about to start call the hook if present */
5418 5419 5420
    if (qemuProcessStartHook(driver, vm,
                             VIR_HOOK_QEMU_OP_START,
                             VIR_HOOK_SUBOP_BEGIN) < 0)
J
Jiri Denemark 已提交
5421
        goto cleanup;
5422

5423
    qemuLogOperation(vm, "starting up", cmd, logCtxt);
5424

5425
    qemuDomainObjCheckTaint(driver, vm, logCtxt);
5426

5427
    qemuDomainLogContextMarkPosition(logCtxt);
5428 5429

    VIR_DEBUG("Clear emulator capabilities: %d",
5430 5431
              cfg->clearEmulatorCapabilities);
    if (cfg->clearEmulatorCapabilities)
5432 5433
        virCommandClearCaps(cmd);

5434 5435
    VIR_DEBUG("Setting up raw IO");
    if (qemuProcessSetupRawIO(driver, vm, cmd) < 0)
J
Jiri Denemark 已提交
5436
        goto cleanup;
5437

5438
    virCommandSetPreExecHook(cmd, qemuProcessHook, &hookData);
5439 5440
    virCommandSetMaxProcesses(cmd, cfg->maxProcesses);
    virCommandSetMaxFiles(cmd, cfg->maxFiles);
5441
    virCommandSetMaxCoreSize(cmd, cfg->maxCore);
5442
    virCommandSetUmask(cmd, 0x002);
5443

5444 5445
    VIR_DEBUG("Setting up security labelling");
    if (virSecurityManagerSetChildProcessLabel(driver->securityManager,
5446
                                               vm->def, cmd) < 0)
J
Jiri Denemark 已提交
5447
        goto cleanup;
5448

5449 5450 5451
    virCommandSetOutputFD(cmd, &logfile);
    virCommandSetErrorFD(cmd, &logfile);
    virCommandNonblockingFDs(cmd);
5452
    virCommandSetPidFile(cmd, priv->pidfile);
5453
    virCommandDaemonize(cmd);
5454
    virCommandRequireHandshake(cmd);
5455

5456
    if (virSecurityManagerPreFork(driver->securityManager) < 0)
J
Jiri Denemark 已提交
5457
        goto cleanup;
5458
    rv = virCommandRun(cmd, NULL);
5459
    virSecurityManagerPostFork(driver->securityManager);
5460

E
Eric Blake 已提交
5461
    /* wait for qemu process to show up */
5462
    if (rv == 0) {
5463
        if (virPidFileReadPath(priv->pidfile, &vm->pid) < 0) {
5464 5465
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Domain %s didn't show up"), vm->def->name);
5466
            rv = -1;
5467
        }
M
Michal Privoznik 已提交
5468 5469
        VIR_DEBUG("QEMU vm=%p name=%s running with pid=%lld",
                  vm, vm->def->name, (long long) vm->pid);
5470 5471 5472
    } else {
        VIR_DEBUG("QEMU vm=%p name=%s failed to spawn",
                  vm, vm->def->name);
5473 5474
    }

5475
    VIR_DEBUG("Writing early domain status to disk");
5476
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
J
Jiri Denemark 已提交
5477
        goto cleanup;
5478

5479 5480
    VIR_DEBUG("Waiting for handshake from child");
    if (virCommandHandshakeWait(cmd) < 0) {
5481
        /* Read errors from child that occurred between fork and exec. */
5482 5483
        qemuProcessReportLogError(logCtxt,
                                  _("Process exited prior to exec"));
J
Jiri Denemark 已提交
5484
        goto cleanup;
5485 5486
    }

5487
    VIR_DEBUG("Setting up domain cgroup (if required)");
5488
    if (qemuSetupCgroup(driver, vm, nnicindexes, nicindexes) < 0)
J
Jiri Denemark 已提交
5489
        goto cleanup;
5490

5491 5492 5493 5494
    if (!(priv->perf = virPerfNew()))
        goto cleanup;

    for (i = 0; i < VIR_PERF_EVENT_LAST; i++) {
5495
        if (vm->def->perf.events[i] == VIR_TRISTATE_BOOL_YES &&
5496 5497
            virPerfEventEnable(priv->perf, i, vm->pid) < 0)
            goto cleanup;
Q
Qiaowei Ren 已提交
5498
    }
5499

5500 5501 5502 5503 5504 5505
    /* This must be done after cgroup placement to avoid resetting CPU
     * affinity */
    if (!vm->def->cputune.emulatorpin &&
        qemuProcessInitCpuAffinity(vm) < 0)
        goto cleanup;

5506 5507 5508 5509
    VIR_DEBUG("Setting emulator tuning/settings");
    if (qemuProcessSetupEmulator(vm) < 0)
        goto cleanup;

5510 5511
    VIR_DEBUG("Setting domain security labels");
    if (virSecurityManagerSetAllLabel(driver->securityManager,
J
Jiri Denemark 已提交
5512 5513 5514
                                      vm->def,
                                      incoming ? incoming->path : NULL) < 0)
        goto cleanup;
5515

5516
    /* Security manager labeled all devices, therefore
J
Jiri Denemark 已提交
5517 5518 5519 5520
     * if any operation from now on fails, we need to ask the caller to
     * restore labels.
     */
    ret = -2;
5521

J
Jiri Denemark 已提交
5522
    if (incoming && incoming->fd != -1) {
5523 5524 5525 5526 5527 5528 5529
        /* if there's an fd to migrate from, and it's a pipe, put the
         * proper security label on it
         */
        struct stat stdin_sb;

        VIR_DEBUG("setting security label on pipe used for migration");

J
Jiri Denemark 已提交
5530
        if (fstat(incoming->fd, &stdin_sb) < 0) {
5531
            virReportSystemError(errno,
J
Jiri Denemark 已提交
5532 5533
                                 _("cannot stat fd %d"), incoming->fd);
            goto cleanup;
5534 5535
        }
        if (S_ISFIFO(stdin_sb.st_mode) &&
J
Jiri Denemark 已提交
5536 5537 5538
            virSecurityManagerSetImageFDLabel(driver->securityManager,
                                              vm->def, incoming->fd) < 0)
            goto cleanup;
5539 5540 5541
    }

    VIR_DEBUG("Labelling done, completing handshake to child");
5542
    if (virCommandHandshakeNotify(cmd) < 0)
J
Jiri Denemark 已提交
5543
        goto cleanup;
5544 5545
    VIR_DEBUG("Handshake complete, child running");

5546
    if (rv == -1) /* The VM failed to start; tear filters before taps */
5547 5548
        virDomainConfVMNWFilterTeardown(vm);

5549
    if (rv == -1) /* The VM failed to start */
J
Jiri Denemark 已提交
5550
        goto cleanup;
5551

5552
    VIR_DEBUG("Waiting for monitor to show up");
5553
    if (qemuProcessWaitForMonitor(driver, vm, asyncJob, priv->qemuCaps, logCtxt) < 0)
J
Jiri Denemark 已提交
5554
        goto cleanup;
5555

5556 5557
    if (qemuConnectAgent(driver, vm) < 0)
        goto cleanup;
D
Daniel P. Berrange 已提交
5558

5559
    VIR_DEBUG("Detecting if required emulator features are present");
5560
    if (!qemuProcessVerifyGuestCPU(driver, vm, asyncJob))
J
Jiri Denemark 已提交
5561
        goto cleanup;
5562

5563
    VIR_DEBUG("Setting up post-init cgroup restrictions");
5564
    if (qemuSetupCpusetMems(vm) < 0)
J
Jiri Denemark 已提交
5565
        goto cleanup;
5566

5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578
    VIR_DEBUG("setting up hotpluggable cpus");
    if (qemuDomainHasHotpluggableStartupVcpus(vm->def)) {
        if (qemuDomainRefreshVcpuInfo(driver, vm, asyncJob, false) < 0)
            goto cleanup;

        if (qemuProcessValidateHotpluggableVcpus(vm->def) < 0)
            goto cleanup;

        if (qemuProcessSetupHotpluggableVcpus(driver, vm, asyncJob) < 0)
            goto cleanup;
    }

5579
    VIR_DEBUG("Refreshing VCPU info");
5580
    if (qemuDomainRefreshVcpuInfo(driver, vm, asyncJob, false) < 0)
J
Jiri Denemark 已提交
5581
        goto cleanup;
5582

5583 5584 5585
    if (qemuDomainValidateVcpuInfo(vm) < 0)
        goto cleanup;

5586 5587
    qemuDomainVcpuPersistOrder(vm->def);

5588 5589
    VIR_DEBUG("Detecting IOThread PIDs");
    if (qemuProcessDetectIOThreadPIDs(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
5590
        goto cleanup;
5591

5592 5593 5594 5595
    VIR_DEBUG("Setting global CPU cgroup (if required)");
    if (qemuSetupGlobalCpuCgroup(vm) < 0)
        goto cleanup;

5596 5597
    VIR_DEBUG("Setting vCPU tuning/settings");
    if (qemuProcessSetupVcpus(vm) < 0)
J
Jiri Denemark 已提交
5598
        goto cleanup;
5599

5600 5601
    VIR_DEBUG("Setting IOThread tuning/settings");
    if (qemuProcessSetupIOThreads(vm) < 0)
J
Jiri Denemark 已提交
5602
        goto cleanup;
5603

5604
    VIR_DEBUG("Setting any required VM passwords");
5605
    if (qemuProcessInitPasswords(conn, driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
5606
        goto cleanup;
5607

5608 5609 5610 5611
    /* set default link states */
    /* qemu doesn't support setting this on the command line, so
     * enter the monitor */
    VIR_DEBUG("Setting network link states");
5612
    if (qemuProcessSetLinkStates(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
5613
        goto cleanup;
5614

5615
    VIR_DEBUG("Fetching list of active devices");
5616
    if (qemuDomainUpdateDeviceList(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
5617
        goto cleanup;
5618

5619 5620
    VIR_DEBUG("Updating info of memory devices");
    if (qemuDomainUpdateMemoryDeviceInfo(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
5621
        goto cleanup;
5622

5623
    VIR_DEBUG("Setting initial memory amount");
5624
    if (qemuProcessSetupBalloon(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
5625
        goto cleanup;
5626

5627
    /* Since CPUs were not started yet, the balloon could not return the memory
5628 5629
     * to the host and thus cur_balloon needs to be updated so that GetXMLdesc
     * and friends return the correct size in case they can't grab the job */
5630
    if (!incoming && !snapshot &&
5631
        qemuProcessRefreshBalloonState(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
5632
        goto cleanup;
5633

5634 5635
    VIR_DEBUG("Detecting actual memory size for video device");
    if (qemuProcessUpdateVideoRamSize(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
5636 5637
        goto cleanup;

5638 5639
    VIR_DEBUG("Updating disk data");
    if (qemuProcessRefreshDisks(driver, vm, asyncJob) < 0)
5640 5641
        goto cleanup;

J
Jiri Denemark 已提交
5642 5643 5644 5645 5646 5647 5648
    if (flags & VIR_QEMU_PROCESS_START_AUTODESTROY &&
        qemuProcessAutoDestroyAdd(driver, vm, conn) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
5649
    qemuDomainSecretDestroy(vm);
J
Jiri Denemark 已提交
5650
    virCommandFree(cmd);
5651
    qemuDomainLogContextFree(logCtxt);
J
Jiri Denemark 已提交
5652 5653 5654 5655 5656 5657 5658
    virObjectUnref(cfg);
    virObjectUnref(caps);
    VIR_FREE(nicindexes);
    return ret;
}


5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689
/**
 * qemuProcessFinishStartup:
 *
 * Finish starting a new domain.
 */
int
qemuProcessFinishStartup(virConnectPtr conn,
                         virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         qemuDomainAsyncJob asyncJob,
                         bool startCPUs,
                         virDomainPausedReason pausedReason)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

    if (startCPUs) {
        VIR_DEBUG("Starting domain CPUs");
        if (qemuProcessStartCPUs(driver, vm, conn,
                                 VIR_DOMAIN_RUNNING_BOOTED,
                                 asyncJob) < 0) {
            if (!virGetLastError())
                virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                               _("resume operation failed"));
            goto cleanup;
        }
    } else {
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, pausedReason);
    }

    VIR_DEBUG("Writing domain status to disk");
5690
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705
        goto cleanup;

    if (qemuProcessStartHook(driver, vm,
                             VIR_HOOK_QEMU_OP_STARTED,
                             VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
}


J
Jiri Denemark 已提交
5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736
int
qemuProcessStart(virConnectPtr conn,
                 virQEMUDriverPtr driver,
                 virDomainObjPtr vm,
                 qemuDomainAsyncJob asyncJob,
                 const char *migrateFrom,
                 int migrateFd,
                 const char *migratePath,
                 virDomainSnapshotObjPtr snapshot,
                 virNetDevVPortProfileOp vmop,
                 unsigned int flags)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    qemuProcessIncomingDefPtr incoming = NULL;
    unsigned int stopFlags;
    bool relabel = false;
    int ret = -1;
    int rv;

    VIR_DEBUG("conn=%p driver=%p vm=%p name=%s id=%d asyncJob=%s "
              "migrateFrom=%s migrateFd=%d migratePath=%s "
              "snapshot=%p vmop=%d flags=0x%x",
              conn, driver, vm, vm->def->name, vm->def->id,
              qemuDomainAsyncJobTypeToString(asyncJob),
              NULLSTR(migrateFrom), migrateFd, NULLSTR(migratePath),
              snapshot, vmop, flags);

    virCheckFlagsGoto(VIR_QEMU_PROCESS_START_COLD |
                      VIR_QEMU_PROCESS_START_PAUSED |
                      VIR_QEMU_PROCESS_START_AUTODESTROY, cleanup);

5737 5738 5739 5740
    if (!migrateFrom && !snapshot)
        flags |= VIR_QEMU_PROCESS_START_NEW;

    if (qemuProcessInit(driver, vm, asyncJob, !!migrateFrom, flags) < 0)
J
Jiri Denemark 已提交
5741 5742 5743
        goto cleanup;

    if (migrateFrom) {
5744
        incoming = qemuProcessIncomingDefNew(priv->qemuCaps, NULL, migrateFrom,
J
Jiri Denemark 已提交
5745 5746 5747 5748 5749
                                             migrateFd, migratePath);
        if (!incoming)
            goto stop;
    }

5750 5751 5752
    if (qemuProcessPrepareDomain(conn, driver, vm, flags) < 0)
        goto stop;

5753 5754 5755
    if (qemuProcessPrepareHost(driver, vm, !!incoming) < 0)
        goto stop;

J
Jiri Denemark 已提交
5756 5757
    if ((rv = qemuProcessLaunch(conn, driver, vm, asyncJob, incoming,
                                snapshot, vmop, flags)) < 0) {
5758
        if (rv == -2)
J
Jiri Denemark 已提交
5759 5760 5761 5762
            relabel = true;
        goto stop;
    }
    relabel = true;
5763

5764 5765 5766
    if (incoming &&
        incoming->deferredURI &&
        qemuMigrationRunIncoming(driver, vm, incoming->deferredURI, asyncJob) < 0)
J
Jiri Denemark 已提交
5767
        goto stop;
5768

5769 5770 5771 5772 5773
    if (qemuProcessFinishStartup(conn, driver, vm, asyncJob,
                                 !(flags & VIR_QEMU_PROCESS_START_PAUSED),
                                 incoming ?
                                 VIR_DOMAIN_PAUSED_MIGRATION :
                                 VIR_DOMAIN_PAUSED_USER) < 0)
J
Jiri Denemark 已提交
5774
        goto stop;
5775

5776 5777
    /* Keep watching qemu log for errors during incoming migration, otherwise
     * unset reporting errors from qemu log. */
5778
    if (!incoming)
5779
        qemuMonitorSetDomainLog(priv->mon, NULL, NULL, NULL);
5780

5781 5782 5783
    ret = 0;

 cleanup:
5784
    qemuProcessIncomingDefFree(incoming);
5785
    return ret;
5786

J
Jiri Denemark 已提交
5787 5788 5789 5790 5791 5792
 stop:
    stopFlags = 0;
    if (!relabel)
        stopFlags |= VIR_QEMU_PROCESS_STOP_NO_RELABEL;
    if (migrateFrom)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;
5793
    if (priv->mon)
5794
        qemuMonitorSetDomainLog(priv->mon, NULL, NULL, NULL);
5795
    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, asyncJob, stopFlags);
5796
    goto cleanup;
5797 5798 5799
}


5800 5801 5802 5803 5804
virCommandPtr
qemuProcessCreatePretendCmd(virConnectPtr conn,
                            virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
                            const char *migrateURI,
5805
                            bool enableFips,
5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816
                            bool standalone,
                            unsigned int flags)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCommandPtr cmd = NULL;

    virCheckFlagsGoto(VIR_QEMU_PROCESS_START_COLD |
                      VIR_QEMU_PROCESS_START_PAUSED |
                      VIR_QEMU_PROCESS_START_AUTODESTROY, cleanup);

    flags |= VIR_QEMU_PROCESS_START_PRETEND;
5817
    flags |= VIR_QEMU_PROCESS_START_NEW;
5818

5819
    if (qemuProcessInit(driver, vm, QEMU_ASYNC_JOB_NONE, !!migrateURI, flags) < 0)
5820 5821 5822 5823 5824 5825
        goto cleanup;

    if (qemuProcessPrepareDomain(conn, driver, vm, flags) < 0)
        goto cleanup;

    VIR_DEBUG("Building emulator command line");
5826
    cmd = qemuBuildCommandLine(driver,
5827 5828 5829 5830 5831 5832 5833 5834 5835
                               NULL,
                               vm->def,
                               priv->monConfig,
                               priv->monJSON,
                               priv->qemuCaps,
                               migrateURI,
                               NULL,
                               VIR_NETDEV_VPORT_PROFILE_OP_NO_OP,
                               standalone,
5836
                               enableFips,
5837 5838 5839
                               priv->autoNodeset,
                               NULL,
                               NULL,
5840
                               priv->libDir);
5841 5842 5843 5844 5845 5846

 cleanup:
    return cmd;
}


5847
int
5848
qemuProcessKill(virDomainObjPtr vm, unsigned int flags)
5849
{
5850
    int ret;
5851

M
Michal Privoznik 已提交
5852
    VIR_DEBUG("vm=%p name=%s pid=%lld flags=%x",
5853
              vm, vm->def->name,
M
Michal Privoznik 已提交
5854
              (long long) vm->pid, flags);
5855

5856 5857 5858 5859 5860
    if (!(flags & VIR_QEMU_PROCESS_KILL_NOCHECK)) {
        if (!virDomainObjIsActive(vm)) {
            VIR_DEBUG("VM '%s' not active", vm->def->name);
            return 0;
        }
5861 5862
    }

5863
    if (flags & VIR_QEMU_PROCESS_KILL_NOWAIT) {
5864 5865 5866 5867 5868
        virProcessKill(vm->pid,
                       (flags & VIR_QEMU_PROCESS_KILL_FORCE) ?
                       SIGKILL : SIGTERM);
        return 0;
    }
5869

5870 5871
    ret = virProcessKillPainfully(vm->pid,
                                  !!(flags & VIR_QEMU_PROCESS_KILL_FORCE));
5872

5873
    return ret;
5874 5875 5876
}


5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915
/**
 * qemuProcessBeginStopJob:
 *
 * Stop all current jobs by killing the domain and start a new one for
 * qemuProcessStop.
 */
int
qemuProcessBeginStopJob(virQEMUDriverPtr driver,
                        virDomainObjPtr vm,
                        qemuDomainJob job,
                        bool forceKill)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    unsigned int killFlags = forceKill ? VIR_QEMU_PROCESS_KILL_FORCE : 0;
    int ret = -1;

    /* We need to prevent monitor EOF callback from doing our work (and
     * sending misleading events) while the vm is unlocked inside
     * BeginJob/ProcessKill API
     */
    priv->beingDestroyed = true;

    if (qemuProcessKill(vm, killFlags) < 0)
        goto cleanup;

    /* Wake up anything waiting on domain condition */
    virDomainObjBroadcast(vm);

    if (qemuDomainObjBeginJob(driver, vm, job) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
    priv->beingDestroyed = false;
    return ret;
}


5916
void qemuProcessStop(virQEMUDriverPtr driver,
5917
                     virDomainObjPtr vm,
5918
                     virDomainShutoffReason reason,
5919
                     qemuDomainAsyncJob asyncJob,
5920
                     unsigned int flags)
5921 5922 5923 5924 5925 5926
{
    int ret;
    int retries = 0;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virErrorPtr orig_err;
    virDomainDefPtr def;
A
Ansis Atteka 已提交
5927
    virNetDevVPortProfilePtr vport = NULL;
5928
    size_t i;
5929
    char *timestamp;
5930
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
5931

M
Michal Privoznik 已提交
5932
    VIR_DEBUG("Shutting down vm=%p name=%s id=%d pid=%lld, "
5933
              "reason=%s, asyncJob=%s, flags=%x",
5934
              vm, vm->def->name, vm->def->id,
M
Michal Privoznik 已提交
5935
              (long long) vm->pid,
5936 5937 5938
              virDomainShutoffReasonTypeToString(reason),
              qemuDomainAsyncJobTypeToString(asyncJob),
              flags);
5939

5940 5941 5942 5943
    /* This method is routinely used in clean up paths. Disable error
     * reporting so we don't squash a legit error. */
    orig_err = virSaveLastError();

5944 5945 5946 5947 5948 5949 5950 5951 5952 5953 5954 5955 5956 5957 5958
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
        if (qemuDomainObjBeginNestedJob(driver, vm, asyncJob) < 0)
            goto cleanup;
    } else if (priv->job.asyncJob != QEMU_ASYNC_JOB_NONE &&
               priv->job.asyncOwner == virThreadSelfID() &&
               priv->job.active != QEMU_JOB_ASYNC_NESTED) {
        VIR_WARN("qemuProcessStop called without a nested job (async=%s)",
                 qemuDomainAsyncJobTypeToString(asyncJob));
    }

    if (!virDomainObjIsActive(vm)) {
        VIR_DEBUG("VM '%s' not active", vm->def->name);
        goto endjob;
    }

5959 5960
    vm->def->id = -1;

5961
    if (virAtomicIntDecAndTest(&driver->nactive) && driver->inhibitCallback)
5962 5963
        driver->inhibitCallback(false, driver->inhibitOpaque);

5964 5965
    /* Wake up anything waiting on domain condition */
    virDomainObjBroadcast(vm);
5966

5967
    if ((timestamp = virTimeStringNow()) != NULL) {
5968 5969 5970
        qemuDomainLogAppendMessage(driver, vm, "%s: shutting down, reason=%s\n",
                                   timestamp,
                                   virDomainShutoffReasonTypeToString(reason));
5971
        VIR_FREE(timestamp);
5972 5973
    }

5974 5975 5976
    /* Clear network bandwidth */
    virDomainClearNetBandwidth(vm);

5977 5978
    virDomainConfVMNWFilterTeardown(vm);

5979
    if (cfg->macFilter) {
5980
        def = vm->def;
5981
        for (i = 0; i < def->nnets; i++) {
5982 5983 5984
            virDomainNetDefPtr net = def->nets[i];
            if (net->ifname == NULL)
                continue;
5985 5986 5987
            ignore_value(ebtablesRemoveForwardAllowIn(driver->ebtables,
                                                      net->ifname,
                                                      &net->mac));
5988 5989 5990
        }
    }

5991
    virPortAllocatorRelease(driver->migrationPorts, priv->nbdPort);
5992
    priv->nbdPort = 0;
5993

D
Daniel P. Berrange 已提交
5994 5995 5996 5997
    if (priv->agent) {
        qemuAgentClose(priv->agent);
        priv->agent = NULL;
    }
5998
    priv->agentError = false;
D
Daniel P. Berrange 已提交
5999

6000
    if (priv->mon) {
6001
        qemuMonitorClose(priv->mon);
6002 6003
        priv->mon = NULL;
    }
6004 6005 6006 6007 6008 6009 6010 6011

    if (priv->monConfig) {
        if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX)
            unlink(priv->monConfig->data.nix.path);
        virDomainChrSourceDefFree(priv->monConfig);
        priv->monConfig = NULL;
    }

J
John Ferlan 已提交
6012 6013 6014
    /* Remove the master key */
    qemuDomainMasterKeyRemove(priv);

6015 6016
    virFileDeleteTree(priv->libDir);
    virFileDeleteTree(priv->channelTargetDir);
6017

6018 6019
    qemuDomainClearPrivatePaths(vm);

6020 6021 6022 6023 6024 6025
    ignore_value(virDomainChrDefForeach(vm->def,
                                        false,
                                        qemuProcessCleanupChardevDevice,
                                        NULL));


6026
    /* shut it off for sure */
6027 6028 6029
    ignore_value(qemuProcessKill(vm,
                                 VIR_QEMU_PROCESS_KILL_FORCE|
                                 VIR_QEMU_PROCESS_KILL_NOCHECK));
6030

6031 6032
    qemuDomainCleanupRun(driver, vm);

6033
    /* Stop autodestroy in case guest is restarted */
6034
    qemuProcessAutoDestroyRemove(driver, vm);
6035

6036 6037
    /* now that we know it's stopped call the hook if present */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
6038
        char *xml = qemuDomainDefFormatXML(driver, vm->def, 0);
6039 6040

        /* we can't stop the operation even if the script raised an error */
6041 6042 6043
        ignore_value(virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name,
                                 VIR_HOOK_QEMU_OP_STOPPED, VIR_HOOK_SUBOP_END,
                                 NULL, xml, NULL));
6044 6045 6046
        VIR_FREE(xml);
    }

6047 6048 6049 6050
    /* Reset Security Labels unless caller don't want us to */
    if (!(flags & VIR_QEMU_PROCESS_STOP_NO_RELABEL))
        virSecurityManagerRestoreAllLabel(driver->securityManager,
                                          vm->def,
6051
                                          !!(flags & VIR_QEMU_PROCESS_STOP_MIGRATED));
6052
    virSecurityManagerReleaseLabel(driver->securityManager, vm->def);
6053

6054
    for (i = 0; i < vm->def->ndisks; i++) {
6055
        virDomainDeviceDef dev;
6056
        virDomainDiskDefPtr disk = vm->def->disks[i];
6057 6058 6059 6060

        dev.type = VIR_DOMAIN_DEVICE_DISK;
        dev.data.disk = disk;
        ignore_value(qemuRemoveSharedDevice(driver, &dev, vm->def->name));
6061 6062
    }

6063
    /* Clear out dynamically assigned labels */
6064
    for (i = 0; i < vm->def->nseclabels; i++) {
6065
        if (vm->def->seclabels[i]->type == VIR_DOMAIN_SECLABEL_DYNAMIC)
6066 6067
            VIR_FREE(vm->def->seclabels[i]->label);
        VIR_FREE(vm->def->seclabels[i]->imagelabel);
6068 6069
    }

6070
    virStringListFree(priv->qemuDevices);
6071 6072
    priv->qemuDevices = NULL;

6073
    qemuHostdevReAttachDomainDevices(driver, vm->def);
6074 6075 6076 6077

    def = vm->def;
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
6078 6079 6080 6081
        vport = virDomainNetGetActualVirtPortProfile(net);

        switch (virDomainNetGetActualType(net)) {
        case VIR_DOMAIN_NET_TYPE_DIRECT:
6082
            ignore_value(virNetDevMacVLanDeleteWithVPortProfile(
6083
                             net->ifname, &net->mac,
6084 6085
                             virDomainNetGetActualDirectDev(net),
                             virDomainNetGetActualDirectMode(net),
6086
                             virDomainNetGetActualVirtPortProfile(net),
6087
                             cfg->stateDir));
6088
            break;
6089 6090 6091 6092 6093 6094
        case VIR_DOMAIN_NET_TYPE_ETHERNET:
            if (net->ifname) {
                ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap));
                VIR_FREE(net->ifname);
            }
            break;
6095 6096 6097 6098
        case VIR_DOMAIN_NET_TYPE_BRIDGE:
        case VIR_DOMAIN_NET_TYPE_NETWORK:
#ifdef VIR_NETDEV_TAP_REQUIRE_MANUAL_CLEANUP
            if (!(vport && vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH))
6099
                ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap));
6100 6101
#endif
            break;
6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112
        case VIR_DOMAIN_NET_TYPE_USER:
        case VIR_DOMAIN_NET_TYPE_VHOSTUSER:
        case VIR_DOMAIN_NET_TYPE_SERVER:
        case VIR_DOMAIN_NET_TYPE_CLIENT:
        case VIR_DOMAIN_NET_TYPE_MCAST:
        case VIR_DOMAIN_NET_TYPE_INTERNAL:
        case VIR_DOMAIN_NET_TYPE_HOSTDEV:
        case VIR_DOMAIN_NET_TYPE_UDP:
        case VIR_DOMAIN_NET_TYPE_LAST:
            /* No special cleanup procedure for these types. */
            break;
6113
        }
6114 6115 6116
        /* release the physical device (or any other resources used by
         * this interface in the network driver
         */
6117 6118 6119 6120 6121 6122 6123 6124 6125
        if (vport) {
            if (vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_MIDONET) {
                ignore_value(virNetDevMidonetUnbindPort(vport));
            } else if (vport->virtPortType == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                ignore_value(virNetDevOpenvswitchRemovePort(
                                 virDomainNetGetActualBridgeName(net),
                                 net->ifname));
            }
        }
A
Ansis Atteka 已提交
6126

6127 6128
        /* kick the device out of the hostdev list too */
        virDomainNetRemoveHostdev(def, net);
6129
        networkReleaseActualDevice(vm->def, net);
6130
    }
6131

6132
 retry:
6133
    if ((ret = qemuRemoveCgroup(vm)) < 0) {
6134 6135 6136 6137 6138 6139 6140
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }
6141
    virCgroupFree(&priv->cgroup);
6142

6143
    virPerfFree(priv->perf);
6144
    priv->perf = NULL;
6145

6146 6147
    qemuProcessRemoveDomainStatus(driver, vm);

6148 6149
    /* Remove VNC and Spice ports from port reservation bitmap, but only if
       they were reserved by the driver (autoport=yes)
6150
    */
6151
    for (i = 0; i < vm->def->ngraphics; ++i) {
6152
        virDomainGraphicsDefPtr graphics = vm->def->graphics[i];
6153 6154
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
            if (graphics->data.vnc.autoport) {
6155 6156
                virPortAllocatorRelease(driver->remotePorts,
                                        graphics->data.vnc.port);
6157
            } else if (graphics->data.vnc.portReserved) {
6158 6159 6160 6161 6162
                virPortAllocatorSetUsed(driver->remotePorts,
                                        graphics->data.spice.port,
                                        false);
                graphics->data.vnc.portReserved = false;
            }
6163 6164
            virPortAllocatorRelease(driver->webSocketPorts,
                                    graphics->data.vnc.websocket);
6165
        }
6166 6167 6168 6169 6170 6171 6172 6173 6174 6175 6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
            if (graphics->data.spice.autoport) {
                virPortAllocatorRelease(driver->remotePorts,
                                        graphics->data.spice.port);
                virPortAllocatorRelease(driver->remotePorts,
                                        graphics->data.spice.tlsPort);
            } else {
                if (graphics->data.spice.portReserved) {
                    virPortAllocatorSetUsed(driver->remotePorts,
                                            graphics->data.spice.port,
                                            false);
                    graphics->data.spice.portReserved = false;
                }

                if (graphics->data.spice.tlsPortReserved) {
                    virPortAllocatorSetUsed(driver->remotePorts,
                                            graphics->data.spice.tlsPort,
                                            false);
                    graphics->data.spice.tlsPortReserved = false;
                }
            }
6187
        }
6188 6189
    }

6190
    vm->taint = 0;
6191
    vm->pid = -1;
J
Jiri Denemark 已提交
6192
    virDomainObjSetState(vm, VIR_DOMAIN_SHUTOFF, reason);
6193 6194
    for (i = 0; i < vm->def->niothreadids; i++)
        vm->def->iothreadids[i]->thread_id = 0;
6195 6196
    virObjectUnref(priv->qemuCaps);
    priv->qemuCaps = NULL;
6197
    VIR_FREE(priv->pidfile);
6198

6199
    /* The "release" hook cleans up additional resources */
6200
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
6201
        char *xml = qemuDomainDefFormatXML(driver, vm->def, 0);
6202 6203 6204

        /* we can't stop the operation even if the script raised an error */
        virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name,
6205 6206
                    VIR_HOOK_QEMU_OP_RELEASE, VIR_HOOK_SUBOP_END,
                    NULL, xml, NULL);
6207 6208 6209
        VIR_FREE(xml);
    }

6210
    virDomainObjRemoveTransientDef(vm);
6211

6212 6213 6214 6215 6216
 endjob:
    if (asyncJob != QEMU_ASYNC_JOB_NONE)
        qemuDomainObjEndJob(driver, vm);

 cleanup:
6217 6218 6219 6220
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
6221
    virObjectUnref(cfg);
6222
}
6223 6224


6225
int qemuProcessAttach(virConnectPtr conn ATTRIBUTE_UNUSED,
6226
                      virQEMUDriverPtr driver,
6227
                      virDomainObjPtr vm,
6228
                      pid_t pid,
6229 6230 6231 6232
                      const char *pidfile,
                      virDomainChrSourceDefPtr monConfig,
                      bool monJSON)
{
6233
    size_t i;
6234
    qemuDomainLogContextPtr logCtxt = NULL;
6235 6236 6237
    char *timestamp;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    bool running = true;
6238
    virDomainPausedReason reason;
6239
    virSecurityLabelPtr seclabel = NULL;
6240
    virSecurityLabelDefPtr seclabeldef = NULL;
6241
    bool seclabelgen = false;
6242 6243
    virSecurityManagerPtr* sec_managers = NULL;
    const char *model;
6244
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
6245
    virCapsPtr caps = NULL;
6246
    bool active = false;
6247 6248 6249 6250

    VIR_DEBUG("Beginning VM attach process");

    if (virDomainObjIsActive(vm)) {
6251 6252
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("VM is already active"));
6253
        virObjectUnref(cfg);
6254 6255 6256
        return -1;
    }

6257
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
6258
        goto error;
6259

6260 6261 6262 6263 6264
    /* Do this upfront, so any part of the startup process can add
     * runtime state to vm->def that won't be persisted. This let's us
     * report implicit runtime defaults in the XML, like vnc listen/socket
     */
    VIR_DEBUG("Setting current domain def as transient");
6265
    if (virDomainObjSetDefTransient(caps, driver->xmlopt, vm) < 0)
6266
        goto error;
6267

6268
    vm->def->id = qemuDriverAllocateID(driver);
6269

6270
    if (virAtomicIntInc(&driver->nactive) == 1 && driver->inhibitCallback)
6271
        driver->inhibitCallback(true, driver->inhibitOpaque);
6272
    active = true;
6273

6274
    if (virFileMakePath(cfg->logDir) < 0) {
6275 6276
        virReportSystemError(errno,
                             _("cannot create log directory %s"),
6277
                             cfg->logDir);
6278
        goto error;
6279 6280 6281
    }

    VIR_FREE(priv->pidfile);
6282
    if (VIR_STRDUP(priv->pidfile, pidfile) < 0)
6283
        goto error;
6284

6285 6286
    vm->pid = pid;

6287
    VIR_DEBUG("Detect security driver config");
6288
    sec_managers = virSecurityManagerGetNested(driver->securityManager);
6289 6290
    if (sec_managers == NULL)
        goto error;
6291 6292

    for (i = 0; sec_managers[i]; i++) {
6293
        seclabelgen = false;
6294 6295
        model = virSecurityManagerGetModel(sec_managers[i]);
        seclabeldef = virDomainDefGetSecurityLabelDef(vm->def, model);
6296
        if (seclabeldef == NULL) {
6297
            if (!(seclabeldef = virSecurityLabelDefNew(model)))
6298 6299 6300
                goto error;
            seclabelgen = true;
        }
6301 6302
        seclabeldef->type = VIR_DOMAIN_SECLABEL_STATIC;
        if (VIR_ALLOC(seclabel) < 0)
6303
            goto error;
6304
        if (virSecurityManagerGetProcessLabel(sec_managers[i],
6305
                                              vm->def, vm->pid, seclabel) < 0)
6306
            goto error;
6307

6308
        if (VIR_STRDUP(seclabeldef->model, model) < 0)
6309
            goto error;
6310

6311
        if (VIR_STRDUP(seclabeldef->label, seclabel->label) < 0)
6312
            goto error;
6313
        VIR_FREE(seclabel);
6314 6315 6316 6317 6318 6319

        if (seclabelgen) {
            if (VIR_APPEND_ELEMENT(vm->def->seclabels, vm->def->nseclabels, seclabeldef) < 0)
                goto error;
            seclabelgen = false;
        }
6320
    }
6321

6322 6323
    if (virSecurityManagerCheckAllLabel(driver->securityManager, vm->def) < 0)
        goto error;
6324 6325 6326
    if (virSecurityManagerGenLabel(driver->securityManager, vm->def) < 0)
        goto error;

6327 6328 6329
    if (qemuDomainPerfRestart(vm) < 0)
        goto error;

6330
    VIR_DEBUG("Creating domain log file");
6331 6332
    if (!(logCtxt = qemuDomainLogContextNew(driver, vm,
                                            QEMU_DOMAIN_LOG_CONTEXT_MODE_ATTACH)))
6333
        goto error;
6334 6335

    VIR_DEBUG("Determining emulator version");
6336
    virObjectUnref(priv->qemuCaps);
6337 6338
    if (!(priv->qemuCaps = virQEMUCapsCacheLookupCopy(caps,
                                                      driver->qemuCapsCache,
6339 6340
                                                      vm->def->emulator,
                                                      vm->def->os.machine)))
6341
        goto error;
6342 6343 6344 6345 6346 6347 6348 6349 6350 6351 6352 6353

    VIR_DEBUG("Preparing monitor state");
    priv->monConfig = monConfig;
    monConfig = NULL;
    priv->monJSON = monJSON;

    priv->gotShutdown = false;

    /*
     * Normally PCI addresses are assigned in the virDomainCreate
     * or virDomainDefine methods. We might still need to assign
     * some here to cope with the question of upgrades. Regardless
M
Martin Kletzander 已提交
6354
     * we also need to populate the PCI address set cache for later
6355 6356
     * use in hotplug
     */
6357
    VIR_DEBUG("Assigning domain PCI addresses");
6358
    if ((qemuDomainAssignAddresses(vm->def, priv->qemuCaps, vm, false)) < 0)
6359
        goto error;
6360

6361
    if ((timestamp = virTimeStringNow()) == NULL)
6362
        goto error;
6363

6364
    qemuDomainLogContextWrite(logCtxt, "%s: attaching\n", timestamp);
6365
    VIR_FREE(timestamp);
6366

6367
    qemuDomainObjTaint(driver, vm, VIR_DOMAIN_TAINT_EXTERNAL_LAUNCH, logCtxt);
6368 6369

    VIR_DEBUG("Waiting for monitor to show up");
6370
    if (qemuProcessWaitForMonitor(driver, vm, QEMU_ASYNC_JOB_NONE, priv->qemuCaps, NULL) < 0)
6371
        goto error;
6372

6373 6374
    if (qemuConnectAgent(driver, vm) < 0)
        goto error;
D
Daniel P. Berrange 已提交
6375

6376
    VIR_DEBUG("Detecting VCPU PIDs");
6377
    if (qemuDomainRefreshVcpuInfo(driver, vm, QEMU_ASYNC_JOB_NONE, false) < 0)
6378 6379
        goto error;

6380 6381 6382
    if (qemuDomainValidateVcpuInfo(vm) < 0)
        goto error;

6383 6384
    VIR_DEBUG("Detecting IOThread PIDs");
    if (qemuProcessDetectIOThreadPIDs(driver, vm, QEMU_ASYNC_JOB_NONE) < 0)
6385
        goto error;
6386 6387

    VIR_DEBUG("Getting initial memory amount");
6388
    qemuDomainObjEnterMonitor(driver, vm);
6389 6390 6391 6392 6393 6394 6395
    if (qemuMonitorGetBalloonInfo(priv->mon, &vm->def->mem.cur_balloon) < 0)
        goto exit_monitor;
    if (qemuMonitorGetStatus(priv->mon, &running, &reason) < 0)
        goto exit_monitor;
    if (qemuMonitorGetVirtType(priv->mon, &vm->def->virtType) < 0)
        goto exit_monitor;
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
6396
        goto error;
6397

6398
    if (running) {
6399 6400
        virDomainObjSetState(vm, VIR_DOMAIN_RUNNING,
                             VIR_DOMAIN_RUNNING_UNPAUSED);
6401 6402 6403
        if (vm->def->memballoon &&
            vm->def->memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO &&
            vm->def->memballoon->period) {
6404
            qemuDomainObjEnterMonitor(driver, vm);
6405
            qemuMonitorSetMemoryStatsPeriod(priv->mon, vm->def->memballoon,
6406
                                            vm->def->memballoon->period);
6407 6408
            if (qemuDomainObjExitMonitor(driver, vm) < 0)
                goto error;
6409 6410
        }
    } else {
6411
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, reason);
6412
    }
6413 6414

    VIR_DEBUG("Writing domain status to disk");
6415
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
6416
        goto error;
6417

6418 6419
    /* Run an hook to allow admins to do some magic */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
6420
        char *xml = qemuDomainDefFormatXML(driver, vm->def, 0);
6421 6422 6423 6424 6425 6426 6427 6428 6429 6430 6431
        int hookret;

        hookret = virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name,
                              VIR_HOOK_QEMU_OP_ATTACH, VIR_HOOK_SUBOP_BEGIN,
                              NULL, xml, NULL);
        VIR_FREE(xml);

        /*
         * If the script raised an error abort the launch
         */
        if (hookret < 0)
6432
            goto error;
6433 6434
    }

6435
    qemuDomainLogContextFree(logCtxt);
6436
    VIR_FREE(seclabel);
6437
    VIR_FREE(sec_managers);
6438
    virObjectUnref(cfg);
6439
    virObjectUnref(caps);
6440 6441 6442

    return 0;

6443 6444
 exit_monitor:
    ignore_value(qemuDomainObjExitMonitor(driver, vm));
6445
 error:
6446 6447 6448 6449 6450 6451
    /* We jump here if we failed to attach to the VM for any reason.
     * Leave the domain running, but pretend we never attempted to
     * attach to it.  */
    if (active && virAtomicIntDecAndTest(&driver->nactive) &&
        driver->inhibitCallback)
        driver->inhibitCallback(false, driver->inhibitOpaque);
6452 6453 6454

    qemuMonitorClose(priv->mon);
    priv->mon = NULL;
6455
    qemuDomainLogContextFree(logCtxt);
6456
    VIR_FREE(seclabel);
6457
    VIR_FREE(sec_managers);
6458 6459
    if (seclabelgen)
        virSecurityLabelDefFree(seclabeldef);
6460
    virDomainChrSourceDefFree(monConfig);
6461
    virObjectUnref(cfg);
6462
    virObjectUnref(caps);
6463 6464 6465 6466
    return -1;
}


6467
static virDomainObjPtr
6468 6469 6470
qemuProcessAutoDestroy(virDomainObjPtr dom,
                       virConnectPtr conn,
                       void *opaque)
6471
{
6472
    virQEMUDriverPtr driver = opaque;
6473
    qemuDomainObjPrivatePtr priv = dom->privateData;
6474
    virObjectEventPtr event = NULL;
6475
    unsigned int stopFlags = 0;
6476

6477
    VIR_DEBUG("vm=%s, conn=%p", dom->def->name, conn);
6478

6479 6480
    virObjectRef(dom);

6481 6482 6483
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_IN)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;

6484 6485
    if (priv->job.asyncJob) {
        VIR_DEBUG("vm=%s has long-term job active, cancelling",
6486
                  dom->def->name);
6487
        qemuDomainObjDiscardAsyncJob(driver, dom);
6488 6489 6490
    }

    VIR_DEBUG("Killing domain");
6491

6492 6493 6494 6495 6496
    if (qemuProcessBeginStopJob(driver, dom, QEMU_JOB_DESTROY, true) < 0)
        goto cleanup;

    qemuProcessStop(driver, dom, VIR_DOMAIN_SHUTOFF_DESTROYED,
                    QEMU_ASYNC_JOB_NONE, stopFlags);
6497

6498
    virDomainAuditStop(dom, "destroyed");
6499
    event = virDomainEventLifecycleNewFromObj(dom,
6500 6501
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
6502

6503 6504
    qemuDomainObjEndJob(driver, dom);

6505
    qemuDomainRemoveInactive(driver, dom);
6506

6507
    qemuDomainEventQueue(driver, event);
6508

6509
 cleanup:
6510
    virDomainObjEndAPI(&dom);
6511
    return dom;
6512 6513
}

6514
int qemuProcessAutoDestroyAdd(virQEMUDriverPtr driver,
6515 6516 6517
                              virDomainObjPtr vm,
                              virConnectPtr conn)
{
6518
    VIR_DEBUG("vm=%s, conn=%p", vm->def->name, conn);
6519 6520
    return virCloseCallbacksSet(driver->closeCallbacks, vm, conn,
                                qemuProcessAutoDestroy);
6521 6522
}

6523
int qemuProcessAutoDestroyRemove(virQEMUDriverPtr driver,
6524 6525
                                 virDomainObjPtr vm)
{
6526
    int ret;
6527
    VIR_DEBUG("vm=%s", vm->def->name);
6528 6529 6530
    ret = virCloseCallbacksUnset(driver->closeCallbacks, vm,
                                 qemuProcessAutoDestroy);
    return ret;
6531
}
6532

6533
bool qemuProcessAutoDestroyActive(virQEMUDriverPtr driver,
6534 6535
                                  virDomainObjPtr vm)
{
6536
    virCloseCallback cb;
6537
    VIR_DEBUG("vm=%s", vm->def->name);
6538
    cb = virCloseCallbacksGet(driver->closeCallbacks, vm, NULL);
6539
    return cb == qemuProcessAutoDestroy;
6540
}
6541 6542 6543 6544 6545 6546 6547 6548 6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562 6563


int
qemuProcessRefreshDisks(virQEMUDriverPtr driver,
                        virDomainObjPtr vm,
                        qemuDomainAsyncJob asyncJob)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virHashTablePtr table = NULL;
    int ret = -1;
    size_t i;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) == 0) {
        table = qemuMonitorGetBlockInfo(priv->mon);
        if (qemuDomainObjExitMonitor(driver, vm) < 0)
            goto cleanup;
    }

    if (!table)
        goto cleanup;

    for (i = 0; i < vm->def->ndisks; i++) {
        virDomainDiskDefPtr disk = vm->def->disks[i];
6564
        qemuDomainDiskPrivatePtr diskpriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
6565 6566
        struct qemuDomainDiskInfo *info;

6567 6568
        if (!(info = virHashLookup(table, disk->info.alias)))
            continue;
6569

6570 6571
        if (info->removable) {
            if (info->empty)
6572 6573
                ignore_value(virDomainDiskSetSource(disk, NULL));

6574 6575 6576 6577 6578 6579
            if (info->tray) {
                if (info->tray_open)
                    disk->tray_status = VIR_DOMAIN_DISK_TRAY_OPEN;
                else
                    disk->tray_status = VIR_DOMAIN_DISK_TRAY_CLOSED;
            }
6580
        }
6581 6582 6583 6584

        /* fill in additional data */
        diskpriv->removable = info->removable;
        diskpriv->tray = info->tray;
6585 6586 6587 6588 6589 6590 6591 6592
    }

    ret = 0;

 cleanup:
    virHashFree(table);
    return ret;
}